circles which, if concentric with the primitive, are famil- to if you wanted to form an image with the 022

to if you wanted to form an image with the 022  reflec-

iar as lines of latitude. (Note, however, that most small tion? (One answer is the [011] zone axis, but why?). For
circles are not concentric with the primitive.) We can the same specimen, if you want to excite the  111 reflec-
then rotate the Wulff net, as we wish, to realign our tion, you could tilt toward the ½011 zone axis keeping
great circles. the 200 reflection excited, not toward the [011] zone axis.

& & You could work this out using equations, but the
We can represent plane normals (also called poles)
and directions on the same projection even if they are stereographic projection tells you what to do while
not parallel to one another. Better still, we can read you are sitting at the microscope.
& If you are working with a non-cubic material, buy a
off the angles between them. Remember that, in
general, the normal to the plane (hkl) is parallel to large Wulff net and construct your own stereo-
the direction [hkl] only for cubic materials. graphic projection; you can buy standard projec-
& The zone axis is always 908 away from any plane tions for cubic materials so, as usual, it’s easier to be
normal that is in its zone. All the plane normals in a a metallurgist than a ceramist or mineralogist.
& Use a program like EMS to help you plot the points,
particular zone, [UVW], will lie on a single great
circle. or download appropriate software from the Web
& The angle between any two planes is the angle (e.g., URL #1).
& Wulff nets and plotting software are also available
between their plane normals, measured along a
great circle using the Wulff net. on-line (e.g., URL #2).
& We can use the same construction to summarize all the
symmetry elements of any particular crystal system.
18.4 INDEXING SINGLE-CRYSTAL DPs
These poles at the UVW zone axis represent the
possible diffracting planes for that zone; [UVW] is the Remember the fundamental relationship in a DP
beam direction. So if [UVW] is in the center of the projec- (Section 9.6.B).
tion, the hkl reflections will be around the circumference
of the projection (the primitive great circle). Now you Rd ¼ lL (18:1)
should appreciate why stereograms can be so useful in
interpreting DPs. Any distance between the direct beam and a specific dif-
Several examples of stereographic projections are fraction spot or the radius of a diffraction ring, R, which
shown in Figure 18.4. Look at the Wulff net and check we measure on the DP, is related to a specific spacing
some simple facts. For example, for the cubic system, between planes in the crystal, d. Since lL is a constant,
check which poles are 908 away from the [001] direction. we can measure several values of R and know that
How large is the angle between ð011Þ and (011)? How
would this angle change if the material were forced to be R1 d1 ¼ R2 d2 ¼ R3 d3 ¼ R4 d4 ¼ . . . (18:2)
tetragonal with c/a > 1? What happens in this case to the
ð111Þ pole or the ð110Þ pole? Now consider the more If you know the lattice parameter of your crystal,
extensive plot shown in Figure 18.5. If the specimen is then you know the allowed reflections and only certain
cubic with the [001] foil normal, what pole would you tilt d-spacings will be associated with diffraction spots.

(A) (B) (C)

FIGURE 18.4. Some standard cubic stereographic projections. The pole in the center defines each projection, so these are 001, 011 and 111.

1 8 . 4 I N D E X I N G S I N G L E - C R Y S T A L D P S ................................................................................................................................................ 287
FIGURE 18.5. The stereographic projection for a cubic foil with a [001] normal, assuming the beam is down [001] also. If you want to form an image with
the 022 reflection, you need to tilt the specimen so the 011 pole rotates until it is on the primitive, i.e., it is 908 from the beam direction. To do this, you need
to tilt about a pole that is 908 from the 022 reflection, such as the [100], [111] or [311] zone axes.

Table 18.1 lists allowed and forbidden reflections for Once you have tentatively identified possible values
some cubic systems. Rules for more crystal systems are for g1 and g2, you need to cross-check your answers
given in Table 16.2. using the angles between the g vectors (i.e., the angles
between the plane normals). The fully indexed patterns
at the end of this chapter (Figures 18.19–18.21) show the
R AND d-SPACINGS
principal interplanar angles and the principal ratios of
The ratio of any two R values gives the inverse ratio
g1/g2. Hence, in practice, you will rarely have to measure
of the d-spacings.
more than two or three spacings in order to index a

288 ....................................................................................... O B T A I N I N G AND INDEXING PARALLEL-BEAM DIFFRACTION PATTERNS

