Introductory Transmission Electron Microscopy Primer

Bob Hafner October 2011

This primer is intended as background for the Introductory
Transmission Electron Microscopy training offered by the
University of Minnesotas Characterization Facility (CharFac).
The primer addresses concepts fundamental to any
transmission electron microscope (TEM). Learning this
material prior to the hands-on training will improve the
effectiveness and efficiency of your TEM imaging.

The Big Picture

To the right is an image of Charfacs FEI Spirit transmission
electron microscope. The microscope column, specimen
chamber, and viewing screen are flanked by the monitor and
panel controls on the table top. As an operator you will need
to understand what is happening inside this black box (microscope column) when a panel or column
control is manipulated to produce a change in the monitor image.
A look inside the black box [Figure adapted from 1] reveals quite a bit of complexity. However, we can
begin to simplify this generic representation by categorizing the components. We have a:
source (electron gun) of the electron
beam which is accelerated down the
column;
series of electromagnetic coils (shifts,
tilts, stigmators) which ensure that the
electron beam is symmetrical and centered
as it passes down the optical column of the
microscope;
series of electromagnetic lenses
(condenser, objective, intermediate,
projector) which act to illuminate the
specimen and focus/magnify the specimen
on the fluorescent screen/ camera;
series of apertures
(micron-scale holes
in metal film) which the beam passes
through and which effect properties of that
beam;
specimen holder
which positions the
sample in the path of the electron beam.
The area of beam/specimen interaction
generates several types of signals that can
be detected and processed to produce an
image, diffraction pattern or spectra. There
are controls for specimen position and
orientation (x, y, z or height, tilt, rotation);
1

fluorescent screen or CCD camera which converts the electron signal to a form we humans can
see;
all of the above maintained at high vacuum levels.
We can provide further structure to this complexity by
comparing the arrangement and function of these
components to something with which we have all had
some experience -- a transmitted light microscope
[Figure adapted from 2]. Both the electron and light
microscope have:
an Illumination system (area above the
specimen) which consists of: a source of radiation
(photons/ electrons) and a condenser lens which
focuses the illuminating beam on the specimen -allowing variations of intensity to be made;
a Specimen stage(black line) situated between
the illumination and imaging systems;
an Imaging system (area below the specimen)
which consists of: an objective lens which focuses
the beam after it passes through the specimen and
forms an intermediate image of the specimen; and,
subsequent lenses (eye piece/intermediate and
projector lenses) which magnify portions of the
intermediate image to form the final image.
an Image recording system (eye, fluorescent
screen or CCD camera) which converts the radiation.

In addition, the traditional ray diagrams of light optics can be

used to display how we magnify/ demagnify and focus the
electron beam and how electrons deriving from a point in an
object plane can come to a well-defined point in the image
plane [Figure adapted from 1].
One additional point needs to be made at this time:
You need to take the concept of a column seriously and
start putting together a picture in your mind of where
components are in relation to one another starting at the
top (gun) and working your way to the bottom (fluorescent
screen). This is essential to understanding alignment of the
microscope as well as everyday operation.

Having mentioned some of the similarities between optical light microscopy and TEM, the Table below
highlights the important differences.

Optical Microscope

Electron Microscope

Information carrier
Wavelength
Medium
Lenses
Aperture angles
Image visibility

Light rays
400-800 nm (visible)
Air
Glass
< 60 degrees
Direct

Contrast producing
mechanism

Light absorption, reflection, spatial

variation in refractive index, scattering,
diffraction, birefringence, fluorescence
200 nm (visible)
52,000 X
Mechanically
< 10 um / < 0.1 mm

Electrons (have mass, charge)

0.0037 nm (100 kV)
Vacuum
Electromagnetic
< 1 degree
Fluorescent screen, CCD
camera
Mass-thickness; diffraction;
phase.

Resolving power
Magnification
Focusing/Alignment
Depth of field/focus

< 1 nm (point to point)

351,000,000 X
Electrically
> 100 um / > 1 meter

Electrons exhibit both particle and wave characteristics (= h/mv). Just as visible light can be treated as
a beam of photons or an electromagnetic waveso too can electrons be characterized as a succession of
particles or treated by wave theory. However, electrons are charged and the Coulomb forces are strong.
The negative charge of the electrons allows them to be:
deflected by means of magnetic or electrostatic fields. This is the basis for how lenses and coils in
the microscope work;
accelerated down the column by experiencing a voltage differential.
The small mass of the electrons necessitate:
high vacuum conditions to negate their deflection by air particles;
ultra thin (<100 nm) specimens to ensure transmission of the beam.
The shorter wavelength of the
electron results in a better
theoretical resolution
(diffraction limit of
resolution).
When light rays emanate from a
point and pass through a lens of
semi-angular aperture , they
form an image which is no
longer a point but with the
intensity spread out in what is
known as an Airy disk.

The distance between the two minima on either side of the main intensity
peak of the Airy disk is:
D = 1.22 / n sin where:
is the wavelength;
n is the refractive index of the material in which the object lies;
is the semi-angular aperture.
[Figures adapted from 3]
When two emitting points of the of the object lie very close together, the
intensity patterns in the image will overlap. The theoretical resolution
of the system is defined as the distance between the maxima when the
maximum intensity from one point is coincident with the first
minimum from the other point. This distance is the diffraction limit of
resolution:
d = 0.61 / n sin
From this equation one can see that increasing the aperture angle and
decreasing the wavelength will enable better resolution. Bottom left
Figure: Small aperture /large wavelength; Bottom right Figure: Large aperture / small wavelength.

In the light microscope the diffraction limit of resolution is approximately 200 nm. The effective (nm)
of an electron beam can be approximated as 1.22 / E1/2, where E is the accelerating voltage (eV). The
wavelength for an accelerating voltage of 100 kV is 0.0039 nm; 0.0022 for 300 kV. The wavelength of
electrons is shorter by orders of magnitude compared to light microscopy, and thus the diffraction limit
of resolution is much improved. Our ability to increase the aperture angle in TEM ends up being
constrained by aberrations in the electromagnetic lenses (more on
this later). Thus the practical, or aberration limited resolution, is
less than the theoretical resolution.
For a given microscope, the attainable resolution depends upon the
accelerating voltage (higher accelerating voltages produce better
resolution) and the design of the objective (focusing) lens. The
objective lens has a spherical aberration coefficient. The smaller
the lens gap the smaller the spherical aberration coefficient and the
better the resolution. However, a smaller lens gap will decrease
contrast (more on all of this later).
Finally, the low mass and negative charge of the electrons means
that they can easily be scattered (deflected) by passing close to
electrons or the nucleus of the specimen atoms.

