3

Elastic Scattering
CHAPTER PREVIEW

Elastically scattered electrons are the major source of contrast in TEM images. They also
create much of the intensity in DPs, so we need to understand what controls this process.
First we’ll consider elastic scattering from single, isolated atoms and then from many atoms
together in the specimen. To comprehend elastic scattering we need to invoke both particle
and wave characteristics of electrons.
Scattering from isolated atoms can occur either as a result of electrons interacting with
the negatively charged electron cloud, which results in angular deviations of only a few
degrees, or by attraction toward the positive nucleus which scatters the electrons through
much larger angles, up to 1808. Such scattering can often be interpreted in terms of billiard-
ball type, particle-particle collisions, cross sections, and mean free paths that we introduced
in the previous chapter. We’ll introduce the Rutherford differential cross section, which
explains the strong dependence of high-angle elastic scattering on the atomic number (Z) of
the atom. Later in the book, we’ll show how to use this Z dependence to form images that
reflect the chemistry of the specimen. When we treat the electrons as waves, their coherency
becomes important. The coherency of the scattered electrons is related to their angle of
scattering (y). As this angle becomes larger, the degree of coherency becomes less and
electrons that are Rutherford-scattered out to high angles are incoherent.
In contrast to Rutherford scattering, electrons that are scattered elastically through
small angles (which we’ll define as <38) are coherent. The intensity of this low-angle
scattering is strongly affected by the arrangement of atoms within the specimen. As we
introduced in the previous chapter, such collective scattering by the atoms is referred to as
diffraction and can only be understood if we treat the electron as a wave. Diffraction is
controlled mainly by the angle of incidence of the electron beam to the atomic planes in the
specimen, the spacing between these planes, and interatomic distances within the planes. So
this small-angle, coherent scattering is invaluable for characterizing the crystallography of
the specimen and is undoubtedly the most significant scattering phenomenon in the TEM.
So as we discuss elastic scattering, you’ll see that we often use the wave-particle duality
simultaneously, because both lines of thought are necessary for a full understanding.

3.1 PARTICLES AND WAVES & They can be scattered through particular angles
(remember our angles are semi-angles).
We have two different ways of looking at how an elec- & The electrons interact with the nucleus and the
tron beam interacts with our TEM specimens. We can electron cloud through Coulomb forces.
consider the beam as a succession of particles or as a & We can relate this process to scattering of other parti-
number of waves. What we want to do is understand the cles such as a particles, so lots of analysis can carry over
relationship between the two approaches and we can from other systems.
summarize the two viewpoints thus:
Electrons are particles so they have the following PARTICLES AND WAVES
properties, which we introduced in Chapter 2. When we discuss X-ray and electron spectrometry
you’ll see that we have to use a particle description.
& They have a scattering cross section and a differen- When we discuss imaging, HRTEM, and DPs you’ll
tial scattering cross section. see that we use a wave description.

3.1 P A R T I C L E S AND W A V E S ....................................................................................................................................................................... 39

 Electrons have a wave nature and the electron beam chance that it will have undergone an inelastic event at
is almost a plane wave, hence: some time during its passage through the specimen.
Despite all this, we’ll continue to ignore any inelastic
& Waves are diffracted by atoms or ‘scattering centers.’ effects in this chapter.
& How strongly a wave is scattered by an atom is The second principal form of elastic scattering occurs
determined by the atomic-scattering amplitude. when the electron wave interacts with the specimen as a
& When we gather atoms together into a solid, the whole. We’ve already mentioned the best-known form of
diffraction process gets much more complicated this interaction, namely, diffraction, which is particularly
but it is central to TEM. important at low-scattering angles. Understanding dif-
& We can relate the process to the diffraction of fraction involves treating the electron beam as a wave,
X-rays, so lots of analysis already exists. rather than as a particle as we did in Figure 3.1. Following
the original approach of Huygens for the diffraction of
visible light, we imagine each atom in the specimen
that interacts with the incident plane wave acts as a
3.2 MECHANISMS OF ELASTIC source of secondary spherical wavelets, as illustrated in
SCATTERING Figure 3.2.

In the previous chapter we simply stated that elec-

trons going through a thin specimen are either scat- INTERFERENCE
tered or not scattered and either lose energy or don’t These wavelets reinforce one another in certain angu-
lose energy. It’s now time to describe the ways in lar directions and cancel in other directions: both
which this scattering occurs and in this chapter we’ll reinforcement and cancellation are extremely useful
confine our attention to elastic events, saving inelas- phenomena.
tic scattering for Chapter 4.
It’s convenient to divide elastic-scattering mechan- Thus, the low-angle, elastic scattering distribution is
isms into two principal forms: electron scattering from modified by the crystal structure of the specimen, and
single, isolated atoms and collective scattering from strong diffracted beams emerge at specific angles. The
many atoms together within the specimen. We’ll start
in the same way we did in the previous chapter by
looking first at the interaction of a single electron with
an isolated atom. In this situation, elastic scattering can
occur in one of two ways, both of which involve Cou-
lomb forces. As shown in Figure 3.1, the electron may
interact with the electron cloud, resulting in a small
angular deviation. Alternatively, if an electron pene-
trates the electron cloud and approaches the nucleus, it
will be strongly attracted and may be scattered through
a larger angle that, in rare cases in the TEM, can
approach 1808 (i.e., complete backscattering).

ELASTIC?
You should be aware that either of these two inter-
actions may not be truly elastic, so our separation of
scattering into elastic and inelastic is a bit of a
simplification.