 TABLE 18.1. The Selection Rules for Cubic Crystal Structures U1 U2 þ V1 V2 þ W1 W2
cos r ¼  1=2  2 1=2 ð18:4Þ
bcc fcc Diamond cubic U1 þ V12 þ W12
2 U2 þ V22 þ W22
h2 + k2 + l2 hkl h2 + k2 + l2 hkl h2 + k2 + l2 hkl
Remember that you can always work out these expres-
2 110
sions for any crystal system using the equation for the
3 111 3 111
dot product of the two appropriate vectors.
4 200 4 200 4 200
6 211
In principle, if we don’t know the crystal structure,
8 220 8 220 8 220 we can still work out the d-spacings of the diffracting
10 310 planes using equation 18.1. However, you should
11 331 11 331 remember that SAD is not the most accurate method
12 222 12 222 for determining the spacing of lattice planes, dhkl, or the angles
14 321 between them, f. SAD is generally very good at distinguish-
16 400 16 400 16 400 ing patterns, but it completely fails when the difference
18 411 between the two patterns is a 1808 rotation, as occurs in
330 some patterns of polar material, like GaAs or GaN.
19 331 19 331
To summarize
20 420 20 420
22 332 &
24 422 24 422 24 422
Tilt your specimen to a low-index pole.
26 431
& Set s = 0 for the innermost reflections.
& Record the SADP.
27 511 27 511
& Repeat the exercise using higher-order reflections
27 333 27 333
30 521 after tilting the specimen to set s ¼ 0. These measure-
32 440 32 440 32 440 ments will be more accurate, but only if you make
sure that s ¼ 0.

particular zone-axis DP. However, if your specimen is The discussion on relrods in Chapter 16 told you
not oriented close to a zone axis, you’ll need to look that both d and f could be seriously in error if reflec-
ahead to Section 18.10. tions are not set to have s ¼ 0, especially since you’ve
probably tilted the specimen.
So far, you have only indexed one DP. You’ll prob-
ably need more than one to determine orientation rela-
WEISS ZONE LAW
tionships. While you’re at the microscope, tilt to pole #2
Check the consistency of your indexing using the
keeping g1 (see Figure 18.6) strongly excited. Repeat the
Weiss zone law. Each hkl reflection must lie in the
[UVW] zone, i.e., hU+kV+lW= 0.

G3 pole #1
The Weiss zone law only holds for diffraction into the G1
zeroth layer of the reciprocal lattice and the diffraction G2
spots that we see here are called the zero-order Laue zone
(ZOLZ). There are circumstances when we can see diffrac- O
tion from higher-order Laue zones (HOLZ) and we’ll talk
a lot more about this phenomenon in Chapters 20 and 21.
All SADPs, particularly those that we talk about and pole #2
index in this and the next chapter, are ZOLZ patterns. G2
The angle between normals to the planes (h1k1l1) and G1
(h2k2l2) is f; the angle between directions [U1V1W1] and
O
[U2V2W2] is r. You can work these out and cross-check
them with your DPs. These are standard equations in O
many texts (e.g., those by Edington and Andrews et al.). pole #3
You’ll probably find that the equations for the cubic
system are the most useful.
FIGURE 18.6. How to confirm your indexing of reflections and poles by
tilting to other poles. Start with g1 and g2 strongly excited at pole #1. Tilt
h1 h2 þ k1 k2 þ l1 l2 to pole #2 keeping g1 strong, then go back to pole #1 and tilt to pole #3,
cos f ¼  1=2  2 1=2 (18:3)
h21 þ k21 þ l12 h2 þ k22 þ l22 keeping g2 strong. Index all the strong reflections each time, measure the
tilt angles between each reflection and estimate the tilt between poles.

1 8 . 4 I N D E X I N G S I N G L E - C R Y S T A L D P S ................................................................................................................................................ 289
 nested spheres. When we intersect these spheres with the
Ewald sphere (which, in the TEM, approximates to a
plane) we will see the rings which are recorded in powder
patterns.

NANO AND DPs

Smaller grains give broader spots in the DP. Nano-
crystals should give the largest spots, or. . .

If a polycrystal is textured, then there is usually one

special plane which is common to nearly all the grains.
Since the grains are small, all the reciprocal-lattice
points will be broadened by the shape effect; so will
the sphere or circles for the polycrystals. This situation
will be exacerbated for nanocrystals.
If we then rotate the reciprocal lattice about the
lattice vector normal to the texture plane, we pro-
FIGURE 18.7. A practical illustration of the procedure described in duce a set of circles in reciprocal space, as shown in
Figure 18.6 for an fcc material, in this case MgO. Figure 18.8. If we are examining cubic materials, the
reciprocal-lattice vector ghkl will be parallel to the direc-
indexing procedure. Go back to pole #1 and tilt to pole tion [hkl] in real space. Otherwise this will not generally
#3 keeping g2 strongly excited. You can repeat this be the case.
indexing as many times as you wish. The important The DP in either of these examples appears as shown
idea is that you now have angular measurements allow- in Figure 18.9A and B which differs because the grain
ing you to cross-check your determination of both g1
and g2 and the zone axes. Of course the task is simple for
an fcc crystal, as you can see in Figure 18.7, which is an [UVW ]
experimental illustration of this procedure. The chal- N = hU + kV + lW
lenge comes when the crystal has less symmetry. If you
already know the crystal structure, then you should plot
out the most important poles, relating their orientations 2
to one another (more on this in Section 18.11 and
Chapter 19) and pay particular attention to the infor-
mation from systematic absences which occur when the
structure factor is zero (go back and check Section 3.9 1
and Chapter 13).

