1

X-ray diffraction
Diffraction techniques can be used to determine the details of the arrangement of ions, atoms, and
molecules in a crystalline solid to high precision. Such techniques are now so well developed that both
the collection of the diffraction data and its interpretation in terms of a structure are automated to a
high degree.

X-ray crystallography
A characteristic property of waves is that they interfere with one another, giving a greater displacement
where peaks or troughs coincide and a smaller displacement where peaks coincide with troughs
(Fig. 1).

According to classical electromagnetic theory, the intensity of electromagnetic radiation is proportional

to the square of the amplitude of the waves. Therefore, the regions of constructive or destructive
interference show up as regions of enhanced or diminished intensities. The phenomenon of diffraction
is the interference caused by an object in the path of waves, and the pattern of varying intensity that
results is called the diffraction pattern. Diffraction occurs when the dimensions of the diffracting
object are comparable to the wavelength of the radiation.

Fig. 1 When two waves are in the same region of space they interfere. Depending on their relative phase, they may interfere
(a) constructively, to give an enhanced amplitude, or (b) destructively, to give a smaller amplitude.

Fig. 2 X-rays are generated by directing an electron beam on to a cooled metal target. Beryllium is transparent to X -rays (on
account of the small number of electrons in each atom) and is used for the windows.

(a) X-ray diffraction

Wilhelm Röntgen discovered X-rays in 1895. Seventeen years later, Max von Laue suggested that they
might be diffracted when passed through a crystal, for by then he had realized that their wavelengths
are comparable to the separation of lattice planes. This suggestion was confirmed almost immediately
by Walter Friedrich and Paul Knipping and has grown since then into a technique of extr aordinary
power.

The mathematical procedures necessary for the determination of structure from X-ray diffraction data
are enormously complex, but such is the degree of integration of computers into the experimental
apparatus that the technique is almost fully automated, even for large molecules and complex solids.
The analysis is aided by molecular modelling techniques, which can guide the investigation towards a
plausible structure.

X-rays are electromagnetic radiation with wavelengths of the order of 1 0−10 m. They are typically
generated by bombarding a metal with high-energy electrons (Fig. 2). The electrons decelerate as they
 2

plunge into the metal and generate radiation with a continuous range of wavelengths called
Bremsstrahlung (Bremse is German for deceleration, Strahlung for ray).

Superimposed on the continuum are a few high-intensity, sharp peaks (Fig. 3). These peaks arise from
collisions of the incoming electrons with the electrons in the inner shells of the atoms. A collision
expels an electron from an inner shell, and an electron of higher energy drops into the vacancy,
emitting the excess energy as an X-ray photon (Fig. 4). If the electron falls into a K shell (a shell with
n = 1), the X-rays are classified as K-radiation, and similarly for transitions into the L (n = 2) and M (n
= 3) shells. Strong, distinct lines are labelled Kα , Kβ , and so on. Increasingly, X-ray diffraction makes
use of the radiation available from synchrotron sources (Further information), for its high intensity
greatly enhances the sensitivity of the technique.

Fig. 3 The X-ray emission from a metal consists of a broad, featureless Bremsstrahlung background, with sharp transitions
superimposed on it. The label K indicates that the radiation comes from a transition in which an electron falls into a vacancy
in the K shell of the atom.

Fig. 4 The processes that contribute to the generation of X-rays. An incoming electron collides with an electron (in the K shell),
and ejects it. Another electron (from the L shell in this illustration) falls into the vacancy and emits its excess energy as an
X-ray photon.

Further information
Synchrotron radiation is generated in a synchrotron storage ring,
which consists of an electron beam travelling in a circular path with
circumferences of up to several hundred metres. As electrons
travelling in a circle are constantly accelerated by the forces that
constrain them to their path, they generate radiation. Synchrotron
radiation spans a wide range of frequencies, including the infrared
and X-rays. Except in the microwave region, synchrotron radiation is
much more intense than can be obtained by most conventional
sources.

