Graef M. Introduction to conventional transmission electron microscopy

Подождите немного. Документ загружается.

248 Getting started

2 mm

score mark

support

(a)

(b)

sample

dimpler

mounting

stub

(c)

(d)

dimple wheel

(e)

(f )

sample

Cu support ring

glue

Fig. 4.6. Illustration of the steps involved in the preparation of a GaAs [001] thin foil.

be around 80–100 µm. This is of course an approximate number, since the accuracy

of the amount of material removed is probably not much better than about 5 microns

per step.

Remove 60

µm using a 1 µm diamond compound (two steps of 30 µm each), and

then continue with three more steps of about 10

µm each, leaving (theoretically)

about 10

µm of material at the center of the dimple. We were rather conserva-

tive on each of the last ﬁve steps, so the true thickness at the dimple center was

probably a little larger than 10

µm. It is possible to estimate the thickness by

means of an optical microscope with a calibrated focusing knob; simply mea-

sure the difference between the in-focus condition at the dimple center and the

in-focus condition of the mounting stub surface, and subtract the thickness of

the mounting wax layer (this assumes that you measured the total thickness,

stub +wax +sample, before you started dimpling). With some experience you

should be able to determine the thickness to within a few microns. After comple-

tion of the dimpling steps you should soak both stub and sample in acetone to

dissolve the mounting wax. After a little while the sample will simply fall off of the

stub.

Since the thin foil you have just produced is mechanically not very strong you

should mount it on a copper support ring (see Fig. 4.6f). Many different types of

mounting rings and disks are available from microscopy supply companies. We used

a copper ring with an inner diameter of 1.5 mm. Use super-glue or two-component

glue to attach the foil to the support ring, taking care to keep the glue away from

the center of the sample. Only a very small amount of glue is needed to attach the

sample to the support ring; typically three or four “dabs” of glue spread out around

the ring will sufﬁce.

†

The dimpled side should face away from the support ring.

Some researchers prefer to attach the sample to a support ring before any dimpling

†

You should use your own judgment on how much glue to apply. If the glue squirts out around the edge when

you place the ring on the sample, then you probably used too much of it. Make sure the sample and support ring

do not get stuck on the bench top or glass slide!

4.3 The study materials 249

is carried out. This may be necessary for materials with high porosity or very low

strength.

It is possible to use double-sided dimpling, i.e. to dimple both surfaces of the disk.

This requires a bit more care in terms of thickness measurements and alignment

of the two dimples with respect to each other. The reader should experiment with

various methods; after a while, you will ﬁnd a procedure that works for many

different materials.

The dimpled foil is now ready for ion milling. The surface is typically exposed

to argon ions (Ar

), which have been accelerated by a potential of ∼1–6 kV. The

ions may reach the sample in a wide beam or in a focused beam, depending on

the type of ion miller you have. We used a Gatan Precision Ion Polishing System

Model 691 (PIPS). Both guns were ﬁrst set at an incidence angle of 10

◦

above the

sample. After about 1 h of milling at a potential of 3 kV, one of the guns was set at

an angle of 10

◦

below the sample surface, and an additional 10 min of milling was

carried out. Finally both angles were reduced to 8

◦

and the potential was dropped

slightly to 2.8 kV for a ﬁnal 7 min polishing step.

The second GaAs sample was oriented along the [110] zone axis, at 90

◦

from

the ﬁrst sample orientation. A bit more work is required to prepare a good thin foil

in this orientation. The method outlined below can also be used for other materials,

in particular for the study of thin ﬁlms.

We cleaved a strip from the GaAs wafer along one of the {110} planes. We cut

the strip into three pieces of roughly equal size, and glued the three strips together

to obtain a 1.8 mm thick stack, which is a bit easier to handle. It is important to

keep track of the orientation of each of the pieces, so that all three are in the same

orientation. Since one of the sides of the stack is parallel to the (110) plane, and

we wish to obtain a thin foil with a [110] foil normal, we can use a diamond saw

to cut a 300

µm thin strip normal to the [110] direction. The ﬁnal sample was then

cut from this strip and had dimensions of 2.0 × 1.8mm

. We mechanically ground

down the sample to a smooth polished ﬁnish, and glued it onto a copper disk with

a central hole (polished side facing the support ring). The remainder of the sample

preparation was identical to that of the [001]-oriented foil.

