X-ray Photoelectron Spectroscopy (XPS) and Auger Electron Spectroscopy (AES) Part 1 (Richard Haasch)
Sep 5, 2023
For more information, visit https://nanohub.org/resources/22627
Richard Haasch
6/3/16
"X-ray Photoelectron Spectroscopy (XPS) and Auger Electron Spectroscopy (AES)" Part 1
I'd like to start first by repeating what Mara mentioned yesterday, that last year we
lost a dear colleague, Nancy Finnegan. And so she was actually in charge of the OJ system,
and I took over that role when she retired. And so I'd just like to mention one thing
about this picture. It's really nice that if we could all wander around the back of
the instrument and look at the user she's working with right here, you'd notice he'd
have a really, has a really huge smile. And that's what it was like to work with Nancy.
She's a complete pleasure and a really great person to work with. We're glad we got to
work with her. Okay, so now let's start talking about surfaces and interfaces. The studies
of surfaces and interfaces actually impact almost every aspect of our life, from transportation,
energy, technology, and even some entertainment. And so what exactly happens in surface analysis?
You can do surface analysis and surface science in a variety of different ways. Usually we'll
have an energetic ion source or electron source or photon source. So we'll have some primary
beam. That beam interacts with your material, and then we'll generate a secondary beam.
And any of these particles can generate any of these other particles. And so what is a
surface? It actually depends on who you talk to, what defines a surface, okay? It could
be anywhere from thousands of nanometers down to a few atoms. And so yesterday we saw Tim
introduce to us this end of the scale, okay, with Rutherford backscattering. And we could
go down roughly about this length, maybe get over here. Then next Scott McLaren went
all the way to this other edge, and we're looking with atoms. And so he was using a
different approach. Tim had an energetic particle. Scott actually had a tip that probed the surface.
And so next, after my talk, Tim is going to tell us how we can actually get from this
regime all the way back to this regime with another energetic ion. For now, though, we're
going to focus in this area with the electrons. And so because the electron, the energies
that we'll be looking at has a wavelength that's very similar to the lattice spacing
of your materials, you have a lot of opportunities for those electrons to interact and interrogate
the material. And from that, you can actually extract a large amount of information. It's
very, very rich. Okay, so on the Evans bubble chart, this is the analytical spot size versus
detection range. So XPS and Auger range from about the millimeter down to the submicron,
with about a tenth to an atomic percent sensitivity. CIMS winds up in the middle and is capable
of having a little bit more sensitivity than the XPS and Auger. Okay, so let's start with
photoelectron spectroscopy. So we'll be using photons as our primary particle, electrons
as our emitted secondary particle. And so Maura already mentioned yesterday, one of
the drawbacks of surface science and electron spectroscopy is that they usually require
a vacuum. And in many cases, it's ultra-high vacuum. So not all samples are compatible
with this sort of environment. Okay, so X-ray photoelectron spectroscopy is also known as
ESTA, electron spectroscopy for chemical analysis. And it's a widely used technique based on
the photoelectric effect discovered by Hertz and then later described by Einstein. And
this was back in the early, late 1800s, early 1900s. And it took until the 1950s for this
technology to turn into a practical analytical tool that is here. And so that work was done
by Professor Uppsala Sweden Kajsikvand. Okay, so in the XPS experiment, we'll irradiate
the sample with an energetic X-ray. That X-ray can penetrate the material to a depth of about
a micron. And then we'll be extracting electrons from some smaller area within that excited
volume. So both the systems that we have here in the facilities employ a static X-ray beam.