Fig. A synchrotron storage ring. The electrons injected into the ring from the linear
accelerator and booster synchrotron are accelerated to high speed in the main ring.
An electron in a curved path is subject to constant acceleration, and an accelerated
charge radiates electromagnetic energy.
 3

Determination of crystal structure:

The method developed by the Braggs (William and his son Lawrence, who later jointly won the Nobel
Prize) is the foundation of almost all modern work in X-ray crystallography. They used a single crystal
and a monochromatic beam of X-rays, and rotated the crystal until a reflection was detected. There
are many different sets of planes in a crystal, so there are many angles at which a reflection occurs.
The complete set of data consists of the list of angles at which reflections are observed and their
intensities.

Single-crystal diffraction patterns are measured by using a four-circle diffractometer (Fig. 5). The
computer linked to the diffractometer determines the unit cell dimensions and the angular settings of
the diffractometer’s four circles that are needed to observe any particular intensity peak in the
diffraction pattern. The computer controls the settings, and moves the crystal and the detector for
each one in turn. At each setting, the diffraction
intensity is measured, and background
intensities are assessed by making
measurements at slightly different settings.
Computing techniques are now available that lead
not only to automatic indexing but also to the
automated determination of the shape, symmetry,
and size of the unit cell. Moreover, several
techniques are now available for sampling large
amounts of data, including area detectors and
image plates, which sample whole regions of
diffraction patterns simultaneously.

Fig. 5 A four-circle diffractometer. The settings of the orientations (φ, χ, θ, and Ω) of the components is controlled by computer;
each (hkl) reflection is monitored in turn, and their intensities are recorded.

An alternative technique was developed by Peter Debye and Paul S cherrer and independently by Albert
Hull. They used monochromatic radiation and a powdered sample. When the sample is a powder, at
least some of the crystallites will be orientated so as to give rise to di ffraction. In modern powder
diffractometers the intensities of the reflections are monitored electronically as the detector is rotated
around the sample in a plane containing the incident ray (Fig. 6).

Powder diffraction techniques are used to identify a sample of a solid substance by comparison of the
positions of the diffraction lines and their intensities with diffraction patterns stored in a large data
bank. Powder diffraction data are also used to determine phase diagrams, for different solid phases
result in different diffraction patterns, and to determine the relative amounts of each phase present in
a mixture. The technique is also used for the initial determination of the dimensions and symmetries
of unit cells.

Fig. 6 X-ray powder photographs of (a) NaCl, (b) KCl and the indexed reflections. The smaller number of lines in (b) is a
consequence of the similarity of the K + and Cl − scattering factors.
 4

(b) Bragg’s law

An early approach to the analysis of diffraction patterns produced by crystals was to regard a lattice
plane as a semi-transparent mirror, and to model a crystal as stacks of reflecting lattice planes of
separation d (Fig. 7).

Fig. 7 The conventional derivation of Bragg’s law treats each lattice plane as a reflecting the incident radiation.
The path lengths differ by AB + BC, which depends on the glancing angle, θ. Constructive interference (a ‘reflection’) occurs
when AB + BC is equal to an integer number of wavelengths.

The model makes it easy to calculate the angle the crystal must make to the incoming beam of X-rays
for constructive interference to occur. It has also given rise to the name reflection to denote an intense
beam arising from constructive interference.

Consider the reflection of two parallel rays of the same wavelength by two adjacent planes of a lattice,
as shown in Fig. 7. One ray strikes point D on the upper plane but the other ray must travel an
additional distance AB before striking the plane immediately below. Similarly, the reflected rays will
differ in path length by a distance BC.

The net path length difference of the two rays is then AB + BC = 2d sin θ where θ is the glancing angle.

For many glancing angles the path-length difference is not an integer number of wavelengths, and the
waves interfere largely destructively. However, when the path-length difference is an integer number
of wavelengths (AB + BC = nλ), the reflected waves are in phase and interfere constructively. It follows
that a reflection should be observed when the glancing angle satisfies Bragg’s law:

nλ = 2d sin θ

Reflections with n = 2, 3, . . . are called second-order, third-order, and so on; they correspond to path-
length differences of 2, 3, . . . wavelengths. In modern work it is normal to absorb the n into d, to write
the Bragg law as
λ = 2d sin θ

and to regard the nth-order reflection as arising from the {nh,nk,nl} planes.