We should point out to the reader that the ease with which GaAs cleaves is

both “a curse and a blessing”; it is a blessing because it is straightforward to

obtain straight cleavage facets in the early stages of sample preparation. On the

downside it is also easy to lose the entire thin area of your foil when the foil

is mounted in the TEM sample holder. The air pressure exerted on the sample

when you move the holder across the room from the mounting table or desk-

top to the microscope column may be sufﬁcient to cause a cleavage fracture of

the entire thin area, leaving a large facetted hole in your carefully prepared thin

foil!

250 Getting started

(a)

(b)

Fig. 4.7. (a) [100] projection of the structure of BaTiO

; the front oxygen atom has been

removed so that the small [001] shift of Ti can be observed. (b) Parallel projection of

2 × 2 × 2 unit cells with oxygen coordination octahedra outlined.

The reader is encouraged to experiment with different specimen preparation

methods to gain experience in sample preparation.

†

One easy experiment is to glue

the pieces of the wafer together with a particular orientation relation between the

pieces. For instance, if the central piece in the stack of three is rotated by 45

◦

with repect to the top and bottom pieces, and the hole in the thin foil is grown

sufﬁciently large so that it overlaps one or both of the glued interfaces, then you

will be able to obtain both [110] and [100] zone axis orientations in the same sample.

Since sample preparation is a time consuming portion of microscopy research, it

usually pays to think carefully about your sample geometry before you start sample

preparation.

4.3.4 Material IV: BaTiO

4.3.4.1 Crystallography

The fourth and last study material, BaTiO

, is a representative of an important

class of structures, the perovskites. The commercial signiﬁcance and application

potential of perovskite-based materials is mind-boggling. Perovskites are used for

their dielectric, piezoelectric, electro-optical, magneto-resistive, catalytic, super-

conducting, or protonic conductivity properties [Roh01]. We have selected BaTiO

because it is easy to obtain in powder form, or you can make it yourself, starting from

BaO and TiO

powders. The room-temperature structure is non-centrosymmetric,

as shown in Fig. 4.7 and Table 4.4. We will use this material to illustrate the image

contrast of defects in a non-centrosymmetric structure.

†

The website will contain a repository for specimen preparation methods for the study materials of this book.

Any reader who has an alternative method of preparing the thin foils can add his/her procedure to the repository.

4.3 The study materials 251

Table 4.4. Material parameters for BaTiO

(taken from [Wyc63]); [barium titanate; Pearson

Symbol tP5 Strukturbericht Notation E2

P4mm

Space group a = 0.39947 nm

Lattice parameters c = 0.40336 nm # 99

Asymmetric unit Ba at 1a (0, 0, 0)

Ti at 1b



, 0.512



Oat2c



, 0, 0.486



Oat1b



, 0.023



The basic perovskite structure is primitive cubic, with CaTiO

as the prototype

structure. The space group is Pm

3m (# 221) and the Strukturbericht notation is

. The Ti atom is octahedrally coordinated by six O atoms, and is located at the

center of the octahedron. The room-temperature structure of BaTiO

differs from

the cubic perovskite only in the small displacement of the positive and negative

centers of charge, which results in a small tetragonal distortion with respect to the

cubic cell. The Ti atom is displaced from the center of the cell by about 7.2pm

along the [001] direction. This small displacement is sufﬁcient to cause the positive

and negative centers of charge to no longer coincide, which in turn creates a perma-

nent electrostatic dipole moment in the unit cell. Regions with identical orientation

of this dipole moment form ferroelectric domains, and in Chapter 8 we will take a

close look at the boundary between two such domains.

4.3.4.2 Thin-foil preparation

We obtained BaTiO

in powder form from Alfa Aesar (99.7% purity), mixed the

powder with a polyvinyl alcohol binder, and pressed a pellet (11 mm diameter, 3 mm

thick) using a 150 MPa uniaxial pressure. The pellet was then heated to 900

◦

Cto

burn off the binder, sintered for 12 h at 1230

◦

C, and then heated to 1430

◦

Cfor3h

to grow the grains. The resulting pellet was light gray in color, and was about 99%

dense. You could also start from the basic oxides BaO and TiO

, mix them in the

proper proportions, and then subject them to the same sequence of processing steps;

you may need to repeat this procedure twice to get a single-phase microstructure

(i.e. after the ﬁrst cycle you crush the pellet to a ﬁne powder and then repeat the

cycle).

A thin foil can be prepared as follows: use a low-speed diamond saw to cut

strips of about 350

µm thickness from the pellet. Use a sharp razor blade to cut

252 Getting started

small squares (2 × 2mm

) from the strips. Mount one such square on a dimple

support stub, using low melting point mounting wax, and remove about 100

µm

of material, using 6

µm diamond paste. Then use a 1 µm compound to remove an

additional 50

µm. After removing the sample from the dimple support, attach a

copper support ring (1.5 mm inner diameter) to the dimpled side of the sample, to

provide mechanical strength.