In fact many electron-electron interactions are

inelastic, as we’ll see in the next chapter. We’ll also see,
for example, that the nuclear interaction may result in
the generation of a bremsstrahlung X-ray or may even
FIGURE 3.1. An isolated atom can scatter a high-energy electron by two
result in the displacement of the atom from its site in the mechanisms. Coulombic interaction within the electron cloud results in
crystal, both of which involve some energy loss for the low-angle scattering; Coulombic attraction by the nucleus causes higher-
electron. Indeed, the higher the angle of scattering of an angle scattering (and perhaps complete backscatter when y > 908). The
electron emerging from the specimen, the greater the potential within the electron cloud is always positive.

40 .................................................................................................................................................................................. E L A S T I C S C A T T E R I N G
 re ¼ e=Vy (3:1)

rn ¼ Ze=Vy (3:2)

where r is the radius of the scattering field of the nucleus

and the electron. The different cross sections for scatter-
ing through angles > y are given by prn2 for the nucleus
and Zpre2 for the scattering by Z electrons in the cloud.
If we sum the two components and (just as back in
equation 2.8) multiply by N0rt/A we’ll get a sense of
the total elastic scattering through a film of thickness t.
This approach is ‘‘defective in many respects’’ as Hall
says but gives you a good qualitative sense of the various
parameters that affect elastic scattering. You can see that
the atomic number Z of the atom controls the elastic
interaction with the nucleus, but the electron-electron
FIGURE 3.2. A plane, coherent electron wave generates secondary scattering is more a function of the incident-beam energy
wavelets from a row of scattering centers (e.g., atoms in the specimen).
The secondary wavelets interfere, resulting in a strong direct (zero-
V (which has to be in esu if the dimensions of these
order) beam and several (higher order) coherent beams scattered (dif- equations are to be correct). We’ll see later in Chapter
fracted) at specific angles. 22 that the strong effect of Z becomes important when
we need to enhance scattering in low-Z materials, such as
diffracted beam scattered through the smallest angle is polymers and biological tissue, in order to get better
called the first-order beam and we’ll discuss these and TEM image contrast. Notice that when the electron
higher-order effects in depth in Chapters 11 and 12. passes close to the nucleus (rn is small) the angle y will
We’ll now go on to examine these two forms of elastic be large. We’ll see in Chapter 22 that this dependence on
scattering in more detail, starting with the billiard-ball y directly relates to TEM-image contrast. The electron-
approach. Then we will briefly describe the scattering of beam energy can also control the image contrast to some
a wave to show how it relates to this particle-based extent. So Z, V, and y all affect image contrast and are
treatment and later we’ll use the wave approach as the the three major reasons why you cannot avoid having to
basis for a full analysis of diffraction. study the physics of electron scattering.

3.3 ELASTIC SCATTERING FROM 3.4 THE RUTHERFORD CROSS SECTION

ISOLATED ATOMS
For the next three sections, we’ll ignore the low-angle,
So let’s look at two possible paths for a beam electron electron-electron scattering and concentrate only on scat-
interacting with an isolated atom as shown in Figure 3.1. tering by the nucleus. The high-angle, electron-nucleus
Whether it interacts more strongly with the nucleus or the interaction is analogous to the backscattering of a parti-
electrons, the electron is scattered through an angle y. cles from a thin metal foil. The first observation of such
backscattering in 1911 by H. Geiger (of counter fame) and
a Manchester University undergraduate, E. Marsden,
enabled their professor, Ernest Rutherford, to deduce
the existence of the nucleus (never overlook undergradu-
SCATTERING ANGLE
ate research results!). Rutherford (1911) described back-
Elastic electron-electron interactions usually result in
scattering as ‘‘the most incredible event that has ever
a relatively low scattering angle, while electron-
happened to me’’ (even though he’d already won the
nucleus interactions cause higher-angle scattering.
Nobel Prize for Chemistry in 1908), and he derived the
following expression for the differential cross section for
this high-angle scattering by the nucleus alone
If we just consider an electron, charge e, accelerated
through a voltage V before being scattered from an e4 Z2 dO
sR ðyÞ ¼ 2
(3:3)
isolated atom, the electron-electron and electron- 16ð4pe0 E0 Þ sin4 y
2
nucleus scattering paths are hyperbolic and can be
given by two simple equations (Hall 1953) which are All the terms in this equation were defined back in
useful because they summarize the principal factors that Chapter 2. The expression ignores relativistic effects
control elastic scattering and assumes that the incident electron does not lose