The primary use of Bragg’s law is in the determination of the spacing between the layers in the lattice
for, once the angle θ corresponding to a reflection has been determined, d may readily be calculated.
 5

Example. 2 Using Bragg’s law

A first-order reflection from the {111} planes of a cubic crystal was observed at a glancing angle of
11.2° when Cu(Kα ) X-rays of wavelength 154 pm were used. What is the length of the side of the unit
cell?
Method The separation of the planes can be determined from Bragg’s law. Because the crystal is cubic,
the separation is related to the length of the side of the unit cell, a, which may therefore be solved for
a.
Answer According to eqn λ = 2d sin θ, the {111} planes responsible for the diffraction have separation

The separation of the {111} planes of a cubic lattice of side a is given by

Self-test Calculate the angle at which the same crystal will give a reflection from the {123} planes.
Answer: [24.8°]

Some types of unit cell give characteristic and easily recognizable patterns of lines. For example, in a
cubic lattice of unit cell dimension a the spacing is given by eqn ,

so the angles at which the {hkl} planes give first-order reflections are given by

The reflections are then predicted by substituting the values of h, k, and l:

{hkl} {100} {110} {111} {200} {210} {211} {220} {300} {221} {310} . . .
h2 +k2 +l 2 1 2 3 4 5 6 8 9 9 10 . . .

Notice that 7 (and 15, . . .) is missing because the sum of the squares of three integers cannot equal 7
(or 15, . . .). Therefore the pattern has absences that are characteristic of the cubic P lattice.

Self-test. Normally, experimental procedures measure 2θ rather than θ itself. A diffraction

examination of the element polonium gave lines at the following values of 2θ (in degrees) when 71.0
pm Mo X-rays were used: 12.1, 17.1, 21.0, 24.3, 27.2, 29.9, 34.7, 36.9, 38.9, 40.9, 42.8. Identify the
unit cell and determine its dimensions. Answer: [cubic P; a = 337 pm]
 6

(c) Scattering factors

The scattering factor is a measure of the ability of an atom to scatter electromagnetic radiation.

The scattering of X-rays is caused by the oscillations an incoming electromagnetic wave generates in
the electrons of atoms, and heavy atoms give rise to stronger scattering than light atoms. This
dependence on the number of electrons is expressed in terms of the scattering factor, f, of the element.
If the scattering factor is large, then the atoms scatter X-rays strongly. The scattering factor of an atom
is related to the electron density distribution in the atom, ρ(r), by

The value of f is greatest in the forward direction and smaller for directions away from the forward
direction (Fig. 8). The detailed analysis of the intensities of reflections must take this dependence on
direction into account (in single crystal studies as well as for powders).

Fig. 8. The variation of the scattering factor of atoms and ions with atomic number and angle. The scattering factor in the
forward direction (at θ = 0, and hence at (sin θ)/λ = 0 is equal to the number of electrons present in the species.

We show in the Justification below that, in the forward direction (for θ = 0), f is equal to the total
number of electrons in the atom.

Justification: THE FORWARD SCATTERING FACTOR

Evaluating the scattering factor in the forward direction

The first step is to note that because sinkr cannot exceed 1, the maximum value of (sinkr)/kr occurs
as k → 0, which corresponds to sin θ → 0 and therefore θ → 0. Therefore, scattering factor has its
maximum value in the forward direction.

The factor (sin kr)/kr is therefore equal to 1 for forward scattering. It follows that in the forward
direction
 7

The integral over the electron density ρ (the number of electrons in an infinitesimal region divided by
the volume of the region) multiplied by the volume element 4πr 2 dr is the total number of electrons, Ne,
in the atom. Hence, in the forward direction, f = Ne.

For example, the scattering factors of Na +, K+, and Cl− are 8, 18, and 18, respectively.

The scattering factor is smaller in non-forward directions because (sin kr)/kr < 1 for θ > 0, so the
integral is smaller than the value calculated above.

(d) The electron density

The problem we now address is how to interpret the data from a diffractometer in terms of the detailed
structure of a crystal. To do so, we must go beyond Bragg’s law.

If a unit cell contains several atoms with scattering factors f j and coordinates (xja, yjb, zjc), where xj is
the coordinate of the atom j in the a direction, expressed as a fraction of the length a, and likewise for
the other coordinates.
Then we show in the Justification below that the overall amplitude of a wave diffracted by the {hkl}
planes is given by

The sum is over all the atoms in the unit cell. The quantity Fhkl is called the structure factor.