Dimple the second side of the disk using 6

µm diamond paste (remove about

100

µm), followed by removal of 75 µm using 1 µm paste. At that point the sample

should become translucent and with an optical microscope (in transmitted light

mode) you may be able to observe a grainy substructure. The sample is now ready

for ion milling.

Place the supported sample in an ion mill holder and mill at an angle of 10

◦

and an acceleration voltage of around 3 kV until a small hole is formed. Once

a hole appears, remove the sample from the mill and place it in a carbon coater.

Most ceramics are poor conductors of electricity, so a thin conducting layer must

be deposited onto the foil to reduce charging due to the electron beam. The exact

evaporation conditions will depend on the type of coater you have. After deposition

of a thin carbon layer your sample is ready to be placed in the microscope.

4.4 A typical microscope session

4.4.1 Startup and alignment

As the reader may have noticed by now, the learning curve for the TEM is quite

a bit steeper than that for, for example, the optical microscope; it may take many

hours of (sometimes frustrating) work to become familiar with the operation of the

TEM. It may take an even longer time to learn how to interpret electron micro-

graphs, whether they be images or diffraction patterns. This is quite normal and

the reader should be patient and persistent. While this book deals with the basics

of TEM operation and image interpretation and leaves out a great deal of advanced

techniques (in particular, all analytical observation modes), the techniques covered

in this and the following chapters are fundamental to all observation modes, in-

cluding analytical methods. A thorough study of this book combined with many

hours on the microscope will provide the reader with a ﬁrm background from

which more advanced techniques can be learned. That this requires a fair amount

of mathematics is unavoidable and the reader should divide his/her time between

the experimental work described in this chapter and the study of the theory of im-

age formation in the following chapters. The reader is encouraged to try out all

examples in this chapter and, since the main illustrations are taken from the four

study materials, it should not be too difﬁcult to reproduce nearly every micrograph

shown.

4.4 A typical microscope session 253

Every microscope session starts with the following basic steps (assuming a thin

foil is available).

(i) Load the sample in the sample holder. Take extreme care not to touch the part of the

holder that will be in the vacuum of the microscope. If you do need to touch the holder,

wear lint-free gloves. Specimen holders are delicate and very expensive. You should

treat the sample and the sample holder as if they were precious gems, and you should

always take your time and use a steady hand when handling them.

(ii) Clean the sample holder (optional). For modern ﬁeld emission microscopes it is rec-

ommended that the sample and sample holder be subjected to a short plasma cleaning

step. This procedure removes most hydrocarbons

from the sample surface

and im-

proves the overall cleanliness of the sample. This is particularly important for analytical

microscopy. More

information on plasma cleaning can be found in [IFO

99].

(iii) Insert the holder into the column. Since top-entry and side-entry machines have differ-

ent standard operating procedures, we refer the reader to those instructions. In general,

the specimen holder is ﬁrst partially inserted to activate the pumping system. Then,

after a ﬁxed pumping delay, the holder is further inserted through an airlock mecha-

nism until it is ﬁrmly seated in position. The pumpdown time can be as short as a few

minutes, or much longer for some top-entry microscopes. One can use this time to

dim

the lights in the microscope room, to give the eyes ample time to adjust to the lower

light level.

(iv) Turn on the acceler

ation voltage.

This again depends on

the particular machine you

are using. On older machines you would typically use push-buttons or a rotary knob to

increase the voltage in steps of 20 kV or so up to the maximum acceleration voltage.

On many modern machines the acceleration voltage is never turned off, so that this

step can be skipped. On higher-voltage machines, such as the JEOL 4000 series, the

voltage is slowly ramped up to the maximum, a process which may take an hour or

longer; for such machines it is a good idea to start them up a few hours before your

session begins.

(v) Turn on the ﬁlament current. To extend the lifetime of the ﬁlament you should always

take

the time to turn the ﬁlament current up from zero to near saturation. For thermal

emission guns, the ﬁlament has to heat up to a temperature where thermal emission

starts (usually in the range 2000–2700 K), and since heating is always associated with

thermal expansion and stresses, it should come as no surprise that slow is better than

fast. As a rule of thumb one could say that the time it takes to heat up the ﬁlament to

near saturation should be proportional to the cost of the ﬁlament: a tungsten hairpin

ﬁlament can be turned on in a matter of a few seconds, whereas LaB

single-crystal

ﬁlaments should be turned on over a period of a few minutes. For hot ﬁeld emission

tips the extraction voltage is often left on and the beam is simply deﬂected away from

the optical axis of the microscope. Opening the gun airlock will bring the beam back

onto the optical axis, and no changes in ﬁlament current are needed.

