Professional Documents
Culture Documents
p897 Chap01
p897 Chap01
p897 Chap01
3. Do they change the structure factors of their unit cells and surrounding
unit cells?
4. Do they interact with external elds?
As is well known, the scattering from atoms of an ideal perfect crystal
add constructively under the BraggLaue conditions, which results in
-function-like intensity maxima localized at reciprocal lattice sites of
the crystal. Real crystals always contain defects, which perturb the ideal
periodic arrangement of the atoms and change local structure factors. The
qualitative nature of diffracted intensity depends on whether it is possible to
relate the atomic positions to an average lattice. If the defects cause atomic
displacements, which are localized in the immediate vicinity of the defects,
then it is possible to relate a single periodic lattice to the crystal; typically,
if the mean square uctuations of the atomic displacements remain nite
and do not exceed the interatomic distance, then it is possible to relate an
average periodic lattice to the crystal with defects. These defects are called
defects of the rst kind (Krivoglaz, 1969, 1996).
Some defects distort the lattice in a manner that makes it impossible to
reference atomic positions to a single periodic lattice. However, it is often
still possible to relate slightly different periodic lattices to relatively large
parts of the crystal (for example, subgrains, dislocation cells, etc.). Such
defects are called defects of the second kind (Krivoglaz, 1969, 1996).
For the analysis of the diffraction, an Ewald sphere representation
is usually used (Fig. 1.1). In this representation, k 2 and k 1 are the
wave vectors along the diffracted and incident beam. In the kinematic
approximation, complex scattering amplitudes from the electron probability
density are integrated with the corresponding phases to determine the
scattering intensity, I(Q), for diffraction vector or momentum transfer, Q.
In electronic units, intensity of scattering by a monochromatic beam is
described by the equation (Krivoglaz, 1969, 1996):
2
iQr
I(Q) = (r)e dr ,
where Q = k2 k1
(1.1)
Fig. 1.1. Ewald representation of the diffraction experiment: Ghkl is a reciprocal lattice
vector, Q is a diffraction vector or momentum transfer and 2 is a scattering angle between
the incident k 1 and scattered k 2 radiation wave vectors with wavelength . The origin of
the reference reciprocal lattice frame, 0, is located at the end of the incident k 1 vector. The
origin of the Ewald sphere with a radius |k 1 | = |k 2 | = 2/ is positioned at E. The ends
of the momentum transfer, Q, and wave vectors k 1 and k 2 , must be on the surface of the
Ewald sphere. If Q = Ghkl , the Ewald sphere passes through one more reciprocal lattice
site and BraggLaue diffraction will occur in the direction of the k 2 vector. If the condition
Q = Ghkl is not satised, as shown in the gure, the BraggLaue diffraction will not occur
for this reection, instead the diffuse scattering will originate from the region of increased
intensity around the (hkl) reciprocal lattice point.
Here, the index s labels the unit cell sites with positions R 0s ; are the atoms
in each unit cell with relative positions r ; F is the structure factor; and fn
is the atomic scattering factor of the nth atom of a unit cell.
Fig. 1.2. Sketch of crystal lattice plane with one point defect in the position t = 6. Only
c6 = 1, for all other lattice sites ct = 0.
(1.3)
In the example given in Fig. 1.2, only c6 = 1 and at all other lattice
sites, ct = 0. In the ideal crystal the lattice site positions without any defects
are characterized by the position vector R0S , and positions of defects by the
vector Rt . Each unit cell will be identied by its number, s. If the unit cell
consists of several atoms then the position of atoms inside the unit cell will
be numbered by the index, . The defect positioned on the lattice site, t,
creates a partial displacement, ust , and partial rotation, mst , of the sth unit
cell. The total displacement, us , and/or rotation, ms , of the sth unit cell
are determined from the sum of all displacements/rotations due to all the
defects:
ct ust , and ms =
ct mst
(1.4)
us =
t
Similarly, the structure factor, Fs , for sth unit cell in the presence of defects
can be written as:
Fs = F +
ct st
(1.5)
t
Fig. 1.3. Impact of the rst and second defect kinds on diffraction: (a, b) -function-like
shape of the intensity positioned in the reciprocal lattice site; (c) peaks of diffuse scattering
and shifted weak -function-like maximum caused by defects of the rst kind; and (d) broad
peak formed in the crystals by defects of the second kind.
1. Defects of the rst kind do not change the -function-like shape of the
intensity distribution, I 0 , of regular Bragg reections. They can shift the
intensity peak position, change the integral intensity of the peak due to
the change in average structure factor and introduce diffuse scattering.
The peak intensity is multiplied by a structural DWF. Examples of such
defects are point defects, small dislocation loops and precipitates.
2. Defects of the second kind cause asymmetric broadening of the peak;
their intensity distribution does not have a -function-like shape even
for an innite crystal. Examples of such defects are dislocations,
disclinations/disconnections, disclination dipoles, stacking faults, twins,
boundaries, long cylinder-like chains of impurities, etc.
Generalizing Eq. 1.2 to include displacement/rotations of the unit
cells and changes in the structure factors, the diffracted intensity can be
written as:
iQ(R 0 R0 )
S
S eT
e
(1.6)
I = |F |2
s,s
Fig. 1.4. (a) Rotations in the vicinity of the edge dislocation with Burgers vector
b = 0.2 nm; and (b) strain eld in the vicinity of the edge dislocation measured along
the X-axes using the beamsize of 50 nm (1) and 300 nm (2).