The structure factor is the net amplitude of a given {hkl} reflection that takes into account
the positions and types of all the atoms in the unit cell.

Justification: THE STRUCTURE FACTOR

Relating the structure factor to the location of the atoms and their scattering factors

Fig. 9 Diffraction from a crystal containing two kinds of atoms. (a) For a (100) reflection from the A planes, there is a phase
difference of 2π between waves reflected by neighbouring planes. (b) For a (200) reflection, the phase difference is 4π. The
reflection from a B plane at a fractional distance xa from an A plane has a phase that is x times these phase differences.

Step 1 Consider the (h00) reflection with h = 1

Consider the crystal shown schematically in Figure. The reflection corresponds to two waves from
adjacent A planes, the phase difference of the waves being 2π. If there is a B atom at a fraction x of the
 8

distance between the two A planes, then it gives rise to a wave with a phase difference 2πx relative to
an A reflection. To see this conclusion, note that, if x = 0, there is no phase difference; if x = 1/2 the
phase difference is π; if x = 1, the B atom lies where the lower A atom is and the phase difference is
2π.

Step 2 Consider the (h00) reflection with h = 2

Now consider a (200) reflection. There is now a 2 × 2π difference between the waves from the two A
layers, and if B were to lie at x = 0.5 it would give rise to a wave that differed in phase by 2π from the
wave from the upper A layer.

Thus, for a general fractional position x, the phase difference for a (200) reflection is 2 × 2πx.

Step 3 Generalize these conclusions

For a general (h00) reflection, the phase difference is therefore h × 2πx. For three dimensions, this
result generalizes to Φhkl = 2π(hx +ky + lz).

Step 4 Formulate the total amplitude of the scattered waves

If the amplitude of the waves scattered from A is f A at the detector, that of the waves scattered from B
with phase difference Φhkl is f B𝑒 iΦhkl .

The total amplitude at the detector is therefore

When there are several atoms present, each with scattering factor f j and phase Φhkl (j) = 2π(hxj +kyj +
lzj), the total amplitude of the (hkl) reflection, the structure factor, is

Example. Calculating a structure factor

Calculate the structure factors for the unit cell of NaCl.

Method The structure factor is defined by the following eqn.

To use this equation, consider the ions at the locations specified in Fig.
Write f + for the Na+ scattering factor and f − for the Cl− scattering factor.
Note that ions in the body of the cell contribute to the scattering with a
strength f. However, ions on faces are shared between two cells
(use ½f ), those on edges by four cells (use ¼f ), and those at corners by
eight cells (use ⅛f ). Two useful relations are

The best way to proceed is to draw up a table showing the weights, positions and phases.
 9

Table for Na+ ions

Atom Weight x y z Φ hkl

1. ⅛ 0 0 0 0
2. ⅛ 1 0 0 2πh
3. ⅛ 0 1 0 2πk
4. ⅛ 1 1 0 2π(h + k)
5. ⅛ 0 0 1 2πl
6. ⅛ 1 0 1 2π(h + l)
7. ⅛ 0 1 1 2π(k + l)
8. ⅛ 1 1 1 2π(h + k+ l)
9. ½ ½ ½ 0 2π(½h + ½k)
10. ½ ½ 0 ½ 2π(½h + ½l)
11. ½ 0 ½ ½ 2π(½k + ½l)
12. ½ 1 ½ ½ 2π(h + ½k + ½l)
13. ½ ½ 1 ½ 2π(½h + k + ½l)
14. ½ ½ ½ 1 2π(½h + ½k + l )

The table for Cl− ions

Atom Weight x y z Φ hkl

1. ¼ ½ 0 0 πh
2. ¼ 0 ½ 0 πk

3. ¼ 1 ½ 0 2π(h + ½ k)
4. ¼ ½ 1 0 2π( ½ h + k)
5. ¼ 0 0 ½ πl
6. ¼ 1 0 ½ 2π(h + ½ l)
7. ¼ 0 1 ½ 2π(k + ½ l)
8. ¼ 1 1 ½ 2π(h + k+ ½ l)
9. 1 ½ ½ ½ 2π(½h + ½k + ½ l)
10. ¼ ½ 0 1 2π(½h + l)

