An Introduction to Scanning Electron Microscopy and Focused Ion Beam (Matthew Bresin)
Sep 5, 2023
For more information, visit https://nanohub.org/resources/22625
Matthew Bresin
6/3/15
"An Introduction to Scanning Electron Microscopy and Focused Ion Beam"
It's after lunch, we've got to keep it entertaining.
Are we, we're good?
Set?
Yeah, all right, cool.
Anyway, thank you.
Welcome to the start of the afternoon session today.
I'm going to be giving you a talk on basically
an introduction or kind of a rush job
on scanning electron microscopy and focused ion beam.
Hi.
I'm Matt, once again.
Just would like to say thank you to my other colleagues
in the electron microscopy core, Honghui, Jim, CQ, and Vasek,
who've all given me a lot of good advice
and valuable contributions to this talk.
So it's kind of a collaboration from all of us.
So a general outline of kind of what
you're going to see in this talk is an introduction
to the scanning electron microscope, the focused ion
beam, what is it?
Then we're going to transition to each one.
How does it work?
We're going to break down the individual components.
And then we're going to go to the fun part,
what can it do for you?
So by the time that we're done with this talk,
I'm going to give you a general idea of what
you can expect as the operator.
So by the time that you are sitting in the command chair
here, you have an idea of what's going to happen on the scope.
So moving along, I'm going to read off the slide real quick
for just one sentence here.
So an SEM is an instrument for observing and analyzing
the surface microstructure of a bulk sample
using a finely focused probe of electrons.
So typically, a SEM can achieve 1 to 4 nanometer resolution,
depending on the instrument type
and some of the components.
We'll break a little bit of that down later.
The primary applications for a SEM,
typically topography and morphology of a sample.
But compositional analysis is a major component of a SEM.
We can also do some crystallography
and some optical and electronic property characterization.
And there are many, many more.
There's tons of techniques.
We'll try to focus on the major ones during this talk.
So before we get into kind of the nitty gritty,
we're going to go back to sort of an overall definition.
Resolution.
So you've heard that maybe a couple times today.
And I just want to take just a few seconds
and say, what is really the definition of resolution
as we term it in an imaging system?
So what we want to be able to do is
tell the difference between two points that are sitting
right next to each other.
So this is the difference between two things sitting
right next to each other versus now two things that
are so close that we can't resolve them
versus two points versus one just larger particle.
And this is what we call an airy function or representation
of the airy function.
It comes from diffraction effects.
And if we can resolve these two points, then it's resolvable.
So when I say 1 to 4 nanometer resolution,
you can tell the difference between two things
that are sitting right next to each other
that are of that size scale.
It does not mean that those two things
are going to be perfectly sharp, well-absorbed, and beautiful.
So some numbers just to throw out.
Some resolution limits based on just diffraction.
So I'm sure you've seen this equation before.
And just throwing out some numbers here.
And it's why do we want to do this
compared to the main technique, optical microscopy.
Opticals, the one thing you're probably all very familiar
with, right?
You've used them all in secondary education,
high school even.
The main advantage that we're getting here
compared to light, using electrons,
is that light, your wavelength, around 400 nanometers or so
is your imaging wavelength.
As compared to an electron, we're
talking the wavelength there is down to very, very, very small.
Actually, I think I might have the wrong number on there.
But that's OK.
So the other difference here is the numerical aperture.
Light's around 1.
Electron's a little bit different
due to the convergence angle of the beam.
But just the hard numbers.
Practical resolution for light microscopy,
you're sitting around about 200 nanometers
of where you can resolve.
And that's going to depend on the instrument, some
of the optics.
For an electron beam, you're looking around 0.3 nanometers.
This is the diffraction limit of the system.
So practical magnifications of those two things,
around 1,000, 2,000 for light, about 500,000
for an electron microscope.
All right.
Now, once again, this is the best possible resolution.
Things will just become worse as we go through the system.
So we're going to have other lenses.
These are going to reduce our resolution.
Just a quick comparison.
So optical versus an SCM-based image, there's two things
you're going to pick out of this.
One, obviously, the resolution is
much better for the SEM image versus the optical.
The second one, which sometimes we don't keep in mind,
is what's called the depth of field.
And that's how much Z, or how much
depth of the image is in focus at any given point.
So we can see here in the optical scope,
the top part of this sample is in focus,
and the bottom part is pretty fuzzy.
And now in the electron microscope,
virtually all of this is in focus at any given point.
And this has to do with the convergence angle of the beam
and, of course, the resolution of the system.
So the major components of a SEM in a quick breakdown.
First thing, we have a vacuum system.
We'll talk about that a little bit later in terms of sources.
Sources require a vacuum.
Your electrons also require a vacuum.
So we have a source, which is going
to produce your electrons, usually
sits on the top of the column.
And then we have a set of optics.
So once we have our electrons, how do we focus them?
How do we make sure we get rid of the aberrations that
are associated with the lenses?
That will all be part of the optics.
Then we move on to a scanning system and a set of detectors
that, once your electrons are hitting your sample,
how do we see stuff?
So moving to electron sources.
This is going to be a quick general breakdown
of how an electron source works.
It's going to contain a filament or a needle,
a set of extraction electrodes or a rough electrostatic lens.
And then you're going to have an anode plate, which
will set your acceleration voltage,
will accelerate the electrons down the column.
So the source is going to contain
these three major components.
What it does is it produces electrons.
It roughly shapes the beam and then sets, once again,
your acceleration voltage.
There's two types of sources, thermionic sources
that use heat to produce electrons.
And then there are field-emitting sources
that use an electric field to produce the electrons.
Each source has advantages and disadvantages.
And now, as we move on, the sources.
First one, thermionic sources, which
is effectively the fancy light bulb of the electron source
world.
Basically, what it comprises of is a tungsten filament
for tungsten.
And we have some lab 6 crystals as well.
You pass a current through this.
You produce heat.
And the idea here is to take electrons
from the top of the conduction band,
heat them up so much that they overcome
the work function of the material,
and they escape into the vacuum level.
So the question now is, how hot is hot?
For tungsten, you're talking 2,700 Kelvin.
For lab 6, you're around 1,800 Kelvin.
Of course, if air hits these while it's
at that high of a temperature, they explode.