(1.7)
Fig. 1.5. (a) Single screw dislocation and (b) statistically stored screw dislocations in
the cylinder in real space; (c) sketch of the disk-shape intensity distribution in reciprocal
space, and relative orientation of the diffraction vector, Q, and Burgers vector, b; and
(d) dependence of the full width at half maximum (FWHM) on the angle , between the
Burgers vector and the diffraction vector.
10
11
Fig. 1.6. Correlated dislocation arrangements within boundaries: (a) incidental dislocation
boundaries of the rst type (without misorientation); and (b) small angle boundaries with
alternating lattice rotations.
12
Gnb2
c
Ln
4
b
(1.10)
13
14
nano- and mesoscale pillars. For example, the inuence of length scale on
deformation was elegantly demonstrated during uniaxial compression of
micron and nanosize gold pillars by Uchic et al. (2004) and Maas et al.
(2007). They argue that size effects are caused by dislocation starvation
hardening, with dislocations leaving the crystal more quickly than they
multiply and leading to the requirement of continual dislocation nucleation
during the course of deformation. This is supported by electron microscopy
studies of deformation that nd fundamentally different behavior in
materials below 500 nm (dislocation free). One explanation is the need
to include a new length scale for plasticity the distance a dislocation
travels before it creates another one or leaves the sample (Pantleon, 2002).
Deformation of crystals smaller than this characteristic size is expected to
be dominated by dislocation starvation while deformation in crystals with
larger dimensions should exhibit conventional dislocation plasticity (Bei
et al., 2005, 2006). One of the unsolved problems is an understanding of
interactions of the plastic events at different length scales. Other problems
relate to the regularities of the collective behavior of the dislocation
ensembles. The new experimental methods described in this book are
specically helpful for studies of materials behavior at the mesoscale.
1.5. Anisotropy of Scattering by Dislocation Ensembles
The scattered intensity distribution due to the presence of dislocations
is sensitive to the kind of dislocations present in the crystal, to their
architecture and correlations (Krivoglaz, 1969, 1996; Krivoglaz et al., 1970;
Wilkens, 1970; Barabash et al., 1976; Mughrabi, 1983; Wilkens et al., 1987;
Ryaboshapka et al., 1990; Klimanek, 1993; Ryaboshapka, 1993; Ungar
et al., 1989, 1999). Around each reciprocal lattice point corresponding to
a reciprocal lattice vector site, Ghkl , the scattered intensity is a function of
the diffraction vector, |Q| = |k2 k1 | = 4/Sin, or of the deviation
vector, q = Q Ghkl . Here, k 1 , k 2 are wave vectors of the incident and the
scattered waves with the directions dened by the unit vectors, k 2 and k 1
(Fig. 1.1). In the presence of dislocations, two directions in reciprocal space
are usually distinguished:
1. The radial direction (q ) parallel to Ghkl . I(q )
= I(2) describes the
shape of the radial intensity distribution of the reection, and depends
15
Here, F is the average structure factor of the matrix atoms, = R0s R0s
is the undistorted distance vector between the lattice cells s and s and the
correlation function, T , is:
T =
c
[1 exp (iQuss t )]
(1.12)
16
T2 =
(1.13)
ti
17
within the range of equal to the average distance between the nearest
dislocations. In that case the product Quss t in Eq. 1.11 for the correlation
function can be written as follows:
Quss t
Qj Rj ij Rss Rss
(1.14)
= 0.5
i,j
n(Qb)2 2
L
XY ln
8
|Qb|XY
(1.15)
l1
(1.16)
18
82 Lh
1
(Q b)2 ni j ij ln
8
(Qb)2
(1.17)
19
condition W > h is valid (Fig. 1.6a). This relatively large width of the
IDB causes additional broadening at the tails of the intensity distribution.
Usually, the contribution of the transition near wall regions (within the
width of the wall) is signicant if the distance, D, between the walls and
width of the wall, W, is of the same order D W, and D
h 5. If the
D
opposite conditions are valid, D W , and h > 5, the inuence of the
wall width on scattering is much smaller.
1.8. GNDs
After plastic deformation, geometrically necessary (polar) dislocations as
well as geometrically necessary boundaries may be formed in a crystal.
These dislocations cause not only random deformation, but also strongly
correlated long-range rotations within the crystal, grain or subgrains. The
polar portion of dislocation content is due to the incompatibility of plastic
deformation and to the curvature of the corresponding crystal lattice. It
appears to maintain the continuity of the crystal lattice. That is why at the
nanoscale volume level of the individual dislocations, all dislocations are
geometrically necessary (Nabarro, 2001; Arsenlis et al., 2003). GNDs
are scale-dependent, while SSDs are not. Thats why it should always be
specied over which so-called representative volume element (RVE) the
GNDs density is measured. For example, for micro-Laue diffraction from
a single crystal the RVE is equal to the size of the probed region. If several
grains of the polycrystalline material are probed, there will be distinct RVEs
for each grain. For GNDs and GNBs the distortion tensor, ij , is related to
the GNDs density, n+ , and slip system (direction, , and Burgers vector,
b). Severe deformation causing grain subdivision (Jakobsen et al., 2006)
can change the RVEs within the grain.
GNDs, unpaired or polar dislocations, may be typically arranged in the
following ways (Fig. 1.7):
Unpaired random GNDs.
Unpaired geometrically necessary walls/boundaries.
Unpaired dislocation walls with random dislocations between the walls.
X-ray scattering near Ghkl by crystals with unpaired dislocations exhibits
distinct intensity distributions in radial and transverse directions.
20
Fig. 1.7. (a) Geometrically necessary (unpaired) dislocations create macroscopic lattice
rotations of the crystal; (b) mutual orientation of the dislocation line, , Burgers vector, b, and
momentum transfer of the reection, Ghkl ; (c) reections from the bent lattice in reciprocal
space turn into intensity streaks with and being the streak long and short axis correspondingly; and (d) sketch of the dislocation network hierarchy formed by GNDs and IDBs.