3.4 T H E R U T H E R F O R D C R O S S S E C T I O N ................................................................................................................................................. 41
significant energy through inelastic processes, so that Z2 l4R dO
the energy of the electrons, E0 (in keV), is fixed. As we’ve sR ðyÞ ¼ (3:6)
64p4 a20   
y20
2
already noted, although strictly inaccurate, these 2 y
sin þ
assumptions can be tolerated in the TEM (at least at 2 4
100 keV or below).
This expression describes the screened, relativistic, differ-
ential Rutherford cross section. One very important
3.5 MODIFICATIONS TO THE effect of incorporating screening into these equations is
RUTHERFORD CROSS SECTION that the cross section does not go to infinity as the scat-
tering angle goes to zero which is an important limita-
You’ll often see the Rutherford differential cross tion of all the simpler equations that we used initially.
section in different, but mathematically similar, forms. The screened Rutherford cross section is the one
For example, equation 3.3 neglects the so-called screen- most widely used for TEM calculations, although it
ing effect of the electron cloud. Screening can be has particular limitations at the highest operating vol-
thought of as making the nucleus appear somewhat tages (300–400 kV) and for heavier elements (Z>30)
less positive to the incident electron (although the over- which scatter electrons through large angles. Under
all charge within the electron cloud is always positive). these circumstances, you should use another cross
So the differential cross section is effectively reduced section, such as that of Mott, for which you should
and the amount of scattering is lowered. Screening is consult the text by Mott and Massey (which we referred
only important when the beam electron passes far from to in Chapter 2) or Newbury (1986).
the nucleus and under these circumstances the scattering So, as we did for the basic cross section back in
angle will be small (say <38). If we wish to account for Chapter 2, we can integrate this expression to obtain
screening, we replace the sin2(y/2) term with [sin2(y/2) the total cross section over specific angular ranges. We
can substitute appropriate values for the various con-
+ (y0/2)2] where y0 is called the screening parameter
stants and integrate the differential cross section from y
given by
to p to obtain the total nuclear cross section (in scatter-
ing events/electron/atom/m2) for electrons elastically
0:117Z1=3 scattered into angles > y
y0 ¼ (3:4)
E0 1=2
 2
Z y
(Here E0 is in keV.) What we are saying is that the snucleus ¼ 1:62  1024 cot2 (3:7)
E0 2
screening parameter can be described by a particular
scattering angle, y0. When the scattering angle is greater (From what we told you in Chapter 2 you ought to be
than y0 we can neglect electron-electron interactions and able to work out the integration necessary to determine
the electron-nucleus interaction is dominant. The value the probability of electrons being scattered into angles
of y0 at 100 keV is only 28 for Cu and less for lighter <y.) Again we see that the beam energy (E0), the angle of
elements, so above a few degrees, all scattering can be scattering (y), and the atomic number (Z) all affect the
approximated to Rutherford high-angle scattering. probability that an electron will be scattered by the nuclei
As we’ve noted, so far all our equations are non- of atoms in the specimen. If you simplify this last expres-
relativistic, which is unfortunate since relativistic sion by assuming that y is small you should be able to see
effects are significant for electrons with energies some parallels with Hall’s less accurate equation 3.2 for
>100 keV (which is the case for most materials inves- nuclear scattering. However, there’s much more to this
tigations in the TEM). Fortunately, we can easily cor- whole scattering process than we have covered here and
rect for relativity to give a more accurate cross section you should read Newbury (1986) and Jones (1992) for
by using lR, the relativistically corrected electron further discussion of these calculations once you really
wavelength (see equation 1.7), and a0, the Bohr radius appreciate their significance.
of the scattering atom The best way to summarize the characteristics of
cross sections is to present some data. Figure 3.3
h2 e0 shows the variation of the screened Rutherford cross
a0 ¼ (3:5)
pm0 e2 section in equation 3.7 with scattering angle for (a) three
different elements and (b) two different beam energies.
where e0 is the dielectric constant. Using the other As you can see for Cu, the cross section decreases by
constants listed in Table 1.1 we find a0 is 0.0529 nm several orders of magnitude from 1022 to 1028 m2
(if you’re old enough you can easily remember this as as the scattering angle increases from 0 to 1808; so, as
0.5 Å). The net result of adding screening and relativity we’ve already told you, scattering is most likely to occur
corrections is that in the forward (y  08) direction and drops off rapidly

42 .................................................................................................................................................................................. E L A S T I C S C A T T E R I N G
 (A)
400 C

–22 Target: Cu Mean free

log10 σ path
(m2) λ (nm)

–24

100 keV
Si
–26
200 keV
200

–28

60 120
Scattering angle, θ (°)
(B)
Cu
–22 Beam energy 100 keV
log10 σ Au Au
(m2) 0
100 200
–24 Cu Incident electron energy, E0 (keV)

FIGURE 3.4. The variation of the mean free paths of elastic scattering for
C four different elements as a function of the beam energy, calculated
assuming a screened, relativistic Rutherford cross section.
–26

if you can make your specimen < 100 nm in thickness.

–28 Within such specimens, most electrons either undergo
a single-scattering event or are not scattered and we’ll
assume that this simplification is a viable approxima-
60 120 tion to what’s actually going on in the microscope
Scattering angle, θ (°) many times throughout this text. This approximation
is the main reason why, as we’ve already noted that in
FIGURE 3.3. The variation of the logarithm of the screened relativistic
almost all TEM studies, the ‘thinner is better’ criterion
Rutherford cross section with scattering angle from equation 3.7, describ-
ing the change in cross section for electrons scattered at angles > y (A) for applies.
different elements at 100 keV and (B) for scattering from Cu at different
accelerating voltages.
3.6 COHERENCY OF THE RUTHERFORD-
above a few degrees. Increasing Z from carbon to gold SCATTERED ELECTRONS
can increase the cross section by a factor of 100, which
is why you need proportionately thinner TEM speci- Up to now, in this chapter, we’ve treated the electron as a
mens if you want to ‘see’ through higher-Z materials. particle, but there is useful insight to be gained if we
Doubling the electron-beam energy can lower the cross examine the wave nature of the scattered electron.
section by a factor of two or three, which is why higher- High-angle Rutherford-scattered electrons are incoher-
energy electrons are less likely to be scattered by your ent: i.e., there is no phase relationship between them.
specimen than lower-energy electrons, all else being This is a tricky concept because we are scattering
equal. Figure 3.4 plots the related mean free paths for particles. Such incoherent scattering is important in
elastic scattering. From this graph you can see that two respects. First, the high-angle, forward scattering
very few high-angle elastic scattering events will occur can be used to form exceptionally high-resolution

