ELECTRON SPECTROSCOPY OF SURFACES

Elemental and Chemical Analysis with X-ray

Photoelectron Spectroscopy

Versuch Nr. 35
F-Praktikum in den Bachelor- und Masterstudiengängen
Physik Department
Lehrstuhl E20, Raum 229
Betreuer: Mateusz Paszkiewicz and Dr. Francesco Allegretti

1 Introduction
Photoelectron spectroscopy (PES) is one of the most established methods for determining
the elemental composition and the chemical state of solid surfaces and thin films. This
experiment focuses on X-ray Photoelectron Spectroscopy (XPS), which exploits X-
ray photons to excite electrons from the core levels of the atoms of a solid into the vacuum,
thus probing the electronic structure of matter with elemental sensitivity and chemical
state specificity.
The purpose of the experiment is to make the students familiar with the fundamental
principles and basic methodology of XPS, also offering an insight into the underlying
physics. A number of different samples will be investigated in order to highlight the main
information that the technique can provide and to show how XPS can be of relevance for
a vast range of systems, not only in condensed matter physics, chemistry, and materials
science, but also in applied physics and biophysics.
The technical basis of the experiment includes:
• basics of ultra-high vacuum technology
• generation of X-ray radiation
• analysis in energy and detection of electrons
• characterization of solid samples (in-situ or ex-situ prepared) by XPS
• computer control of the measurements
• data analysis via computer software
Learning how to understand and interpret the experimental data and drawing up a short
scientific report (in German or English) are further achievements of this FOPRA exper-
iment. The experimental session will typically last about five hours and will be followed
by a colloquium a few days after handing in the report.

2 X-ray Photoelectron Spectroscopy

2.1 Underlying physical principles and fundamental aspects
XPS is based, in principle, on a very simple process. When a solid surface is irradiated
with soft X-ray photons (Fig. 1a), an incident photon of energy ~ω can be absorbed by an
 z
photon e-
ħω in vacuum final state
photoelectron
Ekin
vacuum level
θp Φw
θe Fermi level

y Eb ħω ω
φp φe
sample initial state

x
(a) (b)

Figure 1: Schematic view of the photoemission process: (a) a photon of energy ~ω impinges on a
solid surface, and after absorption of the photon an electron is ejected in a given direction (θe , φe )
with respect to the reference system; (b) corresponding energy balance within the framework of a
single particle (or one electron) picture. In (b) Φw denotes the work function of the material.

electron with binding energy EB below the vacuum level: the entire photon energy is trans-
ferred to the electron, which is then promoted to an unoccupied state above the vacuum
level. As a result, this photoelectron is ejected into the vacuum with kinetic energy:

Ekin = ~ω − EB (1)

and can be detected by an electron energy analyzer.

For a solid surface the binding energy1 (Eb ) is conventionally measured with respect to
the Fermi level rather than to the vacuum level (Fig. 1b), and the previous relationship is
written in the form:
Ekin = ~ω − Eb − Φw (2)
Here, Φw is the work function of the material and represents the minimum energy required
to remove an electron from the solid; it can be seen as an energy barrier that electrons
need to overcome in order to escape from the surface into the vacuum. Obviously, for an
electron to be excited, the photon energy must be greater than Eb with the difference at
least exceeding Φw .

In the simple single particle picture described above, which is based on a single electron
being excited, it is clear that the energy distribution of photoemitted electrons should be a
faithful replica of the electron states in the solid shifted up in energy by an amount ~ω-Φw
(see Fig. 2). In practice, it provides only a distorted replica of the electronic structure,
complicated − for example − by the occurrence of multi-electron processes and by the
fact that the probability of a photon being absorbed is not the same for all electron states.
The ejected electrons can originate from core levels or from the occupied portion of the
valence band (Fig. 2); in standard XPS, however, due to the relatively high photon energy
(typically in the range between 100−200 eV and 2000 eV), the attention is predominantly
focused on core level electrons.

Three main aspects of XPS deserve a special mention:

• the elemental sensitivity

• the chemical state specificity

1
In the following we adopt the notation Eb for the binding energy when the Fermi level is taken as the
reference level, whereas EB is used when the binding energy is referenced against the vacuum level.

2
 ħω

ħω

Eb

Figure 2: Schematic of the photoemission process in the single particle picture (adapted from
Ref. [1]). Electrons with binding energy Eb can be excited above the vacuum level Evac by photons
with energy ~ω > Eb +Φw . The photoelectron distribution I(Ekin ) can be measured by an electron
energy analyzer and is - to a first approximation - a replica of N (Eb ), the occupied density of
electronic states (DOS) in the sample.