THE GOLDEN RULE

Make the task as easy as possible. Record ALL the 0
DPs you might need and how they relate to one
another (draw road maps) while you’re at the
microscope.
–1

18.5 RING PATTERNS FROM Origin

POLYCRYSTALLINE MATERIALS of reciprocal
lattice
Diffraction from polycrystalline specimens (especially
when the grain sizes are at the nanoscale) can be viewed FIGURE 18.8. The generation of a set of circles in reciprocal space by a
textured polycrystal. When the reciprocal lattice is rotated about a partic-
in much the same way as X-ray diffraction from pow- ular direction [UVW] (in this case the normal to the texture plane) each
ders. For a completely random polycrystal, we rotate Laue zone (N ¼ 1, 2, etc.) produces a set of concentric circles for each
the reciprocal lattice about all axes and produce a set of allowed reflection in each zone.

290 ....................................................................................... O B T A I N I N G AND INDEXING PARALLEL-BEAM DIFFRACTION PATTERNS

 (A) CDF image with the arc of diffracted intensity. In Figure
18.10C, these oriented grains are uniformly distributed,
but you might encounter a situation where this is not the
case. Think how the pattern might differ in this case.
Figure 18.10D and E emphasizes that these ring
patterns can be quite varied. In this case the specimen
is a-Ag2Se, which is textured about an axis inclined to
the beam. When the Ewald sphere cuts the circles now, it
produces elongated spots which lie on an ellipse.
Vainshtein et al. point out that all the ‘spots’ on one
ellipse can be indexed with the same hk indices but a
different l, and call such a pattern an oblique-textured
electron DP. You should also be careful in indexing
these textured patterns since not all possible dhkl values
need be present, depending on the texture plane.
There is more information in ring patterns than spot
patterns. Like a powder pattern, you could estimate the
grain size from the width of the rings, but it’s more direct
(and more reliable) to just look at the DF image. You
can see kinematically forbidden rings because you don’t
necessarily have single scattering from each grain but
(B) there are ways around this problem.
Nanocrystalline materials, which fall into our small-
est range of 10-nm grain size, are very challenging
because there’s probably always more than one grain
through the foil thickness. Careful DF imaging com-
bined with HRTEM or CBED with the smallest possible
probe (often called nanodiffraction, see Section 21.8.B),
is probably optimal, but you need to look for clustering
of similarly oriented grains.

18.6 RING PATTERNS FROM

HOLLOW-CONE DIFFRACTION

We can combine the advantages of ring patterns from

small-grained materials with individual spot patterns
from larger-grained materials, As we noted back in
Section 9.3.D, if you have a nano/microcrystalline speci-
men, it is rather inefficient to get a DP from one grain at
a time by carefully tilting your specimen and then get-
FIGURE 18.9. (A) and (B) Ring DPs from polycrystalline foils. In (A) the ting individual DF images of single areas, or of phases
grain size is larger than in (B) so the rings are made up of discrete spots. A within a single grain, by carefully tilting the incident
finer grain size, as in (B) produces a more continuous ring pattern, but the beam. So we can use a computer to continuously change
width of the rings of diffracted intensity in fact becomes broader and can
the beam orientation so it samples many angles of inci-
be used as an inverse measure of the grain size.
dence and gather DPs from all the crystals superim-
posed on one another (i.e., get a ring pattern, even
from a relatively large-grain material). This is the prin-
size is different. A larger grain size gives a more speckled ciple behind hollow-cone (sometimes called ‘conical’)
pattern. diffraction (go back and look at Figure 9.15). In the
You can distinguish the pattern produced by a textured same section we also introduced the idea of hollow-
specimen from one produced by a random polycrystal by cone DF imaging. In earlier TEMs, before computer
tilting. If your specimen is textured, the rings become arcs control of the beam, hollow-cone illumination was
as shown in the pattern in Figure 18.10A together with the achieved by using an annular C2 aperture rather than
Ewald-sphere construction in Figure 18.10B. You can a circular hole. Thus the beam that came through the
locate the grains which give rise to the arcs by forming a annulus irradiated the specimen from a fixed angle

18.6 RING PATTERNS FROM H O L L O W - C O N E D I F F R A C T I O N ........................................................................................................... 291

 (A) (D) Plane of Ewald sphere
Intersection with
Ewald sphere

c*

Point
(B) on
reciprocal
lattice

c*
Rotation axis

Planar
Ewald
sphere

hkl φ
b*
(E)

0
a*
hk0

(C)

FIGURE 18.10. (A) A textured ring pattern where the rings are more intense over a certain angular range. (B) The corresponding interception of
the Ewald sphere (plane) with the reciprocal lattice. (C) A DF image of the textured grains, taken from a brighter potion of one of the hkl rings, showing
an equiaxed structure. In (D) the specimen is textured about a direction at an angle to the beam, so the Ewald sphere creates elongated spots or arcs in the
DP (E).