11. ¼ 0 ½ 1 2π(½k + l)

12. ¼ 1 ½ 1 2π(h + ½k + l)
13. ¼ ½ 1 1 2π(½h + k + l )

The phase factors for the first eight Na atoms in the table are +1, and as they each have a weight of ⅛,
the total contribution to the structure factor is f +. The remaining atoms all have weight ½, and their
contribution to the structure factor is

½ [ei2π(h/2 + k/2) + ei2π(h/2 + l/2) + ei2π(k/2 + l/2) + ei2π(h + k/2 + l/2) + ei2π(h/2 + k + l/2) + ei2π(h/2 + k/2 + l )]

The terms ei2πh , ei2πk, and ei2πl are all +1, so the last three terms can be simplified.
 10

½ [ei2π(h/2 + k/2) + ei2π(h/2 + l/2) + ei2π(k/2 + l/2) + ei2π(k/2 + l/2) + ei2π(h/2 + l/2) + ei2π(h/2 + k/2 )]
A further simplification is to use einπ = (−1)n:

½[(−1)h+k + (−1)h+l + (−1)k+l + (−1)k+l + (−1)h+l + (−1)h+k ]

= (−1)h+k + (−1)h+l + (−1)k+l

The overall contribution of the Na+ ions to the structure factor is therefore

f +[ 1 + (−1)h+k + (−1)h+l + (−1)k+l]

A similar procedure to that used for the Na+ ions gives the following contribution of the Cl − ions to the
structure factor
f −[(−1)h+k+l + (−1)h + (−1)k + (−1)l]

The structure factor is therefore

Fhkl = f +[ 1 + (−1)h+k + (−1)h+l + (−1)k+l]

+ f −[(−1)h+k+l + (−1)h + (−1)k + (−1)l]

Now note that:

• if h, k, and l are all even, Fhkl = f +{1 + 1 + 1 + 1} + f −{1 + 1 + 1 + 1} = 4( f + + f −)
• if h, k, and l are all odd, Fhkl = 4( f + − f −)
• if one index is odd and two are even, or vice versa, Fhkl = 0
The hkl all-odd reflections are less intense than the hkl all-even, and some of the reflections are absent.

Comment. For f + = f −, which is the case for identical atoms, the hkl all-odd reflections have zero
intensity; such a structure would be a cubic P with lattice parameter a/2

Self-test. Which reflections cannot be observed for a cubic I lattice?

Answer. [for h + k + l odd, Fhkl = 0]

Fourier synthesis
Fourier synthesis is the construction of the electron density distribution from structure factors.

The intensity of a reflection is proportional to the square modulus of the amplitude of the wave, which
in turn proportional to the structure factor,|Fhkl |.

If the structure factor is f A + f B𝑒 𝑖𝛷ℎ𝑘𝑙 , the intensity, Ihkl, is

This expression expands to

[∵ eix +e−ix = 2cosx]

The cosine term either adds to or subtracts from 𝑓𝐴2 + 𝑓𝐵2 depending on the value of Φ hkl, which in turn
depends on h, k, and l and x, y, and z. Hence, there is a variation in the intensities of the reflections
with different hkl.

The A and B reflections interfere destructively when the phase difference is π, and in this case the total
intensity is zero if the atoms have the same scattering power.
 11

For example, if the unit cells are cubic I with a B atom at x = y = z = ½, then the A, B phase difference
is (h + k + l )π. Therefore, all reflections for odd values of h + k + l vanish because the waves are
displaced in phase by π.

For a cubic P lattice diffraction is possible for all {hkl}, therefore the diffraction pattern for a cubic I
lattice can be constructed from that for the cubic P lattice by striking out all reflections with odd values
of h + k + l.
Similarly, for a cubic F lattice the missing lines are ones with two out of h, k, and l odd, and the
remaining one even, or two even and one odd.

Recognition of these systematic absences in a powder spectrum can be used to assign the lattice type
(Fig 10).