And that's not fun for anyone, hence the reason
for the vacuum system.
See, I came back to it.
So don't worry about it.
We'll all get there.
All right, now, major advantages of this type of source,
very high probe currents are obtainable,
very stable currents, good for analysis.
And you don't need as complex as a vacuum system.
Mostly, the reason being there is the very short lifetime
of these filaments.
And they tend to be a little bit lower resolution,
too, because they're just larger.
These are sacrificial filaments that we
use in training scopes or other scopes
that need less stringent vacuum requirements.
Now, the sources that are more resolution-oriented
are field-emitting sources.
These sources are a little bit different.
They still use tungsten, but they're
going to come to a sharp, tapered tip, a few nanometers
in curvature.
And instead of using heat to produce the electrons,
we now keep them cold.
And we use an electric field to actually strip electrons off
of the very tip of the emitter.
So you have a first anode, OK, which
is going to strip electrons off of the source.
And then you have a second anode that's
going to accelerate those electrons down into the column.
All right, so instead of going up and over the thermal barrier,
you now go through a new barrier that the electrons
can quantum mechanically tunnel into the vacuum level.
So that's the basis of a field emitter.
And the main point of there is cold.
Major advantages, very high resolution.
I put best with a question mark, because the next source
and this one kind of compete a little bit for best resolution.
Has very good lifetime.
So these last a very long time compared
to thermionic sources, years.
Disadvantages, they don't have very good probe currents.
They tend to have fairly poor stability.
And they need to be flashed periodically.
Because the gun is cold, any molecules
that are in the gun head, if they get attracted to it
and they mess up kind of the tip shape a little bit,
then you're going to get less emission.
So occasionally you have to flash it.
And that gets rid of all those molecules.
And then your current will go back up.
So over time, you're imaging with this source.
And your image gets darker, darker, darker.
And you say, hey, what's going on?
Oh, I need to flash the tip, hit the button,
and normally that takes care of it.
And that's the other reason.
Ultra high vacuum in the gun area.
Need that.
Finally, the last source, the shocky source.
This is your hybrid heating and electric field extraction
source.
Uses the same type of filaments or needle
as a field emitter, sharp tungsten tip.
But it has this additional zirconium oxide coating
right here, which lowers the overall work
function of the surface.
So instead of going purely this way or purely this way,
we now go over a little bit of a hump here.
And we get kind of the best of both worlds
between the sources.
You have very high resolution.
You have very good probe currents, long lifetime,
good stability.
Still requires ultra high vacuum in the gun area.
And of course, since it's being heated,
you now have a bit more of a finite lifetime.
So maybe a couple years compared to just a pure cold
FEG, that's going to be several years at least.
So that's sources.
We've got our electrons.
They're flying through space.
That's great.
What do we do with them?
Lenses.
We need these to focus our electrons
and get them down the column.
So we're going to talk about the lenses and the aberrations
that occur from these lenses.
So the primary lens that we use is an electromagnet.
The reason being is that we can't have the same focusing
optics that an optical microscope has.
If we put a big milled piece of glass,
the electrons will hit that and just scatter all over it,
and there'd be no point.
So electrons have to go through vacuum.
And what we use is a coiled set of wire, copper wire,
around a soft iron pole piece.
And we pass a current through that set of windings.
And that's going to produce a magnetic field.
So how we work with this magnetic field
is that a very small slit is cut into this iron core.
And that allows the magnetic field
to leak out in a very controlled fashion
into the center of the microscope.
And by controlling the amount of current
that we pass through the windings,
that lets us control the magnetic field,
and hence the focusing action of the electrons.
So how this works is that we have a magnetic field.
We have a charged particle that's
moving through this magnetic field.
And it's going to see a force based on the Lorentz equation,
velocity cross the B field, the magnetic field.
And that's always going to give us a inward focusing action.
The lenses.
There are two primary lenses that we
use in electron microscope.
The first one is the condenser lens.
The condenser lens sets the spot size of the probe.
And how it works is that up here we
have our electron source, our anode, condenser lens
right below it.
Based on the convergence of the condenser lens,
we set the spot size.
So when the condenser lens is in a low convergence type point,
it's in a weak lensing condition,
all the beam is going to pass through this limiting aperture.
If we now change the convergence point of the beam,
some part of that beam is going to be
cut by the limiting aperture.
And that gives us less current and a smaller spot size.
So overall, if we have a low convergence,
we get more beam is accepted through this limiting aperture.
This gives you more current.
But your probe size is larger, so you have less resolution.
Good takeaway there.
At a high convergence, we have less beam
accepted through the aperture.
This gives us less current, but it
gives us a higher resolution.
So if you are an SEM operator or a budding SEM operator,
this is one thing to keep in mind,
is that based on what current you set in the instrument,
this will affect the performance, the image
and quality that you get.
The second major lens that we have
is called the objective lens.
That sits right below the limiting aperture here.
And what the objective lens does is
it sets, very similar to an optical scope, the focus.
So based on changing the lens strength,
we can change the point at which the system focuses.
So we have in this one here, this lensing condition
sets us exactly the focal point of the sample.
If we do this too strongly, we have an over-focused condition.
If we go now to a weak lensing condition,
we have an under-focused situation.
So obviously, these two were out of focus.
In this one, we are at focus.
This is commonly called the working distance,
the distance between the objective lens
and your sample or the stage.
This is a good way to read out where your sample is in space.
And you'll typically find this on most SEMs.
So we have these lenses.
Ideally, if we have a point source,
it's going through a lens.
We want that point focus back to a point.
But life doesn't exactly work the way that we want to.
And that's why we can't have nice things.
So the actual lens here is going
to focus us to more of a disk.
It's what we call a disk of confusion.
And several effects are going to contribute
to the size of this disk.
And these are these limiting parameters,
which are our aberrations.
Spherical, chromatic, aperture diffraction, and stigmatism.
Stigmatism is in red here, because that's
something that you as the operator
can correct all the time on a SEM.
So I'm going to step through the aberrations fairly quickly.
The first one is chromatic.
And what happens with chromatic is
that, say, for instance, we have two electrons.
And these two electrons are going
to pass between this path from P to B here
and go through this lens.