This scattering can be [4]:

forward or backward;
coherent or incoherent; and
elastic or inelastic (more on this later).
The scattering results in a non-uniform
distribution of electrons emerging from the
specimen. This non-uniform distribution is the
basis of the major contrast producing
mechanisms in the TEM:

Mass/thickness

Diffraction

Phase

As TEM operators we can selectively

choose which aspects of the scatter will
transmit and thus contribute to an image
[Figure adapted from 4]. One common
way to do this is to determine the
collection semi-angle () through the
use of the objective aperture below the
specimen.
The collection semi-angle is a subset of
the general scattering semi-angle ()
the latter being determined by the nature
of the illuminating electron beam and the
properties of the specimen.
The illuminating beam has several
interrelated properties that we can control:
accelerating voltage (kV); convergence
semi-angle (); diameter and current of the
beam on the specimen.
5

In what follows we will work our way down the column -- building a more detailed understanding as we
go. We will first address bright field and dark field imaging that utilizes amplitude (scattering)
contrast. Amplitude contrast includes both mass/thickness and diffraction contrast. Phase
(interference) contrast will then be explained and this will allow us to introduce the concept of the
contrast transfer function. Those individuals who wish to have a better conceptual understanding of
the alignment of the microscope should consult Appendix A. Appendix B utilizes the contrast transfer
function to characterize how the Moos Tower FEI Tecnai Spirit and the Nils Hasselmo Hall FEI Tecnai
F30 Field Emission Gun microscopes differ from one another in terms of imaging capabilities.

Illumination System
Electron Guns
The purpose of the electron gun is to
provide a stable beam of electrons of
adjustable energy. There are three main
types of electron guns:
Tungsten hairpin
Lanthanum hexaboride (LaB6)
Field Emission
There are three properties of the Tungsten
hairpin and LaB6 filament guns that you
have some control over.
You can:
heat the filament to facilitate the emission of electrons from it (emission current). The goal is to
maximize emission current and minimize filament temperature. The point at which this occurs is
termed the saturation point.
apply a potential difference between the filament (cathode) and the anode of the gun to accelerate the
emitted electrons down the column. The accelerating voltage we choose will have an effect on the
penetrating power of the electron beam as well as its resolving capabilities.
apply a small negative bias between
the filament and the Wehnelt cap
which houses the filament [Figure
adapted from 4]. With no bias (i)
there is high emission current, but
that current has a high divergence
angle or spread. A very high bias
(iii) will actually retard the emission
of electrons. The optimum bias
level (ii) will take the emitted
electrons and produce a crossover
point (gun source) of small diameter
-- a saturated electron cloud.

Producing a small gun source or crossover has several benefits:

An electron beam emanating from a small source size is said to have high spatial coherency. That
is, the waves emanating from the source point are in phase with one another. High spatial coherence
of the illuminating beam will enhance the contrast producing mechanisms in the TEM (we will
discuss other ways of enhancing beam coherence later on).
Also, by using the bias to form a small source
size we are putting the emission current into a
smaller area and thus maximizing the brightness
of the electron beam.
Brightness is the beam current per unit area per
solid angle [Figure adapted from 4]

= 4i / ( d )2 where:
i = beam current;
d = diameter of the source (crossover); and,
= semi-angle of beam divergence.
With greater brightness we can have a smaller diameter beam on the specimen and still have adequate
current for signal production. The result is higher magnification imaging capabilities. Brightness,
unlike current, is conserved down the column. Brightness increases linearly with accelerating voltage.
Electron beams can also be characterized in terms of temporal coherency. A beam with high temporal
coherency will have electrons of the same wavelength. In reality there is a certain Energy Spread
associated with the beam. The energy spread will decrease with: high tension stability; a lower surface
temperature of the emitter; and, a somewhat lower emission current (Boersch effect).
The table below highlights the important differences between the various gun types. There is a
progression of capabilities from Tungsten to LaB6 to FEG in terms of brightness, source size, and
temporal coherency.

Tungsten:
Brightness (A/cm2str)
Lifetime (hrs)
Source Size
Energy Spread (eV)
Current Stability (%hr)
Vacuum (Torr)

JEOL 1200 & 1210

105
40-100
30-100 um
1-3
1
10-5

LaB6:

Tecnai Spirit & T12

106
200-1000
5-50 um
1-2
1
10-7

Shottky FEG:
Tecnai G2 F30
108
>1000
<5 nm
1
5
10-11

Along with these capabilities comes the need for better vacuum systems. The advantages of a more
coherent beam source will be negated if the beam is interacting with molecules on its path down the
column. The table below is provided simply to give you a feeling for what these vacuum levels translate
to inside the microscope. For reference: 1 Torr = 133 Pa = 1.33 mbar. Mean Free Path is the average
distance covered by a moving particle between successive collisions. Time to Monolayer is the length
of time required, on average, for a surface to be covered by an adsorbate.
7

Vacuum
1 Atm (760 Torr)
10-2 Torr
10-7 Torr
10-10 Torr

Atoms/cm3
1019
1014
109
106

Distance between atoms

5 x 10-9 meters
2 x 10-7 meters
1 x 10-5 meters
1 x 10-4 meters

Mean Free Path

10-7 meters
10-2 meters
103 meters
106 meters

Time to monolayer
10-9 seconds
10-4 seconds
10 seconds
104 seconds

We wont spend time here talking about the various kinds of pumps and gauges associated with the
vacuum system since those are maintained by the staff. However, as an operator, you should be very
aware of the vacuum state of the microscope and ensure that:
the microscope is at good vacuum level when you begin your session;
if there is a Gun Valve separating the upper column from the rest of the microscope that it is
closed during sample exchange;
appropriate vacuum levels are achieved prior to engaging and stepping up the high tension;
your hands do not touch any part of the specimen holder that is inserted into the column (under high
vacuum); and
your samples are dry and free of contamination and outgassing.