Fig. 10 The powder diffraction patterns and the

systematic absences of three versions of a cubic
cell: cubic F (fcc; h, k, l all even or all odd are
present), cubic I (bcc;h + k + l = odd are absent),
cubic P. Comparison of the observed pattern
with patterns like these enables the unit cell to
be identified. The locations of the lines give the
cell dimensions.

The intensity of the (hkl) reflection is proportional to |Fhkl |2 , so in principle we can determine the
structure factors experimentally by taking the square root of the corresponding intensities (but see
below). Then, once we know all the structure factors Fhkl, we can calculate the electron density
distribution, ρ(r), in the unit cell by using the expression

where V is the volume of the unit cell. The above equation is called a Fourier synthesis of the electron
density.
Fourier transforms occur throughout chemistry in a variety of guises. The essence of the procedure in
this case is to express the varying electron density in a unit cell as a superposition of sine and cosine
waves.

Example. Calculating an electron density by Fourier synthesis

Consider the {h00} planes of a crystal extending indefinitely in the x-direction. In an X-ray analysis the
structure factors were found as follows:

h: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Fh : 16 −10 2 −1 7 −10 8 −3 2 −3 6 −5 3 −2 2 −3

(and F−h = Fh ). Construct a plot of the electron density projected on to the x-axis of the unit cell.
 12

Method. We need to substitute these values into eqn,

but because the problem is one-dimensional, the sum is over only the index h and only the terms
e−2πihx need to be considered.

The solution. Because F−h = Fh , the sum, rather than running from h = −∞ to +∞, can be written as
running from 1 to +∞:

[∵ eix +e−ix = 2cosx]

We evaluate the sum (truncated at h = 15) for points 0 ≤ x ≤ 1 using mathematical software. The results
are plotted in Fig. (blue line). There are three clear maxima in this function, which can be identified as
the positions of three atoms.

Comment. The more terms that are included

(meaning the more reflections that are measured),
the more accurate is the density plot. Terms
corresponding to high values of h (which
corresponds to short wavelength cosine terms in
the sum) account for the finer details of the
electron density; low values of h account for the
broad features.

(e) The phase problem

A problem with the procedure outlined above is that the observed intensity Ihkl is proportional to the
square modulus |Fhkl|2 , so we cannot say whether we should use +|Fhkl| or −|Fhkl| in the sum in
Fourier synthesis.

In fact, the difficulty is more severe for noncentrosymmetric unit cells because, if we write Fhkl as the
complex number |Fhkl |eiα, where α is the phase of Fhkl and |Fhkl | is its magnitude, then the intensity
lets us determine |Fhkl | but tells us nothing of its phase, which may lie anywhere from 0 to 2π.

This ambiguity is called the phase problem; its consequences are illustrated by comparing the two
plots in Fig. (plot of the electron density), in which the phases of the structure factors have been
changed but the amplitudes kept the same. Some way must be found to assign phases to the structure
factors, for otherwise the sum for ρ cannot be evaluated and the method would be useless.

The phase problem can be overcome to some extent by a variety of methods. One procedure that is
widely used for inorganic materials with a reasonably small number of atoms in a unit cell and for
organic molecules with a small number of heavy atoms is the Patterson synthesis.
 13

Instead of the structure factors Fhkl, the values of |Fhkl |2 , which can be obtained without ambiguity
from the intensities, are used in an expression that resembles Fourier synthesis:

The outcome of a Patterson synthesis is a map of the vector separati ons of the atoms (the distances
and directions between atoms) in the unit cell.

Thus, if atom A is at the coordinates (xA,yA,zA) and atom B is at (xB,yB,zB), then there will be a peak at
(xA − xB, yA − yB, zA − zB) in the Patterson map. There will also be a peak at the negative of these
coordinates, because there is a vector from B to A as well as a vector from A to B. The height of the
peak in the map is proportional to the product of the atomic numbers of the two atoms, Z AZ B.

For example, if the unit cell has the structure shown in Fig. 11a, the Patterson synthesis would be the
map shown in Fig. 11b, where the location of each spot relative to the origin gives the separation and
relative orientation of each pair of atoms in the original structure.

Fig. 11. The Patterson synthesis corresponding to the pattern in (a) is the pattern in (b). The distance and orientation of each
spot from the origin gives the orientation and separation of one atom–atom separation in (a). Some of the typical distances
and their contribution to (b) are shown as R1, etc.