Now, if these two electrons have different incoming velocities
or energy, they will be focused to two different points
on the sample.
So we have one ray that's going to pass where we want it to.
And then we have another ray at different energy
that will now pass to a different position.
And this gives us our defocal disk.
So where this comes into play is that if you're
using a heated source, thermionic source, OK,
or a shocky source, you have an energy spread
that comes off that source.
That energy spread will then translate
into chromatic aberration in your system.
This is particularly pertinent for a thermionic source, which
has the worst spread.
So what this means is that what we
can do is if we increase the electron energy,
the acceleration voltage of the beam,
we can decrease the effects of chromatic aberration
because the chromatic aberration spread,
the energy spread of the electron,
becomes less of the overall acceleration
of that electron itself.
So when you're working with a thermionic instrument,
you typically want to operate a higher acceleration voltage
in order to mitigate your chromatic aberrations.
So that's the one thing you got to keep in mind
when you're working with a thermionic source.
The major limitations, OK, for your other source types
are typically these, spherical aberration
and aperture diffraction.
And they kind of go hand in hand.
What spherical aberration is is that electromagnetic lenses
are not uniform throughout the radius of the column, right?
Focusing action becomes stronger towards the outside.
So ray paths that are passing, for instance, from P to B
here tend to get over-focused.
Whereas ray paths, for instance, P to A,
which are closer to the optic axis,
now get focused approximately where we want them to be.
Where this comes from is actually,
sorry, I'm getting a little out of myself there.
But basically, the idea is that the magnetic field
is not uniform, stronger on the outside,
weaker towards the center.
Off-axis rays tend to be over-focused.
The solution to this is just cut them off.
Just get rid of them.
Why do we need these?
They're reducing our resolution.
So now we come to aperture diffraction.
If we put in a living aperture and say,
let's cut down all the high-angle rays,
well, why can't we just make it as small as possible?
So an electron still acts the same way as light, OK?
The Airy function, ray light criteria
all come into play here, all right?
Diffraction effects will increase
with a smaller aperture.
So if we make it smaller and smaller and smaller,
now your aperture diffraction is going
to increase significantly.
So typically, optimum aperture sizes
are determined by the manufacturer.
They're not going to give you an aperture that's
going to completely destroy your imaging capability.
But generally speaking, as you make that aperture smaller,
you're going to lose some current, OK?
But you'll mitigate the effects of spherical aberration.
So you'll hear a little bit more about this in the TEM talk,
because this is a very important effect
in how we deal with spherical aberration in both systems,
and particularly what methods we use to correct those.
So the last major aberration that we have, OK,
is called astigmatism.
And this is the one that you guys as operators
can correct on a daily basis.
So astigmatism comes from if we have the same origination
point of your ray, and it's passing from, say,
P to the y-axis, P to the x-axis,
these rays are focused to different points.
So this ray particularly is over-focused,
and this ray is under-focused.
And what this creates in the image plane
is an elliptically shaped probe.
And what we'll see in the SEM when
you have this condition is that your image
is going to look stretched, OK?
So we see here this image looks like it's
stretched in this direction.
And then you see it's kind of sharp over on this side.
If we go through the focal condition,
we can then see that now the image here
is stretched in this direction, and now kind of sharp
on this side.
Now where astigmatism comes from is
that we have slight imperfections
in the manufacturing of the pole piece, in the copper windings.
You can have stray magnetic fields.
The operator may not be very good.
Sometimes that tends to happen.
But we can do some things to correct this, all right?
So we have a set of deflectors in the system
that are designed specifically to counteract astigmatism.
And these deflectors, instead of being wound kind of this way,
are wound in this direction.
And what happens here is that now
these electromagnetic lenses are pushing and pulling.
They're kind of stretching the beam in order
to get rid of that elliptical probe.
So what happens here is we have the basis of this quadrupole.
Now if we put a couple of these together,
we get the modern octupole stigmator, all right?
And through a combination of working with this,
we apply our stigmators.
And then what happens here is we change our rate paths.
So we turn our elliptical probe into a circular probe,
and we can eliminate the stretching in the image.
So the transition here is from a stretched beam,
and now we have an image where everything is pretty uniform,
but still fuzzy, right?
So this is what you're looking for as the operator.
If you correct your astigmatism to the point
where everything is uniformly fuzzy, you're good.
Next thing, change the focus of the objective lens,
and now make that circle as small as possible.
And you've got a nice focused image that you'll then take it,
you'll put it on the cover of Nature,
and you'll make everyone famous.
So we've got our electrons.
They've gone through our lenses.
We're focused to a nice probe.
We've gotten rid of as many aberrations
as we can on the system.
We're keeping some of those things in mind,
depending on our source.
How do we form an image?
What does that do?
So say we have our detector.
We've got our sample, and our sample
is kind of sliced into this grid, shall we say.
Now, as the beam is going to be scanned across
with a set of deflection coils, the interaction of the beam
plus the sample will produce some emission.
You have electrons.
You have light that could come off.
And what happens here?
So as the beam scans across this first pixel,
we get five counts, and that's a bright pixel.
Then if it scans across the next section, it picks up two counts,
and that gives us a dark pixel.
So it's effectively a proportional counter.
If you've got a lot of counts, you're
going to get a bright pixel.
If you've got fewer counts, you're
going to get a dark pixel.
Now, I don't want to give you the impression that the beam is
sort of just scanning one little section at a time.
The beam is constantly going.
Scanning electron microscope, beam
is constantly rastering across the sample.
Looks more like this.
If we have our happy smiley face on the surface here,
we've got our beam.
We've got our monitor.
And the beam's going to start scanning.
And it's going to scan lines.
And eventually, there's our little happy smiley face.
There she is.
Very good.
And the beam will scan, scan, scan.
And this is what you start seeing on the screen.
And then as it finishes, what will happen
is the beam will come back, and we'll start scanning again.
So it flies back to its initial point.
It'll start scanning.
And then go over and over and over and over again.
That's your live imaging mode in the SEM.
So as that's kind of wrapping up,
the magnification in a SEM, it's kind of an interesting point,
is defined by the screen size.
So your CRT, your computer monitor, whatever it may be,
versus the raster size of the beam.
So if the raster size is large, your magnification is lower.
