Rutherford Backscattering Spectroscopy (Tim Spila)
Sep 5, 2023
For more information, visit https://nanohub.org/resources/22619
Tim Spila
6/2/15
"Rutherford Backscattering Spectroscopy"
So, let's go back into very deep history, and the ancient Greeks asked a question, and
that was, what would happen if you kept cutting something in half each time, right?
So this is similar to the idea of Atlas holding up the world, right?
So Atlas is holding up the world, but then you ask the question, what's Atlas standing
on?
It turns out Atlas is standing on a pair of elephants.
Okay, but what are the elephants standing on?
Well, the elephants are standing on this gigantic turtle.
Okay, well, what's the turtle standing on?
Now, just, it's turtles all the way down, right?
So that was one of the things the Greeks thought about, but then another one was, okay, what
if you keep cutting material in half and further and further down?
What happens?
And there was one philosopher who said, well, eventually you get down to the smallest particle
that you can without being able to cut in half anymore, and he called that an atom.
He, his ideas were, Aristotle hated his ideas, and so it never really came back up until
much later.
RBS, actually, is one of the techniques that we looked at to figure out what an atom looks
like.
So it came back to the idea of atoms, but then RBS was just trying to determine what
an atom looked like.
So the first idea that was very popular in the 20th century, 1900s, was that an atom
had this positive charge that was just diffuse and everywhere, and then you had little electrons
through it, and they called that the Plum Pudding Model, because I guess that's what
a plum pudding looks like.
But then in 1909, Hans Geiger and Ernst Marston, under the direction of Ernst Rutherford, did
a little experiment by which they put a foil behind a sample, and they sent alpha particles
through it, and as expected, the alpha particles made it all the way through and showed up
on the phosphorus foil.
But then he sent his graduate student off and said, okay, so now I'm going to have you
do the experiment that shouldn't work, and they put the foil behind, on the same side
as the source, and they got signal there.
And so like, well, that doesn't make any sense.
The alpha particles are massive compared to the diffuse masses there, it shouldn't work.
And so what they found out was that their idea was wrong.
And so they came up with the idea of a small, heavy nucleus surrounded by an electron cloud.
And their first idea was like a planetary model.
Of course, as we go more and more through the 20th century, our science figures out
more and more quantum, and so we know how the electrons are moving, or actually we don't,
I guess, as the case may be.
Now, of note with this picture, you know, this isn't even accurate, because this is
way too big for a nucleus.
The nucleus is very, very small compared to the size of an atom, with the electrons.
About ten to the minus fourth of the volume.
So, RBS, if you want your big definition, so RBS is the analytical technique where high
energy ions, and in this case we're talking about two MeV, whenever someone says high
energy, you always ask them, well, what energy?
And are scattered from atomic nuclei in a sample.
And then the energy of the backscattered ions are measured to give information on the sample
composition as a function of sample depth.
Now, most people, one of the things we use to generate these high energy ions is a dendrographic
accelerator, and most people's first experience with a dendrographic is something like this,
where you go to the science fair, you put your hand on the dendrographic accelerator,
and the hair stands up.
I guess it works better if you have long hair that's kind of dry.
No?
Okay.
I don't know.
I've never had long hair.
At least not that long.
But how does it work?
It's actually a mechanical device where you have, you deposit charge on a non-conductive
belt that spins around, and so the charge, because it's non-conductive, the charge just
sits on the surface, it doesn't move.
And so then you send it up here and you deposit it on the, up here on the, in this case it's
a sphere, and that's a conductor.
So you just keep dumping more and more charge up there and you can build up more and more
potential, more electrons go up there, positive charge in this case, and eventually you have
to have some way where it's going to come off.
And so what we do is we'll put that into a helium atom, and then use that to accelerate
the helium atom, and that'll be our source of ions that we're going to do our measurement
with.
This is the system that we have.
It's actually, that Van der Graaff accelerator is inside this pressure vessel.
The pressure vessel sits at about 14 atmospheres, so that we don't have breakdown.
If I remember correctly, it's 30,000 volts per inch is what can be sustained in air.
So if we go to a higher pressure, we can sustain more.
And this is designed to be able to run up to 3 MeV, but if we're doing runs for backscattering
we run it at 2 MeV.
And the beam size tends to be about 1 to 3 millimeters.
The downside of the technique is it really doesn't work well unless you have a flat sample.
If your sample is very particular on what it can handle, you can rotate the sample so
you're hitting different parts of it.
Generally it can be considered non-destructive, but if you have very soft samples, polymers
and that, it can scorch the surface.