• the surface sensitivity

1. A typical XPS spectrum is obtained at a given photon energy by recording the number
of photoelectrons as a function of the kinetic energy and can be plotted as a function of
the binding energy by making use of Eq. (2). The spectrum consists of a succession of
distinct lines, reflecting the sequence of occupied levels - in particular the core levels -
in the system under investigation. Since the binding energies of core levels are different
for (and therefore characteristic of) each chemical element, and no two elements share
the same set of binding energies, photoemission spectra can serve as “fingerprints” of the
respective elements. For example, as illustrated in Fig. 3, a peak located at Eb ∼ 285 eV

Figure 3: Photoemission spectra of the 1s core levels of the second row elements (taken from
Ref. [2]).

3
 (a)

(b)

Figure 4: (a) Chemical shifts in isolated molecules (C 1s and N 1s levels; adapted from Ref.
[2]). (b) A plot of the chemical shift in the C 1s binding energy in different carbon-containing
compounds (from Ref. [3]).

is characteristic of photoelectrons coming from the C 1s core level, while the N 1s photoe-
mission line appears at higher binding energy (∼ 400 eV). This elemental sensitivity
renders PES a very useful analytical tool, making it possible - for instance - to determine
the composition of an unknown sample and to assess the degree of cleanliness of solid
surfaces. Note that XPS can be used for analysis of all elements in the periodic table
except hydrogen and helium.

2. When comparing the binding energies of core electrons of a given element in different
systems (i.e., in different molecules, compounds or solids), one finds that the exact value of
Eb depends on the chemical environment of the atoms and that energy shifts may occur
when inequivalent atoms of the same elemental species are present (Fig. 4). These energy
shifts are called chemical shifts. In general, electrons photoemitted from atoms in a higher
positive oxidation state display a higher binding energy because of the increased Coulomb
interaction with the ion core upon withdrawal of valence charge due to chemical bonding.
By comparison with data from well-known “standard” substances, the values of the bind-
ing energy shifts can often be used to determine oxidation states and to gain insight into
the nature of the chemical bond formed by the atoms. Such a chemical state specificity
is one of the major strengths of XPS for application in chemistry, surface physics and ma-
terials research, and is at the origin of the alternative name ESCA (Electron Spectroscopy
for Chemical Analysis) with which the technique is denoted.

3. XPS and, more in general, all electron spectroscopies are highly surface sensitive,
due to the strong interaction of electrons with matter. The latter entails that electrons
travelling in a solid exhibit a very short inelastic mean free path, λe . This quantity de-

4
Figure 5: Inelastic mean free path of electrons as a function of their kinetic energy for various
elements. While λe depends strongly on the kinetic energy, it is only weakly dependent on the
material: for this reason the general trend λe = λe (Ekin ) is often referred to as universal curve.

scribes the average distance that an electron can travel through the material without
suffering energy losses by inelastic scattering; it therefore provides an indication of the
electron escape depth. For electrons with kinetic energy in the range between 10 and
1000 eV, λe can vary between 5 and 30 Å depending on the material (Fig. 5) and thus
amounts to only a few atomic layers. As a result, in a typical photoemission experiment
only electrons originating from a narrow region at the solid-vacuum interface can reach
the detector without losing energy, thus contributing to the main lines of the measured
spectrum. For this reason, PES and XPS are widely employed to probe the electronic
structure and composition of solid surfaces and thin films.

2.2 Beyond the single particle picture: binding energy and final state
effects
The single particle picture adopted so far to describe the photoemission process assumes
implicitly that when the photoelectron is excited above the vacuum level all other electrons
in the sytem remain unperturbed in the same state as before the photoionization event.
This assumption has an immediate implication on the location of a given core level peak,
the binding energy of which should be simply equal to:

EB (ν) = Ef (N − 1) − Ei (N ) = −ν (3)

Here, Ei (N ) is the energy of the ground state of the atom with N electrons and Ef (N − 1)
the energy of the final state of the ionized atom, whereas ν denotes the energy of the elec-
tron in the state ν below the vacuum level. The previous expression, known as Koopman’s
theorem, is a consequence of the energy conservation ~ω + Ei (N ) = Ekin + Ef (N − 1)
combined with Eq. (1).
In reality, however, the Koopman’s energy is never observed, and this can be mainly
ascribed to the fact that the electron system “relaxes” in response to the creation of
the vacancy in the inner shell (core hole) upon photoemission. This relaxation is an
important many-body effect in the one-hole final state of the electron system. The core hole
represents a positive charge acting on the remaining electrons, which rearrange themselves
to partially shield the electron vacancy and to minimize the total energy of the ionized