If the raster size is, say, very small,
like let's say we scan across just the eye right here,
your magnification is larger.
So you're not particularly changing anything
in the optics to make the magnification different.
What you're changing is the raster scan size.
That will dictate your magnification
because your screen is approximately fixed.
Once we now have the system scanning,
electromatter interactions and the detectors,
that will be looking at these interactions.
So when the primary beam comes in and hits the sample surface,
there are a wide variety of particles and quanta
that will be emitted from the sample.
And particularly, what we're interested in is other electrons
that are emitted from the sample and characteristic X-rays.
Now, of course, we have some detectors
that can deal with light, that can deal
with certain scattering effects.
And we'll kind of talk about those as we go through.
But these are the ones that we really care about,
secondary electrons, backscatter electrons,
and characteristic X-rays.
So what are these things?
Two ones, secondary electrons and backscatter.
A secondary electron is, by definition,
a low-energy electron that's emitted from the interaction
of the primary beam with typically the valence shell
of the target atom that it hits.
And that will, you have your instant electron comes in.
It's going to interact here.
It will eject an electron.
Okay? And now that electron is free to go and do things.
As opposed to, say, a backscatter electron is a
primary electron that comes in.
It interacts, you know, with your target atom here.
Okay? And then it gets scattered over large angles.
These two things are going to carry different forms
of information, and we'll get to that in a few slides.
Before we do, I just want to talk
about characteristic X-ray generation in a SEM.
You have two types of X-rays that you can get.
You can get a continuum X-ray,
sometimes called a Bremsstrahlung
or white radiation.
And what this is, is that you have an instant electron.
It comes in.
It interacts, you know, with your atom here.
And then as an effect of that interaction,
that electron will be slowed down.
It will be braked and do the conservation of energy.
That electron will give off an X-ray
to maintain its momentum.
All right?
Now, on the other hand, okay,
if you have an incident electron comes in,
it knocks out a core shell electron.
An electron from a higher level orbital will jump
down to take its spot.
And when it jumps down, it's going to emit an X-ray.
And that is characteristic to the atom
from which it came from.
So this is how we figure out what do we have inside
of our sample.
So the spectrum looks a little bit like this.
We have our background, which is the blue, you know,
continuum X-rays.
And then we have these sharp peaks,
which are the red characteristic X-rays,
which actually identify the elements
that we're interested in.
So these are the major things that are going to come
off the sample that we're going to use, okay, for our detectors.
So the next part here is the interaction volume
of a primary electron.
So what's happening is that we have a primary beam.
It comes in, and it hits the sample.
And what does it do?
Does it just stop?
Not really the case.
It's got a lot of kinetic energy.
It's going to bounce around, okay,
within some interaction volume here.
And it will eject, you know, electrons, X-rays.
But specifically, we care about where they come from.
First one, for a secondary electron,
their energy being very, very low, their range
when they're created is only about a few nanometers.
So let's say we have a primary electron that comes in.
It hits the surface here and starts rattling
around the interaction volume.
If a secondary electron is generated around this point,
okay, it can only move a couple nanometers before it's going
to be reabsorbed by the sample.
In which case, that secondary is not getting out.
The only secondaries that are getting
out of the sample are the ones that are emitted
from the near-surface region.
Within the first couple nanometers
of most materials, those electrons will come out.
So a secondary electron is surface topography, you know,
oriented.
That's what you're going to get your surface topography
formation out of is an SE-based detector.
Now, a backscatter electron, on the other hand,
kind of same principle.
The primary comes in, it rattles around, gets scattered
over a high angle, and now flies back
out essentially from whence it came.
All right, this carries a little bit of different information.
The scattering is based on the material,
the average Z of that material.
And this lets us do some basic Z separation.
And finally, x-rays.
X-rays are going to come from the total complete volume,
and this is all elemental-based information.
So a quick example of an interaction volume here.
This is 20 kV electrons in PMMA, and it shows you
that this teardrop shape is not just something
that I like to draw.
It is real, and it can be simulated with Monte Carlo
or statistical software.
So as an example of just what you see in the microscope
from a beam energy perspective, you know,
why do you set one energy versus another?
And it depends on the interaction volume
that we just talked with.
Let's take, for instance, this sample, which has, you know,
gold flakes, and then there's some gold particles sitting
underneath of those flakes.
At a high acceleration voltage, around 15 kV, what you see here
under this flake, you can clearly tell
that there's some particles sitting underneath of it.
The mechanisms that you're seeing here
from an image formation standpoint is
that you're getting secondary electrons from the flake.
You're getting some backscatter electrons that are coming
from underneath of that flake.
So you're seeing a lot of depth-based information
at a higher acceleration voltage.
Now, as we start stepping down that acceleration voltage,
the amount of material that you see underneath the flake starts
becoming smaller and smaller and smaller, to the point now
at 1 kV, you're getting essentially all surface-based
information because your interaction volume is confined
to the near surface.
So you have to be a little careful as an operator
to know kind of what you're seeing
and whether you have kind of a defect in your sample
or if what you're really seeing is kind of on your surface.
Moving on to the detectors, and we're going to start getting
to the fun pictures, I promise.
An SE-based detector,
which is commonly called an Everhart-Thornley detector,
it uses a biased grid, typically a couple hundred volts positive,
which are used specifically
to attract the low-energy secondary electrons.
So as we see in this schematic here,
the low-energy secondaries are attracted to this biased grid.
It will collect some backscatter electrons just
from the solid angle.
So whenever you're using an SE detector,
you're getting all the SEs, plus you get a little bit
of backscatter information.
Those electrons hit a scintillator,
which transforms it into light, hits a photomultiplier tube,
and that gets read out through your preamplifier.
That's the basis of an Everhart-Thornley detector.
Now, the contrast mechanism
for the secondary electron generation itself is really
based on the topography of the surface.
So as the surface curvature increases, okay,
the effective tilt of the sample towards the beam,
you get more secondary electrons generated.
So based on this chart here, the number of SEs is going
to increase as the tilting increases, and it's based
on effectively an inverse cosine.
So what happens here, if you have spheres, okay,
on your surface, what you see is that right around the edges
where the curvature is the highest, those are bright,
okay, towards the top of your sphere,
where now your surface is leveled out,
that's going to give you a more duller contrast, okay,
or a darker type contrast.
