Fundamentals of

Crystallography

R.M. Sweet
Brookhaven Biology
 Outline for the Lecture
• Remind you how much you already know -- lenses, crystals
• Show why crystals give diffraction spots.
• Develop the idea of “The Reciprocal Lattice”
• Give some idea how we might actually measure diffraction
data
• Show how, given a crystal, we can calculate the diffraction
pattern
• Conversely, show how to calculate the structure from the
diffraction
• Describe the importance of symmetry to diffraction
• Outline the structure-solving methods -- heavy atoms and
MADness
 Outline for the Lecture
• Remind you how much you already know -- lenses, crystals
• Show why crystals give diffraction spots.
• Develop the idea of “The Reciprocal Lattice”
• Give some idea how we might actually measure diffraction
data
• Show how, given a crystal, we can calculate the diffraction
pattern
• Conversely, show how to calculate the structure from the
diffraction
• Describe the importance of symmetry to diffraction
• Outline the structure-solving methods -- heavy atoms and
MADness
 Creation of a molecule’s image from a crystal has
similarities to creating an image with a lens
LIGHT Detector

Fourier
X-Rays Computer
Synthesis

Object Image
Lens
Visible

Image
plane
You already understand a little about how lenses work

LIGHT Detector

Fourier
X-Rays Computer
Synthesis

Two rays leaving from the same

point end up at the same place
Object Image
Lens
Visible

Image
plane
 Maybe you didn’t know …

LIGHT Detector

Fourier
X-Rays Computer
Synthesis

All rays leaving in

the same direction… …end up together in this plane
Object Image
Lens
Visible

Diffraction Image
plane plane
 We use a crystal to give us diffraction, and
computation to do the rest of the work of the lens.
LIGHT Detector

Fourier
X-Rays Computer
Synthesis

Object Image
Lens
Visible

Diffraction Image
plane plane
We’ll see that the diffraction pattern
gives information about the
dimensions and periodicity of this
one view of the object.
 Why do we use x-rays?

• The features we’re trying to see are on the

order of the distance between atoms: 10-10
meters.
• To “see” the atoms, we need to use light
with a wavelength that is near to this distance.

• X-Rays (x-ray light) have a suitable

wavelength.
 What is a crystal?
• A crystal is a periodic arrangement of objects
(molecules) repeating in two or three dimensions.
• The repeating unit is a parallelepiped (in 3-D) or a
parallelogram (in 2-D).
• A crystal of a typical protein will be half a mm on
a side and contain 1015 molecules.
Here’s one choice of repeating unit
in this crystal made of apple trees

Parallelograms
defining crystal
repeat.
We could make a different
choice of repeating unit

Other
parallelograms
defining crystal
repeat.
Why do we use crystals when we’d
like to see one molecule?
• We can’t focus enough x-rays into a small enough
volume to “see” a molecule.
• Even if we could, the x-rays would burn up the
molecule.
• Even if that would work, we don’t have a lens for
the x-rays.
• The crystal amplifies the signal, and gives us a
way to get the phase information back.
Let’s return to our crystal made of
apple trees, and define “planes” in
that crystal.
We can slice the crystal at lattice points:
all planes pass through the same apple
Or like this, at an angle. As before, all planes pass by
the same apple in trees along the oblique row.
And at other angles. Notice:
• planes all pass by the same apple;
• the “stuff” between pairs of planes is always the same.
 Outline for the Lecture
• Remind you how much you already know -- lenses, crystals
• Show why crystals give diffraction spots.
• Develop the idea of “The Reciprocal Lattice”
• Give some idea how we might actually measure diffraction
data
• Show how, given a crystal, we can calculate the diffraction
pattern
• Conversely, show how to calculate the structure from the
diffraction
• Describe the importance of symmetry to diffraction
• Outline the structure-solving methods -- heavy atoms and
MADness
 Diffraction
Let’s do a thought experiment. Send a beam of x-rays at a
pair of single atoms, suspended in space. If the angle is just
right for the wavelength and distance between the atoms,
the scattered x-rays will be in phase, and they will interfere
constructively.