Electromagnetic Lenses

(Coils [shifts, tilts, stigmators] will be addressed in the alignment section of the primer)
It is useful to reason by analogy with glass lenses used for focusing
light to help one understand the operation of electromagnetic
lenses. Analogies also have regions of non-correspondence which
are equally important to understand. The diagram to the right
shows a perfect optical lens. The rays emanating from a point in
the object plane (a distance p from the lens) come to one
common well defined point in the image plane ( a distance q
from the lens). Parallel rays entering the lens are focused in the
back focal plane (a distance f from the lens). The optical lens
has a fixed focal point and the object is in focus at the image plane.
The magnification of the lens is q/p (demagnification =
1/magnification)

In light optics we exchange lenses to change magnification and adjust

the height of a given lens along the optic axis to focus. Electromagnetic
lenses in contrast are stationary -- but we can vary their strength by
altering the amount of current running through them. Thus
electromagnetic lenses have variable focal points. The diagram to the
left shows a stationary lens which is stronger (larger). In this instance
we have actually demagnified the image.
An electromagnetic lens [Figure from 5] consists of a coil of copper
wires inside an iron pole piece. A current through the coils creates a
magnetic field (symbolized by red lines) in the bore of the pole pieces
which is used to converge the electron beam.
8

When an electron passes through an electromagnetic lens it is subjected

to two vector forces at any particular moment: a force (HZ) parallel to
the core (Z axis) of the lens; and a force (HR) parallel to the radius of the
lens. These two forces are responsible for two different actions on the
electrons, spiraling and focusing, as they pass through the lens. An
electron passing through the lens parallel to the Z axis will experience
the force (HZ) causing it to spiral through the lens. This spiraling causes
the electron to experience (HR) which causes the beam to be compressed
toward the Z axis. The magnetic field is inhomogeneous in such a way
that it is weak in the center of the gap and becomes stronger close to the
bore. Electrons close to the center are less strongly deflected than those
passing the lens far from the axis.
So far weve mentioned that electromagnetic lenses are unlike optical
lenses in that they are: stationary; have variable focal points; and cause
the image to be rotated. The latter is corrected for in modern TEMs.
Electromagnetic lenses also differ in that: the deflection of the electron
within the lens is a continuous process (no abrupt changes in the
refractive index); only beam convergence (not divergence) is possible; and the convergence angle with
respect to the optic axis is very small compared with optical light microscopy (less than one degree!).
Finally, it is important to keep in mind that electron lenses, compared to glass lenses, perform much
more poorly. Some have compared the quality of electron optics to that of imaging and focusing with a
coke bottle. This is mainly due to the fact that aberrations are relatively easily corrected in glass lenses.

Lens Aberrations
Up to this point, all of our representations depict a perfect lens. That is, all rays emanating from a point
in the object plane come to the same focal point in the image plane. In reality, all lenses have defects.
The defects of most importance to us here are spherical aberration; chromatic aberration and
astigmatism. Rather than a clearly defined focal point, we end up with a disk of minimum confusion
in each instance.
Spherical aberration: The further
off the optical axis (the closer to the
electromagnetic pole piece) the
electron is, the stronger the magnetic
force and thus the more strongly it is
bent back toward the axis. The result
is a series of focal points and the point
source is imaged as a disk of finite
size (dS).

To reduce the effects of spherical aberration, apertures are introduced into the beam path. Apertures are
circular holes in metal disks on the micron scale. The net effect of the aperture is to reduce the diameter
of the disk of minimum confusion. However, that positive effect comes at the price of reduced beam
current. Also, a very small aperture will display diffraction effects. The diameter of the aperture used
will also affect the convergence angle of the beam and this in turn will affect its coherence as well as
image properties such as depth of focus.
Chromatic aberration: The electron beam generated by the gun
will have a certain energy spread. Electrons of different energies at
the same location in the column will experience different forces.
An electromagnetic lens will bend electrons of lower energy
more strongly than those of higher energy. As with spherical
aberration, a disk of minimum confusion (dC) is produced.

Astigmatism: Finally, the

electromagnetic lenses used in
the TEM can not be machined to
perfect symmetry. If the fields
produced by the lenses were
perfectly symmetrical, a
converged beam would appear
circular (looking down the
column). A lack of symmetry
would result in an oblong beam:
the narrower diameter due to the
stronger focusing plane; the
wider diameter due to the weaker
focusing plane. The net effect is
the same as that of the
aberrations abovea disk of
minimum confusion rather than a
well defined point of focus.

10

Illumination system lenses and apertures

Although an understanding of beam -- specimen interactions is necessary to inform user control of the
illumination system, we can begin to put that picture together. The illumination system can be set up in
two basic modes: Convergent beam and parallel beam. The latter will be our focus in this primer.
Convergent beam
In convergent beam mode we focus a reduced image of the
gun crossover on the specimen to make a small bright probe
which would be useful in analytical STEM work or
convergent beam diffraction.
The diagram to the right shows the demagnification of the
gun crossover to produce a small probe on the specimen. The
actual demagnification is as follows:
the size of the gun source at d1 = d gun (p1/q1);
the size of the probe on the specimen is d2 = d1 (p2/q2).
If we wanted a small diameter beam probe we would excite
condenser lens 1 (C1) causing q1 to decrease. Then we
would focus condenser lens 2 (C2) to form the smallest spot
on the specimen. Smaller C2 apertures would eliminate
aberrations also enhancing a smaller spot size. The size of
the C1 aperture is usually fixed. If you have used an SEM,
you should be familiar with this arrangement. The cost of a
smaller diameter beam; however, is the loss of beam current.
(black lines in the diagram). So the operator needs to find a
compromise of these factors. Although the principles just elaborated certainly apply, the lens
configurations in the FEI TEMs that you will be using are more complex. Option A in the Figure below
is a more realistic depiction of those microscopes in convergent beam mode.
Parallel beam
In parallel beam mode we provide a
nearly parallel illumination at low
intensity with high coherency that would
be useful in bright field and dark field
imaging, phase contrast imaging, and
selected area diffraction. To achieve this
we spread the beam on one side of focus
(crossover) using the C2 lens. Option B
in the Figure to the right shows C2
spread in the underfocus direction;
Option C the overfocus direction.
[Figure adapted from 4]
The beam is more parallel, and thus
more coherent, as it approaches the
sample when C2 is spread in the
overfocus direction. On the FEI TEMs
11