Some examples, okay, of SE-based imaging.
Dendritic golds here,
we can see some nice surface microstructure.
And then just one kind of resolution-based image.
Note the scale bar here, 10 nanometers, about that.
And we can see these small, sputtered AU PD particles
on magnetic tape.
So the vast majority
of these particles are sub-10 nanometers.
All right, typically speaking, you'll be able to resolve things
around the 2-3 nanometer range in an FCM with an SC detector.
Some other examples of things you're typically going
to be doing in a SEM, okay, you see things, you measure things.
All right, so let's say, for instance,
we got some zinc oxide particles on your surface,
and we can measure, you know, all the little facets, okay,
what size these particles are.
For instance, here, this was some work that I did a while ago
where we were cutting out the coating on some
of these carbon nanotubes, and we were trying to measure, well,
how big's the tube, you know, how big's the coating.
Whatever you see on the screen, typically, you're going to be
able to measure it right on the screen
and get some instant feedback on what your particle sizes are,
what your film sizes are.
And lastly, you can do things, observe your world, okay,
be a little artistic, take a fun image every once in a while,
okay, this was a mixture of silver, some surfactants, okay,
this was a carbon nanotube bundle image
that I took down at Oak Ridge.
It's fun.
You get to see all the things that you never,
ever get to see on the nanoscale, and that's the major,
kind of, sticking point of SEM.
If I were to say anything else,
you get to see a lot of cool stuff.
So, let's move on to backscatter detectors, okay.
How a backscatter detector works is
that you get increased scattering
of your primary beam with the sample Z. All right,
so if we look at this plot here,
the backscatter coefficient will increase, not linearly,
but always increase as a function of the atomic number.
So, typically, a backscatter detector is going
to sit directly under the pole piece,
so that's this piece right here.
Here's your sample, okay, the quadrants are right in here.
Actually, this is the cage of the Everhart-Thornley
off to the side there.
I forgot that was in there.
So, with this backscatter detector, you can do things
like phase separate, okay, or Z separate in this case.
Two examples, one, a chip here.
In this SE image, you see very little contrast.
It's a flat smooth surface with some buried lines.
Under the backscatter, what it picks out is the lines, okay,
the interconnects that are in there.
So, chip manufacturer uses quite a bit to show
where your buried lines are,
if things are connected, disconnected, et cetera.
Another good example, okay, SE image on this side.
This was a sample that a company gave me, and they wanted
to know, well, I've got this material,
does it have fillers in it?
Does it have things that are kind of keeping the structure,
you know, making it stronger, making it more durable?
And you really can't see these structures except for a surface
and a bunch of cracks in it in the SE-based image.
But as soon as we turn on the backscatter detector,
now all of a sudden, these filler particles light
up like light bulbs, all right?
The background is low Z. These fillers are very high Z-based
materials, so I can tell exactly where the source
of that is coming from and then use things like EDS,
which is the next detector we're going to talk about to figure
out what these particles are.
So, good transition.
X-ray detection and energy determination.
How does this work?
Generate an X-ray, okay, through the natural interaction
of an electron with the sample, okay?
That X-ray is going to hit the detector, and when it does,
it's going to, through inelastic collisions,
generate a bunch of electron hole pairs.
And what happens here is that we apply a bias to this detector
so that these electron hole pairs can't recombine.
All right?
And they are going to be swept, okay,
to the different electrodes of this detector, all right?
And then they're going to be counted.
So, every electron hole pair that's generated is counted
with the detector, and then these are all summed
up based on the energy of a silicon electron hole pair,
and that is reconstituted into one X-ray count
from the X-ray source that it came from.
And that's going to give you the individual peaks
that are characteristic to that element.
So, things that we can typically do
with an EDS system is we can do some point and shooting.
We can do mapping.
There's a lot of good stuff we can do with the EDS system.
But just to get a little out of myself here,
mostly this is a qualitative elemental analysis technique.
Typically, beryllium and up is what we can look at.
Sensitivity is around .1 weight percent.
Quantitative analysis is a little tricky with an EDS.
Really requires standards, really requires a homogeneous,
polished, well-prepared sample, and the vast majority
of samples that you guys have are not that.
And we typically don't have a standard
that matches exactly what you're working with.
Sometimes we do.
Sometimes we don't.
So, if you want to do quantitative analysis
that requires a little bit more thought into your experiment,
I highly recommend talk to us, and we can work
on getting you quantitative analysis.
We have point line mapping scans, which is a good example
if mapping is down here, where we can take an SE image,
a backscatter image, and then we can separate
out the individual contributions of the elements
that comprise those images.
So, particularly, backscatter
and the elemental analysis is very powerful.
Because the backscatter shows you
in your electron beam image kind of what it looks like.
Okay? And then we can overlap, you know,
these individual elements on top of the picture itself,
and you can have a very good representation
of where the signal is coming from in your sample.
That analysis, of course,
is throughout the entire interaction volume.
So, you have to be a little bit careful
about buried structures, okay, or heterogeneous structures.
We have a resolution of about 130 EV.
This is an industry standard.
And this is important because it's going to come up next.
So, 130 EV, this is why we have some width in these lines.
Now, the complementary technique
to this is called wavelength dispersive spectroscopy.
And you want to use this
when you just don't have enough resolution in your EDS.
How it works is that your electron beam comes in,
hits a sample, generates X-rays.
They're going to fly up into this flat crystal here.
And the crystal is used to separate
out X-rays very specifically.
And this is going to give us much more resolution
than we had previously.
So, 130 for the EDS, about 5 EV.
Okay, for the WDS systems.
What this does, okay, it highly minimizes element overlapping.
So, let's take, for instance, this spectrum down here.
Okay, this is, the red is the EDS.
And then the blue is the WDS.
So, we can see within this one silicon peak.
Now, with the WDS, we can pick
out a very small tungsten inclusion here.
And we can have the individual peaks for tungsten.
And we have our peak for silicon.
On top of this, the detection limit
for your WDS is a little bit better.
And so, those are your two main advantages.
Okay, you've got less overlapping.
Okay, and you've got better sensitivity to the sample.
So, moving on to now crystallography.
All right, this is some more specialized techniques in CEM.