There are instruments available in the market today that can actually offer a focused probe
X-ray that scans. We have a static beam. Okay, so in the XPS experiment, photon comes in,
it's fully absorbed by your material, and that energy can be used to excite not only
valence level electrons, but core level. So mostly when we're in the XPS technique with
an X-ray as a photon, we'll be looking at the core levels. Now following this process
of ejecting the photoelectron, we'll have an ion that's in an excited state. So one
of the ways the ion can relax is by having an upper shell electron fall and fill that
hole and have the Auger electron emitted, conserving energy. The complementary process
that can take place is the emission of an X-ray photon. Now this will be the basis of
another technique you'll see later on this afternoon known as EDS. So in the case of
the photoelectron, this photoelectron will come out with a kinetic energy that's equal
to how much energy we put in, which is a known quantity, minus how tightly it is held, which
is usually what we're interested in finding. So we'll solve this equation for binding energy,
and this binding energy is characteristic depending on Z. Now the Auger process, because
it's a conservation of energy process, we'll be looking at the kinetic energy of the electrons
that's representative of the differences in the orbitals that are involved in the process.
So what does that all mean? So let's say our material is a city, and our atoms are
different buildings of different sizes. Now being good electron spectroscopists, we're
not going to look at the city this way. We're going to look at it this way. And that's so
that we can define a convenient energy scale that is representative of the chemical potential
of an electron at a surface, and this can be defined according to temperature. We'll
call that level the Fermi level. So the binding energy then will be the depth of each of these
individual levels with respect to the Fermi level. In the case with Auger spectroscopy,
what we'd be looking at is just basically the differences between the levels. How deep
we can look depends upon the energy of the source that we use. So primarily in XPS, we'll
be using soft X-rays of either magnesium or aluminum K-alpha, so they go from 1200 to
1400 eV. Okay, I'll show you a trick in the next half of the talk, a little bit later
this morning, that'll show how we can extend this range a little bit more. So with the
X-ray photon, we'll not be able to access all of the orbitals of the heaviest elements.
And so for ultraviolet photoelectron spectroscopy, where we use a UV photon, we can access even
fewer orbitals. In this case, we'll be concentrating mostly in electrons that reside within the
valence band. Okay? And so because of the wavelength of the electron matches quite nicely
with the lattice spacing of your materials, there's a lot of chances for interaction.
And that interaction with the electron exiting your sample with the material is often referred
to as elastic, inelastic mean-free path, which is the mean distance an electron can
travel between inelastic scattering events. And you can see right here that this follows
a general trend of increasing with kinetic energy, but if you look really closely, there's
no real relationship with the element that's emitting the electrons, or the material that
the electron is traveling through. That's because the inelastic mean-free path is also
related to the electronic structure of the material. So it's sensitive mainly to the
number of valence electrons that is in that element. Because you can see up here, cesium
and lithium are actually pretty close, but they're very, very different in mass. And
so this is about the range of energies we'll be accessing with XPS. Because OSA usually
involves an energetic source that's about twice the kinetic energy of the X-ray photon,
we can access a little bit higher energy OSA lines. But the bottom line here is electrons
only travel a few nanometers through a solid. And that distance can be described by an equation
known as the Beer-Lambert relationship, which describes the intensity of the electrons being
emitted from your solid as an exponential decay as a function of depth. And so if we
look at I over I0 down to values of 5%, we'll call this the 95% level. And this is usually
what people define as the analysis depth of XPS. Of course it's infinite, but we have
to cut it off somewhere so that it makes it a little bit easier to describe what you're
looking at. So the bottom line is 95% of the signal will come from within 5, well I'll
call it from 5 to 10 nanometers of the surface of us. So what does that mean? What is a nanometer?
It's kind of hard for a lot of people to get their hands around the value of a nanometer.
And so it's good to have a common unit of measure. That's a foot-long sub. So I'm sure
you all recall a few years back there was a big controversy in the news where people
were starting to sue Subway because their foot-long subs weren't coming out exactly
a foot long. And then Subway came back with a statement that said basically, guys relax,
it's a sandwich, not a ruler. But if it were exactly a foot long, we could actually cut
that sub into roughly 305 million slices, which is a pretty small amount of material.