5
 (a) (b)
shake-up shake-off shake-up structure

CuO
Cu 2p1/2
Cu 2p3/2

970 960 950 940 930

binding energy (eV)

Figure 6: (a) Schematic of the shake-up and shake-off transitions; (b) Shake-up loss structure in
the Cu 2p spectrum of copper (II) oxide (CuO), adapted from Ref. [5].

atom, thus making more energy available for the outgoing photoelectron. Therefore, the
relaxation of the eletron system reduces the apparent binding energy EB measured for the
photoelectron. Typical relaxation shifts are of the order of 10-20 eV [4].
The screening of the core hole typically comprises an intra-atomic and an extra-atomic
contribution. The former is due to the relaxation of the electrons (essentially, the outer
shell electrons) of the excited atom itself and is the only contribution in isolated atoms.
For atoms bound in molecules or solids the electrons of the neighboring atoms are also
polarized by the creation of the core hole, although to a lesser extent, thus giving rise
to the extra-atomic component of the relaxation. Because the screening of the core hole
depends on the electronic environment2 , relaxations shifts differ from system to system
for a specific core level line. This difference can contribute to the chemical shift described
in §2.1, although in many cases the effect of the modified physical and chemical local
environment on the initial electron state is considered to be dominant.
The relaxation of an electron system in response to the creation of a core hole is an
important final state effect. Other final state effects that involve many-body processes
and can lead to a modified photoelectron’s kinetic energy are, for example, multi-electron
excitations such as plasmon losses and shake-up and shake-off transitions. In all these
processes, electronically excited final states are created upon photoemission, so that less
kinetic energy is available for the photoelectron: this leads to satellite lines shifted to
higher binding energy relative to the main core level line. Shake-up and shake-off satellites
occur when an additional electron is promoted from an occupied energy level into a bound
unoccupied state or into a state in the continuum above the vacuum level, respectively
(Fig. 6). Plasmon loss satellites, instead, are the result of the excitation of quantized
plasma oscillations, namely collective oscillations of the free electron gas density with
respect to the positively charged ion cores. These plasmon satellites are displaced to
higher binding energy relative to the “elastic” line by amounts n(~ωp ) + m(~ωs ), where
n and m are integers and ~ωp and ~ωs are the energy of a bulk and a surface plasmon,
respectively (Fig. 7).
2
For example, the screening ensured by electron rearrangement is more effective in metals than in ionic
solids, thanks to the presence of valence electrons (the conduction electrons) that are free to move inside
the material. Accordingly, the magnitude of the relaxation shift is larger in metals than in ionic materials.

6
 plasmon losses

Mg 2s

-ħωs

Relative binding energy (eV)

Figure 7: String of plasmon satellites in the Mg 2s XPS spectrum, adapted from Ref. [6].

2.3 Decay of a core hole

The ejection of an electron from an inner shell creates an electronically excited state with
a core hole in that shell. This excited state can decay via two distinct de-excitation routes:

• The core hole is filled by an electron from an outer shell with simultaneous emission
of a photon with energy equal to the difference in binding energy of the two levels
involved in the electron transition. This is the well-known X-ray emission process
or X-ray fluorescence (Fig. 8, path (a)).

• The core hole is filled by an electron from an outer shell as in the previous case,
but the excess energy is transferred to another outer shell electron, which is emitted
(Fig. 8, path (b)). The process, called Auger decay, involves three electrons: the
electron knocked out in the initial photoionization event, the electron that decays
to fill the inner shell vacancy, and the emitted electron from the outer shell (Auger
electron). As a consequence, the final state is a doubly ionized state.
Auger transitions are denoted with the notation XY Z, where X, Y and Z refer to the
three shells involved in the decay process. For example, in a KLL transition the
electron that leaves behind the initial core hole is from the K shell (1s state), while
the other two electrons come from the L shell (2s or 2p levels).

X-ray fluorescence and Auger electron emission are competitive processes, the branching
ratio of which is governed by the respective decay cross-sections. For the binding energy
range below 1.5 keV, as in the present experiment, and especially for elements with low
atomic number, the non-radiative decay by far prevails.