T2 = C2 (Qb)2 ln L
(1.18)
Here, C1 , C2 are the contrast factors for imaginary and real parts of the
correlation function, L is the size of the probed subgrain (or the cut-off
radius, whichever is smaller) and is the orientation factor for each
dislocation system. T1 is linear with respect to the density of GNDs, n+
and goes to zero when n+ = 0. From Eq. 1.18, it follows that GNDs have
the same scattering intensity in the radial direction as SSDs, because in the
radial direction the cross product [Q R ij ] = 0 and T1 vanishes.
A random GND distribution results in a Gaussian shape of the intensity
prole in the radial direction with the FWHMrad 2 depending only on
2 C2 n
21
(1.19)
If unpaired dislocations are grouped within walls, the intensity prole has a
Lorentzian shape with 2 depending on the average distance, D, between
walls.
In the intermediate case of dislocations distributed both in tilt walls
and throughout the crystal, the radial intensity prole corresponds to a
convolution of a Laue function for coherently scattering regions between
walls and a Gaussian function corresponding to the effect of the randomly
distributed dislocation walls (Krivoglaz et al., 1970; Barabash, 2001; Balzar
et al., 2004; Tomota et al., 2004; Huang et al., 2008, 2010).
1.8.2. Orientation space (qGhkl )
For GNDs (T1 = 0), the scattered intensity distribution becomes much
broader in orientation space and is described by a more complicated
function. For crystals with a random distribution of GNDs, the FWHM
q , the transverse intensity distribution scales linearly with the density of
GNDs, n+ , and the length of the probed region, L:
q Ln+
(1.20)
22
Here the Levi-Civita tensor has elements 123 = 231 = 312 = 1; 213 =
132 = 321 = 1; and all other ijk = 0. The index, , in species
the crystallographic direction of the dislocation line, and indicates the
Burgers vector direction. Based on Nyes denition (1953) for a particular
type of GND, = n+ b . For example, an edge GND slip system
with dislocation density, n+ , a line direction parallel to the [lm] plane
normal, and Burgers vector [lm], the GNDs density tensor may be
written as:
+
n ll
n+ lm
n+ l
(1.22)
= n+ ml n+ mm n+ m
n+ l
n+ m
n+ n
23
involves multiple slip (Mughrabi, 1983; Ungar et al., 1984; Wilkens et al.,
1987; Mughrabi and Obst, 2005; Larson et al., 2007, 2008). When different
dislocation systems are simultaneously activated in the probed region, the
correlation functions for radial direction and orientation space, Eqs 1.18
and 1.20, are sums over all activated slip systems, . The individual type
of GND or GNB is identied by their slip plane, N , Burgers vector, b,
and the dislocation line direction, . Under multiple slip the correlation
function, T , depends on the dislocation density tensor, ij , and effective
strain gradient tensor, lmk . The element of a GNDs density tensor
using Nyes notations (Nye, 1953; Arsenlis and Parks, 1999) can be
written as:
ik = i bk (r)
(1.23)
Following this approach (Gao et al., 1999; Huang et al., 2000) in the
framework of strain gradient plasticity, the total GNDs and GNBs density
tensor, ij , relates to the strain gradient tensor, lmk , with the following
equation:
ilm lmk = ik
(1.24)
24
AScrew =
n+
Screw {RS [Q Rij ] 2( RS )( [Q Rij ])}
(1.25)
(1.26)
25
plane (Patterson, 1959; Rollett, 1965; Busing and Levy, 1967; Chung and
Ice, 1999). Any vector in the Cartesian coordinates vu can be transformed
to a vector with unit cell coordinates v by vu = Av, where:
a1
A = 0
a2 cos 3
a2 sin 3
a3 cos 2
a3 sin 2 cos 1
(1.27)
1/b3
Here the {bi } and {i } are the sets of reciprocal lattice parameters and angles
for the unit cell, respectively.
A position of any cell in a crystal can change either due to rigid body
translation of the crystal, distortion of the crystal lattice or both of these
(Noyan and Cohen, 1987). Comparison of the measured unit cell parameters
with the undistorted unit cell ones allows the determination of the strain
of the unit cell. The matrix, AMeas , converts a measured position vector v
into the measured crystal Cartesian coordinates. Let A0 be the matrix for
an unstrained unit cell that converts v into the Cartesian reference frame of
the measured (strained) crystal. Let the origins of these two unit cell vectors
coincide. The transformation matrix, T, relates the matrices AMeas and A0
and maps, A0 = T AMeas . Both distortion and rotation are included in the
transformation matrix. For unstrained crystals T = I. The strain tensor, ij ,
in the crystal coordinate reference system can be written as:
ij = (T ij + T ji )/2 = I ij
(1.28)
Equation 1.28 takes into account that the strain tensor is symmetric,
ij = ji , and removes the anti-symmetric rotation terms (Noyan and Cohen,
1987). The strain tensor components in the crystal reference frame can
be rewritten in sample reference coordinates if the rotation matrix, R, is
known:
Sample
ij
= R ij R 1
The strain tensor components (Eq. 1.28) contain both a hydrostatic strain
(dilatation) and a shear (deviatoric) strain term. These two components can
be written separately subtracting the isostatic strain, /3 or the mean strain
26
components as follows:
12
11
3
ij =
22
21
3
31
32
ij = ij + D
ij
13
23
+ 0
33
0
3
0
3
0
3
(1.29)
where = 11 + 22 + 33
Deviatoric strain tensor components can be measured with micro-Laue
diffraction using polychromatic radiation (see Section 1.15), while dilatational strain is measured with monochromatic radiation in different diffraction methods (Chapters 4, 5 and 6).