this corresponds to turning the Intensity knob clockwise from crossover. If we spread the beam too
far overfocus we will loose intensity on the screen. In practice we spread the beam to: just cover the
fluorescent screen on the microscope; obtain a proper image histogram from a CCD camera; or obtain
the sharpest diffraction pattern.
A smaller C2 aperture and/or a more strongly
excited C1 lens (small spot size) will also
increase beam coherence. The down side is
less beam current illuminating the specimen.
Again, the operator needs to find a
compromise between these factors. Use the
smallest C2 aperture/spot size for which you
have adequate illumination at the
magnifications being used. One needs both
adequate signal and good contrast producing
mechanisms. It should be noted that the angles
are greatly exaggerated here for pedagogical
purposes.
Spherical aberration of the illumination lenses, while important in the convergent beam mode, plays
little role in parallel illumination mode.
So, weve seen how beam current, probe size, and convergence angle interrelate. What accelerating
voltage (kV) should you use? Your choice will, again, depend on knowledge of beam-specimen
interactions; however, here are some points to consider concerning higher accelerating voltages:
brightness increases;
wavelength decreases and thus resolution capabilities increase;
some contrast mechanisms will decrease (more on this later);
beam penetration will increase;
the heating effect of the beam is smaller (especially with low Z materials).

Beam-Specimen Interactions
The scatter which results from the electron
beam interacting with the specimen can be
[Figure adapted from 4]:
forward or backward. We will focus on
forward scatter;
coherent or incoherent. Beam electrons
undergoing coherent interactions
maintain phase relations and those
undergoing incoherent interactions do
not. We will utilize both coherent and
incoherent scattering;
elastic or inelastic. Beam electrons
undergoing inelastic interactions with
the specimen atoms show measurable
energy loss and those undergoing elastic
12

interactions do not. We will focus on elastic scatter. Elastic scatter is the dominant component of
the total scatter occurring in the TEM. It is also the principle source of contrast in TEM images and
intensity distributions in diffraction patterns.
Inelastic interactions are the basis for a wide range of signals (X-rays, EELS, secondary electrons,
cathadoluminescence) as well as beam damage (bond breaking, knockout damage). They are
incoherent and have a low scattering angle () usually less than one degree. The low scattering
angle means that they represent only low resolution/structural information. Inelastic scatter which
passes through the objective aperture creates a background fog and does not contribute optimally to
the in-focus image. This is due to the energy losses and thus increased chromatic aberration in the
objective lens.
Elastic scattering can be understood from a particle view (scattering from isolated atoms) or a wave
view (scattering from the specimen as a whole) [4].
From the particle view: beam electrons can interact with the electrons or the nucleus of a specimen
atom through Coulomb forces.
Electron-electron interactions result in relatively low and are primarily a function of accelerating
voltage more interactions at lower accelerating voltages.
Electron-nucleus interactions result in higher which is incoherent (termed Rutherford scattering).
Above approximately 5 degree , all of the elastic scattering from a particle view can be considered
Rutherford incoherent. The probability of this type of interaction increases with:
higher atomic number elements;
lower accelerating voltages;
greater specimen thickness; and
lower .
From the wave view: the electron wave can
interact with many atoms together within a
crystalline specimen resulting in collective
coherent scattering (diffraction).
Each atom in the specimen (E and E) acts
as a source of secondary spherical
wavefronts. These new wavefronts are
coherent and will therefore interfere -reinforcing one another in certain angular
directions and canceling in others. The
points on the wavefront marked A, A are in
phase and will reinforce. This will give rise
to the powerful central part of a diffraction
pattern. AB and AB reinforce (due to
interference between wavefronts separated
by one wavelength) to give the two separate
patterns on either side of the central band
called first order spectra. Second order
spectra are due to interference AC and
AC.. and so on.
13

Diffraction is controlled by the angle of incidence of the

electron beam to the atomic planes in the specimen and
the spacing of those planes. Diffraction occurs when
Braggs law is met.

Braggs Law: n = 2d sin where:

is the wavelength of the incident beam;
d is the spacing between atomic planes; and
is the angle of the incident beam to the atomic plane.
In the diagram to the right, the extra distance the bottom
beam travels is 678. If this extra path length is some
multiple of the wavelength (n), the two beams at 3 and
10 will constructively interfere to produce a diffraction pattern. The distance 67 = d sin; the
distance 678 = 2d sin. Thus, diffraction will occur when: n = 2d sin. [Fig courtesy of 6]

Imaging System
The forward elastic scattering taking
place within the specimen results in
an emerging non-uniform distribution
of electrons which is the basis of the
major contrast producing
mechanisms.
[Figure adapted from 4]
In imaging mode, the scatter is
focused by the objective lens to the
first intermediate image plane which
subsequently acts as the object plane
for the magnifying lenses.
Magnification can be calculated in
the same manner as we did for the
illumination system in convergent
beam mode.
Rays emanating from a given point
in the specimen plane come to a
focused point in the image plane. The
adjacent Figure shows rays coming
from two different points of the
specimen as dark and light blue.

14

An objective aperture is situated within the beam path just below the objective lens. The objective
aperture is important for several reasons. The aperture will:
allow for signal selection (this will depend upon what we see in diffraction modesee below);
provide for contrast within the image ( a lower accelerating voltage will also increase contrast);
decrease objective lens aberrations, spherical and chromatic, which will degrade image resolution.
The objective aperture is optimally chosen as a compromise to limit both spherical aberration and
diffraction error defects. However, the energy range of electrons leaving the sample can be
significant (15-25 eV). Thus, a smaller objective aperture may be necessary to alleviate chromatic
aberration effects. Ensuring a thin specimen will also help;
affect depth of field in the image a smaller aperture giving better depth of field.
In diffraction mode, rays emanating from the specimen that are parallel to one another come to focus in
the back focal plane of the objective lens. The previous Figure shows parallel rays with different colors:
dark blue, light blue, and red. The diffraction pattern which occurs in the back focal plane of the
objective lens arises simply as a consequence of image formation in the image plane of that lens. When
we enter diffraction mode on the TEM, we are adjusting the strength of the intermediate lens so that this
objective back focal plane becomes its object plane. We remove the objective aperture and insert
another aperture further down the columna selected area diffraction (SAD) aperture, to select a
portion of the sample from which the diffraction pattern arises. That SAD aperture acts as a virtual
aperture.