All right, first one's called electron backscatter diffraction.
And what happens here is that if the beam hits your sample
at a very high angle, okay,
we can get these things called Kikuchi bands.
And Kikuchi bands are formed by the scattering off
of each individual plane of the surface.
So, you see here, one plane, the red line, okay,
one other plane, the yellow line,
and the blue plane, blue line.
And what we get is different patterns, okay,
based on the orientation of your sample to the beam.
Based on these patterns, we can now do some mapping
and determine, you know, what orientation is coming
from where on your sample.
And so, what EBSD typically looks
like is something like this.
You're going to get this type of pattern, all right.
Software is designed nowadays
that will automatically index these patterns for you
with a high accuracy.
And then we take this data, we apply it per pixel,
and we can get things like orientation, okay.
We can see grain boundaries, and you can pick
out the grain size, okay, of your sample.
One thing I want to point out, okay,
is this gives you the orientation,
the microtexture of your sample.
It's not exactly giving you phase, though, okay.
If you want phase, go talk to Mauro, XRD lab, okay.
He's going to love the plug, all right.
He's going to love that.
XRD can do something better than the CEM.
So, go talk to him.
If you want phase, go talk to Mauro, XRD.
Now, the next one, photoluminescence, okay,
in the CEM.
What you're having here is that the electron,
when it interacts with your sample,
it's going to give off some light, okay.
We collect that light with a parabolic mirror,
and it tends to look like this.
So, here's your objective lens, okay.
It's going to go right through this little slit, okay,
on the mirror, hit your sample.
All the light's going to reflect off of that mirror,
and it's going to go to your spectrograph, all right.
So, we have a fairly large range, optical range.
You get decent spatial resolution.
The primary applications
of this semiconductor defect analysis,
quantum wells, plasmons.
So, some examples, okay, defects in gallium nitride.
So, we can see point defects, line defects here.
And then, another decent example is emission
from gallium nitride pyramids.
So, we can isolate out the exact emission points
from this type structure.
So, one thing I really haven't discussed is what do you need
to do about samples?
You're injecting charge, okay, into your system.
You have a beam current, okay.
When the beam hits the sample, you have a backscatter current,
you have a secondary electron current,
and you've got a specimen current.
If you're using a conductor, all is well, all right.
Everything will dynamically compensate, and things are fine.
But what happens when you're dealing with something
like an insulator or something that's just really not very
happy under the electron beam, all right?
You have a little bit of an issue in terms of charging.
When you have charge buildup, you start getting things
like distortions or other effects in your SCM.
So, this is a very good example of charging, where you'll see is
sort of like evanescent waves,
or you see really strong contrast.
So, see, you have this really bright like banding structures.
This is a classic example of charge buildup in a CEM.
To the point that if you really charge your sample,
the beam comes in, it looks like your sample goes, nope,
not happening, and goes straight back up to the point
where you start imaging the top of your column.
So, if you look at this image, we can pick
out all the little individual parts.
Here's the pole piece.
Here's the Everhart-Thornely detector right there.
I'm pretty sure that's the ion beam for the fib,
getting a little ahead of things.
And I'm pretty sure that's the platinum needle.
So, yeah, this is always fun.
You take a little bead, you charge it up,
and then we do the reflection.
And that lets you see what's going on the top
of your chamber, and sometimes that's useful.
So, what can we do about this?
You've got two main methods to deal with a charging sample.
All right, the first one is,
as we've seen this image before, sputter-coded.
So, we have a couple of sputter coders right downstairs.
Typical ones, gold and palladium,
or you have carbon coding.
You put a very thin layer of either of these two things,
and now your sample goes from charging
and insulating to now conductive.
And everything is fine.
So, as long as your coding does not exceed the things
that you're looking for, so if you're trying to look
at 5 nanometer pores, coding is probably not going
to be the best way of going about it.
But if you have larger features, okay,
things like soils or sands, coding is a really good way
of dealing with your charging sample.
The next way of going about it is,
let's say you can't code your sample.
You can do something which is called E2 point matching,
which is basically we match the beam voltage to the emission
that you get from that particular voltage
from the sample.
Now, we have here two points, E1, E2.
E1, the acceleration voltage, that's too low for most SEMs.
All right, it's going to be sub 100 EV or so.
Now, E2, on the other hand, typically is much more
in the range of your SEM.
So, what happens here is that I've got a list of, you know,
some common materials and the E2 point that you would want
to match on your SEM.
The idea here is that if you match this point exactly,
your sample will emit just as many electrons
as you insert into it.
So, your insulator becomes essentially a metal,
effectively a metal.
So, the only downside to this is
that if you're using a thermionic source, okay,
this is right at the point
where chromatic aberration totally takes over.
So, in a thermionic source, you do not want
to E2 match at any point.
But, if you're working on a cold field emitter or a shocky,
a good chance you can do this and not have
to code your precious sample that you want to use
for other types of analysis
or if you just really don't feel like it.
The last method is to use a variable pressure
or an environmental SEM.
So, we talked about before, remember, you want a vacuum,
okay, inside your SEM.
That's true up to a point.
So, there's a major advantage
or a major technological advance in the past, probably 15,
20 years or so is the development
of variable pressure or full ESEM based systems.
And, what they do is they isolate the gun and just
in the chamber will introduce a gas,
typically air or water, all right.
And, what happens here is the primary electrons will ionize
this gas that you introduce.
It will create charged particles.
Those charged particles will dynamically compensate the
charging that builds up on your surface.
Hence, everything will once again image very similar
to a metal, okay.
So, this is good for samples that cannot be coded.
Typically, a VP SEM is going to run
around the two-tor pressure.
A couple of examples here.
These were taken in our 6060.
So, I found this handsome specimen sitting
on a window sill somewhere.
Stuck him in the SEM, fairly decent image here.
I think he's a mosquito, but not sure.
Anyway, not a big deal.
Next one, this was a plant, okay.
This came off of a magnolia in my backyard.
I find a lot of wacky stuff to throw into the SEMs.
But, these are small stoma, okay, on the back of the plant.
And, just once again, none of these are coded, okay.
They just go directly into the SEM.
Turn on the variable pressure.
Put a little bit of gas into the chamber.
Start imaging, and everything images just the way you'd
expect it to.