2.4 Linewidth of core level lines

The width and shape of a photoemission peak results from the convolution of differ-
ent effects. One important contribution is related to instrumental factors, primarily the
linewidth and lineshape of the X-ray line used for the excitation and the resolution of the
electron energy analyzer employed to analyze in energy the distribution of photoelectrons.
Another essential contribution, related to the photoemission process itself, is the peak
broadening due to the finite lifetime of the core hole. Thermal braodening by coupling

7
 Photoionization
photoelectron

XYZ Auger electron

(a) (b)

Y
fluorescence X
photon

X-ray Fluorescence Auger Emission

Figure 8: Schematic of (a) X-ray fluorescence and (b) Auger decay following a photoionization
event. Adapted from Ref. [3].

with intra-molecular vibrations and phonons may be non-negligible in some cases. In ad-
dition, many-body effects, such as the excitation of electron-hole pairs close to the Fermi
level in metals or shake-up satellites that cannot be resolved due to limited instrumen-
tal resolution, can cause additional asymmetric broadening with a tail on the low kinetic
energy side of the photoemission peak.
Instrumental and vibrational broadening yield Gaussian lineshapes. The spectral line-
shape due to the core hole lifetime τ is instead a Lorentzian function with full-width at
half-maximum (FWHM) 2Γ, the so-called intrinsic or natural linewidth, which is related
to the mean lifetime through the relationship [7]:

~ 6.58 × 10−16 eV · sec

2Γ = = (4)
τ τ [sec]

2Γ is typically larger for inner shell orbitals, because these can be filled rapidly by electrons
from the outer shells. Therefore, for a given element 2Γ tends to increase with increasing
binding energy. In a similar way, for a given core level orbital (e.g., 1s) 2Γ increases as the
atomic number Z increases, due to the higher valence electron density in heavier atoms.
The contributions of intrinsic and instrumental effects to the experimentally observed
linewidth can be combined to a first approximation into a quadratic sum [8]:
2
F W HMtot = (2Γ)2 + F W HMX
2
+ F W HMA2 (5)

with the second and third term on the right side due to the X-ray source and analyzer
energy resolution, respectively.

2.5 Spin-orbit splitting of core level lines

In the photoemission from p, d and f inner subshells the spectral lines are split into two-
component peaks (Fig. 9a). This splitting arises from spin-orbit coupling effects in the final
state [8]. While in the initial state the inner subshells are completely filled (for example,
the configuration of the inner shells in gold is: (1s)2 (2s)2 (2p)6 ... (4f )14 ... ), in the final

8
 (a) (b)
Au 4f 7/2 Total angular Ratio of the
Subshell momentum j degeneracies
Au 4f 5/2 components (g = 2j +1)

s 1/2 -
Counts

p 3/2, 1/2 4/2

d 5/2, 3/2 6/4

f 7/2, 5/2 8/6

Binding Energy (eV)

Figure 9: (a) Spin-orbit coupling leads to a splitting of the 4f photoemission line of gold into a
doublet [2]. (b) Table listing spin-orbit split components and intensity ratios for different subshells.
The notation n[l-subshell]j is commonly used (e.g., 4f7/2 ).

state one electron has been removed and an unpaired spin is left behind. The spin can be
oriented up or down, and if the core hole belongs to an orbital with non-zero orbital angular
momentum (l > 0) there will be a coupling between the unpaired spin and the orbital
angular momentum: the two states j+ = l + 1/2 and j− = l − 1/2 are not degenerate and
a spin-orbit split doublet is observed in the photoelectron spectrum. The intensity ratio
of the two spin-orbit components is dictated by the ratio of the respective multiplicities:
(2j+ + 1)/(2j− + 1) = (2l + 2)/2l (Fig. 9b), which determines the relative probability
of transition to the two states upon photoionization. Generally, in the photoemission
spectrum from a given subshell the core level subpeak with maximum j is detected at
lower binding energy. Obviously, no spin-orbit splitting occurs in the photoemission from
s core levels.

2.6 Summary − spectral features in a photoemission spectrum

We can now understand all the main spectral features that can be observed in a typical
XPS spectrum of a solid (as the one illustrated in Fig. 10) and reflect the characteristic
excitation processes in condensed matter.

1s

plasmon Fermi level

losses 2s
KLL Auger
2p

shake-up valence
background of secondary electrons band
and inelastically scattered electrons
binding energy kinetic energy

Figure 10: A typical XPS overview spectrum from a solid surface. Adapted from Ref. [9].

As noted above, XPS spectra are in general dominated by sharp peaks correspond-
ing to core level photoemission. These peaks may contain chemically shifted components