Non-linear least-square tting of the peaks is typically used to
determine the local crystallographic orientation and strain. Stephen (1999)
proposes a phenomenological model of strain-induced anisotropic peak
broadening in powder diffraction. However, his method is not valid for
strain measurements in orientation space.
1.10. GNBs
At a higher structural level, dislocations can organize into strongly correlated arrangements including walls and sub-boundaries. In real crystals
individual GNDs tend to group into walls to reduce the stored energy
(Krivoglaz, 1969, 1996; Mughrabi, 1983; Ungar et al., 1984; Nabarro,
1987; Hansen, 2001; Pantleon, 2002). Consider the structure with geometrically necessary randomly distributed tilt dislocation boundaries (GNBs)
(Fig. 1.6b). Each wall provides a rotation between two neighboring mosaic
blocks around the line of the wall (Fig. 1.6b). The unit vector parallel
to the rotation axis of each wall coincides with the unit vector, along
the dislocation lines in the case of a tilt boundary. We consider pure tilt
boundaries formed by equidistant edge GNDs (so-called thin walls).
These boundaries in the innite crystal do not produce long-range elastic
strain but generate subgrain rotations. The remaining stresses in the GNB
region are periodic with a period, b = b csc /2. These stresses are only
appreciable at distances within b from the GNB (Nabarro, 1987). If h
27
is the distance between these dislocations in the wall, one can consider
the boundary as a single defect producing the local rotation eld, . The
misorientation angle across the boundary is dened by the equation:
b
= 2 sin
h
2
(1.30)
where hb for small angle GNBs. If GNBs are not correlated and the
number of + and GNBs probed by the beam is equal, the broadening
of the transverse intensity distribution increases with the misorientation
angle through each GNB and with the total number of GNBs probed by the
beam.
1.11. Cumulative GNBs Misorientation
To characterize this model quantitatively, we dene the average distance,
D, between GNBs, and write the number of GNBs per unit length as D1 .
The total density of GNDs grouped in the GNBs is denoted by:
1
(1.31)
Dh
We further dene an X-axis perpendicular to the plane of the wall, and
a Z-axis parallel to the direction of dislocation lines in the wall. For this
coordinate system (Figs 1.7a, b, c), two non-zero components of the mean
deformation tensor are equal: xy = yx = x/D. This means that the
deformation tensor results in pure rotations about the Z-axis. The rotation
increases with displacement x. The number of GNBs per unit length and the
length of the probed region along the Burgers vector direction determines
the total measured rotation.
Generally, real crystals contain a hierarchy of dislocation structures.
Some of the GNDs are distributed randomly, and the rest may form
different non-random arrangements: IDBs and GNBs. It has been found that
dislocation boundaries evolve within a regular pattern of grain subdivision
on two scales. The smaller scale is related to usual cell boundaries socalled IDBs. The larger scale is related to long and continuous so-called
GNBs. Usually, GNBs separate volume elements which deform by different
slip system modes with different strain amplitudes (Hughes et al., 1997;
Hughes and Hansen, 2000; Hansen, 2001; Pantleon, 2002). Typically there
n+ =
28
Intensity
Disk
n
Dipole L
d
Shoulder
Rst
Ghkl
rd
Real Space
Reciprocal
Space
(a)
(b)
(c)
(d)
Fig. 1.8. (a) Randomly distributed dislocation dipoles on real space; (b) disks of intensity
formed at each reciprocal lattice point are oriented perpendicular to the direction; (c)
polarized dislocation walls composed by grouping of dislocation dipoles within the walls;
(d) dislocation multipoles formed by correlated groupings of three parallel dislocations.
are many IDBs separating ordinary dislocation cells between two cell-block
boundaries formed by GNBs (Fig. 1.7d).
1.12. Dislocation Pairs and Dipoles
Under certain conditions, plastic deformation results in the formation of a
large number of pairs of parallel straight dislocations with opposite Burgers
vectors (Krivoglaz, 1996; Kaganer and Sabelfeld, 2010). These dislocation
pairs are termed dislocation dipoles. Within each dipole, dislocations are
located at a relatively small distance, the dipole shoulder, Ld , compared
to the distance between the dipoles, rd , so that Ld rd (Fig. 1.8a).
The center point between the dislocations within the dipole in the plane
perpendicular to their lines is referred to as t. The presence of dislocation
dipoles essentially changes the diffracted intensity distribution. In the sth
unit cell, located at large distance, Rst , compared to the dipole shoulder,
Rst Ld , the distortions from opposite sign dislocations in the dipole
mutually compensate each other, and the resulting displacement of the
sth cell from its ideal position linearly decreases with distance from the
dipole:
Qust =
QbLd
2Rdt
R st Ld
(1.32)
Here the function depends on the specic type of the dislocation dipole.
The type of the dislocation dipole is characterized by the unit vector parallel
29
30
Fig. 1.9. (a) Dislocation loops in real space; possible shapes of diffuse scattering in
reciprocal space from small loops; (b) single drop; (c) double drop.
bR20
R2st
(1.33)
where the orientation factor, |C|, depends on the mutual orientation between
the Burgers vector, loop plane normal and radius vector to the specic lattice
point R st . The strongly distorted volume around the loop is of the order of
R20 . Close to the reciprocal lattice point for q R 1QB , strong so-called
0
Huang scattering is observed (Barabash et al., 2012), given by:
IH = NL |f |2 e2M
b2 R40 Q2
H
2 q2
(1.34)
The factor H depends only on the relative orientation of the loops plane and
Burgers vector to the diffraction vector and direction in reciprocal space. It
characterizes the angular dependence of the intensity. For example, for facecentered cubic (FCC) crystals with dislocation loops in the {111} planes
and <110> Burgers vectors, factor H for (h00) type reections is equal to
(Trinkaus, 1972; Krivoglaz, 1996):
2
c12
c12
2
H=
for q running along [100] direction.