And so let's quickly just summarize a little bit some of the advantages and disadvantages
of the electron spectroscopies. First advantage is extremely surface sensitive. So what's
its main disadvantage? It's extremely surface sensitive. Sometimes it's a little bit too
surface sensitive and just the interaction of the material with the environment around
it can alter the nature of the surface enough that what we're looking at initially is not
actually what we're interested in studying. And in those cases, we can combine the XPS
experiment with a technique known as ion sputtering, where we'll direct an energetic ion beam roughly
of the half to 5 keV range. And that ion will impart momentum to the surface. And by a collision,
a series of collisions, a collision cascade will develop where that momentum is eventually
redirected towards the surface and then an atom from the surface is ejected. Sort of
like peeling away layers of an onion, for example. Unfortunately, it's not quite that
simple. With a single ion beam of these energies, there is a fair amount of interaction with
the solid and you will do a little bit of physical damage. And I'll show you a couple
of examples of that a little bit later. Okay, here's an example of an X-ray photoelectron
spectrometer. You start with an X-ray source that irradiates the sample here. And you can
think about the rest of this system as basically a part of an electron microscope, where we're
going to project an image of the electrons to the plane right here. We call that the
selected aperture plane, selected area aperture plane. And there is where we define the analysis
area of the electron, of the sample, I should say. In the next plane right here, we'll call
the entrance aperture of the analyzer. And in this case, for spectroscopy, we'll close
that aperture down to a very narrow slit. So then when we project the image of the detection
plane, we'll be doing a transform. So we'll basically be transforming a spatially dispersed
image here to an energy dispersed image here, or a spectrum. And then we'll scan the binding
energy scale to analyze our elements. And this is what the instrument looks like in
real life. X-ray sources here. Sample sits inside here. This analyzer is this dome in
the back right there. Okay, so in XPS, we're looking at the binding energy of the electron
with the nucleus. And so we're actually looking at this interaction right here. And so for
the most part, if we follow along a certain core level, okay, the trend is that the binding
energies for that element will, for those elements, will increase according to atomic
number. And we can see that in this chart right here. Initially, for the 1s's, you can
see the changes are quite large. And then they decrease as you go higher and higher
in core level. And that's sort of repeated here. So you'll notice the general trend for
the 2p orbitals for the first row transition metals. And then you'll also notice that the
3p's shift. But not only the 2p's and the 3p's shift, they do shift with respect to
one another. And this is how we're able to actually use Auger. So the Auger differences
will also be characteristic of the elements that we're looking for. Okay, here's an example
of some survey spectra from a series of first row transition metal nitrides. And you'll
see the common element in all of these is nitrogen. So this peak and this energy scale
appears to be stationary. The core levels, as you go up in atomic number, move deeper.
Okay, you'll also notice that the Auger lines shift the lower binding energies. That's because
Auger lines, we look at those usually on a kinetic energy scale. And the way we define
the binding energy, the binding energy is decreased in this direction, while the kinetic
energy is increased in that same direction. So for the kinetic energy scale, zero is over
here. Binding energy scale, zero is over here. So this shift towards higher binding and lower
binding energies actually represents an increase in kinetic energy. Okay, so as you can see
allows us to identify different elements. Now, if we were able to use a fairly high
energy resolution analyzer, we could look for very small shifts in the binding energies
of the electron. And then we can use that information and relate that to changes in
chemical environment. Okay, we call that a chemical shift. And so chemical shifts in
the simplest sense can be thought of as due to electronegativity of the bond. Okay, so
in this case, we're looking at a carbon-oxygen bond. Oxygen is slightly more electronegative
than carbon. So it's going to attract electron density away from the carbon atom. And that's
going to decrease the amount of nuclear screening for the carbon nucleus. And so that's going
to result in having the 1s electrons pulled in a little bit more tighter to balance the
electron-neutron, electron-proton charge. And that's how we'll actually, the physics
that we'll use to relate the binding energy to the chemical state. So here's an example
of an expansive polymethylmethacrylate. You'll see in the survey spectrum, we'll see a carbon
and oxygen signal identifying the elements. But if we look really closely, we do notice
that the oxygen line is split into two states. That's because polymethylmethacrylate actually
occurs in a hybrid resonance structure where one of the oxygens will have a slightly positive
charge with respect to the other one. And that splits the two lines. Also, there's four
different types of carbons you'll see that you'll split the line into four states. Unfortunately
though, XPS does not allow you a very long range of information. We're only usually sensitive
to the nearest neighbors. These are the direct bond or the neighboring atom bond. And so
we'll get small pieces of information. So without having additional information, I may
or may not be able to identify that as specifically polymethylmethacrylate. Usually we'll have
to rely upon some other information. So this is the bottom line of this whole workshop
basically is complementary methods in a study are usually always essential.