Now, the next part of it is that you've got a VP SEM.
We also have full, what's called environmental or E-SEMs, all right.
The major advantage of an E-SEM is that they always will come
with a cold or heating stage.
Not always, but typically.
And, they can operate at very high pressures.
So, two torr, less than two torr for a VP.
For E-SEM, up to 20 torr inside of the chamber.
What this allows us to do is create a 100% relative humidity
condition, okay, inside of the chamber.
So, for things that are very hydration sensitive,
hydration sensitive, okay, this is the scope for you.
To the point that we can do things
like image water inside of the microscope.
So, this is some work that I did as a post-doc.
And, we were trying to form water droplets and move them
around and stuff inside the SEM.
And, do some chemistry with them.
So, we formed these directly inside the SEM,
not inject them or anything else.
So, what SEMs do we have at MRL?
We have four, okay.
The Joel 6060, which is I'm in charge of, is a thermionic source.
Okay, roughly 10 nanometer resolution.
Thermionics, not quite so good for the res.
Really good for things like geological samples,
chip imaging, powder soils, most geological type stuff.
Very good currents for EDS, chemical analysis and mapping.
Has a backscatter detector.
Has the low vacuum variable pressure mode.
That's the major advantage of this instrument.
The Hitachi 4700 is our other trainer system.
Cold field emission source, very good resolution.
Has an EDS detector and a backscatter detector.
Good for particles and anything
with really fine surface detail.
This is the SEM of choice for that.
We have two advanced SEMs in the center.
The first one is a Joel 7000.
Has a shocky source.
Very good resolution.
And, this is our quantification or chemical SEM.
It's got EDS, EBSD, CL, WDS, everything you got on this system.
Now, as opposed to that, we have the Hitachi 4800.
It's got none of that.
This is our straight,
you want the best resolution possible at the center.
4800 is the microscope for you.
It still has several of the electron beam detectors,
but doesn't have any of the chemical analysis based stuff.
You can get to these SEMs once you've spent some time
on the 6600 or 4700.
Demonstrate that you are a very good user.
That's not going to break our microscopes.
And, you get to use the fun ones.
Now, where does SEM fit kind of in the realm
of all the techniques that you've been hearing about today?
I tend to think that the SEM is sort of a,
is mostly just a versatile system.
That's its main advantage.
Gets you information quickly.
Doesn't need a lot of fuss for your sample prep.
It's sort of your jack of all trades.
So, resolution, very good resolution
over a wide magnification range.
But, in the ultimate aspect of resolution,
not as good as a surface probe technique or the TEM,
which is what's going to be coming up next.
For element identification, it's a really good
for what is it, where is it on your sample,
through EDS and WDS.
But, in terms of knowing exactly what the composition
of your sample is, maybe not the best technique for it.
It can tell you approximately what your composition is.
You can do better with standards.
Okay? You tend to have a difficult time
with low Z materials.
You might be better off with a complementary technique
like RBS, XPS, or Auger.
Okay? Those things might be better for you.
If you want better composition analysis.
Surface orientation or texture can be done with EBSD.
But, once again, does not give phase.
Go talk to Maura.
Maura wants to talk to you about XRD
and have everyone use XRD and have that be the only thing
that we ever use in the center.
So, all right.
Now, the fun part.
The focused ion beam, my favorite.
Think of the FIB like this.
If you've ever played pool, if you've ever bowled, okay,
you've got a projectile.
You're going to fire that projectile
at other poor things that are going to get knocked around.
So, you saw the Sims talk this morning by Tim,
which was really nice.
And, he had those images where he's showing you the sputter
of all the atoms that go flying all over the place.
Same principle.
So, we're going to take an ion.
We're going to accelerate it at our sample.
Okay? Take a focused beam here.
And, we're going to mill our sample and cut it to pieces.
So, we can use this as a cutting tool.
We can use it as an imaging tool.
How does it work?
We have several components that are a little bit different
from your typical SEM.
All right?
First of all, the source is different.
It's called a liquid metal ion source.
It looks like a shocky source, okay,
except that it has this reservoir of gallium on it.
And, what we do is we heat a little bit of this.
The gallium wicks onto the tip.
And then, we use a set of lenses to extract the gallium
from the sharp tip, okay, and then send that down the column.
Now, instead of using an electromagnetic lens,
we use an electrostatic lens for a FIB.
The reason being is that since the ion is so massive, okay,
the velocity doesn't really change as much
with the acceleration voltage.
So, the V cross B term of the Rem's force isn't going
to do much for you.
But, the charge aspect of the particle moving
through the magnetic field will do a lot for you.
So, regardless of the mass of the ion, okay,
we can use an electrostatic lens, okay, to change the focus.
This is just a good example
of a typical electrostatic or ion cell type lens.
Path comes in.
We have, you know, some difference
between these individual plates.
And, that will give us focusing action on the ion beam.
All right, what does it look like inside the chamber?
And, kind of what's the orientation of the things
that are pertinent here?
So, we notice that it gets kind of cluttered
and crowded inside the chamber.
We've got an electron column, okay,
on our FIB and dual beam systems.
We've got a FIB column that's sitting
at 52 degrees off axis.
We've got injection needles, our specimen stage over here,
and then a variety of detectors.
So, we've got a lot of stuff kind of sitting
around this sort of central point here.
If we kind of simplify this a little bit more, if we look
at a rough schematic of what's happening, our electron beam
and our ion beam are designed to converge around a point.
So, that regardless of what tilting we use on our sample,
the point that you're looking at is not going
to move in the tilt plane.
Okay, and the reason why we want this is
because we want the electron beam and the ion beam to look
at the exact same point so we can do things to our sample.
Now, why do you want to tilt?
In this case, if you're using your ion beam to cut,
if your sample's at zero degrees,
you're not cutting on an angle, okay?
If your sample is now tilted to the axis of the ion beam,
you're now cutting perpendicularly into your sample.
So, this is good for things like cross-sections.
The last thing that the focused ion beam is designed
to do is to deposit materials, okay?
This is called beam-induced deposition.
Sometimes you can think of this as a localized CBD-type process
where you have a gas injector, a needle.
It's going to have a precursor gas flows through that needle.