Heavy atoms dominate the scattering because their scattering factors are large, of the order of their
atomic numbers, and their locations may be deduced quite readily. The sign of Fhkl can now be
calculated from the locations of the heavy atoms in the unit cell, and to a high probability the phase
calculated for them will be the same as the phase for the entire unit cell. To see why this is so, we have
to note that a structure factor of a centrosymmetric cell has the form

where f heavy is the scattering factor of the heavy atom and flight the scattering factors of the light atoms.
The f light are all much smaller than f heavy, and their phases are more or less random if the atoms are
distributed throughout the unit cell. Therefore, the net effect of the f light is to change F only slightly
from f heavy, and we can be reasonably confident that F will have the same sign as that calculated from
the location of the heavy atom. This phase can then be combined with the observed | F | (from the
reflection intensity) to perform a Fourier synthesis of the full electron density in the unit cell, and
hence to locate the light atoms as well as the heavy atoms.

Modern structural analyses make extensive use of direct methods. Direct methods are based on the
possibility of treating the atoms in a unit cell as being virtually randomly distributed (from the
radiation’s point of view), and then using statistical techniques to compute the probabilities that the
phases have a particular value. It is possible to deduce relations between some structure factors and
sums (and sums of squares) of others, which have the effect of constraining the phases to particular
 14

values (with high probability, so long as the structure factors are large). For example, the Sayre
probability relation has the form

sign of Fh+h', k+k', l+l' is probably equal to (sign of Fhkl) × (sign of Fh ′k′l′)

For example, if F122 and F232 are both large and negative, then it is highly likely that F354 , provided it
is large, will be positive.
In the final stages of the determination of a crystal structure, the parameters describing the structure
(atom positions, for instance) are adjusted systematically to give the best fit between the observed
intensities and those calculated from the model of the structure deduced from the diffraction pattern.
This process is called structure refinement. Not only does the procedure give accurate positions for all
the atoms in the unit cell, but it also gives an estimate of the errors in those positions and in the bond
lengths and angles derived from them. The procedure also provides information on the vibrational
amplitudes of the atoms.

.
CONCEPTS
1. A reflection refers to an intense beam emerging in a particular direction and arising from
constructive interference.

2. The glancing angle, 2θ, is the angle through which a beam is deflected.

3. Bragg’s law relating the glancing angle θ to the separation of lattice planes is λ = 2d sin θ,
where λ is the wavelength of the radiation.

4. The scattering factor is a measure of the ability of an atom to scatter electromagnetic

radiation.

5. The structure factor is the overall amplitude of a wave diffracted by the {hkl} planes and atoms
distributed through the unit cell.

6. Fourier synthesis is the construction of the electron density distribution from structure
factors.

EQUATIONS

• Bragg’s law λ = 2d sin θ d is the lattice spacing, 2θ is the glancing angle.

• Scattering factor

spherically symmetrical atom

• Structure factor

• Fourier synthesis

V is the volume of the unit cell

 15

Discussion questions
1. What is meant by a systematic absence? How do they arise and how they can be helpful in
identifying the type of unit cell?
2. Discuss what is meant by ‘scattering factor’. How is it related to the number of electrons in the
atoms scattering X-rays?
3. Describe the consequences of the phase problem in determining structure factors and explain
how it may be overcome.

Further information

X-ray crystallography of biological macromolecules

X-ray crystallography is the deployment of X-ray diffraction techniques for the determination of the
location of all the atoms in molecules as complicated as biopolymers. Bragg’s law helps us understand
the features of one of the most seminal X-ray images of all time, the characteristic X-shaped pattern
obtained by Rosalind Franklin and Maurice Wilkins from strands of DNA and used by James
Watson and Francis Crick in their construction of the double-helix model of DNA (Fig.). To interpret
this image by using the Bragg law we have to be aware that it was obtained by using a fibre consisting
of many DNA molecules oriented with their axes parallel to the axis of the fibre, with X-rays incident
from a perpendicular direction. All the molecules in the fibre are parallel (or nearly so), but are
randomly distributed in the perpendicular directions; as a result, the diffraction pattern exhibits the
periodic structure parallel to the fibre axis superimposed on a general background of scattering from
the distribution of molecules in the perpendicular directions.