This brings us to the most important part of the startup procedure: the alignment

step. Nothing will be more frustrating to the microscopy student (or for that matter

254 Getting started

to an experienced user) than an incorrectly aligned microscope. It is important that a

routine alignment be carried out at the beginning of every session, even when others

have used and aligned the instrument before you begin your session. The details

of this routine alignment will depend on the particular microscope being used, but

there are a few general steps which should be part of any alignment. While the more

modern machines often have the built-in computer prompt the user for the various

alignment steps, many of the older machines do not have an on-board computer

and the user has to execute the steps from the user manual. Alignments can vary

from purely electronic or software-driven steps to partially mechanical alignment,

and this depends mostly on the age of the instrument. The following list itemizes

the most common alignment steps.

(i) Gun alignment. Use the gun tilt controls to obtain a symmetric ﬁlament image, as

shown in Fig. 4.8(a) for a tungsten source and a LaB

source. If no undersaturated

image can be obtained, then the gun tilt controls should be adjusted so that maximum

brightness is observed on the viewing screen.

(ii) Condensor aperture alignment. The electron beam should expand and contract con-

centrically when the second condensor lens current is changed. If the center of the

(a)

(b)

(c)

(d)

spotsize = 2

1 µm

Fig. 4.8. Undersaturated ﬁlament tip images for (a) tungsten and (b) LaB

ﬁlaments;

of severe astigmatism on the beam shape: top and bottom are over-focus and under-focus

images, respectively, while the center image is in-focus (all images are taken at the same

magniﬁcation).

4.4 A typical microscope session 255

beam moves upon changing this current, then the condensor aperture should be

translated until no further movement occurs for a wide range of beam diameters.

An application which needs a high beam current should use a large-diameter con-

densor aperture, whereas observations which need a high beam coherence should

sacriﬁce some current by selecting a smaller diameter aperture.

(iii) Illumination stage alignment. This step ensures that the electron trajectory is aligned

with the optical axes of both ﬁrst and second condensor lenses. The alignment pro-

cedure is usually an iterative procedure whereby the spot size (ﬁrst condensor lens)

is alternated between two settings, and the beam displacement on the screen is min-

imized by means of the beam and gun shift deﬂection coils. The beam alignment

is satisfactory when no substantial beam shifts are observ

ed when the spot size is

changed.

†

Figure 4.8(c) shows a typical series of spot sizes for a JEOL 2000EX

microscope.

(iv) Condensor astigmatism correction. The incident electron beam should have a circular

cross-section for all settings of the second condensor lens. Condensor lens astigma-

tism can be recognized readily as an elliptical distortion of the beam, which changes

major axis when the second condensor lens current goes from an under-focus to an

over-focus condition, as shown in Fig. 4.8(d). The condensor stigmator coils must

be used to correct the beam

shape. Condensor astigmatism can also be corrected

in the in-focus condition (smallest beam diameter) by ensuring that the ﬁlament tip

produces the sharpest possible image (under-saturated condition).

(v) Filament saturation. When all four previous steps have been carried out properly,

the ﬁlament current (in the case of a thermal emission gun) can be increased to the

saturation level, often indicated on the microscope console by the position of a so-

called beam-stop, a metal rod which prevents the user from turning the current up

any further.

(vi) Sample eucentricity. The height of the sample inside the microscope column is an

important variable. The primary tilt axis is a ﬁxed axis with respect to the microscope

column; when the primary tilt is changed, the entire goniometer stage rotates around

this axis. The sample holder

axis, i.e. the center line along the holder rod, does not

necessarily coincide with the primary tilt axis, as illustrated in Fig. 4.9(a). When the

goniometer is tilted, a feature on the sample, indicated by a black circle, will also

rotate around the primary tilt axis, and its projection on the viewing screen will move

side

ways.

‡

The eucentric position is then obtained by adjusting the height of the sam-

ple, the z-control, until there is no further lateral movement during tilting (Fig. 4.9b).

This adjustment could be manual or electronic, depending on the microscope model.

Eucentric alignment is important because it facilitates sample tilting while in diffrac-

tion mode (the region of interest will stay on or near the optical axis). The eucentric

†

Note that the beam diameter will change when changing the spot size; this is due to the fact that the object

plane of the second condensor lens changes with spot size, as discussed in Chapter 3. One must then modify the

second condensor lens current to obtain the same beam diameter on the screen as before.