So how does it work?
So if you're doing sputtering, which we'll talk about later, you're actually working
down here in these types of energies.
And so when you're doing that, you actually will see the electron clouds of atoms.
And so it looks very dense when you come in and you knock things out.
But with RBS, we're talking about working in the MeV range.
And so in that case, we're only looking at nuclei.
And so it's a lot of open space.
So lots of area for ions to come in, go deep into a sample, knock off a nuclei, and come back.
The effect of the electrons does...the electrons do have an effect.
They actually produce like a drag force on the ions moving through, and we'll talk about
that a little bit more later.
So of course you have the drag effect, but we also...the main effect that we talk about
is the atomic nuclei interacting.
And that can be thought very simply as an elastic two-body collision.
And so the math for that's all well worked out.
You know, you've got your conservation of energy and conservation of momentum.
So if an instant ion coming in hits a target, the instant ion gets scattered, the target
gets recoiled.
And so the energy of the scattered atom is related to the masses that are involved.
And so we can actually define a constant here known as the kinematic factor, which has all
these fun terms, but if you plot it out, it looks like this.
And from this graph, we can see some interesting things.
First off, the kinematic factor is high if you have a large target mass from helium.
And so that means that we have good sensitivity.
We can measure lower concentrations.
But down here at lower masses, it changes very quickly.
And so that tells us that we actually...at lower masses, we actually have good...the
ability to tell what we're looking at.
So that would be good selectivity.
If you do some more fancy math, which I'm not going to do now, you can actually figure
out...we have a Coulomb interaction between the things, between the nuclei, so we can
make an exact expression.
Thus, we have a quantitative method here.
We can define the scattering cross-section that is a function of energy and angle and
get this result, which is the atomic number divided by the energy squared.
So the scattering cross-section is proportional to that.
And so if we were to do a measurement, you can kind of do a conceptual measurement.
So here's our energies and our...here.
And we have like a platinum-silicon pseudo-alloy, so just 50% of each.
When you have your helium coming in, it'll scatter off of an atom somewhere in the sample.
If it scatters off the front surface, depends on what it scatters off.
If it scatters off of platinum, it's only going to drop a little bit.
If it scatters off of silicon, it'll drop a lot.
And so you'll get some...a detection with a certain energy, and that would be here or
here, depending on what it's scattered off.
Now, as it moves through, I'm talking about like drag on the ions, on the helium, it basically
lowers the energy a little bit until where it scatters.
So if it came all the way through and scattered off of platinum, it's going to drop and then
come all the way back through and scatter, and that'll give you a reading of this energy.
So if it's somewhere in the middle, it scatters and comes back here.
Note that the amount that it drops is dependent on what it's scattering off of.
So if you were to come all the way through and scatter off of silicon, it would drop
to a much lower energy and then come back up.
So the width of these two can give you information on how thick you are, and how much it scatters
can give you information on what you're looking at.
So this is an example of some simulated data.
So we made another hypothetical alloy.
It's got gold, 20% gold, 20% indium, 20% titanium, 20% aluminum, 20% oxygen, one carbon, carbon
substrate.
These were picked kind of because you're jumping up, kind of by doubling the atomic number.
So, here's our graph of the kinematic factor again.
Again, you have high sensitivity at large Z and high selectivity at low Z.
So, if I read the energy, the RBS spectra, this is what one looks like.
If you look at it from high energy going down to low energy, the gold scatters first, and
it's very sensitive to it, so we have lots of signal.
But, we don't have good selectivity, so even though we're dropping by about half the Z
number to indium, it's very close together.
So now it's got lower, so it's dropping in signal, even though it's the same amount in
the alloy.
And then as we go to titanium, it drops lower, aluminum drops lower, oxygen drops lower.
Carbon's higher because it's 100% and it's a substrate.
So now, but you can see we have wider spacing between the oxygen and aluminum, so it's easier
to determine what we're looking at.
So now, if we increase the thickness of our film from 10 modulators to 100, you'll see
our total counts go up, so we have more counts of everything, and we have the gold net relative
to the substrate, because we're shielding more of the substrate, we don't get to see
as much of the substrate.
If we go even thicker, now to 1000, you'll see that we've actually got some width to
these peaks, so from that you can actually get a thickness of your layer.
And here in the indium and gold, because those were the two that were closest together to
start with, they're actually merging together, and it's just a summation.
So here's the signal from indium, here's the signal from gold, and this is the signal from
the indium plus gold.