+3 2
1+2
9
c11
c11
H=
2
3
(1.35)
31
not pass through reciprocal lattice points but surround them. For some
specic types of dislocation loops for example, prismatic loops with b
<100> the scattered intensity has a double-drop shape around (h00) and
(hh0) reections at small q (Fig. 1.9c). However, even for these reections,
at larger q the intensity shape changes to a single drop (Fig. 1.9b).
Scattered diffuse intensity strongly depends on the loop radius as R40 . Each
dislocation loop increases diffuse intensity (R0 /d)4 times more than a
single vacancy or interstitial. That is why grouping of vacancies/interstitials
into NL loops increases the diffuse intensity. The sign of the product b n
determines the direction of the shift of the intensity compared to the ideal
Bragg position but does not inuence the amplitude of the diffuse intensity.
The shift of the intensity maximum is proportionate to NL R20 . That is why
measuring the shift of the diffuse maxima NL R20 , static DebyeWaller
factor M NL R20 and the integrated diffuse intensity NL R40 allows one
to determine the type of the loop interstitial or vacancy and separately
identify the density of loops (number of loops per unit volume), NL , and
their radius, R0 .
For large hkl Miller indices of the reection there is an interval of
q where the main input to intensity is coming from the coupling of
interference from the two strongly distorted regions at large distances from
the loop and the diffuse intensity follows the dependence ID 1/q4
with oscillating terms. Oscillations are most pronounced for the product,
ID q4 . If R0 rL and Qb 1, the diffracted intensity averaged over the
orientation of different crystallites (for example, in powder diffraction) has
a Lorentz shape. Full width at half maximum, 2, and shift of the maxima,
0 ), for prismatic interstitial loops in {111} planes with isotropic
2(m m
elastic moduli are described by the following dependencies (Barabash and
Krivoglaz, 1978, 1982; Krivoglaz, 1996):
2
= 2.9(h00),
NL bR20 tg
0)
2(m m
= 0.86(h00),
NL bR20 tg
5.01(hh0),
5.2(hhh)
(1.36)
0.92(hh0),
0.87(hhh)
32
1
Nl R0 (Qb)2
2
(1.38)
33
It is important to note that variation of the half width due to the presence
of large dislocation loops is similar to that for random straight dislocations
with slightly different coefcients due to correlation between different parts
of the dislocation loop. At the tails, the diffracted intensity decreases as
ID q2 , similar to the case of straight dislocations.
1.14. Stacking Faults, Twins and Partial Dislocations
Stacking faults, twins and partial dislocations are often formed during
plastic deformation and/or crystal growth. They change the order of the
atomic planes and cause an abrupt phase change at certain lattice planes
resulting in a broadening of those (hkl) reections for which the phase
change is not multiple of 2. They inuence diffraction similar to antiphase
boundaries separating ordered domains in ordered alloys. Stacking faults
displace atoms at certain vectors and rotate parts of the crystal at certain
angles (Fig. 1.10). For example, in FCC crystals the {111} most closely
packed planes are also the dislocation slip planes and the twinning planes.
Let us indicate the possible positions of the closely packed planes as
A, B, C. In FCC crystals the two neighboring planes cannot have same
indices like AA. The undeformed sequence of layers for FCC structures is
ABCABCABC. . . Twin formation corresponds to the rotation by 180 in
{111} plane, or mirror image of the lattice planes sequence in {111} plane.
Fig. 1.10. (a, b) Sketch of the two different orientations of twins in real space; (c) rodlike intensity distribution corresponding to twin 1 in reciprocal space; (d) multiple rod-like
intensity distribution corresponding to simultaneous presence of twins 1 and 2.
34
(1 Cos); qm = Sin.
(1.40)
D
D
They depend on the average distance, D, between the stacking faults and
on the (hkl) type of the reection.
For example, in FCC crystals with stacking faults parallel to {111}
planes:
d1 + d3 2d2
hi h2
; = (h1 + h2 h2 )/3 =
u=
6
3
q=
(1.41)
In FCC lattices constructive intensity interference takes place only for hi
being all odd or all even. That is why for i hi = 3m, where m is an
integer, parameter 2 and stacking faults do not inuence diffracted
intensity of these reections. If i hi = 3m 1, then differs from 2
by 2
i hi = 3m + 1,
3 and rod-like intensity maxima will appear. For
the reection maximum will be shifted away from the zero point of the
35
reciprocal lattice, and for i hi = 3m 1, the maximum will be shifted
towards it. If there are several types of stacking faults, then several intensity
rods perpendicular to the stacking faults planes will appear near reections
with:
hi = 3m 1
(1.42)
i
36
Fig. 1.11. (a) Misorientation vector m in micro-Laue diffraction shows the difference
between the unit vector parallel to exact momentum transfer for (hkl) reection and an
arbitrary direction in its vicinity; (b) a probed volume of reciprocal space (shaded in gray)
is bounded by the short and long wavelength limits of Ewalds sphere.
