Analyzing composition with x-rays and electrons

Todd C. Hufnagel
September 2, 2016
We can make use of the fact that atomic energy levels are quantized
to identify the presence, and amounts, of specific atoms in solids. In
this handout we discuss four ways of doing so: energy-dispersive
analysis of x-rays in the electron microscope, x-ray fluorescence, x-ray
photoelectron spectroscopy, and Auger electron spectroscopy. In each
technique, incident radiation (either x-rays or electrons) ionizes the
atoms in the specimen, and the energy of radiation emitted by the
specimen (again either x-rays or electrons, depending on the technique)
is analyzed. All of these techniques are thus forms of spectroscopy. The
various techniques all have different strengths and weaknesses, largely
determined by the interactions between the material and the incident
and emitted radiation.

Electron (and x-ray) interactions with matter

Lets begin by considering all of the possible interactions that can
occur when an electron (or x-ray) interacts with an atom. Well assume that the energy of the incident radiation is sufficient to ionize
the atom (that is, to remove an electron from either a valence shell or
one of the inner shells). The different possibilities are illustrated here:
Figure 1: An incident electron of sufficient energy can ionize an atom (top),
with the incident electron losing some
energy and being scattered. The atom
is left in a high-energy state and relaxes by emitting a photon (middle).
At this point there are two possibilities;
for most atoms the most likely event
is emission of this photon as x-ray
fluorescence (bottom right), but from
low-Z elements it is most likely that
the photon will be internally absorbed
with the energy carried away by an
Auger electron (bottom left). (Figure
from Goldstein et al., Scanning Electron
Microscopy and X-Ray Microanalysis.

The incident electron1 is scattered in this process; because it loses

energy (given up to ionize the atom), this is referred to as an inelastic

Or x-ray; the discussion here pertains

to either kind of incident radiation.
1

analyzing composition with x-rays and electrons

scattering event.2 We can analyze the energy lost by the incident

electron to learn about the structure and chemistry of the material, a
technique called electron energy-loss spectroscopy (EELS) but which
is not the focus of this handout.
The techniques we consider here all center on analyzing the three
kinds of radiation that can be emitted by the atom: The ejected orbital electron, or the emitted Auger3 electron or x-ray photon. Considering these three, with the choice of either x-rays or electrons as
the incident radiation, gives us a total of six potential analytical techniques. Not all of the combinations are widely used; the common
ones are as follows:

Technique
X-ray fluorescence (XRF)
Electron microprobe
X-ray photoelectron spectroscopy (XPS)
Auger electron spectroscopy (AES)

Incident
radiation
X-rays
Electrons
X-rays
Electrons

Emitted
radiation
X-rays
X-rays
Electrons
Electrons

Elastic scattering, in which the incident

particle doesnt lose energy, gives rise to
diffraction, which we will study later in
the course.
2

Pronounced oh-ZHAY

Table 1: Common analytical techniques

for studying composition of materials.

The different combinations of incident and emitted radiation give

each of these techniques characteristics that make them well (or
poorly) suited to particular characterization tasks.

X-ray fluorescence (XRF)

As its name implies, x-ray fluorescence involves irradiating the specimen with x-rays and analyzing the energies of the x-ray photons that
are emitted. The absorption and emission process looks like this:

Continuum

N
M

Continuum

Continuum

4
3

L2, L3
L1

2p
2s
Eincident

Efluorescent

1s

(a)

(b)

(c)

The atom is initially in its ground state (Fig. 2(a)). The fluorescence
process begins (b) with absorption of an incident photon by the atom,
simultaneously ejecting a core-shell electron. This requires that the

Figure 2: (a) Schematic electron energy levels for an atom in its ground
state. The 2s and 2p levels are shown
separately, but higher shells are shown
schematically as a single energy. (Actually, there is also a split in the 2p
levels that is not shown.) (b) The atom
absorbs an incident photon and ejects
a core-shell electron, removing it from
the atom. For this to happen, we must
have Eincident EK . (c) The atom relaxes
from its high-energy state by dropping
an electron down from a higher level,
emitting a fluorescent photon. The
energy of the fluorescent photon is
Ffluorescent = EL2 EK .