3.6 C O H E R E N C Y OF THE R U T H E R F O R D - S C A T T E R E D E L E C T R O N S .................................................................................................... 43

images of a crystalline specimen in which the image dsðyÞ
contrast is due solely to the value of Z, not the jfðyÞj2 ¼ (3:8)
dO
orientation of the specimen (as is the case for low-
angle coherent diffraction). Such Z-contrast images, What we will now do is to highlight some of the impor-
as we’ll see in Chapter 22, provide qualitative atomic- tant features that lead to this result by outlining the
resolution elemental analysis, in addition to showing basic arguments.
atomic-resolution detail at interfaces between regions & f(y) is a measure of the amplitude of an electron wave
of different Z. Compared to other image-contrast
scattered from an isolated atom.
mechanisms, Z-contrast imaging is a relatively new & |f(y)|2 is proportional to the scattered intensity.
technique for most microscopists but, particularly
since the availability of Cs-correctors, it has consis-
From these two statements and given the importance of
tently held the record for the highest-resolution
scattered-electron intensity in images and DPs, you can
images and analysis in the TEM (e.g., Varela et al.
appreciate why f(y) is such an important parameter in
2005) and is already beginning to revolutionize our TEM.
understanding of the atomic-level structure and chem- The scattering-factor approach is complementary to the
istry of crystal defects. Rutherford differential cross section analysis, because it is
most useful for describing the low-angle (i.e., <38) elastic
scattering where the Rutherford model is inappropriate.
COHERENCY Usually, f(y) is defined in the following manner
Coherency of the scattered electron is a wave prop-  
erty. If the scattered electron waves have a phase E0 0 12
1þ 2
relationship they must be coherent. m0 c @ l A
fðyÞ ¼ ðZ  fx Þ (3:9)
8p2 a0 sin y2

Second (but much less important), the high-angle All the terms have been previously defined (note
backscattered electrons (BSEs) can be used to form that we’ve dropped the screening term, so remember
images of the beam-entrance surface of the specimen, what this implies). If you need a more detailed approach
in which the contrast is not only due to differences in Z, you could consult the physics-based text by Reimer.
but also to changes in surface topography of the speci- Because we’re now thinking in terms of waves, we need
men. BSE images are rarely used in the TEM because the wavelength l (controlled of course by the beam energy
the BSE signal is small. If you go back and look at the E0), and fx is the scattering factor for X-rays, which is well
Monte Carlo simulation in Figure 2.4 you’ll see that known. The most widely referenced source of electron-
out of 103 incident electrons in Cu only about three scattering factors for TEM was usually the classic work of
(0.3%) were backscattered. Therefore, the quality of Doyle and Turner (1968), but you can now find values in
this signal is very poor, the images are noisy, and the software packages (see Section 1.6) and you can even do
contrast is low. The contrast is much better for bulk your own calculations using the free software on the
specimens in an SEM in which many more electrons are NIST database (also in Section 1.6 and URL #1). The
backscattered (e.g., about 30% in Cu) and BSEs pro- appearance of fx in equation 3.9 is a reminder that f(y) is a
vide a stable, high-contrast imaging technique in SEMs fundamental result of the wave nature of the electron.
in which you can discriminate between the signals from
adjacent elements in the periodic table. In principle, Z-
contrast should be able to do the same in the TEM. f(q)
The atomic-scattering factor f(y) depends on l, y,
and Z.
3.7 THE ATOMIC-SCATTERING FACTOR
We can plot this angular variation for a single iso-
The classic Rutherford differential cross section cannot lated atom. Figure 3.5 summarizes graphically what we
be used to calculate the cross section exactly, because it already know about the magnitude of elastic scattering
ignores the wave nature of the electron beam. A full (see equations 3.1 and 3.2)
treatment involves wave mechanics and is well beyond
the scope of this text. Perhaps the most familiar aspect & It decreases as y increases (y ¼ 08 for the incident-
of the wave approach to cross sections is the concept of beam direction).
the atomic-scattering factor f(y), which is related to the & It decreases as l decreases (i.e., as the accelerating
differential elastic cross section by a simple equation voltage (V) increases).
(more on this in Section 3.8) & It increases with Z for any value of y.

44 .................................................................................................................................................................................. E L A S T I C S C A T T E R I N G
 To find the total elastic-scattering cross section,
we have to integrate ds/dO. Note that this is a particle
12
model, but you should also be aware of how the wave
f (θ) nature of the electrons is brought in. We can consider
the wave nature by looking at Figure 3.6 (which
(Å) you should realize is closely related to Figures 2.3
and 2.12).
8 We can describe the incident beam as a wave of
Au
amplitude c0 and phase 2pkr

c ¼ c0 e2pikr (3:10)
Cu
4
In this definition of phase, k is the magnitude of the
wave vector and r is the distance that the wave has
Al propagated, as we’ll discuss in detail later in Chapter
11. When the incident plane wave is scattered by a
0 point charge, a spherical scattered wave is created
0.2 0.4 0.6 which has a different amplitude csc but keeps the
(sin θ)/λ (Å–1) same phase apart from a p/2 addition which we return
to in a moment
FIGURE 3.5. Change in the atomic scattering factor f(y) with scattering
angle y (calculated from equation 3.9) showing that elastic scattering
e2pikr
decreases with angle away from the incident beam direction (y ¼ 08) and csc ¼ c0 fðyÞ (3:11)
increases with Z. r

In this equation, f(y) is the amplitude we would have

This expression (equation 3.9) for f(y) contains
components of both elastic nuclear scattering (the Z if c0 ¼ 1, i.e., it is the atomic-scattering amplitude.
term) and elastic electron-cloud scattering (the fx
term). We’ll see later in the chapters on diffraction Incident beam
in Part 2 that the f(y) approach is used exclusively
and, if we neglect the fx term, then it can be shown that
|f(y)|2 is mathematically equivalent to the high-angle
Rutherford differential cross section, as we defined it
in equation 3.6. So now we’ve tied together the Plane wave
particle and wave approaches to elastic scattering.
k

ANGLE VARIATION
The important point to remember is that both the
differential cross section and the scattering factor are
simply measures of how the electron-scattering inten- Constructive
sity varies with y. interference
Radial
scattered
spherical θ
3.8 THE ORIGIN OF f(q) wave
dθ
Since f(y) relates to the amplitude of a scattered wave,
we’ll consider briefly how it arises. The following anal-
ysis is not intended to be completely rigorous, but only FIGURE 3.6. Generating a scattered wave by the interaction of a plane
to give the fundamental ideas behind the meaning of f(y) wave (horizontal line, wavelength l) with a point charge. The circles
and its relation to the differential scattering cross represent the scattered spherical wavefronts which are in phase and retain
the original l. The in-phase, constructive interference between the plane
section. You can safely delay studying this topic until and spherical waves is shown by the dark areas. The angles y and dy are
curiosity wins, then you can go and read the really thick the same as in Figure 2.3.
physics textbooks.