I0 (k)I(q)dk;
(1.43)
37
Here, |k hkl | is the radius of the Ewald sphere that passes through Ghkl. ,
A is a constant, and mrad and m are components of the misorientation
vector, m, along and perpendicular to the direction of the scattered beam,
k hkl , respectively. The probed volume of reciprocal space is bounded by the
short and long wavelength limits of the Ewalds sphere, shaded in gray in
Fig. 1.11b. The misorientation vector, m, characterizes the Laue intensity,
I L (m), assuming the spectral distribution varies smoothly near the energy
corresponding to the Bragg reection. The second part of Eq. 1.43 is valid
when |kk0 | 1. In the rst approximation, the misorientation vector, m, is
perpendicular to k hkl (Fig. 1.11a).
Using GND density tensor components as input parameters, the
intensity distribution of each Laue spot can be calculated numerically from
Eq. 1.43.
There is a small 1mrad convergence in the incident beam due to
specics of micro-focusing optics (Chapter 2). This convergence does not
practically affect the length of the streak, but for the narrow direction of the
streak it should be taken into account. I L (m) should therefore be convoluted
with the experimental angular resolution function.
1.15.2. Origin and analysis of streaks at the Laue patterns
In Sections 1.8 and 1.9 it was shown that at the mesoscale GNDs
cause lattice curvature. Micro-Laue diffraction provides direct information
about the lattice curvature in the probed region. To describe the intensity
redistribution near a Laue spot due to a particular set of GNDs, the mutual
orientation of two planes are important (Figs 1.7b, c):
The plane perpendicular to the direction of dislocation line, , in real
space.
The plane perpendicular to the direction of reciprocal lattice vector,
Ghkl .
The line of intersection of these two planes naturally denes the direction
of the streak axis, :
= Ghkl /| Ghkl |
(1.44)
38
Fig. 1.12. Distinct streaking of the Laue spots for an FCC crystal in cubic axes with 12
different GND slip systems. Simulation performed for the back Laue geometry where the
incident beam is perpendicular to the sample surface. One frame is indexed for convenience.
39
AI0 (k0 )
Q /k0
1
GN
+ d2 (m )/( ) + (m ))
d (
1
( GN + d2 (m )/( ) + (m ))
d1
d3 (m )
Exp 2
(1.45)
( )
= n b L 1 ( ) ; =
(1.46)
k0
8(1 )
Here, n+ is the GNDs density for an activated slip system; n is the
total (GN + SS) dislocation density; is a unit vector along the diffraction
vector Q; L is the size of the probed region in the direction of the Burgers
vector, b; parameter l is l 1; Q is a projection of the diffraction vector Q
along the dislocation line direction; d1 , d2 , d3 are parameters depending on
the elastic constants and on the mutual orientation between the dislocation
slip system and momentum transfer for each Laue spot. Using Eqs 1.45
and 1.46, the Laue pattern from crystals with specic GND distribution can
be modeled and compared to the experimental ones. An example modeling
for 12-edge GND slip systems typical for an FCC crystal is presented in
Fig. 1.12 for the geometry where the incident beam is perpendicular to the
sample surface. As was noted in Section 1.10, GNDs broaden orientation
space and cause streaking of the Laue spots when their portion in the total
dislocation population is more than 1%. In the simulations it was assumed
that the portion of GNDs is 5% of the total dislocation density. One
frame is indexed for convenience. It is seen that Laue intensity streaks
rotate simultaneously as the GNDs slip system changes. It is not only
the orientation of the streaks, but also the length and width of each Laue
streak which changes corresponding to their contrast factors (Eq. 1.46).
Some Laue spots are practically not streaked, while others demonstrate
40
long streaks. Compare, for example, the (1, 1, 7), (1, 1, 7) and (1, 1, 7)
Laue spots at the indexed frame in Fig. 1.12. Multiple examples of the
streak analysis of Laue patterns can be found in Barabash et al. (2001,
2003a, 2003b, 2004, 2009), Ice and Barabash (2007), Ohashi et al. (2009)
and Wang et al. (2011).
1.15.3. Insensitivity of micro-Laue diffraction to GNDs in specic
directions
Condition 1.44 and contrast factor Eq. 1.46 dene the case when plastic
deformation will not be visible with micro-Laue diffraction: if GNDinduced overall rotation axes are parallel to the momentum transfer Ghkl ,
the corresponding Laue spots will not be affected by plastic deformation.
Insensitivity of micro-Laue diffraction to deformation in specic directions
was experimentally conrmed byYang et al., (2003) with 3D depth-resolved
measurements. The Cu micro-compression study (Chapter 5) demonstrates
an example when micro-Laue diffraction is not sensitive to specic kinds
of deformation.
A comparison between the experimental and the modeled Laue pattern
for hexagonal crystal is shown in Fig. 1.13 for the measurement geometry
described in Chapter 2 with the incident beam intercepting the sample
surface at 45 . The comparison between the experimental and the modeled
Laue patterns (Figs 1.13a, b) demonstrates that the best t with experiment
Fig. 1.13. Comparison between the (a) experimental and (b) modeled Laue pattern
demonstrates that (1101)[1213] GND slip system is activated in a deformed grain of a
Ti polycrystal; (c) sketch of the active slip system orientation in the unit cell.
41
mcb
(1.47)
Fig. 1.14. Partial grouping of GNDs within the dislocation boundaries splits Laue spot.
Simulated results for the Laue streak and intensity prole along the streak. The portion of
dislocations grouped within the walls increases from (a, d) 0.5% to (b, e) 50% and (c, f)
90%. Different numbers of walls are modeled: (a, b, c) three walls and (d, e, f) seven walls.
Contour maps of the streak are shown on top of the streak intensity prole for each gure.
42
43
sharp rotations of the cell blocks with an abrupt rotational phase variation.