In phase
In phase
Now, let’s think of the stuff between the lattice planes
as being like those two atoms, and try to write a law
that will show conditions to get diffraction.
Now get rid of
the orchard...
Braggs’ Law describes diffraction as reflection from planes

Wavelength

The wave travels exactly one wavelength to

take the little detour
Watch what happens as we go from maximum
to minimum diffracting position and back.

See: www.journeysunysbedu/ProjectJava/Bragg/home.html
 Now let’s use a similar diagram to learn
something new about diffraction from a
crystal

sin + sin ' = 2 sin[ ( + ')/2 ] cos[ ( - ')/2 ]

cos - cos ' = 2 sin[ ( + ')/2 ] sin[ ( - ')/2 ]

Notice that some of the path length
differences are an integral number of
wavelengths:
If we make this substitution, to define
a reflecting plane:
We discover that the reflecting plane must pass
through lattice points; it is a general lattice plane!
Sometimes we call them Bragg planes.
We have a way to “index” planes in a paralellepiped to
give a unique description of them.

The indices are

the number of
times each
plane cuts the
axis of the
“unit cell” of
the crystal --
the smallest
repeating unit
that makes up
the crystal.
(2, 1, 2)
 Outline for the Lecture
• Remind you how much you already know -- lenses, crystals
• Show why crystals give diffraction spots.
• Develop the idea of “The Reciprocal Lattice”
• Give some idea how we might actually measure diffraction
data
• Show how, given a crystal, we can calculate the diffraction
pattern
• Conversely, show how to calculate the structure from the
diffraction
• Describe the importance of symmetry to diffraction
• Outline the structure-solving methods -- heavy atoms and
MADness
To relate the planes in the crystal lattice to the points in
the diffraction pattern, we make Ewald’s construction.
We have that sin θ = (OA/2)/(1/λ) = λ×OA/2, or λ = 2 sin θ / OA. If we take
1/OA as being equivalent to d, this represents Braggs’ Law: λ = 2d sin θ.
Notice the reflection plane, and that OA is perpendicular to it. The Ewald
construction exists in a space with dimensions of reciprocal distance!

This defines
Reciprocal
Space! The vector
of length 1/d is
perpendicular to
the reflecting plane
that lies θ from the
“rays.”
And Bragg’s Law is obeyed – diffraction will occur – when
that vector of length 1/dhkl, that is perpendicular to the lattice
plane (hkl), touches the Ewald sphere of radius 1/.

Because we
have this tool,
the Ewald
sphere and
reciprocal space,
we never need
to think about
lattice planes
and Bragg’s law
again.
How can we define this vector that is perpendicular to
the Bragg plane, and has a length that is the reciprocal
of the distance between the planes?
We’ll define the edges of a unit cell with three vectors.
Start with a and b. We know that the cross product of two
vectors lies perpendicular to the plane of the two vectors.
This is the direction we want. The amplitude of a×b is the
area of the parallelogram defined by the vectors:

‌a×b‌ = ab sinα

a
α
We’ve described the base of the unit cell of the crystal by
two vectors a and b, and the area of the base is the amplitude
of the cross product of a and b.
Now we’ll include the third vector c. We want the spacing
d(001), between the ab planes [the (001) lattice planes]. It must be
the projection of c on the vector a×b. We know that we get the
product of the projection of one vector on another with the
vector dot product: (a×b)∙c.

a×b

c
d(001) (001) plane
b

a
α
So a×b∙c, known as a vector triple product, is the area of ab
times d(001), the spacing between the planes. That, of course is
the Volume of the unit cell. If we divide this quantity into the
area, we get the reciprocal of the spacing, which is what we
want!!
1/d (001) = a×b / a×b∙c = c*

a×b

c
d(001)
b

a
α
So the reciprocal lattice vector that represents the (001)
planes is
s001 = a×b / a×b∙c and ‌s001‌ = 1/d(001)
We define each axial reciprocal lattice vector as a
reciprocal unit cell axis:
s100 = a* s010 = b* s001 = c*

a×b

c
d(110)
b

a
shkl = ha* + kb* + lc*
The implication of this is that we
never think about Bragg planes
again, we think only of reciprocal-
lattice vectors.