Contrast
A high resolution signal is worthless if adequate contrast is absent. Contrast can be defined as (S2 S1) /
S2 where S2 is the signal from the feature of interest; S1 is the background signal; and S2 > S1. For most
users the minimum useful image contrast level is about 5 % (contrast lower than this can be enhanced
via image processing).
The figure to the right shows a simplified screen image
under different brightness and contrast conditions. Below
each screen image is an intensity profile across that image.
The X axis corresponds directly to distance; the Y axis
represents the dynamic range of the signal. The base level
of the signal is the brightness; the spread of the signal up
from this baseline is the contrast. The contrast should span
as much as possible of the dynamic range because this
produces the most useful and pleasing images.
In what follows we will take a closer look at the amplitude
contrast mechanisms (mass/thickness and diffraction) and
how they contribute to image formation. Differences in
intensity due to amplitude (scattering) contrast result
from the scattered electrons being intercepted by the
objective lens/aperture and thus not contributing to image
formation. Next we will discuss phase contrast.
Differences in intensity due to phase (interference)
contrast arise from interference effects between scattered and unscattered electrons which pass through
the objective lens/aperture. A slight amount of underfocus will enhance the phase contrast in our
images.
15

Bright and dark field images showing mass/thickness contrast

Consider an amorphous biological or polymer sample. We can neglect collective coherent elastic
scattering due to the lack of crystallinity. The principle scattering mechanism in our sample will be
Rutherford incoherent elastic scattering. That scattering is enhanced by atomic number, density and
specimen thickness and the contrast produced is thus termed mass thickness contrast. Given their
composition, biological and polymer samples are often stained with heavy metals to enhance this type of
contrast.
Those areas of the sample that are thicker or of
higher atomic number will proportionally scatter
more. If we introduce an objective aperture directly
below the specimen then that scatter will not be
allowed to travel down the column and those areas of
scatter will appear darker in the image. A smaller
objective aperture will intercept more of this scatter
and thus will enhance massthickness contrast.
The images to the left are of a filamentous carbon
membrane with 15 nm gold particles attached. The top
image is referred to as bright field; the bottom
dark field. When we go to diffraction mode with
the beam on the specimen we get a diffuse diffraction
pattern with an intense central spot (upper right image
insets). If we position our objective aperture (dark
circle) around this high intensity, low scatterthen a
bright field image is produced when we subsequently
go back to imaging mode. Only the scatter within this
circle will contribute to image formation. Thus the
gold particles appear darker as compared to the carbon
membrane in the bright-field image due to the fact that
the particles have higher density (atomic number) and
the beam travels through 15 additional nanometers of
this material. Both of these properties result in
proportionally more higher-angle scatter which does
not make its way through the objective aperture. A
third property of the gold, its polycrystalline nature,
also contributes to the higher angle scatter (next
section).
If we somehow (next section) allow only the high
scatter to contribute to the image-- then a dark field
image is produced. Just remember that those areas in
an image that are darker correspond to scatter which
was intercepted by the objective aperture and not
allowed down the column.

16

Bright and dark field images showing diffraction contrast

Consider a specimen which has a crystalline nature.
The incident electron beam interacting with such a
specimen will result in coherent elastic as well as
incoherent elastic scattering. The former type of
scatter will occur at special (Bragg) angles
controlled by the crystal structure and the
orientation of the specimen with respect to the
incident beam. At these angles the specimen,
whether mono (left) or polycrystalline (right), will
produce a diffraction pattern.
To obtain a bright field image of a crystalline specimen we will:
ensure the beam is on the specimen;
enter diffraction mode;
position the objective aperture around the transmitted beam
(shown blue in the ray diagram) in the diffraction pattern;
enter imaging mode.
[Figures adapted from 4].

A dark field image results from allowing

one of the higher order diffraction spots
(shown red in the ray diagrams) to constitute
the signal for the image. There are two
ways of doing this. We can:
move the objective aperture off-axis so
that it selects one of the higher order
diffracted beams as the signal. This
situation, which we term displaced
aperture dark field imaging,
enhances off-axis lens aberrations and
thus is not desirable;
keep the objective aperture stationary on
axis and tilt the beam to achieve the
same result. This is termed centered
dark field imaging and is the choice of
microscopists.

17

The coherent elastic scattering from this type of sample gives rise to diffraction contrast. Diffraction
contrast will co-occur with massthickness contrast, but we can select/accentuate the contrast arising
from particular specimen planes by selecting the corresponding diffraction spot as the source of our
image.

Phase Contrast and the Contrast Transfer Function

The phases of elastically scattered electrons are shifted as they pass through the specimen. The amount
of shift is proportional to the electron density of the specimen atoms. This is analogous to how the
refractive index affects a light wave. Phase contrast in images arises from differences in phase
between scattered and unscattered rays in different parts of the image and interference between these
rays.
Using a monocrystalline sample as an
example, a bright or dark field image
requires that we select a single beam
using the objective aperture. A phasecontrast image requires the selection of
more than one beam. Generally the
more beams collected, the higher the
resolution of the image. [Figure from
7]
In general, both amplitude and phase contrast will coexist when imaging most samples. However,
amplitude contrast is dominant for structures with large mass thickness, while phase contrast increases
in importance for small and thin structures and becomes the dominant source of contrast for very small
objects of low atomic number.
Phase contrast results in the formation of strong Fresnel fringes about any part of the specimen where
there is a rapid change in mass thickness. The fringes serve to enhance these specimen areas by
delineating them with a bright line when the objective lens is in the under-focused position and a dark
line in the over-focused position. The Fresnel fringes virtually disappear at the point of "exact" or
minimum contrast focus. The relative amount of amplitude contrast can be determined by focusing
the specimen to minimum contrast focus where phase contrast is at a minimum. The image below is
of a hole in a thin amorphous carbon film at under-focus, minimum contrast true-focus, and over-focus
respectively.