So here's an example of the use of chemical state analysis. And this is some work done
with John Rogers' group here at MRL in the study of a new class of materials called transient
electronics. And these were a study of different types of metals that dissolve. So this is
the idea behind these that have some metals and electronic circuitry that you can implant
in a body and it will dissolve with time. And so XPS was a convenient method to use
in order to solve the dissolution chemistry. So we can look at the bonding energy shifts
and watch the changes of the oxidation state in tungsten as it goes to higher oxidation
states and eventually forms a soluble ion and then dissolves away from the electrode.
Okay here's another example of the use of chemical shifts. But in this case, the shift
in the nitrogen is not entirely due to the oxidation state. In all these cases, nitrogen
should not only be minus three in these metal nitrides, but there is still a shift between
the two. And that's because the bonding energy is actually related to the degree of the electron
transfer between the metal and the nitrogen. Okay, you'll notice that scandium nitride
has the lowest bonding energy of all the species. And that indicates a more ionic bond. Titanium
nitride has the highest bonding energy, that means it's a little bit more covalent. And
as you remember all your chemistry, all bonds are mixtures of both. So XPS is a nice technique
to study how these electrons are actually interacting between the metal and the nitrogen.
Okay so you've seen a couple of ways we can use XPS, identify elements, identify chemical
states. Now we can also study a little bit of electronic structures. Okay here's another
way you can look at the electronic structures. And so not only is that electron going to
carry information from the element it was ejected from, it's also going to interrogate
the materials traveling through. And so here's a couple of examples of a technique that's
very similar to something you'll see this afternoon known as EELS, Electron Energy Loss
Spectroscopy, that's commonly found on TEMs. In this case, instead of having a primary
source of electrons, we'll think of the silicon 2P as our internal electron source. So that's
going to emit electrons through your sample. Those electrons interact and can excite resonances
known as plasmons. And so we can use that to actually look for small differences in
electronic structures. So here's an example of a silicon with a native oxide as compared
to silicon with a thick oxide. And the structures you can see are very different. So they're
different enough that at a low resolution somebody could show me the spectrum without
any labels and I could tell you which is the native oxide and which is the oxide. Okay
here's another example of a way we can get some additional electronic structure information.
So in most cases, well I should go back. So the cartoon I showed at the beginning shows
the photoelectron process and the orbitals are fairly static. But that's not truly the
case. When the electron leaves, there'll be some coulombic potential that can actually
cause the electrons in that atom and surrounding atoms to relax and rearrange themselves in
energy. In most cases, the atoms will relax producing one final state. In some cases,
you can have several discrete final states. And the case of the first transition of metal
nitrides is one of those instances. And here's an example of some work done a few years ago
where we use that information to actually identify the degree of electron transfer between
a titanium nitride single crystal and a ultra-thin layer of silicon nitride. Okay these final
states that occur in the spectrum are caused by screened final states and unscreened final
states. And we can see that these different conditions of growth of silicon nitride actually
increase the amount of unscreened final states in the titanium signal. And this suggests
that we're actually withdrawing electrons away from the titanium nitride towards the
silicon nitride. So we're losing the capability of those electrons to screen the titanium
2p final state. And you can see as we go through some other films, we can actually suppress
that satellite feature. So by depositing a very, very thin layer of titanium on top of
the titanium nitride, we can actually fully suppress it. So we're now in a case where
we have a fairly fully screened final state. Okay this is one brief example I'll show of
UPS. I'll add one later on too. So here's basically what the valence band spectra will
look like. They're not quite as simple to look at as core levels, but they're useful
in doing electronic structure calculations because you can actually use that to verify
your models. So here's an example of a calculation with the actual physical measurement. Okay
So like most techniques, you'd like to have the capability of the signal that you achieve
be related to the quantity of a material that's in your system. Okay. And that can be done
with XPS. And so with XPS signal, we'll have a signal that relates to electrons per unit
volume and volume. Some of these terms will be sample dependent, others will be instrument
dependent. By assuming though that our concentration that we measure is actually a relative ratio
of atoms, we can neglect the terms that depend only on the instrument. And so that makes
it a lot more convenient to do the calculation. We don't have to do a lot of modeling. And
so we can lump these terms into another term that we call the relative sensitivity factor.