Fig. 1. The X-ray diffraction pattern obtained from a fibre of B-DNA. The black dots are the reflections, the points of maximum
constructive interference, that are used to determine the structure of the molecule.

Fig. 2. The origin of the X pattern characteristic of diffraction by a helix. (a) A helix can be thought of as consisting of an array
of planes at an angle α together with an array of planes at an angle −α. (b) The diffraction spots from one set of planes appear
at an angle α to the vertical, giving one leg of the X, and those of the other set appear at an angle −α, giving rise to the other
leg of the X. The lower half of the X appears because the helix has up–down symmetry in this arrangement. (c) The sequence
of spots outward along a leg of the X corresponds to first-, second-, . . . order diffraction (n = 1, 2, . . . ).
 16

Fig. 3. The effect of the internal structure of the helix on the X-ray diffraction pattern. (a) The residues of the macromolecule
are represented by points. (b) Parallel planes passing through the residues are perpendicular to the axis of the molecule. (c)
The planes give rise to strong diffraction with an angle that allows us to determine the layer spacing h from λ = 2h sin θ

Fourier series
A Fourier series is a linear combination of sines and cosines that replicates a periodic function:

--------------- (1)

A periodic function is one that repeats periodically, such that f(x + 2L) = f(x) where 2L is the period.
Although it is perhaps not surprising that sines and cosines can be used to replicate continuous
functions, it turns out that—with certain limitations—they can also be used to replicate discontinuous
functions too. The coefficients in eqn. (1) are found by making use of the orthogonality of the sine and
cosine functions

------------- (2a)
and the integrals

-------------- (2b)

where δmn = 1 if m = n and 0 if m ≠ n.

Thus, multiplication of both sides of eqn. (1) by cos(kπx/L) and integration from −L to L gives an
expression for the coefficient ak, and multiplication by sin (kπx/L) and integration likewise gives an
expression for bk:

---------------- (3)
Fourier transforms
The Fourier series in eqn. (1) can be expressed in a more succinct manner if we allow the coefficients
to be complex numbers and make use of de Moivre’s relation

-------------- (4)
for then we may write

--------------- (5)

This complex formalism is well suited to the extension of this discussion to functions with periods that
become infinite. If a period is infinite, then we are effectively dealing with a nonperiodic function, such
as the decaying exponential function e −x.
 17

We write δk = π/L and consider the limit as L → ∞ and therefore δk → 0: that is, eqn. (5) becomes

---------------- (6)

In the last line we have anticipated that the limits of the integral will become infinite. At this point we
should recognize that a formal definition of an integral is the sum of the value of a function at a series
of infinitely spaced points multiplied by the separation of each point:

-------------- (7)
Exactly this form appears on the right-hand side of eqn. (6), so we can write that equation as

--------------- (8)

At this stage we can drop the prime on x in the expression for 𝑓̃ k. We call the function 𝑓̃ k the Fourier
transform of f(x); the original function f(x) is the inverse Fourier transform of 𝑓̃ k ( ).

Illustration. A Fourier transform

The Fourier transform of the symmetrical exponential function f(x) = e−a|x| is

The original function and its Fourier transform are drawn in the following Figure.

Figure (a) The symmetrical exponential function f(x) = e−a|x| and (b) its Fourier transform for two values of the decay constant
a. Note how the function with the more rapid decay has a Fourier transform richer in short-wavelength (high k) components.
 18

The physical interpretation of eqn.8 is that f(x) is expressed as a superposition of harmonic (sine and
cosine) functions of wavelength λ = 2π/k, and that the weight of each constituent function is given
by the Fourier transform at the corresponding value of k. This interpretation is consistent with the
calculation in illustration. As we see from Fig., when the exponential function falls away rapidly with
time, the Fourier transform is extended to high values of k, corresponding to a significant contribution
from short-wavelength waves. When the exponential function decays only slowly, the most significant
contributions to the superposition come from low-frequency components, which is reflected in the
Fourier transform, with its predominance of small-k contributions in this case. In general, a slowly
varying function has a Fourier transform with significant contributions from small-k components.