The two common programs that are used for stimulation and fitting of the RBS spectrum
are SIM-NRA and ROM.
That's an example of what the SIM-NRA will look like if you're trying to define your
experimental setup and your target, your sample.
The analysis of the data requires good information on the scattering cross-section as a function
of angle for the different elements, but since these have all been well measured, and we
have good data for that, we can actually get quantitative results from our simulations.
The key is to know a little bit of something about your sample, especially if you have
high Z, because it's hard to determine what you're looking at.
So, different important fitting parameters that are needed are the instant and exit angle
of your sample, the geometry of your setup basically, energy calibration and offset,
so that tells you how your detector, information on your detector.
And finally, you have to define your layer thicknesses and concentrations.
This is basically an alloy, so it's just one layer, but from that then, we can fit
the data by modifying this, or have a computer do it, and then we can get the information
out of it.
So in this case, we have 20% half PM, 1.8% zirconium, 13% aluminum, 50% nitrogen, and
7% oxygen.
Again, going back to that kinematic factor, this is a huge height and peak, 27%, that's
great.
It's good to have something you expect to be large.
This is 1.8%, okay, we expect that to be small, but this is 13% aluminum, it's smaller than
the zirconium.
Of course, it's also a much lower Z, and so that's why it's going to be smaller.
To go a little bit deeper into thickness effects, so of course when you're fitting this, so
for like a 300 nanometer film of titanium nitride on MgO, on magnesium oxide, the surface
comes up at the value that you'd expect, and then you should also see nitrogen from the
surface, so you'll have a peak there.
Of course, the magnesium's bigger than the nitrogen, bigger Z than the nitrogen, so it's
going to show up first, even though it's at the interface.
And then this is from the oxygen.
If we go thicker, what does that do?
You kind of think of it as the film growing up, but effectively when we're fitting it,
we're pushing the substrate down further.
And so you still have the titanium at the surface, it's the same energy, and the nitrogen
at the surface is the same energy, but now we've pushed the substrate further down such
that the magnesium at the interface and the nitrogen from the surface actually have the
same energy.
And if you go even thicker, then you get the nitrogen, and then the magnesium comes in
later.
But this width here from the surface to the interface, you can get information on how
thick your film is.
So, if you have a problem like the one where we had where the nitrogen from the surface
and the magnesium from the interface are both showing up at the same spot, what you can
do is you can change your scattering angle from this one, which would be the standard
setup, to one that's more surface sensitive.
So that means that actually the helium has to travel, alpha particle has to travel a
longer distance to get out of the film.
It has to travel more, so there's more effect from the drying force.
And so then we can actually resolve the nitrogen from the surface, from the magnesium, from
the interface.
Another real life example, this is a multi-layer of titanium aluminum nitride and titanium
niobium nitride.
So you can see we get good signal from the niobium, high Z, relative to the effect that
we get from the aluminum and the titanium.
Nitrogen, barely see, even though that's 50%.
Downside with the way this was measured, it's a multi-layer, so we don't see the individual
layers because they're very, very small.
So you can have that problem with using RBS if you're looking at very thin layers.
It is good for measuring trace amounts on a surface though, but if it's a buried layer,
you're going to have more trouble.
This is an interesting experiment that was done and that used the RBS.
So we have a titanium nitride on silicon dioxide.
As deposited, it looks like this, so this is a nice titanium signal.
Here's the signal from silicon, signal from oxygen here, nitrogen's back up this way.
So now what they did was they took the sample out, they oxidized it, put it back in.
So what happens is the titanium actually decreases at the surface, and that makes sense because
you're going from Ti one-half of the unit cell to one-third Ti versus, TiN versus TiO2.
And then we also pick up some signal here from the oxygen that's in the film at the
surface.
It goes back down and has to come back up later when it gets to the silicon dioxide.
And so you see both of these coming up, and from the measurement you can actually see
how efficient you were at changing the titanium nitride into titanium oxide.
So just to summarize, here, RBS, we have an incident ion coming in, helium coming in,
and we measure it coming back out, the scattered helium.
It's a quantitative technique for elemental composition.
Requires flat samples, beam size is about one to three millimeters.
It's non-destructive.
The detection limit varies from 0.1 to 10 to the minus 6, so 10% to 10 to the minus
6, very dependent on the Z of the element.
So therefore, it's very good at measuring heavy elements on a light matrix.
So, for instance, when I was a graduate student, we looked at germanium and silicon, because
the germanium is heavier than the silicon.
And then you can get depth information from monolayers up to microns.
And finally, thanks to our sponsors.