9
and be spin-orbit split. Their binding energies always incorporate the contribution of the
relaxation shift and their linewidth can be affected by many-body effects, instrumental
resolution and natural broadening. In many cases, from the relative intensity of core level
peaks one can estimate semi-quantitatively elemental ratios, in order to assess the relative
surface coverage of adsorbates and contaminants or the stoichiometry of a compound. To
this purpose it is necessary to correct the relative intensities by the respective photoion-
ization cross-sections, which express the probability of creating a photoelectron in each
core level at the photon energy ~ω.
Together with core level lines, electron emission from the valence band is also observed.
This part of the spectrum thus represents a projection of the occupied electronic states to
energy above the vacuum level, and the photoemission intensity drops to zero above the
Fermi level (Fig. 10, right end of the spectrum).
Since Auger electrons are also detected by the electron energy analyzer, Auger lines
are often observed. If it is possible to change the photon energy for the acquisition of the
spectrum, an Auger line can be easily distinguished from a core level line, because the
former appears at fixed kinetic energy regardelss of the photon energy, while the latter
moves linearly with the photon energy according to Eq. (2).
Satellites corresponding to multi-electron excitations, such as multiple plasmon losses
and shake-up satellites, may also show up in the XPS overview spectrum. Finally, if X-
ray emission lines corresponding to different photon energies are used for the excitation,
replicas of the same core level peak at different kinetic energy will be observed. Note
that all photoemission features appear superimposed on a featureless background3 , arising
from inelastically scattered electrons (which undergo energy losses before escaping from
the surface) and secondary electrons excited in “cascade” processes.

2.7 Sampling depth of XPS and collection geometry

As anticipated in §2.1, the surface sensitivity of XPS stems from the short inelastic mean
free path of electrons. While X-rays interact only weakly with matter and consequently
can penetrate deeply into a solid (1000 nm or more at ~ω ∼ 1 keV), electrons with energy
in the range 5-1500 eV readily lose energy by inelastic scattering. The XPS experiments
are mainly concerned with the intensity of emitted photoelectrons that have suffered no
energy losses: therefore, the XPS sampling depth refers to a characteristic length over
which electrons can travel without undergoing inelastic scattering events. This quantity is
the inelastic mean free path λe plotted in Fig. 5, which is defined as the average distance
that an electron of given kinetic energy Ekin travels between two successive inelastic col-
lisions [8]. In other terms, the probability that the electron travels a distance z through
the solid without undergoing inelastic scattering can be written as an exponential decay:
z
− Z
P (z) ∝ e λe =⇒ < z >= zP (z)dz = λe

If one considers for simplicity the electrons emitted in the direction perpendicular to a
solid surface, a small element of thickness dz at a distance z from the surface (z = 0
denotes the surface plane) will give a contribution dI to the measured intensity:
z
−
dI ∝ e λe dz (6)
The total intensity I coming from a thin film of thickness t right underneath the surface
is obtained by integrating Eq. (6) between 0 and t, and it thus amounts to:

I = I0 [1 − e−t/λe ] (7)
3
For the correct determination of peak areas, peak positions and peak widths this background needs to
substracted.

10
The sampling depth of the XPS experiment is often defined as t = 3λe , which means that
95% of the photoemission intensity of the core level line comes from the corresponding
region (3λe thick) below the surface [8].
When detecting photoelectrons emitted at an angle θe away from the surface normal
(see Fig. 1a), one has to consider the effective path z/cosθe of the electrons in the solid,
and Eq. (6) becomes:
z
dI ∝ exp[− ] · dz (8)
λe · cosθe
Hence, in grazing emission geometry the sampling depth is reduced to 3λe cosθe and the sur-
face sensitivity of the measurements is strongly increased (e.g., θe ∼ 80◦ =⇒ cosθe ∼ 0.17).

3 Adsorption on surfaces
The term “adsorption” indicates the process of trapping and binding of foreign species
(atoms or molecules) onto a surface by condensation from the gas phase. The energy
released in this process is called energy of adsorption Uads or also binding energy (not
to be confused with the binding energy in PES). The term “substrate” refers to the solid
surface onto which adsorption occurs, while “adsorbate” is used to indicate the species that
are adsorbed onto the substrate (Fig. 11). Adsorption can be associative (if a particle
or molecule is adsorbed as a single unit) or dissociative (if the particle or molecule is
dissociated upon adsorption), and moreover adsorption processes may be reversible or
irreversible.
The reverse process, the removal of adsorbed species from the surface, is called “desorp-
tion”. Desorption may be stimulated by heat (thermal desorption), by electrons (electron
stimulated desorption), or by photons or ions (photon or ion stimulated desorption).

3.1 Types of adsorption

Depending on the strength of the interaction between the adsorbate and the substrate, a
distinction is typically made between two limiting cases:

• Physisorption
Physical adsorption (= physisorption) is governed by weak Van der Waals interac-
tions with the surface; no chemical bonds are formed. The adsorption energy Uads
is small, typically < 30 kJ/mol or 0.3 eV/particle. Formation of multilayers (con-
densates) is possible at sufficiently low temperatures.
Example: He on noble metal surfaces at cryogenic temperatures.

adsorp on
desorp on
adsorbate

substrate

Figure 11: Schematic of adsorption onto (and desorption from) a solid surface.