292 ....................................................................................... O B T A I N I N G AND INDEXING PARALLEL-BEAM DIFFRACTION PATTERNS

around the optic axis, as shown schematically back in materials, but we should first say what we mean by
Figure 9.15A. It’s much easier and more flexible to use ‘amorphous.’
the computer to control the pre-specimen scan coils
An amorphous material is one where the locations
and bring the beam onto the specimen at a fixed (or a
of the neighboring atoms are defined by a prob-
range of) conical scan angle and a typical conical-scan
ability function such that the probabilities are
DP is shown in Figure 9.15B. (If you look ahead, you
never unity.
can see another example of a hollow-cone DF image
from an amorphous DP in Figure 18.14C.) Of course, This idea is best illustrated by a plot of the probability
we index the rings in exactly the same way as we which we call the radial distribution function (RDF). The
described in Section 18.5 for ring patterns obtained RDF, r(r), is the probability, per unit element of volume,
from a polycrystalline specimen where the grain size that an atom will be found at a distance r from another
is much smaller than the beam size. atom. The first example in Figure 18.11A compares the
With hollow-cone illumination, we can create a DF curves for liquid sodium and crystalline sodium; the
image of all the crystals in a film that are diffracting numbers on the crystalline curve remind us that in the
from their {hkl} planes, rather than imaging only that crystal each sodium has eight nearest neighbors, etc. The
small fraction diffracting into the part of the diffraction second plot, Figure 18.11B, shows the RDF for vitreous
ring that happens to be selected by the aperture. The silica. This time the peaks are associated with distances
difference in information content between these two
approaches is shown back in Figure 9.15B and C. The
hollow-cone technique is widely used in CBED (see (A)
Chapter 21) and has more recently been used as the Average +
basis for precession diffraction (see Section 18.8.) 25
density
4πr ρ(r)
2
+
+
+ RDF for
20 liquid Na
18.7 RING PATTERNS FROM +

AMORPHOUS MATERIALS
+
15
24 24
+
Amorphous materials used to represent somewhat of a + Neighbors in
fringe field in materials science but the rapid growth in 10 + Crystalline Na
+
glass technology over the last decade, fueled by the 12
optical-communications bubble, the development of 8 8
5 6 6
bulk metallic glasses and the increased awareness of
the role of glassy films at interfaces (particularly the 0 2 4 6 8 10
gate oxide in semiconductors) have all conspired to r in Å
crystallize the following question: is the material really
amorphous or is it (sub) nanocrystalline? Actually this
question is still debated when discussing both amor- (B)
Si¥Si
phous materials and, more intensely, oxide and metallic ∑Km4πr2ρ(r) Si¥0
glasses because there’s still no broadly accepted defini- 0¥0
tion of when grains become small enough that a deter- 10,000 Si¥0 Si¥S i
minable crystal structure no longer exists. At best we 0¥0
should think of a continuum of states from nanocrystal-
line to amorphous.
6000

DPs AND AMORPHOUS MATERIALS

The DP from an amorphous material looks similar to Vitreous SiO2
that from polycrystalline material but the rings are 2000
broader and there is no speckle.
0 2.0 4.0 6.0
Rudee and Howie showed that electron scattering r in Å
from regions of 1.5 nm diameter could be coherent.
FIGURE 18.11. (A) Radial distribution functions for liquid Na and the
Graczyk and Chaudhari proposed modeling these average density curve superimposed on the distribution of the nearest
materials as random networks. If we are careful, we neighbors in crystalline Na (vertical lines). (B) The RDF for vitreous SiO2
can learn quite a lot about the structure of amorphous is peaked at a distance that represents spacings between Si and O atoms.

18.7 RING PATTERNS FROM A M O R P H O U S M A T E R I A L S ..................................................................................................................... 293

between different pairs of Si and O atoms. The features to
notice are

& The two curves both show definite peaks.

& The two curves are different.

Some diffraction theory. Since these materials are so

different, we’ll give a brief introduction to the theory of
scattering from amorphous materials. We make the
assumption that the electron beam is only scattered
once; this is kinematical, but it’s more realistic than for
crystals at the Bragg condition. Following Howie, we
express the kinematical intensity, I(k), by the expression
X
IðkÞ ¼ jfðkÞj2 ei2pkðri rj Þ (18:5)
i;j

Here we assume that there are N identical atoms con-

tributing to the scattered intensity and they are located
at the different positions (ri or rj).
The f(k) terms are the atomic scattering amplitudes,
with k reminding us that there is an angular dependence
to f. If the material is isotropic, we can simplify equation FIGURE 18.12. An intensity profile across an energy-filtered DP from
18.5 as follows amorphous Si obtained by scanning the pattern across the entrance slit to
  a serial EELS spectrometer and recording only the elastic (on-axis)
2 FðkÞ electrons.
IðkÞ ¼ NjfðkÞj 1 þ (18:6)
k

where
X
FðkÞ ¼ ei2pkðri rj Þ (18:7)
i6¼j

ð
FðkÞ ¼ k rðrÞei2pkr dV (18:8)