shkl = ha* + kb* + lc*

We can use these figures from Taylor and Lipson to teach us something about
crystal diffraction. Notice how the diffraction maxima appear perpendicular to
the Bragg planes. If the “crystal” is single atoms, we get just a lattice of
diffraction.

crystal Coordinates of points diffraction

are “indices.”
Plate 2 from Taylor and Lipson -- Optical Transforms
 Do we understand the real/reciprocal lattice idea?
Confirm that the
vectors
Crystal – perpendicular to
Real Lattice the Crystal-
Lattice planes are
parallel to the
Reciprocal
Diffraction – Lattice vectors,
Reciprocal and that the
Lattice reciprocal
distances make
sense.
 Outline for the Lecture
• Remind you how much you already know -- lenses, crystals
• Show why crystals give diffraction spots.
• Develop the idea of “The Reciprocal Lattice”
• Give some idea how we might actually measure
diffraction data
• Show how, given a crystal, we can calculate the diffraction
pattern
• Conversely, show how to calculate the structure from the
diffraction
• Describe the importance of symmetry to diffraction
• Outline the structure-solving methods -- heavy atoms and
MADness
Remember the geometry -- if the Bragg
planes lie angle  from the incident x-ray
beam, the total diffraction angle will be 2.
We can make an instrument to exploit that
geometry.
And Remember the objective – We must
view the molecule from every direction to
recreate a three-dimensional image:

• We must obtain diffraction from all of the

Bragg planes;
• We must sample all of the reciprocal lattice.
My first data were collected with a Weissenberg
Camera
 The Weissenberg photograph gives a
wonderfully distorted view of reciprocal space.

(h0l) data
Martin Buerger devised a camera geometry
that preserved the shape of reciprocal space.
 The precession
photograph
allows us to
view the
diffraction
pattern of the
crystal lattice as
an undistorted
pattern of spots.

An antique precession photo of Chymotripsin, courtesy of David M Blow

 Notice the
counterweights, the
beam stop, and the
screen to select only
one reciprocal lattice
layer.
Cameras very similar
to this were used 40
yrs ago to solve the
first protein crystal
structures.

Bram Schierbeek, Bruker-AXS

Data for the first protein structures were measured
on precession cameras with x-ray film. (t-40)

Bram Schierbeek, Bruker-AXS

The automated Eulerian cradle decreased the labor, but
still one measured reflections one at a time. (30yrs ago)
 Simple rotation
geometry
produces a
complicated pattern
that requires good
software to
interpret. Modern
CCD-based
detectors with four-
circle
diffractometers
record such
patterns and
measure every spot
intensity.
An antique rotation photograph of B-Phyoerythrin -- real x-ray film.
Uli Arndt and Alan Wonacott invented the automated
rotation camera. Still x-ray film, but very much more
efficient.
(~23yrs ago)

Bram Schierbeek, Bruker-AXS

Another Uli Arndt invention was a video-based detector

The screen was small, but it was very sensitive and could read out
continuously – the x-tal just kept rotating as images came out.
(20yrs ago)

Bram Schierbeek
Bruker-AXS
 An important advance was photoluminescent imaging plates.
MAR research, followed by Rigaku, made a successful camera
that worked like electronic x-ray film, but much better. (19 yrs)
 The advance that made possible our modern detectors
was made by Ed Westbrook, Sol Gruner, and others:
bonding of a charge-coupled device to a fiber-optic taper
with an x-ray sensitive phosphor in front. (13yrs ago)
PHO SPHO R
SCREEN F IB E R O P T IC T A P E R

CCD

25 m m

1 2 k e V X -ra y 6 5 0 lp h 3 5 lp h 10 e-
Several of these can be bonded together to make a large
detector… (9yrs ago)
Like this one, made
for us by Walter
Phillips
And the modern commercial versions are large,
fast, and very accurate. (4yrs ago)
Detectors like these are the basis for
modern, high-throughput crystallography!
The planes of
spots in reciprocal
space appear as
circles of spots on
an area-sensitive x-
ray detector (film,
IP, CCD-based,
etc.)
Z. Dauter
As the
crystal is
rotated, the
circles are
extended into
“lunes”