incident photon have enough energy to lift the K shell electron from
its ground state at EK to E = 0 (because we arbitrarily set our zero
of energy to correspond to an electron removed from the atom but
without any extra kinetic energy). Photons of lower energy cannot
be absorbed by this process of photoionization, so if one plots the
absorption of the atom as a function of x-ray energy there is a large
jump, or absorption edge, just at EK , as shown in Fig. 3. It is also
possible for electrons to be ejected from other shells, so the complete
absorption spectrum for a given atom shows an absorption edge for
each shell.
After absorption of the incident photon the atom is in a high energy state, with a vacancy in one of its core orbitals. The atom can
reduce its energy by allowing an electron from a higher shell to drop
into the vacancy, simultaneously emitting a photon with energy equal
to the difference between the two electron energy levels involved
(Fig. 2(c)).
Of course, the electron making the transition can come from a
number of different shells, so fluorescent photons of various energies
can be emitted.4 To keep these straight, there is a (fairly) simple notation best illustrated by example. The transition in Fig. 2(c) is called a
K transition, the K coming from the fact that it is a K-shell vacancy
being filled, and the from the fact that the electron making the transition comes from one shell higher (L). If the electron dropping in
the K-shell vacancy had come from the M shell instead, a K photon
would have been emitted. Similarly, an L shell vacancy filled from the
M shell would result in emission of a L photon.
There are a few subtleties to be aware of. First, not all transitions
between electron levels are allowed; for example, the L1 K transition is forbidden. Second, the energy levels in Fig. 2 can be further
subdivided according to the value of the magnetic quantum number
ml . The 2p level, for example, has azimuthal quantum number l = 1
and the electrons in this subshell can have ml = -1, 0, or 1. The two
sublevels with ml = 1 have slightly higher energy than the sublevel
with ml = 0. A fluorescent photon making the jump from a ml = 1
sublevel, called a K1 photon, thus has slightly higher energy than a
K2 photon, which comes from the ml = 0.
So how is this used to characterize the chemical composition of
materials? The electron energy levels, and especially the differences
between electron energy levels, are characteristic of particular elements. For example, a K2 fluorescence photon from Zr has an
energy of 15.775 keV. If we bombard a specimen with x-rays and see
photons with energy of 15.775 keV being emitted, its a pretty good
bet that the specimen contains Zr. (Its not a guarantee, though, because in some cases fluorescence lines from different elements can

Absorption

analyzing composition with x-rays and electrons

K absorption edge
18.0 18.2 18.4 18.6 18.8
Eincident/keV

Figure 3: X-ray absorption spectrum for

Zr. Note the large increase in absorption at Eincident ' 18 keV, corresponding
to absorption of a photon and ejection
of a K-shell electron. The oscillations in
absorption above the edge are known
as the x-ray absorption fine structure, analysis of which can be used to
discern details of the structure of the
material.

Not all transitions are allowed, due to

quantum mechanical selection rules. For
example, transitions from one s orbital
to another are forbidden.
4

Figure 4: Typical XRF spectrometer. The

tube generates x-rays tos illuminate the
specimen, and a solid-state detector
measures the energies of the fluorescent
photons emitted by the specimen.

analyzing composition with x-rays and electrons

overlap.) So the basic idea behind x-ray fluorescence (XRF) spectroscopy is to illuminate the specimen with high-energy x-rays, and
measure the energies of the x-ray photons that are emitted. Figure 4
is a schematic of a typical XRF spectrometer.
There are two basic ways in which the energies of the fluorescent photons can be measured. The one illustrated in Fig. 4 is called
energy dispersive spectroscopy (EDS5 ), which uses a semiconductor
detector capable of recording a wide range of photon energies simultaneously. In the other approach, the fluorescent photons are
diffracted from a crystal into a detector; by rotating the crystal different wavelengths are diffracted at different angles, so that only a
single wavelength enters the detector at a given time. This wavelength
dispersive spectroscopy (WDS) approach is much slower than EDS,
but has the advantage that its energy resolution (the ability to discriminate between peaks that are closely-spaced in energy) is much
superior.
To give you an idea of what the data look like, heres an energydispersive XRF spectrum taken from a marine diatom, T. weissflogii:

Sometimes called EDAX, for energy

dispersive analysis of x-rays

Figure 5: XRF spectrum from a marine

diatom. The K peaks from major
elements present in the specimen are
indicated. (Data from B.S. Twining et al.
Anal. Chem. 75, 3806 (2003).

X-rays have penetration depths on the order of 10-100 m in most

materials (depending on both the material and the x-ray energy),
so XRF samples to approximately this depth. This may seem like a
fairly thin surface layer, but as these things go such depths qualify
as bulk characterization. This can be either an advantage or a disadvantage, depending on whether youre interested in knowing the
bulk composition of your specimen, or if you seek information about
the surface in particular. Other advantages of XRF are that the spectra are rather simple to interpret and not very sensitive to chemical
effects (since the electrons involved come from core levels they are
not sensitive to the surrounding atoms) and that it has good precision and accuracy over a wide range of concentrations. Perhaps the
biggest limitation of XRF is that it has trouble detecting very light

analyzing composition with x-rays and electrons

(low atomic number) elements; the precise limit depends on both instrument and sample, but as a rule of thumb elements lighter than Ti
(Z = 13) can be difficult to see with common instruments.