3.8 T H E O R I G I N OF f ( y) ............................................................................................................................................................................. 45
 So obviously we need to know f(y) and an accep- Now if f is small, then cos f  1 and sin f  f; cos y is
table model is essential to make the problem manage- always the same as sin (y + p/2), hence
able. Up to this point, our treatment has been quite 
rigorous and ideally, the model would distinguish p
ctot ¼ c0 sinð2pkzÞ þ c0 f sin 2pkz þ (3:15)
between a neutral atom in a metal, a covalently bonded 2
atom, and an ion. If you’re desperate, the quantity f(y) The p/2 term would arise if we used the exponential
can always, in principle, be calculated from the rather than the sine to denote phase, so we can now
Schrödinger equation. In practice, however, we usually write equation 3.15 as
use a simple approximation which we’ll now describe.
If we write down the expression for the scattering ctot ¼ c þ icsc (3:16)
process shown in Figure 3.6, then we have
  This equation has the same form as that given in equa-
2pikI  r e2pikr tion 3.12.
csc ¼ c0 e þ ifðyÞ (3:12)
r

You should note first of all that, as usual for Huygens 3.9 THE STRUCTURE FACTOR F(q)
wavelets, there is a 908 phase shift (shown by the
inclusion of ‘i’ in the second term) between the incident The next introductory step in discussing electron scatter-
and scattered beams and secondly, that f(y) can be ing is to take the idea of individual atoms scattering
expressed as electrons (i.e., f(y)), which we’ve just discussed in some
detail, and consider what happens when the atoms are
stacked together regularly in a crystal structure. (We can,
fðyÞ ¼ jfðyÞjeiZðyÞ ¼ jfðyÞjðcos ZðyÞ þ i sin ZðyÞÞ (3:13)
in principle, also do this for an amorphous solid but we’ll
stick to crystals for simplicity.) We will deal with this
which means that the phase, Z(y), of f(y) also depends approach in great detail in Chapter 13, but for now we
on the angle of scatter, y. can introduce the structure factor F(y), which is a measure
First aside: In writing equation 3.12, we have of the amplitude scattered by a unit cell of a crystal
introduced two wave-propagation parameters: structure. Because F(y) is an amplitude like f(y), it also
the vector kI for the incident plane wave and has dimensions of length. We can define F(y) as the sum of
the scalar k for the spherical scattered wavelet. the f(y) terms from all the i atoms in the unit cell (with
By writing the 2p factor separately as part of the atomic coordinates xi yi zi) multiplied by a phase factor.
phase term, we have implicitly defined k to be 1/l. The phase factor takes account of the difference in phase
Many physics textbooks include the 2p in k so they between waves scattered from atoms on different but
have k given by 2p/l. Just be careful when you parallel atomic planes with the same Miller indices (hkl).
compare similar formulas in two textbooks. The scattering angle y is the angle between the incident
and scattered electron beams. So we can write

X
1
1/l AND 2p/l FðyÞ ¼ fi e2piðhxi þkyi þlzi Þ (3:17)
Sometimes k ¼ 1/l and sometimes k ¼ 2p/l and it’s i
sometimes difficult to find out which definition is
being used. The amplitude (and hence its square, the intensity)
of scattering is influenced by the type of atom (f(y)),
the position of the atom in the cell (x,y,z), and the
Second aside: The 908 phase change for specific atomic planes (hkl) that make up the crystal
the scattered-wave component in equation 3.13 structure. None of this is very surprising, but it turns
can be easily understood if you consider out that this equation predicts that in certain circum-
the following. If the amplitude of the wave stances the amplitude of scattering is zero. This beha-
is initially c0sin(2pkr) then, after it has passed vior is intrinsic to the scattering process, is implicit
through the specimen, it will be ctot. After scatter- back in Figure 3.2, and is often a very useful diagnostic
ing, the phase is increased by f, so we can express test when determining crystal structures in the TEM.
the new ctot as
ZERO SCATTERING
ctot ¼ c0 sinð2pkz þ fÞ ¼ c0 sinð2pkzÞ cos f Under specific conditions, electrons scattering in a
crystal may result in ZERO scattered intensity. Why
þc0 cosð2pkzÞ sin f (3:14) might this occur?

46 .................................................................................................................................................................................. E L A S T I C S C A T T E R I N G
 We’ll return to this point in Chapter 13 in much (A)
more detail.
Scattered
amplitude
3.10 SIMPLE DIFFRACTION CONCEPTS f (θ)
As we mentioned earlier, electron diffraction is by
far the most important scattering phenomenon in the
TEM. The reason for this importance, as we’ll show
you in Chapters 11 and 12, is that we can use
diffraction to determine the spacing of planes in
crystals and, as you’ll see later in Chapters 20, 21,
and in the companion text, there is a whole field
termed electron crystallography which gives an
unprecedented amount of crystallographic informa-
tion from space-group symmetry data right down to
the dimensions of single unit cells. At the most basic
level, the interplanar spacings in different crystal
structures are characteristic of that structure.
We’ll see that the positions of the diffracted beams
of electrons are determined by the size and shape of Scattering angle θ
the unit cell and the intensities of the diffracted beams
are governed by the distribution, number, and types of (B)
atoms in the specimen. We’ll also show you in Part 3 Scattered
how diffraction leads to contrast in TEM images amplitude
which is controlled by the orientation of a crystal
f (θ)
with respect to the electron beam and which you can
control simply by tilting your specimen.