Z. Dauter
 Rotation sweeps out a strangely-
shaped volume. However...
• Many r.l. points will be
recorded during a single short
rotation.
• Contiguous rotations will cover
much of the reciprocal lattice.
• The “camera” is simple: an
axis, a film, and a shutter.
• It’s easy to substitute a range of
detectors.
Let’s look at a series of images from a CCD-
based detector, each representing one degree
of crystal rotation
 Outline for the Lecture
• Remind you how much you already know -- lenses, crystals
• Show why crystals give diffraction spots.
• Develop the idea of “The Reciprocal Lattice”
• Give some idea how we might actually measure diffraction
data
• Show how, given a crystal, we can calculate the
diffraction pattern
• Conversely, show how to calculate the structure from the
diffraction
• Describe the importance of symmetry to diffraction
• Outline the structure-solving methods -- heavy atoms and
MADness
 Now we use the Taylor and Lipson figures to see how the
contents of the crystal relate to the diffraction pattern.
Notice (1)
The
symmetry,
and (2) how
the
continuous
diffraction
pattern of
one molecule
(b) is
“sampled” by
the lattice of
diffraction
points.
crystal Coordinates of points diffraction
are “indices.”
Plate 2 from Taylor and Lipson -- Optical Transforms
Here’s another (2D) example with an asymmetric motif

Each spot
represents the
intensity of
reflection
from one set
of planes
cutting
through the
crystal

Note the
inversion
symmetry

Plate 26 from Taylor and Lipson -- Optical Transforms

 Remember that we can use an x/y graph to
represent the phase and amplitude of a wave:

i

Ai
Ao Phase
Ao angle

0  2 Ar r

And then we describe the “wave” as a complex number:

f = Ao{cos + i sin } and
f = Ao ei
• The amplitude of scattering depends on the
number of electrons on each atom.
• The phase depends on the fractional distance it
lies from the lattice plane.

Randy Read

Atomic structure factors

Scattering from
add as complex numbers,
lattice planes
or vectors.
 We can write an expression to describe this
diffraction from atoms in a crystal
The scattering amplitude (the structure factor)
for an individual atom is going to be: The hkl describe
the Bragg Planes

The 2 and the fractional

The scattering power of the coordinates xi take care of
atom, ~ the number of electrons the phase angle

And the structure factor for a crystal of atoms will be:

The strength of scattering from each atom

Does this expression for the
Structure Factor make sense?
Try it with an example: a crystal with
three atoms. What are the phases of
scattering from each atom? Use this

For these planes, (h, k) = (3, 2)

For atom 1. x, y = 2/3, 0: So 2π(hx + ky) = 2π(3 x 2/3 + 2 x 0) = 4π = 0

The atom is on the plane, so this makes sense.

For atom 2. x, y = 0, 1/2: So 2π(hx + ky) = 2π(3 x 0 + 2 x 1/2) = 2π = 0

Again, the atom is on the plane, so this makes sense.

For atom 3. x, y = 1/3, 1/4: So 2π(hx + ky) = 2π(3 x 1/3 + 2 x 1/4) = 3π = π

The atom lies half-way between two planes, so this makes sense.
We can see how the structure factors
from individual atoms add up.
See also:
http://www.ysbl.york.ac.uk/~cowtan/
sfapplet/sfintro.html Structure Factor Tutorial
fourier/fourier.html Book of Fourier
 Outline for the Lecture
• Remind you how much you already know -- lenses, crystals
• Show why crystals give diffraction spots.
• Develop the idea of “The Reciprocal Lattice”
• Give some idea how we might actually measure diffraction
data
• Show how, given a crystal, we can calculate the diffraction
pattern
• Conversely, show how to calculate the structure from the
diffraction
• Describe the importance of symmetry to diffraction
• Outline the structure-solving methods -- heavy atoms and
MADness
Q: How do we perform the second
interference step in the functioning of the lens
-- to reconstruct the image of the original
object?
A: We will have to calculate it.