11
 • Chemisorption
Chemical adsorption (= chemisorption) involves the formation of a chemical bond
between the adsorbate and the surface, e.g. transfer of electrons and/or formation
of covalent bonds (hybrid molecular orbitals). In most cases the adsorption energy
is much larger than 30 kJ/mol, ≥ 100 kJ/mol or 1 eV/particle. Chemisorption
is often dissociative and may be irreversible; activation barriers are possible, and
chemisorbates form typically only a single layer (= direct contact of surface atoms
and the chemisorbate is required). Moreover, specific binding sites may be favoured
because of the highly corrugated surface potential.
Examples: CO and oxygen chemisorption on transition metals at room temperature.

3.2 Adsorption order

• First order adsorption
One particle from the gas phase yields one particle of the adsorbate, i.e., the im-
pinging particles are molecules that do not dissociate upon adsorption. The reverse
process is first order desorption.
Examples: CO/nickel (CO chemisorbs on Ni surfaces molecularly with Uads ∼ 120
kJ/mole); Ar/Ruthenium (Ar physisorbs on Ru with Uads ∼ 10 kJ/mol).

• Second order adsorption

A molecule dissociates upon adsorption. One particle in the gas phase yields two par-
ticles on the surface. The reverse process is second order (or associative) desorption.
Example: H2 /Nickel (H2 dissociates into two H atoms on Ni, Uads ∼ 90 kJ/mole).

3.3 Nomenclature
Other important concepts encountered in adsorption studies in the field of surface science:
Nads
• Surface (relative) coverage Θ: Θ≡
Nsub
with Nads = number of adsorbate particles per unit area; Nsub = number of particles
per unit area in the substrate surface layer. Θ is typically expressed in monolayers
or as a fraction of monolayer (ML), 1 ML corresponding to an adsorbate surface
density equal to the surface density of substrate atoms.

• Collision rate ν:
it is defined as the number of particles hitting a surface per unit time and unit area.
The Hertz-Knudsen equation following from the gas kinetic theory states that:
p
ν=√ [particles · s−1 · m−2 ]
2πM kB T
with p = pressure [N· m−2 = Pa]; M = particle mass [Kg]; T = temperature [K];
kB = Boltzmann constant.

• Sticking coefficient s:

dNads dNcoll
s= 0≤s≤1
dt dt
probability that a particle colliding with the surface is adsorbed and stays on it.

• Exposure:
measure of the amount of gas that a surface has been exposed to; more specifi-
cally, the product of the gas pressure and the time of exposure. Normal unit is the
Langmuir, where 1L = 10−6 Torr ·s (1 Torr = 133.3 Pa = 1.333 mbar).

12
4 Experimental set-up
The set-up that will be used for the experiment is schematically depicted in Fig. 12 and
is based on an ultra-high vacuum (UHV) chamber on which the XPS apparatus and all
equipments for sample insertion, preparation and manipulation are mounted.

4.1 Vacuum system

The UHV chamber is maintained at a base pressure lower than 1×10−9 mbar (1×10−7 Pa).
This low pressure is required in order to permit electron detection (avoiding scattering of
the photoelectrons from residual gases between the sample surface and the detector) and
to ensure that atomically clean surfaces can be prepared and maintained free of contami-
nants during the experiment. The chamber is pumped by an ion pump directly mounted
inside the chamber. Auxiliary pumping by a titanium sublimation pump is possible. In
addition, an external turbomolecular pump (with attached a backing roughing pump)
can be connected to the UHV chamber through a gate valve. This turbomolecular pump
is also used to pump the load-lock system adjacent to the main chamber. The UHV
regime can only be reached after baking out the chamber for about 24 hours (or longer) at
temperatures exceeding 160-180◦ C. This procedure allows to eliminate residual molecules
(predominantly water) adsorbed on the stainless steel walls of the chamber, which other-
wise would be released at a low rate thus increasing the pressure. A brief introduction to
the UHV technology, instrumentation and related protocols can be found in Refs. [10, 11].

hemispherical
electron energy
analyzer (HSA)

detector

UHV chamber

pressure gauge

e- leak
X-ray source holder valve
O2 line

W sample

load
-lock
Mg evaporator syst
em
gate valve

turbomolecular
pump
titanium +
sublimation roughing pump
pump

ion pump

Figure 12: Scheme of the UHV chamber with the XPS apparatus and ancillary equipments.