ð1
FðkÞ ¼ 4p rðrÞ sinð2pkrÞr dr (18:9)
0

The term r(r) is the RDF. Equation 18.9 can be inverted

to give an expression for r
ð
1 1
rðrÞ  r0 ¼ FðkÞ sin 2pkr dk (18:10)
r 0
FIGURE 18.13. A computer plot of the diffracted intensity distribution
This equation means that, in principle, the RDF can be from an amorphous structure, showing diffuse rings of intensity. The
obtained directly from DPs. This process is enhanced if direct-beam intensity is off-scale.
the patterns are energy-filtered to remove inelastic con-
tributions (see Chapter 37 and the work of Cockayne
et al.) as shown graphically in Figures 18.12 and 18.13. showed clearly that the structure correlation can extend
Compare these figures with Figure 2.13A. Other ways to to 1.5 nm or more. Life gets more interesting as we
determine the RDF include extended X-ray absorption explore the structure out beyond the nanometer
fine structure (EXAFS) studies in a synchrotron X-ray level and Treacy et al. review a new imaging technique
and, less expensively but more noisily, via EXELFS (see called fluctuation microscopy which is a hybrid imaging/
Chapter 40). Alternatively, we can rearrange equation diffraction technique sensitive to the presence of me-
18.6 again to give a ‘reduced-intensity function’ as illu- dium-range order in amorphous materials. We’ll discuss
strated by the work of Graczyk and Chaudhari who this technique more in Chapter 29. If you get the idea that

294 ....................................................................................... O B T A I N I N G AND INDEXING PARALLEL-BEAM DIFFRACTION PATTERNS

(A) (B) (C) 18.8 PRECESSION DIFFRACTION
Precession is a relatively new approach to performing
electron diffraction; it is similar to the X-ray method
which goes by the same ‘name’ (although in XRD the
crystal, rather than the beam, is precessed) and it is
closely related to hollow-cone diffraction that we just
talked about. The main advantage that precession brings
is to remove strong dynamical effects from the patterns
leaving high-quality kinematical data, eliminating, e.g.,
extra spots due to dynamical effects. Precession diffrac-
tion can also be used for crystal structure and full sym-
metry determination. While the method was first tried in
a TEM by Vincent and Midgley, recent advances in both
CCD cameras and (you’ve guessed it) aberration correc-
FIGURE 18.14. (A) BF image of amorphous carbon. (B) DF image from tion make both the operational aspects and the interpre-
the diffuse diffracted intensity taken with a defocused beam and (C) tation of precession DPs much easier.
hollow-cone image showing more structure.
In precession diffraction we double-deflect the inci-
dent beam (either parallel or convergent) using the usual
DF scan coils in a circular hollow cone (radius G and
the study of amorphous materials is at the frontiers of angle C) about a centered zone-axis direction (Figure
TEM materials-structure determination, you are right! 18.15) and de-scan the beam onto the plane of the DP.
To summarize this discussion, the scattering theory So it is really a double conical beam-rocking system.
is well known but the capability for routinely removing
the inelastic contribution is only now becoming avail-
able and is still not commonplace. Probably the best
way to answer whether a material is nanocrystalline or
amorphous will come from a combination of SAD and
EELS. A BF image of amorphous material is generally
uninformative (Figure 18.14A), but if you try to form a Beam-tilt
DF image you will see a speckle of white spots against a coils
dark background, as shown in Figure 18.14B. The size
of the speckle increases as the defocus increases, so be
wary of interpreting the image in terms of the size of G
regions in the amorphous structure. Hollow-cone DF
Objective
imaging, as shown in Figure 18.14C, gives even more
pre-field
and finer ‘structure’ in the image. The fact that you can
produce this type of speckled contrast is important C
because you may well want to study small particles
(e.g., catalysts) supported by an amorphous film. In Positions of Eucentric
such a case, you need to know what the image of the specimen z specimen
support film looks like before you add a new plane
component.
Glass at interfaces and grain boundaries. Another De-scan coils
area where it is important to know whether or not an
amorphous material is present, occurs in the analysis of
grain boundaries in ceramic materials or gate oxides in
semiconductor devices. Answering this problem
severely taxes the limits of HRTEM when the film
thicknesses are < 1 nm, which is the case for the gate
oxide in the latest semiconductor devices. For somewhat
thicker films we can use conventional HRTEM and
I
another technique, known as diffuse-dark-field (DDF)
imaging, which essentially forms an image from the
II
region in the SADP where the amorphous ring would
be, if glass were present. We’ll return to DDF imaging in
Chapter 29. FIGURE 18.15. Schematic ray diagram for precession diffraction.

1 8 . 8 P R E C E S S I O N D I F F R A C T I O N ............................................................................................................................................................. 295
(A) (B)

FIGURE 18.16. Si <130> DP taken (A) with a small precession angle (18) and (B) with a large precession angle (38). The kinematically forbidden
reflections 002, 006, 0010 disappear at the larger precession angle. Note also the appearance of strong differences in the intensity of individual hkl maxima.