Q: How will we represent that object?

A: The x-rays are scattered from electrons in
the atoms of the crystal.

Therefore: for us, the “image” is going to be

a representation of the electron density.
The structure factor and the electron density
function are Fourier inverses of one another

• Note that the electron density is real but the structure factor
is complex.
• The phase information must be included in the Fourier
synthesis that produces the electron density!
• This has to be recovered, because the diffraction
experiment measures the intensity of diffraction, which is
the square of the structure factor:
I = F2
 How does Fourier synthesis work?
Can we produce a trial structure and see how
waves can be summed to give this structure
back?
In the Fourier Synthesis, just a few waves suffice to give a
reasonable approximation to the original pattern
What is the concept of
“resolution?”
Here is the Fourier
synthesis function:

When the limits of the

summation are not so
great, information is
lost in the synthesized
structure.
We say that the
“resolution” equals the
d-spacing of the
smallest Bragg planes.
Resolution: The d-spacing of the highest
order Bragg planes included in the Fourier
synthesis. Small is good.

1.0 Ång 1.8 Ång

3.0 Ång 4.0 Ång Graphics by

Phil Evans
Another
example.
The famous Taylor
and Lipson rubber
ducky.
 Outline for the Lecture
• Remind you how much you already know -- lenses, crystals
• Show why crystals give diffraction spots.
• Develop the idea of “The Reciprocal Lattice”
• Give some idea how we might actually measure diffraction
data
• Show how, given a crystal, we can calculate the diffraction
pattern
• Conversely, show how to calculate the structure from the
diffraction
• Describe the importance of symmetry to diffraction
• Outline the structure-solving methods -- heavy atoms and
MADness
Symmetry of crystals
We’ll take some of our examples
from David Blow’s book.
Symmetry: An operation of
rotation, translation, inversion,
mirroring, or some combination
of these that takes an object back
into itself.
• The simplest symmetry in a crystal
is repetition.
• The repeated motif may have its
own symmetry.
Symmetry Groups
Biological
molecules are all
chiral, or “handed,”
so only rotation
and translation
symmetry are
permissible.
Here are the
combinations
(groups) of
symmetries one
finds in
macromolecular
crystals.
 What is a Group?
Elements in a group must obey certain properties:
• There must be the identity element.
• The combination of any two elements must generate
an element of the group. This is called closure.
• Number of elements = number of objects repeated =
order of the group.
• Every element in the group must have an inverse.

3 × 32 = I
32 = 3-1
3 × 2 = 2’
Point Group is 32
 Simple crystal symmetry

The simplest crystal

would contain a single
asymmetric object
repeated by translational
repetition only, like our
apple orchard.
 More complicated crystal symmetry
A crystal could contain a symmetric object, also
repeated by translational repetition.

Three new
symmetry
operators are
generated
Notice:
exactly two
ducks in the
unit cell
 Now let’s try it in three dimensions

Space Group P2: P = “primitive,” 2 = two-fold rotation axis.