13
4.2 X-ray source
A conventional X-ray tube provides characteristic Al Kα radiation. This photon source
consists of a cathode filament, which emits electrons upon direct current heating (filament
current: 4.5 A), and an anode, onto which the electrons are accelerated by applying a high
voltage (11-12 kV). The typical emission current (i.e., the current of electrons leaving the
cathode and reaching the anode) is 20-22 mA. Water cooling is required during continuous
duty operation of the X-ray source, in order to prevent melting of the anode target. The
anode is an Al film, some 10 µm thick, supported on a copper rod. During operation,
the electrons from the cathode collide with the atoms of the Al target, causing vacancies
in inner electron shells. The subsequent de-excitation by fluorescence decay (see §2.3)
produces X-ray radiation of characteristic energy.

L3 2p3/2
L2 2p1/2
L1 2s
energy

K α1
K α2 Al K α1,2
K 1s ħω = 1486.6 eV
FWHM = 0.78 eV

(a) (b)
Figure 13: Characteristic radiation from an Al anode source: (a) schematic energy level diagram
explaining the production of the individual Kα1 and Kα2 lines; (b) emission spectrum showing
the convolution of the Kα1 and Kα2 lines into a composite Kα1,2 line centered at 1486.6 eV with
0.78 eV FWHM. Adapted from Ref. [12].

As illustrated in Fig. 13, the Kα1 and Kα2 lines are generated, corresponding to L3 → K
and L2 → K transitions, respectively, where L3 , L2 and K denote the 2p3/2 , 2p1/2 and
1s levels. The L2,3 subshell is split due to spin-orbit coupling in the final state, such that
the relative intensity of the two lines reflects the L3 to L2 population ratio of 2 : 1. As a
consequence, the Kα1,2 line is in principle a doublet with an energy splitting of 0.43 eV.
However, since each line has a natural width of 0.47 eV caused by lifetime broadening,
the doublet cannot be resolved and the resulting composite line has a linewidth of 0.78
eV (FWHM) and is centered at ~ω = 1486.6 eV. In addition, a Kα3,4 satellite is also
produced, attributed to emission from doubly ionized atoms [12]. This satellite line is
centered at about 1496 eV and its intensity is ∼ 10% of that of the main Kα1,2 emission.
Additional emission lines can arise from contamination (e.g., the O Kα line centered
at ~ω = 524.9 eV if the anode material is partly oxidized) or other modifications of
the anode (e.g., when the Al film becomes too thin, the Cu Lα line from the copper rod
beneath may be excited, ~ω = 929.7 eV).

4.3 Electron energy analyzer

An electrostatic 180◦ hemispherical analyzer (HSA) is used to analyze in energy the elec-
trons emitted from the sample. This electrostatic device, which consists of two metallic
hemispheres concentrically arranged as in Fig. 14, enables to “disperse” the electrons as a
function of their kinetic energy similarly to a prism that disperses light depending on its

14
V(R2) = - V2

electron
optics
V(R1) = - V1 detector

Ep R R
∆V = V(R1) - V(R2) = e R 2 - R 1
1 2

Figure 14: Schematic of a concentric hemispherical analyzer. Both hemispheres are at negative
potential, with V (R2 ) < V (R1 ) (→ V2 > V1 ). Adapted from Ref. [3].

wavelength. For this purpose, a potential difference ∆V is applied between the inner and
outer hemispheres, so that the trajectory of the incoming electrons is bent into a curve4 .
An electrostatic lens between the sample and the entrance slit S of the analyzer focuses
the photoelectrons from the sample onto S. At a given ∆V , ideally only the electrons
with a well-defined kinetic energy Ep are able to complete their path along the median
trajectory of radius R0 = (R1 + R2 )/2 (where R1 and R2 are the radii of the inner and
outer hemisphere, respectively) and emerge at the exit slit F . Ep is called pass energy and
its typical values range between 5 and 50 eV for XPS measurements.
Importantly, the choice of the pass energy determine the energy resolution of the
analyzer, which quantifies the ability to separate (→ to resolve) peaks that differ in energy
by only small amounts. The lower the pass energy, the better the resolving power. As
discussed in §2.4, the energy resolution of the analyzer contributes to the overall FWHM
of a core level peak (Eq. (5)). For the HSA the following expression holds [11, 12]:

∆E WS + WF α2
= + (9)
Ep 4R0 2

in which ∆E is the FWHM contribution from the analyzer (F W HMA ), WS and WF are
the widths of the entrance and exit slit, respectively, and α is the angular spread of the
electron beam. During the XPS measurements, the analyzer is operated at constant pass
energy to yield a constant absolute energy resolution ∆E.
In order to enable scans over arbitrary kinetic energy ranges, the electrons are retarded
(or accelerated) to the kinetic energy Ep by a potential difference R inside the lens system
between the sample and the entrance slit. A channel-type electron multiplier (channeltron
4
To prevent perturbation of the electron trajectories by stray magnetic fields, the analyzer region is
shielded by a µ-metal shield inside the vacuum vessel.