18

In order to image a phase object, we need to shift the phases of the scattered electrons additionally so
that they will be 180 degrees out of phase and can thus interfere destructively with unscattered beam
electrons. This additional phase shift of the scattered electrons is caused by the objective lens spherical
aberration coefficient (fixed microscope value) and the amount of defocus of the objective lens (variable
value under control of the operator)
This phase shift as a function of spatial frequency is: X (k) = Cs 3 k4 + f k2 where:
k = spatial frequency;
Cs = spherical aberration coefficient of the objective lens;
= electron wavelength defined by accelerating voltage;
f = defocus value of the objective lens.
sin X(k) is termed the Contrast Transfer Function (CTF). The CTF is displayed simply and
directly in the Fourier Transform (FT) of the image in the back focal (diffraction) plane of the
objective lens. The FT corresponds to a frequency analysis of the image. The image is considered to be
the sum of a set of sinusoidal waves of varying frequencies: the higher the frequency of the wave, the
finer the detail it refers to. In FT reciprocal space, high spatial frequency (resolution) corresponds to
long distances from the optic axis.
The image to the right is the FT of an amorphous thin carbon film taken at
under-focus on our Tecnai T12. The FT is the CTF with envelopes
(more on envelopes later). Actually, since only intensity is displayed, the
CTF is sin2 X(k).
The figure below shows a simulated CTF from the program CTF Explorer
[8]. This simulated CTF is of the Tecnai T12 under the conditions which
generated the FT to the right. Also shown below (upper right inset) is a
simulated FT corresponding to this CTF.

The Y-axis displays contrast and the X-axis

spatial frequency. The CTF is oscillatory. When
it is equal to zero, there is no contrast
(information transfer) for that spatial frequency.
Those spatial frequencies without contrast are
characterized as Thon rings on the FFT.
Likewise, there are passbands where there is
good transmittance of spatial frequencies. The
oscillation (and thus contrast) eventually dies off
due to envelope functions associated with
temporal and spatial coherency.
Negative contrast values correspond to the type
of contrast we are used to seeing with scattering
or amplitude contrast, ie a subtraction from the
background electron intensity over regions of
high mass density; positive values reverse this
intuitive relation.
19

If we map intensity, the CTF looks as follows.

The relation between the FT and this
characterization of the CTF is more direct and
thus understandable:
Bright concentric areas of the FT map to
CTF spatial frequency ranges displaying
high intensity contrast;
FT Thon rings correspond to CTF spatial
frequencies of zero transmittance;
Higher spatial frequencies occur farther to
the right on the CTF X-axis and farther
from the FT center.

The best CTF is the one with the fewest zeros

and with the broadest passband of good
transmittance (where CTF is close to -1).
Scherzer suggested an optimum defocus
condition occurring at:
f = - (Cs )
The resolution at this defocus, now called "1
Scherzer", is:
r = 0.66 Cs1/4 3/4

"Extended Scherzer" (1.2 Scherzer) provides

even broader band of transmittance with CTF
still close enough to -1. The crossing of the kaxis at "extended" Scherzer defocus corresponds
to point-to-point resolution of the microscope.
The point where envelope functions damp the
CTF to zero corresponds to the information
limit of the microscope.
The envelope function acts as a virtual aperture.
The envelope shown in our figures combines a
temporal coherency envelope and a spatial
coherency envelope. The former is caused by
chromatic aberrations, focal and energy spread,
instabilities in the high tension and objective lens
current; the latter by the finite incident beam
20

convergence
Phase contrast images are directly interpretable only up to the point resolution. If the information
limit is beyond the point resolution limit, one needs to use image simulation software to interpret any
detail beyond point resolution limit. On LaB6 filament instruments such as the Tecnai Spirit there is
relatively little information limit beyond the point resolution.

Operator control
Specimen thickness: Thinner specimens are better! Increased specimen thickness results in the
elastically scattered electrons being inelastically scattered as well. As a result, image resolution is
reduced due to chromatic aberration of the objective lens and there is a loss of contrast. The scatter
angle of inelastically scattered electrons will also increase with specimen thickness draining out higher
resolution detail
Beam coherence: Contrast mechanisms are enhanced when we use a parallel illuminating beam which
is itself coherent. Weve seen that we can increase the spatial coherency of that illuminating beam by:
lowering the magnification when possible; using a smaller spot size; using a smaller C2 aperture; and by
spreading the beam in the under-focus direction. The down side of increasing beam coherency by these
means is the reduction in beam current on the specimen and thus signal/illumination. This can be
compensated to some extent with longer camera exposure times. So, you are balancing beam current
and the degree of coherency of the beam. FEG sources provide a very strong improvement in spatial
coherency extending the information limit to very high frequencies
Accelerating voltage: A higher accelerating voltage has many advantages including: greater
brightness, increased resolution; and better beam penetration. The down side is that proportionally the
amount of scatter decreases with increasing accelerating voltage. Samples with inherently little mass
thickness contrast (many biological and polymer samples) may require lower accelerating voltages to
increase the angle and amount of scatter and thus the necessary contrast. So, you are balancing
resolution and contrast when deciding upon the accelerating voltage to use. Use the highest accelerating
voltage for which you have adequate contrast.
Objective aperture: The size of the aperture you use will depend on a number of factors. If your
sample is adequately thin and has high inherent mass-thickness contrast, then an intermediate size
aperture that balances aberration and diffraction effects would be appropriate. If your sample is thicker
(prone to producing scatter of a high energy spread thus displaying chromatic aberration) or has less
mass-thickness contrast, then you may wish to use a smaller aperture. Larger aperture sizes (or no
aperture) may be required when doing high resolution phase contrast imaging. Higher spatial frequency
information is farther from the central diffraction spot axis and thus may be excluded from producing
phase (interference) contrast in the image. In other words, the objective aperture acts as an envelope
function in the CTF. If you are interested in high resolution imaging, you must also understand the
nature of the spatial and temporal envelope functions on your microscope as they may attenuate the
contribution of higher spatial frequency information. The relationship between objective aperture size
and spatial frequency (maximum obtainable resolution [r]) can be approximated by:
r = 2 (focal length / aperture diameter). The Table below shows corresponding values for the Tecnai
Spirit
Aperture Diameter (microns)
r (nanometers)

20
2.04

40
1.02

70
0.58

150
0.27
21

Focusing and spatial frequency: As mentioned

before, most images are a combination of phase
and amplitude contrast. The degree of amplitude
contrast can be assessed by focusing to minimum
contrast with the assistance of a Fast Fourier
Transform (FFT). Fresnel fringes may also be
used for this purpose but their accuracy is not as
good.
Minimum contrast is a reference point from which
to determine your optimal under-focus. As shown
in the image to the right, a slight amount of underfocus (here Scherzer) cancels the effect of
objective lens spherical aberration and will
increase the point resolution and enhance phase
contrast over a broad range of spatial frequencies.
You need to keep in mind that amplitude contrast
is not being represented in this CTF image. The
apparent lack of transmittance at lower spatial frequencies will be compensated by amplitude contrast at
those frequencies. So, for many samples, amplitude contrast is enhanced with phase contrast at slight
under-focus. The images below of gold nanoparticles on amorphous carbon where taken close to
minimum contrast (left) and slight under-focus (right).