DP PLUS IMAGE
We can distinguish different crystal structures by
observing and measuring DPs. The combination of
the DP and the electron image(s) is a most powerful
tool for characterizing crystals and particularly their
defects.

It’s easy to see, in a qualitative manner, how diffrac-

Scattering angle θ
tion modifies the distribution of the low-angle scatter-
ing, described by f(y), and shown for a single atom in FIGURE 3.7. Change in f(y) with y for (A) an amorphous specimen and (B)
Figure 3.5. When we consider the effect of the arrange- a crystalline specimen. The amplitude (and therefore the intensity) of scatter-
ment of atoms in the specimen, then Figure 3.5 has to be ing generally decreases with increasing y but the smooth decrease is modified
at certain scattering angles (compare these curves with the intensity variation
modified. For an amorphous specimen, the atoms are
along a radius of the DPs in Figure 2.13A and C, respectively).
almost (but not quite) randomly arranged. A random
arrangement would result in a plot similar to Figure 3.5,
but there are certain interatomic spacings that tend to intensity variation across the DPs in Figure 2.13A
occur in an amorphous structure (e.g., first- and sec- and C, respectively, and thus emphasizes the strong
ond-nearest neighbor spacings are usually relatively relationship between f(y) and diffracted intensity.
well defined). As a result, the amplitude (and hence We’ll describe this important relationship mathemati-
cally in Section 3.10.B below.
the intensity) of diffraction is stronger at some angles
than at others, so we see diffuse, bright rings on the
TEM screen. If the specimen is crystalline, then the
intensity of the diffracted beams is a maximum at 3.10.A Interference of Electron Waves; Creation
specific angles because the interplanar spacings are
of the Direct and Diffracted Beams
very well defined. The variation of f(y) with y plotted To interpret low-angle elastic scattering (which is pri-
in Figure 3.7A and B is equivalent to the radial marily from the electron cloud) it is best to think in

3.10 S I M P L E D I F F R A C T I O N C O N C E P T S ................................................................................................................................................... 47
terms of electron waves and not in terms of particle- assistants, and Knipping tested the idea experimentally
particle interactions that characterize high-angle by irradiating a copper sulfate crystal and became the
Rutherford scattering. If you go back and look at first to observe diffraction from crystal planes. In fact it
Figure 3.2 you see a periodic one-dimensional array was a remarkable stroke of luck that the CuSO4 dif-
of scattering centers (slits), and a monochromatic fracted the X-rays at all because of the strict equations
wave (i.e., fixed l) is advancing toward these centers. that govern diffraction.
Each center acts as a new source of a wave of the same Von Laue used the well-known light-optics
l. Thus many new waves are created and, when more approach to argue that the diffracted waves are in
than one wave is present, the waves can interfere with phase if the path difference between waves scattered by
one another. This process happens from even the adjacent scattering centers is a whole number of wave-
thinnest specimens and is entirely a wave phenom- lengths, hl (h is an integer). Thus, as shown in Figure
enon that doesn’t need concepts such as cross section, 3.8, if the scattering centers (B and C) are spaced some
which we apply when we think of the electron as a distance a apart and the incident beam (wavelength l)
particle. makes an angle y1 with the line connecting the scattering
A rule of wave theory is that waves reinforce one centers and is diffracted at an angle y2, then the path
another (this is constructive interference) when they are difference AB – CD is
in phase. Waves also cancel one another (destructive
interference) when they are out of phase. What you see aðcos y1  cos y2 Þ ¼ hl (3:18)
in Figure 3.2 is that the diffracted waves are in phase
with one another only in certain directions. There is Now in three dimensions, two more Laue equations
invariably a zero-order wave that proceeds in the same can be written for two more distances b and c and
direction as the incident wave, which in the TEM we’ll appropriate angles yn
refer to as the direct beam of electrons, as we defined at
the start of Chapter 2. There are also higher-order waves bðcos y3  cos y4 Þ ¼ kl (3:19)
that propagate in forward directions that are at some
fixed (but very small) angle to the incident wave and cðcos y5  cos y6 Þ ¼ ll (3:20)
we’ll call these the diffracted beams.
So diffraction creates many electron beams trav- These three simultaneous equations bear von
eling at specific angles relative to a single mono- Laue’s name and for his original suggestion and the
chromatic incident beam. In the chapters on analysis of the experiments of Sommerfeld’s students,
diffraction in Part 2, we’ll find ways to measure he received the Nobel Prize in Physics in 1914 (nice
these angles and relate them to the spacing of the work: three equations). We’ll show in Chapter 11 that
scattering planes.
Inc
A ide
nt p
DIRECT AND DIFFRACTED lan
ew
The direct beam consists of electron, scattered in the ave
same direction as the incident beam. Often in TEM
terminology these electrons are called the transmitted
beam but this term is ambiguous since, in fact, all C
forward-scattered beams are ‘transmitted’ through
the specimen. θ2
a
3.10.B Diffraction Equations
θ1
Here we’ll introduce the mathematical relationships
that describe the diffraction process. The idea of using
diffraction to probe the atomic structure of materials B D
was credited to von Laue (1913) in Germany, although
others such as Ewald were working on similar ideas at
Scattered plane wave
the same time. Von Laue’s crucial idea was that much
shorter electromagnetic rays than light would cause diffrac- FIGURE 3.8. The approach used by von Laue to calculate the path differ-
tion or interference phenomena in a crystal. Although his ence for a wave (wavelength l). In this one-dimensional figure the wave is
colleague Sommerfeld, with whom he discussed the idea incident at an angle y1 and scattered at an angle y2 from two atoms (B and
while skiing, disagreed, Friedrich, one of Sommerfeld’s C) spaced distance a apart. The path difference between scattered waves is
AB – CD.