We call this type of crystal monoclinic. Order = 2.
The Screw Axis
This symmetry
operation is a rotation
followed by a
translation.
The translation is a n/m
translation along one of
the major crystallo-
graphic directions,
where m is the order of
the major rotation axis.
Here, it’s written 21 to
represent the two-fold
screw axis, and the
translation is ½.
Centered Lattice
To make a new
monoclinic lattice, shift
the motif at the origin
along a diagonal to a new
spot by a major fraction of
the unit cell edges.
The lattice is “centered”
because a new motif
appears in the center of a
face or of the body of the
unit cell.
Subtle and beautiful rules
govern what centering
operations are allowed.
 And higher symmetry
If one has two-fold axes in more than one direction, it must be
three directions, and the axes must be perpendicular. We call
this orthorhombic.
 And finally …
A three-fold axis will produce a trigonal crystal.
Notice how the first three-fold axis creates two others
with different environments.
 The Seven Crystal Systems
The combination of symmetry elements yields only these forms
Crystal Bravais External Minimum Unit Cell
System Types Symmetry Properties
Triclinic P None a, b, c, al, be, ga,
One 2-fold axis, parallel b (b
Monoclinic P, C a, b, c, 90, be, 90
unique)
P, I,
Orthorhombic Three perpendicular 2-folds a, b, c, 90, 90, 90
F
Tetragonal P, I One 4-fold axis, parallel c a, a, c, 90, 90, 90
Trigonal P, R One 3-fold axis a, a, c, 90, 90, 120
Hexagonal P One 6-fold axis a, a, c, 90, 90, 120
P, F, Four 3-folds along space
Cubic a, a, ,a, 90, 90, 90
I diagonal
 The Bravais
Lattices
Here are the 14 ways
crystal lattices can be
formed in the seven
crystal systems.
 How many space groups?
• There are 230 space groups possible
• Only 65 of these employ only rotational symmetry
(suitable for chiral molecules)
• Here are the most abundant observed in macromolecular
structures, 65% of the total:
Space
group % of total
symbol
P212121 24.2
P3221 &
15.2
P3121
P21 13.8
C2 6.1
P43212 5.4
And finally the icosahedral symmetry of
spherical viruses
 How does symmetry affect a
diffraction pattern?
Symmetry affects a diffraction pattern in at least
three ways:
Friedel’s Law – There’s an inversion centre in
reciprocal space.
Laue Point Group – Diffraction has symmetry like
that of the crystal.
Systematic absences – some of the symmetry
operations erase some reflections.
Friedel’s Law: Bragg reflection from the
front of the planes is the same as from the
back.
(h, k, l)

(-h, -k, -l)

Laue Point Group: The diffraction will adopt
some of the symmetry of the crystal.
(-h, k, -l)

a Let’s say the

(1, 5, 2) and
the (-1, 5 –2)

c
(h, k, l)
The Laue Point Group for a crystal is
the rotational or mirror symmetry of the
space group, plus Friedel’s Law. For
example:
P2 or P21 → 2/m
Produces a two-fold, a mirror
perpendicular to it, and an inversion
centre in the diffraction pattern /
reciprocal space.
 Bijvoet
pair

Implication of this for the experiment:

One will need only to record 1/12 of reciprocal space
to get complete data. Sometimes one can record
anomalously-related reflections on the same
image.
 Outline for the Lecture
• Remind you how much you already know -- lenses, crystals
• Show why crystals give diffraction spots.
• Develop the idea of “The Reciprocal Lattice”
• Give some idea how we might actually measure diffraction
data
• Show how, given a crystal, we can calculate the diffraction
pattern
• Conversely, show how to calculate the structure from the
diffraction
• Describe the importance of symmetry to diffraction
• Outline the structure-solving methods -- heavy atoms
and MADness
How we solve structures? We must
somehow estimate phases so we can
perform the inverse Fourier transform.
• Isomorphous Replacement with heavy atoms
• MAD/SAD, a variant of IR
• Molecular replacement if we have a decent model.
 Perutz’s Fundamental Idea:
Isomorphous Replacement

FP = Σ Fatoms FPH = FP + FH FH

We find that, for some things, we can

approximate |FH| with |FPH - FP|. This often
suffices for us to solve for the positions of the
heavy atom as if it were a small-molecule
structure.
 So for some particular reflection and a
particular heavy atom, we can begin to find the
phase:

Knowing the position of the heavy atom allows us to

calculate FH. Then we use FP = FPH + (-)FH to show that the
phase triangles close with a two-fold ambiguity, at G and
at H. There are several ways to resolve the ambiguity.
One way to resolve the ambiguity is to use a
second isomorphous heavy-atom derivative.
A second technique involves use of anomalous
(resonant) scattering from a heavy atom.