Unlike standard TEM hollow-cone where we are seek- Precession is an area of electron diffraction that is
ing to get multiple (i.e., ring) patterns sampling many seeing a rapid growth in applications across many fields.
grains, in precession mode we are usually trying to Commercial retrofits are available for any TEM (e.g.,
average out the diffraction conditions within a single URL #3) but, if you want to do it on your own, then
grain and obtain a single-crystal pattern in which dynam- Own et al. give good detailed instructions.
ical effects are reduced. If the hollow-cone angle is
large enough, all the diffraction data correspond to a
two-beam condition and show reduced dynamical dif- 18.9 DOUBLE DIFFRACTION
fraction because not many reflections are simulta-
neously excited off the zone-axis condition. When we Double diffraction occurs when a diffracted beam trav-
precess the beam, we integrate the diffracted intensities eling through a crystal is rediffracted either within the
through the Bragg condition, so small specimen-tilt same crystal or when it passes into a second crystal. If
changes are averaged out and we can more reliably the initial diffraction vector of the beam is g1 and it is
interpret crystal-symmetry information in the DP. Con- rediffracted by reflection g2 , then the resultant diffrac-
tinuous changing of the precession angle over several tion vector of the double-diffracted beam is (g1–g2). If g2
degrees can help with point and space-group determina- is not an allowed reflection in the first crystal, the dou-
tion (see Chapter 21, the companion text and Morniroli ble-diffracted beam is characteristic of neither the first
et al.’s paper). Figure 18.16A and B shows, respectively, nor the second crystal.
the differences between small and large precession-angle Reflections attributable to double diffraction are a
DPs from a Si single crystal. The larger angle removes common feature of DPs recorded from two-phase mate-
the kinematically forbidden reflections that are present rials exhibiting epitaxy or topotaxy including, e.g., oxi-
in the smaller-angle DP because the multiple-diffraction dized metallic specimens. Quite complicated patterns
paths to them are unlikely to occur during the preces- may be formed, requiring careful analysis to distinguish
sion movement. Most of the time, only the direct beam the ‘real’ reflections from the double-diffraction reflec-
and one diffracted beam are strongly excited as the tions. Double diffraction is directly responsible for the
beam precesses. This does not mean that the dynamical moiré effect in the electron images that we will discuss in
interactions disappear, they just occur between the Chapter 23. As an example of this effect, we’ll consider
direct beam and one diffracted beam (as in the two- small a-Fe2O3 (hematite) islands grown on a single-
beam case). crystal a-Al2O3 (alumina or sapphire) substrate as

296 ....................................................................................... O B T A I N I N G AND INDEXING PARALLEL-BEAM DIFFRACTION PATTERNS

 shown in Figure 18.17A. The position of the double-
(A)
diffraction spots relative to the hematite and alumina
reflections actually changes depending on whether the
islands were on the top or bottom surface of the speci-
men. This particular top–bottom effect can be derived
from simple geometry; however, dynamical diffraction
effects must also be considered when the materials are
thicker.
Figure 18.17B shows the [0001] SADP recorded
from one of these a-Fe2O3 particles. The closest reflec-
tions to the direct beam are the six f11 20g reflections.
The next closest reflections are the six f3 300g reflec-
tions, only four of which are visible in the figure.
Double-diffraction spots are visible around each of
(B) these primary reflections. They also surround the direct
beam, although they are hidden by the flare from that
beam in Figure 18.17B.
Figure 18.17C and D shows enlargements of regions
near the f1120g reflections in the [0001] SADPs
recorded when the hematite island was on the top sur-
face in (C) and on the bottom surface of the sapphire in
(D). Both g and g reflections are shown for the two
cases. In (C) the ring of six double-diffraction spots
surrounds the Al2O3 reflection while in (D) the double-
diffraction spots surround the Fe2O3 reflection.
The same observation can be made for the { 3300}
regions of the SADPs as shown in Figure 18.17E and F. In
(C) this case, an inner ring of double-diffraction spots (small
filled circles) with the same spacing and orientation as the
double-diffraction reflections in Figure 18.17C and D are
still visible, as are the outer rings of spots (large filled
(D) circles). In general, the outer ring of double-diffraction
spots is more intense than the inner ring.
This top-bottom effect in particular, and double
diffraction in general, can be explained by the simple
geometric analysis we show in Figure 18.18; the bottom
crystal is Al2O3, which has the smaller lattice param-
eter and therefore has the larger reciprocal-lattice vec-
(E) tors. Double-diffraction spots can be formed around
the primary hematite reflection, gH, by two different
routes

(F)
& 2gH þ gA (A: alumina, H: hematite) giving the
double-diffraction spot just inside gH.
& gH þ 2gA gives a double-diffraction spot just outside
gH.