48 .................................................................................................................................................................................. E L A S T I C S C A T T E R I N G
in a TEM specimen, when all three Laue equations are waves reflected from the upper and lower planes in Fig-
satisfied simultaneously a diffracted beam is produced. ure 3.9 is (AB + BC). Thus, if the ‘reflecting’ hkl planes
We’ll also show you in Chapters 11 and 12 that the are spaced a distance d apart and the wave is incident and
letters hkl are the indices of the diffracted beam and reflected at an angle yB, both AB and BC are equal to
are equivalent to the Miller indices (hkl) of the diffract- d sin yB and the total path difference is 2d sin yB. So we
ing crystal plane (or some multiple thereof). can write what is known as Bragg’s (although gramma-
Usually in TEM, we use a simpler approach to tically and historically it should be Braggs’) law
describe diffraction. Von Laue’s approach was simpli-
fied by the family team of Sir William H. (obviously the nl ¼ 2d sin yB (3:21)
dad) and Mr. W. Lawrence Bragg (the son) in England
who proposed (Bragg and Bragg 1913) that the waves We’ll reserve yB for the Bragg angle, which is the
behaved as if they were reflected off atomic planes as most important scattering angle (remember we really
shown in Figure 3.9. mean semi-angle) in TEM and you’ll come across it
In parallel with von Laue’s optical approach, the many more times in this text. The Braggs also received
Braggs argued that waves reflected off adjacent scattering a Nobel Prize in Physics a year after von Laue but this
centers must have a path difference equal to an integral time for only one equation (even nicer work!) and
number of wavelengths, if they are to remain in phase. despite the fact that the idea of reflected electrons,
So, in the TEM the path difference between electron while mathematically correct, is physically wrong.
We’ll continue to use the term Bragg reflection to
describe diffraction in the TEM because everyone
does so, even though it’s inaccurate, and because it is
extremely useful. However, we’ll demonstrate to you,
Incident in a rigorous fashion, the mathematical equivalence of
plane wave Scattered the Bragg and von Laue approaches in Chapter 12.
plane wave It is simple to see from the Bragg equation that
θ θ atomic planes which are closer together give rise to
larger angles of scatter. This reciprocal relationship (d
is proportional to 1/y; see Chapter 12) is very impor-
d θ θ tant in diffraction-pattern interpretation. So, if you
A C know l for the incident electron (which you control
B by choosing the accelerating voltage) and you can
measure y experimentally, you can work out the inter-
FIGURE 3.9. The Bragg description of diffraction in terms of the planar spacings in your specimen. It is this crystallo-
reflection of a plane wave (wavelength l) incident at an angle y to
atomic planes of spacing d. The path difference between reflected
graphic information that makes diffraction such an
waves is AB + BC. important aspect of the TEM.

CHAPTER SUMMARY
What should you remember from this chapter? Until you have time to study this material
very carefully you may find it difficult, so here are a few suggestions:
& Know the words! In particular, we can describe the scattering process by three param-
eters
s(y) the scattering cross section
dsðyÞ
the differential scattering cross section
dO
f(y) the atomic-scattering amplitude

In particular, don’t be put off because ‘differential scattering cross section’ sounds
difficult. All three terms are very important in different parts of TEM.
& The relationships between f(y) and s(y) are very important (as a principle, but not

much used in practice).

CHAPTER SUMMARY ........................................................................................................................................................................................... 49

 & The relationship between f(y) and the intensity in a DP is very important.
Remember that, although we often write s(y) as s, there is an angle involved in any s

& The fact that the electron is a charged particle is critical to the whole scattering process.
& The strength of the scattering, f(y), depends inversely on the scattering angle, y.

Yes, a really rigorous treatment of scattering would take into account the wave nature of the
electron (wave mechanics), relativity, and the electron charge, all at the same time. Because
we’re good guys we won’t inflict this on you or ourselves. Fortunately, if required, we can do
very well using compiled tables of cross sections and scattering data, which are available on
the web (e.g., URL #1).
We can describe the effect of the crystal structure on the electron scattering by one more
parameter, the structure factor F(y)

& F(y) is a measure of the amplitude scattered by a unit cell and jFðyÞj2 is proportional to
the scattered intensity.

The diffraction process from a TEM specimen is usually described by the Bragg equation
which tells us the important reciprocal relationship between atomic-plane spacings and
scattering angles.
A final point to think about: remember that f(y) is the property of a ‘scattering center.’ We
usually think of this center as being an atom. What happens if the scattering center is an ion (i.e.,
if it is charged)? Is the scattering process affected by how this atom is bonded to its neighbors?
What changes if the atom has a covalent rather than a metallic bond? These are important
questions (otherwise we wouldn’t ask them) and we’ll teach you the answers as we go on.

DIFFRACTION AND SCATTERING

Andrews, KW, Dyson, DJ and Keown, SR 1967 Electron Diffraction Patterns The Institute of Physics
Bristol UK. The original text focused on diffraction in the TEM.
Bragg, WH and Bragg, WL 1913 The Reflection of X-rays by Crystals Proc. Roy. Soc. Lond. A88 428–438.
The paper that led to the Nobel Prize.
Doyle, PA and Turner, PS 1968 Relativistic Hartree-Fock X-ray and Electron Scattering Factors Acta
Crystallogr. A24 390–397.
Mott, NF and Massey, HSW 1965 The Theory of Atomic Collisions Oxford University Press New York.
Reimer, L 1997 Transmission Electron Microscopy; Physics of Image Formation and Microanalysis 4th
Ed. Springer New York. Rigorous thorough treatment of the scattering process, especially as used
in Section 3.8.
Rutherford, E 1911 The Scattering of a and b Particles by Matter and the Structure of the Atom Phil. Mag.
21 669–688. His scattering.
von Laue, M 1913 Kritische Bemerkungen zu den Deutungen der Photoframme von Friedrich und Knipping
Phys. Z. 14 421–423. Paper that led to the Nobel Prize.
Wang, ZL 1995 Elastic and Inelastic Scattering in Electron Diffraction and Imaging Plenum Press New
York. Much more detailed than the approach used here.