The choice of under-focus depends on the resolution expected from a specimen and becomes more
critical as the resolution improves. The optimum defocusing can be specified as that which puts the first
zero of the transfer function at a value somewhat greater than the highest spatial frequency or resolution
of interest. This will produce maximum overall contrast with enhancement of higher resolution details.
Maximum phase contrast for a specimen spacing of r will occur at defocus: f = r2 / 2. The table
below shows point resolution and the spatial frequency of maximum phase contrast for various
underfocus values on the Moos Tower Tecnai Spirit TEM.
22

Underfocus (nanometers)
Spatial Frequency of Maximum
Phase Contrast (nanometers)
Point Resolution (nanometers)

-174
1.08

-250
1.3

-500
1.8

-1000
2.6

-2000
3.7

-3000
4.5

0.45

0.81

1.26

1.81

2.58

3.15

The Figures below show TEM images and corresponding CTFs and FFTs at -174, -250, -500 and -1000
underfocus on the Tecnai Spirit.

23

Appendix A: Alignment of the Microscope

As an operator you should expect the following from the microscope. The:
brightness and resolving capabilities of the beam should be maximized;
field of view should stay centered on the screen when changing magnification;
illumination should stay centered and even on the screen when you vary it;
image should not move on the screen when you try to focus it; and
illumination should not be lost when switching from one mode of operation to another.
These expectations are met only when the microscope is well aligned. Microscope alignment is
multilayered: some aspects are so fundamental and detailed that they are done only by service
engineers; others advanced users will need access to. We will focus here on basic alignment capabilities
that the beginning user should master.
Think in terms of an optic axis. That axis can be defined by (travels through) the field center of the
objective lens and the center of the viewing screen. We want the gun, all other lenses and apertures
centered with respect to that optic axis. We will be using deflection coils to ensure proper alignment
since the lenses are obviously stationary. We also want the beam to be symmetric as it passes down the
column. Thus there are additional coils associated with the various lenses that correct for lens
astigmatism.
If you give it a little thought, hopefully it is apparent why you would want to start alignment at the top of
the column and work your way down. Thats what we will do -- align the illumination system first and
then the imaging system. The former does not require the specimen to be in the beam path.

Gun alignment
Earlier we talked about maximizing the
brightness of the gun by adjusting the bias
at saturation. To take advantage of that
setup the electron beam should be adjusted
so that it travels directly down the optic
axis of the column.
A: the gun is shifted horizontally off-axis.
B: the gun is tilted off-axis.
C: the gun is misaligned with respect to
both shift and tilt.
D: proper gun alignment.

24

Now we dont move the gun physically. Instead

we use a pair of electromagnetic coils (gun shift
and tilt) just below the gun to achieve the same
result. If we want to horizontally shift the beam,
we apply current to the coils in such a manner that
the tilt angle created by the first coil is
compensated by the tilt angle produced by the
second coil. Tilting the beam requires an
additional angular tilt applied to the second coil.
Gun alignment involves bringing the beam to its
smallest spot (crossover), de-saturating the beam to
reveal some filament structure, ensuring symmetry
of illumination with gun tilt, and centering the
beam on the screen with gun shift. Remember that
there is no specimen in the beam path so that what
we are imaging here is the gun source.

Spot size alignment

We want the beam to stay centered
when we change current to the C1
lens in order to change spot size
(smaller numberbigger spot.). Spot
size alignment uses an additional set
of electromagnetic coils (beam shift
and tilt) located beneath the C2 lens
[Figure adapted from 5]. These coils
operate in the same manner as the gun
coils above. The gun shift and tilt
coils align the gun with C1. The beam
shift and tilt coils align the beam
exiting C2 with specimen plane and
the rest of the electron microscope.
When the C1 lens strength is weak
and thus the spot size is large (spot 1),
it magnifies any error in the gun
position. Because we know that the
phosphor screen is reasonable well
lined up with the C2 lens, then getting the illumination back to the center of the screen is accomplished
with the gun shifts.
When the spot size is small (spot 5), the cross-over below C1 lens is close to the center of that lens. The
C1 spot acts as the source for C2 lens and we align it with the screen center through the use of beam
shifts. We go back and forth between spot sizes, manipulating gun and beam shifts, until the beam
crossover remains centered on the viewing screen.
25

C2 astigmatism correction
If the C2 lens is astigmatic, the beam will elongate as
you move from crossover in either the underfocus or
overfocus direction. Again, there is no specimen in the
beam path so we are viewing the gun source on the
viewing screen. The adjacent image shows an
exaggerated (hopefully!) astigmatic beam condition on
the viewing screen.
If astigmatism exists then we apply current differentially
to a series of stigmator coils around the C2 lens to
compensate--thus creating a symmetrical beam

C2 aperture centering

When we insert the C2 aperture we want it to be

positioned so that the beam travels through its
center. Changing the focus of C2 will move the
focal point of the lens vertically up and down.
The beam will remain centered and expand
uniformly when the C2 aperture is centered
[Figures adapted from 5].