‡

This is actually useful, because it provides a simple way to identify the orientation of the primary tilt axis on

the viewing screen.

256 Getting started

optical axis

primary

tilt axis

sample

holder

z-control

image moves

during tilting

(a)

optical axis

primary

tilt axis

sample

holder

z-control

image stationary

during tilting

z adjustment

(b)

Fig. 4.9. (a) A sample point (black circle) that does not lie on the primary tilt axis will

move laterally during tilting. (b) By adjusting the height of the sample in the column, the

lateral movement can be removed completely; this is the eucentric position.

position is also the position for which the objective lens current generates the smallest

possible spherical aberration coefﬁcient.

(vii) Voltage centering. To ensure that the electron beam travels along the optical axis of the

microscope, we can superimpose an oscillating voltage on the acceleration voltage.

This will result in

a sinusoidally varying electron energy and hence wavelength. This,

in turn, means that the total rotation angle of the electron trajectory will change

and the resulting image will rotate around a point that does not necessarily coincide

with the optical axis. Use the beam tilt controls

(in bright

ﬁeld mode) to bring the

center of rotation onto the optical axis (center of the screen). Alternatively, one could

superimpose an alternating current on the objective lens current and bring the apparent

rotation center to the center of the screen; this is known as current centering. Current

and voltage centering should in principle give rise to the same position of the rotation

center, but in practice there may be a small difference between the two.

(viii) Diffraction centering. When you switch to diffraction mode the central beam may not

be located precisely in the middle of the viewing screen. You can use the projector

lens alignment to bring the central beam to the

center. In older microscope models

the projector alignment is a mechanical alignment, using two pairs of screws near

the bottom of the column. In more modern instruments the projector alignment uses

deﬂection coils. Since the projector lens is the ﬁnal lens in the microscope, changing

its alignment will only cause a lateral movement of the entire image or diffraction

pattern.

Now the microscope is ready to begin observations. Insert the sample into the

beam path if it is not already there, and use the specimen translation controls to

locate the edge of the thin foil. In the next section we will discuss a few basic

techniques to obtain images and diffraction patterns. Then we will record a series

of micrographs illustrating contrast features which one can expect to observe in

any defect-free crystalline material.

4.4 A typical microscope session 257

4.4.2 Basic observation modes

4.4.2.1 Going from image to diffraction mode

Assuming that you have prepared a foil from one of the four study materials, now

is the time to insert it in the column. When the alignment procedure has been

completed, you should perform the following steps.

Make sure that both diffraction and selected area apertures are removed from the column;

set the magniﬁcation to somewhere between 5000×and 10 000×.Donot move or remove

the condensor aperture, as you have just completed its alignment!

Use the specimen translation controls to bring the edge of the sample (or any other area

of interest) to the center of the screen. You may need to reduce the magniﬁcation to ﬁnd

the edge ﬁrst.

Change the height of the sample (using the z-control) to bring it to the eucentric position;

this can be done by tilting the sample back and forth around the primary tilt axis, and

minimizing the lateral shift of the sample on the screen.

Insert the selected area aperture and select an aperture size that covers the area of interest

in the sample. Center the aperture on the screen.

If necessary, adjust the position of the

sample so that the area of interest is visible inside

the aperture.

Focus the image inside the aperture.

†

This ensures that the specimen is conjugate to the

SA aperture plane.

Change to diffraction mode.

The procedure described above is one of the basic procedures for transmission

electron microscopy. You will perform this routine over and over again, every time

you switch from image to diffraction mode. The ﬁrst three steps may not always

be necessary, but the remaining ones should become second nature to you. It is a

bit like changing gears in a car with manual transmission; this involves a number

of coordinated moves, but after a while you do not really have to think about them.

Going from image to diffraction mode (and then back, which is described in the

next section) should become part of your microscope driving skills, so you may

want to spend some time practicing the individual moves.

Before moving on with the next step we should point out a few image charac-

teristics that you may have observed while carrying out the above routine. Before

you inserted the SA aperture you may have noticed that the image contrast was

rather poor. If you go back to image mode and remove the aperture, you will only

observe contrast if you change the OL focus (either over-focus or under-focus).

At the in-focus condition, you actually have the lowest contrast, and you may not

even be able to see the sample at all. When you change the OL focus, you may

also notice bright features moving back and forth with changing focus. There may

†

Some older microscopes have a special focus knob to focus the edge of the SA aperture independently.