All right, the ion beam is going to hit the surface, okay,
where that gas is generated.
It deposits or breaks down these molecules right
in the vicinity here, and you get some part
of the precursor that stays and some part of it
that flies away into the vacuum system.
So, we can build things.
We can mill things, a lot of stuff that it can do.
This is just an example here of, here's our platinum needle.
Here's our surface.
This is very close.
So, the distance between the needle and the substrate,
50 to 100 microns or so, typically.
So, that's a lot of fun
when someone smashes the sample into the needle.
I don't really like that.
I've got to be very careful when using fibs.
So, what can it do?
It can build things.
It can destroy things, and it can do that stuff on a variety
of different length scales, okay,
and it can do things very accurately.
All right, so you see the larger ones here, around 3 micron,
4 micron scale bar, and then we have some
of the smaller matched images right
into about the 500 nanometer range.
So, each of these lines is sitting
around 50 nanometers or so.
I particularly like this one.
That's a lot of fun.
This was done on our Helios system, but really,
if you're not playing with it,
what do you really use a fib for?
Cutting through materials and seeing cross sections.
So, if you want to know the exact thickness
of a deposited layer, okay, this is what fib is good for.
In the IC industry, if we're cutting through and we want
to look at the chips, okay, look at individual layers,
see if things have defects, if they're connected, disconnected.
One of the major early applications of fib was
for the semiconductor and chip analysis.
Another good example here, this is something
from the Rogers group where they wanted
to know what these internal lines look like.
I tend to feel that sometimes these are Kit Kats, but, yeah,
not sure, so, yeah, pretty small for a Kit Kat,
not very satisfying, but this also demonstrates
if the length scale here, I think that's 50 microns,
yeah, 50 microns.
So, you can make some fairly big trenches, okay,
in your materials.
Some other examples, controlled etching and deposition.
So, everyone that's got a fib is always going
to put their, you know, their center name,
the university logo, but you can do a lot
of other fun stuff, too.
So, we have dot arrays that can be used
as like other masks or emitters.
This is, I don't remember who did this, per se, but, you know,
platinum flower that they built.
Now, you can do a lot of other cool stuff, too,
and the fib can be used for sort of functional things
and maybe not so functional things.
So, first example, functional,
this was a platinum basket that was created to dip
into a microfluidic stream and to pull out individual cells.
All right, so, this was the kind of test that they were doing
where they had styrene spheres
in a microfluidic stream, dip that in, okay,
and grab them and take it out.
So, these are, fib's really good for designing kind
of one-off, you know, structures to see, well, what can we do?
Now, a fib can also be used for things
that are a little different.
So, this is the not so functional 1 billion scale
Starship Enterprise, which I guess they didn't really do a
very good job with this part here,
a little bit of damage to the system.
But, anyway, you can do a lot of really cool stuff, all right.
If you've seen some of the news recently,
MIT's got a really nice program where they do kind
of fib sculptures, they make like sandcastles,
and so it's pretty neat, check it out.
Okay, now, the last major thing that a fib is designed
to do is make TEM samples, and you're going to hear a lot more
about that in one of the next sessions.
All right, I'm going to bring you through kind
of the major steps that you need to make a TEM sample.
So, the first one is locate something
that you're interested in making a TEM sample out of.
We put down a platinum protection layer on that spot
that we're interested in, okay, then we're going to mill
out the outsides around that protection layer,
thin it down a little bit.
Okay, this is the electron beam view, so we now have kind
of a thin, like, wedge sitting in here.
Okay, then we're going to bring in a needle, okay,
we're going to weld it using the platinum to that wedge,
we're going to make some relief cuts in here, okay,
we're going to slice this out, go, there we go, and then we lift
out the small part, okay, weld it to a grid,
and then we thin it down,
so we can see all the internal structure,
and you'll see why this is really important during the
TEM talk, but this lets you basically take
out very small portions of your sample, okay,
and investigate the, you know, the interfaces, you know,
between individual layers, and do a lot of really cool stuff
that Vasek and CQ will tell you about in the next section.
Some other really cool stuff you can do, serial slicing
and 3D reconstruction, so you can take the fib, you can slice
out, you know, individual parts of this,
and then reconstruct all the data to show you things
like the pore sizes and the shell structure.
You can also use this to do things
like elemental reconstruction,
or say channeling contrast reconstruction,
or more appropriately, like EBSD reconstruction,
if you want to know, say, the crystallographic orientation
of your structure in a three-dimensional way.
So, fibs are capable of doing that.
We have some limited capabilities on this,
but if you're interested, please talk to us.
So, we have two fibs in the center.
First one is our FEI DB235, has seven nanometer resolution
at 30 kV, has pretty good beam currents, has an omni-probe
for TEM lift-out, has a Zyvex multi-probe station
for doing in-situ electrical characterization,
has three GIS sources, so we have a platinum source,
we have a carbon source, and we have a selective carbon mill.
So, if you have things like you want to cut
through heavy plastics,
or other carbon-containing structures like diamond graphite,
that's what you would use this selective carbon milling for.
Our other system is a Helio 600i.
This is not ours, it's a stock image.
If you've seen on the tours,
ours is a little torn apart right now.
But it has a little bit better resolution,
30 kV on the ion beam, very high currents are obtainable.
The major advantage here is
that we have some advanced milling patterning capabilities
which aren't available on our 235.
Also has a charge neutralizer, omni-probe for TEM lift-out,
so that's its major use, has a platinum GIS,
and it also has a retractable stem
and BSC detector all inside the system.
So, to wrap this up, fib pros and cons.
The fib is really good for patterning small areas, okay,
one-off fabrication for functional devices,
cross-sections, and precision TEM samples.
What the fib is not good for is it's not going
to replace your E-beam or optical lithography, all right,
because those things will do large areas.
The fib is not designed for that, okay?
And it will also not make hundreds
of micron-sized TEM samples.
So, occasionally I get that request,
make me a 500-micron TEM sample, and I go, no.
You can try that, have a good time.
So, if you want small, locally precise,
that's what a fib is good for.
Think of it like a scalpel, and that's what it will do for you.
So, with that, I'd like to thank our sponsors very much,
letting us put on, you know, this wonderful AMC,
and at that point, I will wrap up,
and thank you very much for your attention.