However, when the walls are not well developed as in the case of 0.5% of
the total number of dislocations grouping within the walls (Figs 1.14a, d),
the condition, h
= D, is true. Here each cell block still contains many
randomly distributed dislocations and has a large mcb . This results in
overlapping of the diffracted intensity and formation of an intensity
cloud with spikes of intensity on top of it. With the increase of the
number of boundaries the cloud can be observed even when 50% of
dislocations are grouped within the walls (Fig. 1.14e). Such intensity
clouds are observed by Levine et al. (2006) using micro-Laue diffraction
and Jakobsen et al. (2006) using monochromatic 3D X-ray diffraction
(3DXRD) (see Chapter 9). The total length of the streak, m , is large.
With an increasingly large number of walls, the scattering from each
cell block blends together. For higher-resolution measurements, a larger
number of separate spots can be detected within a streak as observed
experimentally. This simulation illustrates that observed splitting of a white
beam reection (hkl) into spots depends on the following parameters:
density of unpaired dislocations inside the wall, misorientation angle
created by the dislocation sub-boundary, number of walls in the irradiated
volume, size of a cell block, size of scattering volume and experimental
resolution function. Based on the above analysis we can make the following
conclusions:
The Laue pattern is mainly affected by the GNDs and/or GNBs(related
to correlated deformations of the lattice), if n+ L 0.1 nl. For
example, this condition is valid if L = 1 m, n = 1011 cm2 , n+ = n.
In this case m m .
When statistically stored dislocations dominate in the dislocation
ensemble, m m , the Laue spots are sharp and almost isotropic.
The intensity of the Laue spot mainly relates to the total dislocation
density.
Micro-Laue diffraction is not sensitive to lattice rotation and plastic
deformation if the axis of an overall mesoscale rotation is parallel to
momentum transfer.
When the Laue intensity splits along the streak (K > 1), the dislocation
structure must be predominantly characterized by GNBs.
44
45
ij = ij + ije
(1.48)
46
Fig. 1.15. Indexed Laue pattern of the NiAl/Mo composite with growth direction along
[001] direction (left) and enlarged region with [00h]-type reections for individual Mo bers
and NiAl matrix (right).
47
Fig. 1.16. Depth-resolved near-surface strain gradients in (a) individual Mo bers and (b)
NiAl matrix.
The enlarged central (00h) Laue spot shows distinct inputs from
NiAl and different Mo bers. Depth-resolved measurements of the (006)
reection were performed with DAXM and monochromatic radiation
within the energy range 18.42018.700 keV for NiAl and 17.0917.47 keV
for Mo bers with 2eV step (Fig. 1.16) in the as-grown state. Both phases
demonstrate strong near-surface gradient of the inverse lattice parameter,
Q = 2/d. Mo bers are in compression along the growth direction and
NiAl matrix is under tension. The as grown strain in these composites is
caused by the difference between the thermal expansion coefcients and
lattice mismatch. Near the surface these as-grown strains tend to relax to a
strain-free surface layer. The detailed micromechanical analysis of the strain
gradient in composites can be found elsewhere (for example, see Bei et al.
(2008) and Barabash et al. (2010a, 2010b, 2011a, 2011b)). Interfaces put
additional constraint on the process of strain relaxation. A non-monotonic
strain gradient in NiAl matrix (Fig. 1.16b) is caused by characteristic strain
uctuations at the interfaces between the matrix and the bers (see Barabash
et al. (2011a) for details). Other examples of depth-resolved measurements
are described in Chapter 2 and Chapter 10.
1.17. Conclusions
Hierarchical mesoscale dislocation arrangements displace atoms in the
crystals and change diffraction conditions. Statistically stored dislocations
create locally uctuating overlapping strain elds, and result in the
broadening of reections along the radial direction of the momentum
transfer. Full width half maximum of the broadened distribution scales
48
Acknowledgments
This research was supported by the U.S. Department of Energy, Ofce
of Basic Energy Sciences, Materials Sciences and Engineering Division.
The use of the APS was supported by U.S. Department of Energy, Ofce
of Basic Energy Sciences, and the Scientic Users Facilities Division.
Measurements were performed at 34 ID-E of the Advanced Photon Source.
The authors wish to thank Dr. J. Vitek (ORNL) for carefully reading the
manuscript.
References
Arsenlis, A. and Parks, D.M. (1999). Acta Mater., 47, 15971611.
Arsenlis, A., Parks, D.M., Becker, R. and Bulatov, V.V. (2003). J. Mech. Phys.
Solids, 52, 1213.
Balzar, D., Audebrand, N., Daymond, M.R., Fitch, A., Hewat, A., Langford, J.I.,
Le Bail, A., Louer, D., Masson, O., McCowan, C.N., Popa, N.C., Stephens,
P.W. and Toby, B.H. (2004). J. Appl. Crystallogr., 37, 911924.
Barabash, R.I. (2001). Mater. Sci. Eng., A309A310, 4954.
49
Barabash, R.I., Bei, H., Gao, Y.F. and Ice, G.E. (2011a). Scripta Mater., 64,
900903.
Barabash, R.I., Bei, H., Gao, Y.F. and Ice, G.E. (2010a). Acta Mater., 58,
67846789.
Barabash, R.I., Bei, H., Gao, Y.F., Ice, G.E. and George, E.P. (2010b). JMR, 25,
199206.
Barabash, R.I., Bei, H., Ice, G.E., Gao,Y.F. and Barabash, O.M. (2011b). JOM-US,
63, 3, 3034.
Barabash, R.I., Demin, S., Krivoglaz, M. and Chuistov, K. (1990). J. Appl.
Crystallogr., 23, 458561.
Barabash, R.I., Ice, G.E., Karapetrova, E.A. and Zschack, P. (2012). Metall. Mater.