And then so basically we could use this to correct the areas and take the corrected areas
as a percentage of the total corrected area. And so the concentrations you'll get from
XPS are usually referred to in the atomic percent style. And so XPS is considered to
be a semi-quantitative technique. Okay. Here's an example of the relative sensitivity factors.
Our XPS will scale roughly three orders of magnitude. And here's an example of the use
of XPS for quantitative analysis. And you will see for the transition metal nitrides
that we're trying to grow in situ, we were able to get pretty close. And actually in
fairly good agreement with Rutherford backscattering. So as you can see, XPS without any corrections
can be fairly quantitative. One thing I'll notice here, or you should notice here, is
that this second row right here is that same material after atomic or ion bombardment,
ion sputtering that is. And you'll notice right away that the metal, the nitrogen metal
peak ratio decreased after sputtering. That comes from pure kinematics that you saw yesterday
that Tim explained where the lighter elements are preferentially removed from the material
than the heavy elements. And this will cause a change in the stoichiometry. It also will
alter the electronic structure of the material causing the metal to reduce. Okay. Another
nice thing you can use XPS for is we can go back to the Beer-Lambert law and use that
as a model to estimate film thicknesses. So if you take this equation and we assume
only an elastic scattering, we can integrate that. And we'll come up with an equation here
that we can relate to different aspects of our material. And then use that, actually
I'll go back a little bit, use that to determine the thickness of our film. Here's an example
of where that is being studied. So the 2D materials are getting a lot of attention right
now. And here's an example of a layer-by-layer transfer of graphene to a silicon dioxide
surface. And so in this study it was desired to know whether or not the graphene was actually
transferring layer-by-layer. And of course XPS would be a good technique to look at that.
And so with a two-layer model, here's what the data looks like actually. Here's the as-grown
surface in black. And the carbon 1s after each subsequent transfer, you see the graphene
decreases, the silicon carbide substrate increases. And then by using that equation that we derived
earlier, we can estimate the thickness. And so we'll look at how many layers were moved.
And it looks like the layer transfers are not occurring as single sheets at one time.
Okay, let me see. I probably will, I'll introduce one more concept and probably we'll cut it in
half here. Okay, so another thing we can rely upon, take advantage of the fact that electrons
can actually only travel a short distance to the solid, is we could do an experiment known
as angle-resolved XPS, where we vary the sample's angle with respect to the analyzer. And then in
fact we're varying the volume of the sample that we're actually analyzing. So gracing angles,
we're very sensitive to the surface. More normal angle, we're sensitive to bulk. So if we have an
alloy, we should not be able to see any change. If we actually have a layered system, we'll actually
see an increase in signal as we increase the emission angle. And here's an example of silicon
with a native oxide, where we're looking at a zero degree emission and 75 degrees. So you can
see right away that the SiO2 signal is enhanced quite a bit as we go to the low takeoff angles,
so the high emission angles. And with that information, we can actually now go back to
the two-layer model again, and study that as a function of angles, and get a linear relationship
here where you actually determine the thickness of the SiO2 layer. And we find it to be about 1.7
nanometers for the native oxide. I think that is a good place to take a break for the first half of
the talk. So what I'd like to do real quickly is jump to the end and thank our sponsors.