FIGURE 18.17 (A) BF on-axis image of a particle of a-Fe2O3 on a-

These two routes at first appear to be equivalent.
Al2O3. (B) [0001} SADP from a-Fe2O3 showing double-diffraction spots
around the f11 20g and f3300g reflections. (C) Enlargements of regions However, if we take into account the curvature of the
near the f11 20g reflections when the hematite island is on the top surface. Ewald sphere, then the deviation parameters of the two
(D) Enlargements of regions near the f1120g reflections when the hema- routes are very different. In the case of diffraction through
tite island is on the bottom. (E) Enlargements of regions near the f3300g the upper crystal, the deviation parameter of the 2g beam
reflections when the hematite island is on the top surface. (F) Enlarge-
ments of regions near the f3300g reflections when the hematite island is on
is slightly more than twice that of the g beam. This differ-
the bottom. ence will not significantly affect the intensities from a very

1 8 . 9 D O U B L E D I F F R A C T I O N .................................................................................................................................................................... 297
(A) plus g (lower). So the double-diffraction spot, which
occurs on the same side of the diffraction spot from the
upper crystal as the diffraction spot from the lower crys-
tal, will be more intensely excited than the double-diffrac-
tion spot which occurs on the opposite side. In two
dimensions, for thin films, the strongest double-diffrac-
tion spots will always be those arranged symmetrically
around the diffraction spot from the lower crystal.
For thicker layers, the relative intensity of the 
g and
2g beams will vary as dynamical diffraction effects
occur. We can simulate the DPs from these structures
using the MacTempas program (see Chapter 30 and
(B) Section 1.6.B) The top-bottom effect is evident in the
case of 2.7 nm of hematite on 13 nm of alumina, but only
just discernible for the case of 2.6 nm of alumina on
13.5 nm of hematite. In the latter case, the dynamical
diffraction effects are stronger.
We will meet this topic in Chapter 23 when we dis-
cuss moiré fringes. We have made this analysis a little
more complicated than usual since we have considered
the details of where the spots will actually be found. You
can make this process simpler
& Trace the patterns from each crystal (if you know
what tracing paper is).
& Then construct a new pattern using each diffracted
beam from the upper crystal as an incident beam for
the lower crystal.
FIGURE 18.18. Top-bottom effect in double diffraction. The pattern
depends on which of the two crystals is on top. In this case a-Fe2O3 The extent of the moiré pattern gives you an idea of
particles are on top of the Al2O3 and the two (non-equivalent) paths (A) just how strong dynamical scattering is, even for thin films!
and (B) for double diffraction are shown. NOTE: gA is for alumina and gH
for hematite.
18.10 ORIENTATION OF THE SPECIMEN
thin epilayer due to streaking of the reciprocal-lattice spots
parallel to the beam direction (the shape-factor effect). Once you have identified three g vectors g1, g2 and g3 in
Now we can analyze the effects of diffraction a single-crystal DP, you can calculate the direction of
through the lower crystal the beam B. You can actually estimate B to within about
108 from the vector cross product as follows
& Draw the reciprocal lattice with the origin of the 2 3
Ewald sphere at 2gH for the first case and on gH in i1 i2 i3
6 7
the second. B ¼ g1  g2 ¼ 4 h1 k1 l1 5 (18:11)
& Keep the radius of the Ewald sphere unchanged h2 k2 l2
since only elastic interactions are considered.
& The incident beam for the lower crystal is in the 2g or ¼ ðk1 l2  k2 l1 ; l1 h2  l2 h1 ; h1 k2  h2 k1 Þ (18:12)

g directions for the two cases.
& The height of the ZOLZ is slightly different in the For the three-beam case, you can determine B with an
two cases since the deviation parameter at the origin accuracy of 38. You first need to make sure that the
must be zero. three vectors are taken in the correct order. Draw a
circle through these three reflections: if O is inside the
You can see from Figure 18.18A that the deviation circle, then the g vectors should be numbered counter-
parameter for 2gA is approximately zero, whereas for gA clockwise; if O is outside, number them clockwise.
it is of the same order as gH. The total deviation param- Check your labeling; the determinant of the matrix in
eter is thus much smaller for the second route than the equation 18.13 should be positive
first. A similar analysis for the inverted structure is shown 2 3
h1 k1 l1
in Figure 18.18B. In both cases, the deviation parameter 16 7
for the route g (upper) plus 2g (lower) produces a much g1 ðg2  g3 Þ ¼ 4 h2 k2 l2 5 (18:13)
V
smaller deviation parameter than the route 2g (upper) h3 k3 l3

298 ....................................................................................... O B T A I N I N G AND INDEXING PARALLEL-BEAM DIFFRACTION PATTERNS

A B

C D


FIGURES 18.19. Four standard, indexed, DPs for bcc crystals in the [001], [011], ½111 
and ½112 beam directions. Ratios of the principal spot spacings
are shown as well as the angles between the principal plane normals. Forbidden reflections are indicated by x.

Now we can write a weighted-average expression for B OUR CONVENTION

The vector B points up the column. It is normal to the
g2  g3 g3  g1 g1  g3 emulsion side of a photographic negative. The electron
B¼ 2
þ 2
þ (18:14)
jg 1 j jg2 j jg3 j2 beam travels along the direction –B.

18.10 ORIENTATION OF THE S P E C I M E N ................................................................................................................................................ 299

A B

C D

 and ½112
FIGURES 18.20. Four standard, indexed, DPs for fcc crystals in the [001], [011], ½111  beam directions. Ratios of the principal spot spacings are
shown as well as the angles between the principal plane normals. Forbidden reflections are indicated by x.