Trans. A, 43A, 14131422.
Barabash, R.I., Ice, G.E., Kumar, M., Ilavsky, J. and Belak, J. (2009). Int. J.
Plasticity, 25, 20812093.
Barabash, R.I., Ice, G.E., Larson, B.C., Pharr, G.M., Chung, K.-S. and Yang, W.
(2001). Appl. Phys. Lett., 79, 749.
Barabash, R.I., Ice, G.E., Tamura, N., Valek, B.C., Bravman, J.C., Spolenak, R.
and Patel, J.R. (2003b). J. Appl. Phys., 93, 9, 57015706.
Barabash, R.I., Ice, G.E., Tamura, N., Valek, B.C., Bravman, J.C., Spolenak, R.
and Patel, J.R. (2004). Microelectron. Eng., 75, 1, 2430.
Barabash, R.I., Ice, G.E. and Walker, F. (2003a). J. Appl. Phys., 93, 3, 14571464.
Barabash, R.I. and Krivoglaz, M.A. (1978). Fiz. Met. Metalloved., 45, 718.
Barabash, R.I. and Krivoglaz, M.A. (1982). Metalloz. Nov. Tech., 4, 310.
Barabash, R.I., Krivoglaz, M.A. and Ryaboshapka, K.P. (1976). Fiz. Met.
Metalloved., 41, 3343.
Bei, H., Barabash, R.I., Ice, G.E., Liu, W., Tischler, J. and George, E.P. (2008a).
Appl. Phys. Lett., 93, 071904.
Bei, H. and George, E.P. (2005). Acta Mater., 53, 6977.
Bei, H., Shim, S., George, E.P. and Pharr, G.M. (2008b). Acta Mater., 56,
47624770.
Breuer, D., Klimanek, P. and Pantleon, W. (2000). J. Appl. Crystallogr., 33,
12841204.
Busing, W.R. and Levy, H.A. (1966). Acta Crystallogr. A, 22, 457464.
Chakravarthy, S.S. and Curtin, W.A. (2011). PNAS, 38, 1571615720.
Chung, J.-S. and Ice, G.E. (1999). J. Appl. Phys., 86, 5249.
Gao, H., Huang, Y., Nix, W.D. and Hutchinson, J.W. (1999). J. Mech. Phys. Solids,
47, 12391263.
Greer, J.R., Oliver, W.C. and Nix, W.D. (2005). Acta Mater., 53, 1821.
Greer, J.R., Oliver, W.C. and Nix, W.D. (2006). Acta Mater., 54, 1705.
50
51
Larson, B.C., Tischler, J.Z., El-Azab, A. and Liu, W. (2008). J. Eng. Technol., 130,
021024-110.
Larson, B.C., Yang, W., Ice, G.E., Budai, J.D. and Tischler, J.Z. (2002). Nature,
415, 887890.
Larson, B.C. and Young, F.W. (1987). Phys. Status Solidi A, 104, 273286.
Lauridsen, E.M., Jensen, D.J., Poulsen, H.F. and Lienert, U. (2000). Scripta Mater.,
43, 561.
Levine, L.E., Larson, B.C., Yang, W., Kassner, M.E., Tischler, J.Z., Delos-Reyes,
M.A., Fields, R.J. and Liu, W. (2006). Nat. Mater., 5, 619.
Maas, R., Van Petegem, S., Van Swygenhoven, H. and Derlet, P.M. (2007). Phys.
Rev. Lett., 99, 145505.
Margulies, L., Winther, G. and Poulsen, H.F. (2001). Science, 291, 23922394.
May, C., Klimanek, P. and Magerl, A. (1995). Nucl. Instrum. Meth. A, 357,
511518.
Mughrabi, H. (1983). Acta Metall. Mater., 31, 9, 13671379.
Mughrabi, H. and Obst, B. (2005). Z. Metallkd., 96, 7, 586695.
Mughrabi, H. and Ungar, T. (2002). Long-range internal stresses in deformed
single-phase materials: The composite model and its consequesces, in
Nabarro, F. R. N. and Hirth, J.P. (eds), Dislocations in Solids, Elsevier,
Chapter 60, p. 345.
Nabarro, F.R. N. (1987). Theory of Crystal Dislocations, New York: Dover
Publications.
Nabarro, F.R. N. (1994). Scripta Metall. Mater., 30, 10851987.
Nabarro, F.R. N. (2001). Mater. Sci. Eng. AStruct., 317, 12.
Needleman, A. (2000). Acta Mater., 48, 105124.
Nix, W.D. and Gao, H. (1998). J. Mech. Phys. Solids, 46, 411425.
Noyan, I.C. and Cohen, J.B. (1987). Residual Stress, New York: Springer-Verlag.
Nye, J.F. (1953). Acta Metall. Mater., 1, 153162.
Ohashi, T., Barabash, R.I., Pang, J.W. L., Ice, G.E. and Barabash, O.M. (2009).
Int. J. Plasticity, 25, 920941.
Pantleon, W. (2002). J. Mater. Res., 17, 9, 24332441.
Patterson, A.L. (1959). International Tables for X-ray Crystallography, Vol. II,
Birmingham, UK: Knoch, p. 61.
Rebonato, R., Ice, G.E., Habenschuss, A. and Bilello, J.C. (1989). Phil. Mag. Lett.,
A60, 571.
Rollett, J.S. (1965). Computational Methods in Crystallography, Oxford, UK:
Pergamon Press.
Romanov, A.E. and Vladimirov, V.I. (1992). In Nabarro, F.R. N. and Hirth, J.P.
(eds), Dislocations in Solids, Elsevier, Vol. 9, pp. 191402.
52