Electron microprobe
If you imagine doing XRF using electrons instead of x-rays to photoionize the atoms, youve got an electron microprobe. Electrons have
one big advantage over x-rays: They can be focused easily, so one
can readily generate beams of 1 m or smaller diameter to probe
local chemistry.6 Among other things, this allows you to generate a
composition map of your sample by rastering the electron beam over
the surface. Since this is pretty much how a scanning electron microscope (SEM) already works, by far the most common approach to
chemical analysis in materials science is simply to outfit an SEM with
an EDS detector, allowing the instrument to be used for imaging and
chemical analysis.
There is also a dedicated instrument called the electron microprobe; because such an instrument is used solely for chemical analysis it is usually outfitted with a WDS detection system, which provides better energy resolution.
One disadvantage to using electrons as the probe is that they are
not very penetrating, so most of the chemical information comes
from the top micron (or less) of material near the surface. This makes
XRF a better choice for true bulk analysis.

In XRF the beam spreads out over

square millimeters or even centimeters.
6

X-ray photoelectron spectroscopy (XPS)

The physics of x-ray photoelectron spectroscopy (XPS7 ) are quite similar to those of XRF (Fig. 2) but instead of detecting and analyzing the
fluorescent x-ray, we detect the ejected photoelectron and measure its
kinetic energy. The kinetic energy of the photoelectron is given by
Ekinetic = h Ebinding ,

(1)

where h is the energy of the incident x-ray photon, is the work

function (the energy required to remove the electron from the specimen) and Ebinding is the binding energy of the electron (e.g. the
electron energy level in Fig. 2). Experimentally, we choose the x-ray
energy, and is usually known for a given material. So when we
measure the kinetic energy of the emitted photoelectron we can immediately calculate its binding energy.
One important consequence of the fact that XPS detects electrons
instead of x-rays is surface sensitivity. As mentioned above the penetration depth of x-rays is several microns (at least) in most materials.

Sometimes called electron spectroscopy

for chemical analysis (ESCA).

analyzing composition with x-rays and electrons

But electrons are scattered much more strongly than x-rays, so a

photoelectron generated several microns below the surface would
certainly lose energy before being able to escape the sample, if indeed
it would be able to escape at all. So the only electrons that can contribute to a peak with the energy characteristic of a particular energy
level (Ebinding above) are those that are generated close enough to
the surface that they can escape without experiencing any inelastic
scattering events. Although the details depend on the material and
electron energy, as a rule of thumb almost all of the information in
XPS comes from the top 10 nm or so of the material. This makes XPS,
together with Auger electron spectroscopy, a true surface analysis
technique.
Heres a typical wide XPS spectrum, in this case from a sample of
gold:
Figure 6: XPS spectrum from Au, plotted as intensity v. kinetic energy of
the photoelectron. Often, XPS data
are plotted against the binding energy instead. The peaks correspond to
particular binding energies, as noted.
The stepped background comes from
electrons that have lost some energy
due to inelastic scattering before escaping from the surface. Figure from
http://www.casaxps.com.

One very powerful feature of XPS is that, because it can provide

excellent energy resolution and can examine outer-shell electrons as
easily as core-shell electrons, it is possible to study chemical effects.
For example, here is a high-resolution XPS scan from a mylar specimen; the bottom axis is not labeled, but the peaks are separated by
1-2 eV:
Figure 7: High-energy-resolution XPS
scan from mylar showing the C 1s peak.
The binding energy of the 1s electron
changes slightly, depending on the
bonding environment of the C atom.
In this case we can clearly distinguish
the C atoms that are bonded to O from
those that are not. Figure courtesy of
Mike Kelly, Stanford University.

analyzing composition with x-rays and electrons

Finally, it is common for XPS instruments to be equipped with

an ion beam gun, for etching the specimen surface. This allows information about chemical composition as a function of depth to be
obtained, by sequentially etching the surface and analyzing it with
XPS. This is called depth profiling.

Auger electron spectroscopy (AES)

Besides XPS, Auger spectroscopy is the other workhorse technique
for surface chemical analysis. Although closely related to the other
techniques described here, its physical basis is slightly different:

Continuum

Continuum

Auger e-

Figure 8: The Auger process is similar,

conceptually, to x-ray fluorescence.
But instead of an x-ray photon being
emitted, the transition of an electron
from a higher shell into the vacancy
is accompanied by the emission of an
Auger electron.

Incident e-

(a)
(b)
As before the process begins with ionization of an atom, in this case
by an incident electron, and again an electron drops from a higher
shell to fill the vacancy. But rather than energy of this transition
being carried off by an x-ray photon (Fig. 4(c)), a second electron
called an Auger electronis emitted instead. As you might expect,
the energy of the Auger electron is characteristic of the energy levels
in the atom. The Auger process competes with x-ray fluorescence; a
given ionized atom can relax by either process. Fluorescence dominates for high-Z materials, but Auger emission dominates for light
elements.8 In fact, AES can be used to detect every element except H
and He.
As with XPS the depth from which the Auger electrons can emerge
without losing some of their energy is quite small, with most of them
coming from depths of 1-2 nm. This makes AES, like XPS, an ideal
tool for studying the chemistry of surfaces.

This is why, as discussed above, XRF is

not very sensitive to light elements.
8