SCATTERING APPLIED TO EM
Hall, CE 1953 Introduction to Electron Microscopy p 229 McGraw-Hill New York.
Jones, IP 1992. Chemical Microanalysis Using Electron Beams Institute of Materials London.
Newbury, DE 1986 Electron Beam-Specimen Interactions in the Analytical Electron Microscope
in Principles of Analytical Electron Microscopy p 1 Eds. DC Joy, AD Romig Jr and JI Goldstein
Plenum Press New York.
Varela, M, Lupini, AR, van Benthem, K, Borisevich, AY, Chisholm, MF, Shibata, N, Abe, E and
Pennycook, SJ 2005 Materials Characterization in the Aberration-Corrected Scanning Transmission
Electron Microscope Annu. Rev. Mat. Sci. 35 539–569. Review of Z-contrast imaging.

50 .................................................................................................................................................................................. E L A S T I C S C A T T E R I N G
 URLs
(1) www.nist.gov/srd/nist64.htm NIST Standard reference database #64 provides values of the differential
elastic-scattering cross sections, total elastic-scattering cross sections, phase shifts, and transport cross
sections for elements with Z ¼ 1 to 96 and for beam energies from 50 eV to 300 keV (in steps of 1 eV).

SELF-ASSESSMENT QUESTIONS
Q3.1 What are the primary causes of elastic scattering?
Q3.2 What do we mean by the term ‘wave-particle duality’?
Q3.3 What forces act on an electron as it interacts with atoms?
Q3.4 What term describes the strength of the scattering process?
Q3.5 What factors control the interference between waves?
Q3.6 How is the scattering amplitude related to the intensity of the scattered beams that we see in the microscope?
Q3.7 What are the two principal forms of elastic scatter?
Q3.8 Relate the general form of the Rutherford differential cross section (equation 3.3) to the equation
describing the cross section for nucleus scattering proposed by Hall (equation 3.2).
Q3.9 What is a screening parameter and why do we need to incorporate it in the equations that describe scattering?
Q3.10 Why is it important to include a screening parameter in the Rutherford cross section?
Q3.11 Why do elastic electron-electron interactions usually result in a relatively low scattering angle, while
elastic electron-nucleus interactions cause higher-angle scattering?
Q3.12 From your answer to the previous question describe the different information that might be contained in
low-angle and high-angle scattered electrons and how you might obtain that information
Q3.13 How thin should your specimen be so that scattering within it approaches the ideal of a single event per electron?
Q3.14 What is the relationship between the atomic scattering factor f(y) and the structure factor F(y)?
Q3.15 Why do crystalline and amorphous specimens give rise to different scattering distributions?
Q3.16 What are the fundamental differences between the von Laue and Bragg approaches to diffraction and
what are the similarities?
Q3.17 Put some reasonable values for d and l into equation 3.21 and calculate a typical Bragg angle in a TEM.
Q3.18 Why is the Bragg approach fundamentally incorrect?
Q3.19 What do we mean by the term ‘scattering center’?
Q3.20 What is the relationship between the spacing of the lattice planes and the angle of scatter?

TEXT-SPECIFIC QUESTIONS
T3.1 In Figure 3.1, why are the electrons interacting with both the nucleus and the electron cloud shown to
deviate in the same directions (i.e., both are bent through an angle y) when the nucleus and the electron
clouds in fact have opposite electrical charges?
T3.2 Why do we show the electron close to the nucleus in Figure 3.1 as being turned around rather than being
pulled directly into the (highly positively charged) nucleus?
T3.3 Look again at Figure 3.1 and explain why elastic electron-electron interactions usually result in a
relatively low scattering angle, while elastic electron-nucleus interactions cause higher-angle scattering.
T3.4 In Figure 3.2 why don’t we see a third-order scattered beam?
T3.5 Relate Figure 3.3 to equations 3.1 and 3.2.
T3.6 Can you show that the data in Figures 3.3 and 3.4 are consistent? (Hint: assume that y is small (i.e.,  08)
for elastic scatter.)
T3.7 Relate Figure 3.5 to equations 3.1 and 3.2.
T3.8 Is Figure 3.5 plotted for a screened or unscreened atomic potential? Explain your answer.
T3.9 In Figure 3.6 why don’t we show constructive interference of waves going back in the direction of the
incident beam?
T3.10 Figures 3.3, 3.5, and 3.7 all have the same general form. Why is this?
T3.11 Discuss the advantages and disadvantages of the Rutherford cross section for elastic scattering. Put in
some values into equation 3.7 and determine the value of the cross section.
T3.12 Write down concise definitions of coherent, incoherent, elastic, and inelastic as we use them. (Hint: first
take a look at Webster’s.)
T3.13 Explain in a paragraph the relationship between scattering cross section and atomic scattering factor
mentioning the important factors that influence them.
T3.14 In Figure 3.6, if this process were Bragg diffraction, how would the Bragg angle relate to y?
T3.15 Use equation 3.21 to determine the value of y if n =1 and d = 2 for each of the wavelengths in Table 1.2.
Thus, discuss whether or not relativistic corrections are important.
T3.16 Copy Figure 3.8 and draw on it where other atoms in the diffracting planes might be positioned. (Hint:
look at Figure 3.9.)
T3.17 Explain why we talk about the Bragg angle in Figure 3.9 as being a semi-angle of scattering.

CHAPTER SUMMARY ........................................................................................................................................................................................... 51