If the C2 aperture is slightly off then

the beam will touch one end of the
viewing screen before another as it is
spread. In the case of extreme
misalignment, the whole expanding
beam will move off the screen.
The beam is brought to crossover,
centered with the beam shifts, and
spread. If the beam does not spread
uniformly on the screen, the C2
aperture has X and Y adjustment
knobs which can be used to center it.
26

Eucentric (Z) height adjustment

The adjacent figure shows that the height of the
specimen within the column will affect the
magnification level of the objective lens.
It is common practice to adjust the specimen
height (Z-shift) so that the middle of the
specimen lies on the rotation axis of the
specimen loading arm. This has the convenience
that the point you are observing remains
stationary when the specimen is tilted. In this
condition the specimen is said to be at the
eucentric height.
Many aspects of the performance of the
microscope depend upon the eucentric height of
the specimen:
magnification and spot size are indicated
for this height;
measuring is calibrated for this height;
objective and selected area apertures are located optimally;
maximum tilt can be obtained.
There are a variety of ways of ensuring that the specimen is at the eucentric height and your training
sessions will provide the specifics. But you should remember to set the eucentric height for each of your
specimens/grids.

Pivot point of beam tilts and shifts

Each pair of coils in the microscope (gun, beam, imaging) needs to be aligned by setting pivot points for
shift and tilt. The gun and imaging pivot points are set by advanced users; however, you will need to
become adept at setting the beam pivot pointspecifically tilt. In Rotation Centering (below) we will
be aligning the whole illumination system with respect to the field center of the objective lens. We will
use the beam coils to do this and thus ensuring the correct pivot point of the beam tilts is critically
important.
A: Depicts beam tilt with
different colored rays.
Here the pivot point is set
at the specimen level. Rays
representing different tilts
converge at the same point
on the specimen. Rays
emanating from that point
on the specimen will focus
to one point in the image
plane of the objective lens.
27

B: What happens if we change the height of the specimen? The beam tilt is off due to the fact that the
different beam tilt settings put the beam at different positions on the specimen. This will correspond
directly to different positions in the image planean undesirable scenario.
We want to ensure that our beam tilt pivot point is on the specimen. There is a button on each of the
microscopes for pivot points which makes the beam jump (wobble) between two tilt settings. If the
pivot points are wrong, you see two beams (converged to crossover) separated laterally on the screen. It
is then a simple matter to adjust a couple of correction knobs until the two beams are coincident. The
correction knobs adjust the ratio of excitation of the two sets of deflection coils.

Rotation centering
To this point we have the
entire Illumination system
aligned we can think
about it as one collective
entity. We also have the
specimen at eucentric
height, and the beam tilt
pivit points set for that Zheight position.
Now we need to align the
transmitted signal from the
specimen with the field
center of the objective
lens. If the field is off
center then the electrons
will move off axis and the
image will rotate about a
position off axis as you
change focus with the
objective lens [Figure
adapted from 4]. A
common means of rotation
centering involves varying
(wobbling) the voltage to
the gun. The beam tilts are then adjusted to ensure that electrons remain on axis through the lens as their
energy varies.

28

Objective aperture centering

In order to center the objective
aperture we:
ensure that the beam is on the
specimen;
put the microscope in diffraction
mode, and focus and center the
central diffraction spot;
insert the objective aperture;
center the aperture around the
central diffraction spot with the
aperture X and Y knobs; and
return to imaging mode.

Objective lens astigmatism correction

Objective lens astigmatism correction is best done utilizing a CCD camera at high magnification (>>100
kX) on all three microscopes. As with the C2 lens, there are a series of stigmator coils around the
objective which we can provide current to differentiallythus compensating/correcting for the
astigmatism present. There are a number of methods for correction:
We can use a pure carbon film specimen which contains holes of widely varying sizes. The holes have
clean edges and show clearly defined Fresnel fringes when the objective lens is slightly under-focused (a
white fringe) or over-focused (a black fringe).

Over-focus: Dark Fresnel fringe uniform.

No astigmatism

True focus: Neither dark nor light fringes.

Lack of contrast
29

By observing the symmetry of the fringes while the objective lens is adjusted from under-focus to overfocus, the degree of astigmatism present can be determined. The astigmatism can then be corrected by
adjusting the direction and strength of the objective lens stigmator. With a correctly adjusted lens, the
black over-focused fringe will appear in its entirety around the inside edge of a hole as the objective lens
is adjusted from under-focus to over-focus. The fringe will be of a constant width. With an uncorrected
objective lens, the black fringe appears at two opposite parts of the circumference of a hole first, then as
the objective lens is further over-focused, the remaining parts of the hole show a black fringe. However
the fringe is not of constant width.

Under-focus: Light Fresnel fringe uniform.

No astigmatism

Astigmatism: Dark Fresnel fringe not uniform.

Alternatively, you can compensate for astigmatism by selecting an area of the holey carbon film that
shows uniform density (no holes) and ensuring that a live Fast Fourier Transform is spherical at a high
magnification

Under-focus; Astigmatism

Under-focus: No astigmatism

Focus: No astigmatism

You can also focus on specimen fine structure at very high magnification. Astigmatism will show itself
as image stretching on both sides of focusmuch like it does in the SEM.
30

Appendix B: Using the CTF to compare the Moos Tower Tecnai Spirit
and the Nils Hasselmo Hall Tecnai F30 Field Emission Gun TEMs.

The above CTFs were simulated at 148K magnification and extended Scherzer underfocus. The CTFs
graphically show the differences in point resolution and information limit.
Figure A: the Moos Tower Tecnai Spirit.
Figure B: the Moos Tower Tecnai Spirit if it had the focusing lens pole piece of the NHH FEGTEM.
Figure C: the Moos Tower Tecnai Spirit if it had the focusing lens pole piece of the NHH FEGTEM as
well as 300 kV accelerating voltage.
Figure D: the NHH FEGTEM: higher resolution pole piece of the focusing lens; 300 kV; and better
spatial and temporal coherency associated with the field emission tip.

References

[1] J. J. Bozolla and L. D. Russell, Electron Microscopy: Principles and Techniques for Biologists
[2] JEOL LTD., Invitation to the SEM World.
[3] http://www.matter.org.uk/tem/
[4] David B. Williams and C Barry Carter, Transmission Electron Microscopy: A Textbook for
Materials Science.
[5] http://www.rodenburg.org/guide/index.html
[6] Greg Slowinski, Warsaw University of Technology, Poland
[7] Leonid A. Bendersky and Frank W. Gayle, Electron Diffraction using Transmission Electron
Microscopy, J. Res. Natl. Stand. Technol., 106, 997-1012, 2001.
[8] http://www.maxsidorov.com/ctfexplorer/index.htm
31