In this case the resonance

between the electrons on the
heavy atom and the x-rays
cause a phase and amplitude
shift. The symmetry of
diffraction (from the front vs
back of the Bragg planes) is
broken. Friedel’s Law is
broken! This can be measured
and used.
 Scattering Power
fo

imaginary
f’
f”

Excitation Scans
real
Onethe
We can observe way to measuring
f” by represent
this
the absorption resonance
of the x-rays byisthe
plots of
atom. We measure an in
the shifts “excitation”
the real part (
spectrum. Often we imaginary
f’) and us the part (f”)
fluorescence of the absorbing atom as
of the scattering of x-rays
a measure of absorptivity.
as a function of the
photon energy.
From Ramakrishnan’s study of GH5
 Scattering Power
fo

imaginary
f’
f”

How to get f’?

real
One waycomponent
The “real,” dispersive to representis
thisf” resonance
calculated from is plots of
by the Kramers-
Kronig relationship.
the shiftsVery roughly,
in the it’s (
real part
the negative f’)
firstand
derivative of f”.part (f”)
imaginary
of the scattering of x-rays
as a function of the
photon energy.
From Ramakrishnan’s study of GH5
 Scattering Power
fo

imaginary
f’
f”

real
One way to represent
this resonance is plots of
the shifts in the real part (
f’) and imaginary part (f”)
of the scattering of x-rays
as a function of the
photon energy.
From Ramakrishnan’s study of GH5
The tunability of the synchrotron source
allows us to choose precisely the energy
(wavelength) we need.
One can see how to choose wavelengths to get
large phase contrast for MAD phasing

* *
*
* *
* * * Maximum
Real Signal

Maximum
* Imaginary
Signal
*

Spectrum from Phizackerly, Hendrickson, et al. Study of Lamprey Haemoglobin.

This Multiwavelength Anomalous Diffraction
method often gives very strong phase
information and is the source of many new
structures.
 How do we find the heavy-
atom positions that allow us
to do MIR or MAD phasing?

There are generally two methods:

• Patterson-function methods
• Direct-phasing methods
 Lindo Patterson saw that to interpret a diffraction
pattern, he could correlate the electron density with
itself:

This is the cosine transform of intensity!

About the same time (all of this happened only a “short”
time ago, in the ’50s) David Harker saw a neat way to
approach “solving” the Patterson function:

This method is the basis of software

such as HEAVY (Terwilliger) and
CNS (Grosse-Kunstleve)
 Direct Phasing Methods
During the 1950s and ‘60s Sayre, Hauptmann, and the Karles
learned to determine crystal structures from the diffraction
intensities directly. They made use of two principles:
• The first was that the sum of phases of three Bragg planes that
form a closed triangle is invariant to the choice of the origin of the
crystallographic unit cell.

• Secondly, when the crystal is comprised of discreet atoms and all

three structure factors from these Bragg planes are large, this sum
of three phases is near to zero.
• Notice that the indices of the three sets of planes in the figure
sum to zero. This particular set of reflections is called a
"triplet," for obvious reasons.

h+k+l=0

• Here we have that the sum of phases h + k + l = const.

• It's not so hard to show that this is true: multiply three Fs and the
three phases end up in a sum.
 h + k + l 0

Finally, it's not so hard to see that if the three Fs are large, the sum
of phases should be near zero, as follows. If the only three atoms
in the unit cell were at the corners of the colored triangle in Figure
5, firstly, all three structure factors would be large since all atoms
lie only on the planes, and secondly, since the atoms are on the
planes, the phases would be zero. One can see that it makes sense
that this sum of phases might be constant. If one moves an atom
from one vertex of the colored triangle in the figure to the next
along the green line (the -2,1 plane), one can see that the other two
phases, for the red and blue sets of planes, would shift smoothly
by +2 and -2 respectively, keeping the total constant.
 General Scheme for Applying Direct
Methods
• Assign a few phases arbitrarily (this sets the origin).
• Find lots of triplets, where h+k+l = 0.
• Use the sum-of-phases  0 expression to propagate
phases
The modern schemes are much more sophisticated
(Weeks) and deal with the probabilities that the sum is
not precisely zero.
 The End
You’ve seen the fundamentals of
crystallography. That’s all there is to it.
I’ll take a few exposures to it for you really to
wrap your mind around all of this.
For example,
• I didn’t really understand the Fourier
transform until I did EM on 2D x-tals
• I didn’t figure out how to explain direct
methods until the 21st Century.
But I can tell you this, if you really
want to learn it…

Teach It!