300 ....................................................................................... O B T A I N I N G AND INDEXING PARALLEL-BEAM DIFFRACTION PATTERNS

 A B
61.38° 61.38° 57.79° 57.79°
0112 0002 0112 2112 0002 2112

0111 N 0111 L
28.62° 28.62° 32.21° M 32.21°
M
L N
0110 0000 0110 2110 0000 2110

0111 0111

0112 0002 0112 2112 0002 2112

N =1.09 M =1.139 B =[2110] N =1.587 M =1.876 B =[0110]

L L L L
foil plane (2110) foil plane (0110)

D
C 30° 30°
56.67° 56.67°
30° 1210
30° 2201 0111 2021
2110 1100 0110
1120
L L L M
30° 30° 33.33° 33.33°
L N
1010 0000 1010 2110 0000 2110

1120 0110 1100 2110

2021 0111 2201

N =1.520 M =1.820 B = [0112]

B = [0001] foil plane (0001) L L

E F 26.03° 26.03°
40.5° 49.48°
1212
2110 1101 0112 0111 1101 2111

1121
40.5° L M 49.48° 63.97° M M 63.97°

L L
1011 0000 1011 1010 1010

2111 1101 0111 1121

0112 1101 2110

M =1.299 B = [0111] M =1.139 B = [1213]

L L

FIGURES 18.21. Six standard, indexed, DPs for hcp crystals in the ½2110;
 ½0110;
 ½0001; ½0112;
 ½0111
 and ½121 3
 beam directions. Ratios of the principal
spot spacings are shown as well as the angles between the principal plane normals. Forbidden reflections are indicated by x.

18.10 ORIENTATION OF THE S P E C I M E N ................................................................................................................................................ 301

 In Figures 18.19–18.21, we illustrate some of the & Notice that if you used the stereographic technique
most useful DPs for bcc, fcc and hcp crystals. You can with a non-cubic material, you would locate a direc-
extend these patterns as far as you wish using vector tion, not a plane normal.
addition; remember the reflections correspond to recip- & You can make the determination of B more accurate
rocal-lattice vectors. For example, in Figure 18.19C by making s ¼ 0 for each reflection you use and then
estimating your deviation from this idealized orien-
   
121 ¼ ð110Þ þ 011 (18:15) tation. If the specimen is thicker, use Kikuchi lines
(Chapter 19).
You can extend the patterns in this way and then
apply the selection rules to find the corresponding
patterns for Si, etc., using the specific examples as a 18.11 ORIENTATION RELATIONSHIPS
guide.
Once you’ve learned how to index a DP and determine
& bcc real space —> fcc reciprocal space. B, you can determine orientation relationships (ORs),
& fcc real space —> bcc reciprocal space. which are one of the most useful aspects of diffraction in
the TEM for the materials scientist and nanotechnolo-
Take the example used by Edington, as shown in gist because the orientation between different grains,
Figure 18.22 for an fcc crystal. Measure the distances phases or crystals controls many properties of engineer-
to the reflections x, y and z. Since the material is fcc, we ing materials. For example, the OR will determine how
can ratio d 2 values to find suitable indices or use a well the atomic planes in two different phases or grains
calibrated camera length. Thus we find that plane fit together and this controls the nature of the interphase
interface (e.g., coherent or incoherent) or grain bound-
A ¼ (420), B ¼ (111) and C ¼ (331); check that the
ary (special or random) which, in turn, governs the ways
angles are correct using
dislocations interact with the planar defect, thus con-
trolling the mechanical behavior of the material. There
gA gB are many other such effects of ORs; for example, we
cosðfAB Þ ¼ (18:16) often want to know how a fiber is oriented to the sur-
jgA jjgB j rounding matrix, a nanoparticle or a thin film to its
supporting substrate. So the OR between two different
and so on for fBC and fCA. You should immediately crystals is important in many materials and can be
recognize that this is the ½123 pole, but continue. Now described in one of two ways
you can plug pairs of these indices into equation 18.11
or all of them 18.14 to show that B ¼ ½123. & Two directions or plane normals (or two sets of
Finally, use the [001] stereographic projection. parallel planes) can be parallel in the two crystals
Draw a great circle that passes through the (111), (the parallel-plane/direction relationship). We use
ð210Þ and ð331Þ points using your Wulff net: they all this description for precipitate-matrix (b – a) orien-
lie on one great circle because they are in the same zone. tation relationships where the crystal systems may be
Now identify the zone axis directly by measuring 908 different.
from all the poles. The result is, of course, the same in & The two crystals have a common direction (axis) so
each case. that one crystal can be rotated through some angle

(A) (B) (C)

FIGURE 18.22. (A) An fcc pattern, indexed in (B) with the major indexed poles plotted on a stereographic projection in (C), identifying the pole of the
great circle as 
1
23 which is therefore the beam direction for the pattern in (A).

302 ....................................................................................... O B T A I N I N G AND INDEXING PARALLEL-BEAM DIFFRACTION PATTERNS