15
 e-

Ep Ekin,F
E vacuum,analyzer
hω
Φanalyzer
E F,analzyer
E vacuum,sample
R
Φw
E F,sample

Eb Ekin,F = R + Ep + Φanalyzer
analyzer
Ekin,vacuum = R + Ep (measured)
sample
Core Levels Ekin,vacuum = R + Ep + (Φanalyzer - Φw )

Sample Analyzer
Figure 15: Photoelectron detection by an electron energy analyzer: schematic diagram of relevant
energy levels and voltages.

detector) is situated behind the exit slit of the analyzer and counts the emerging elec-
trons. Therefore, by scanning R one can record a spectrum of the photoelectron intensity
as a function of the kinetic energy, the measured kinetic energy being simply R + Ep .
When R = 0, the Fermi levels of sample and analyzer are aligned (in practice, both
objects are grounded), whereas they are separated by R when the retarding voltage is
applied. However, due to the different values of the work function, the vacuum levels
of sample and analyzer do not coincide even for R = 0. From the energy level diagram
illustrated in Fig. 15 it then follows that:
analyzer
Eb = ~ω − (Ep − Φanalyzer − R) = ~ω − Ekin, vacuum − Φanalyzer (10)
Eq. (10) shows that the work function of the analyzer, rather than the work function of the
sample, needs to be taken into account for calibrating the binding energy scale. Φanalyzer
enters the equation because in the experiment the kinetic energy scale is referenced to the
vacuum level of the analyzer. To keep Φanalyzer constant, all electrodes are covered with a
chemically inert material (gold or graphite; in the present case graphite).

4.4 Samples
Four samples will be investigated during the experiment. Initially, a tungsten (W) ribbon
will be studied. It can be heated up to 2500 K by direct current flow, in order to remove
contamination from the surface. The W ribbon is mounted on a four-degree-of-freedom
manipulator enabling translational motion along three orthogonal directions and rotation
around the axis of the manipulator. In this way, the polar angle θe between the electron
emission direction and the surface normal (see also Fig. 1a) can be changed - ideally -
between 0◦ and 90◦ , thus controlling the surface sensitivity of the XPS measurements.
Further, the W ribbon will serve as substrate for the growth of a relatively thick Mg
film that will be deposited in-situ under UHV conditions. Here, Auger electron emission

16
from the KLL decay will enable to study the excitation of plasmons in the metal film.
Moreover, the subsequent oxidation of magnesium at room temperature will provide a
clear illustration of the concept of chemical shift and of the information that can be
gained by changing the surface sensitivity.
An additional sample may be introduced through the custom-made load-lock system
(Fig. 12). This system allows fast insertion of ex-situ prepared samples without breaking
the vacuum of the main chamber. For example, a gold substrate covered by an organic
self-assembled monolayer (SAM) can be investigated. The SAM is prepared by immersion
of the substrate into an ethanol solution containing thiol molecules. Thiols are endowed
with a SH functional group, which provides a strong anchoring to gold upon breaking of
the S-H bond. Alternatively, a sample of carbon nanotube powders or a graphite substrate
covered by organic molecules (such as metalloporphyrins or metal phthalocyanines) could
be studied. The elemental and chemical state analysis will provide valuable insight into
the composition of the system, nature of the adsorbed molecules, bonding interactions etc.
Students are welcome to select a sample that is of interest for their course of
studies, communicating their choice to the advisor (Betreuer) 3-4 days before
the experiment.

4.5 Magnesium evaporator

The evaporator consists of a small high-melting-point crucible containing Mg grains. The
crucible is heated by passing current through a conducting wire coiled around it. Prior
to evaporation, the Mg grains need to be outgassed repeatedly, in order to ensure that a
clean Mg film is deposited onto the tungsten ribbon during the experiment.

4.6 Gas handling line

A gas line filled with molecular oxygen is connected to the main chamber through a high-
precision leak valve. In this way, O2 can be admitted into the chamber in a controllable
fashion during the experiment.

5 Important concepts
The following concepts should be clear before coming to the experiment:

• Fundamental principles of XPS. Photoelectrons and Auger electrons. Core level

lines. Binding energy and work function. Spin-orbit splitting. Chemical shifts.

• Surface sensitivity of XPS. Importance of the electron detection geometry.

• Operating principle of the X-ray tube. Emission lines from the Al anode.

• Hemispherical electron energy analyzer. Pass energy and retarding voltage. Energy
resolution.

17
6 Instructions for the experimental session
The goal of the experiment is to learn how, in practice, XPS is used for elemental
and chemical analysis. Insight into instrumentation, methodology for data acquisition
and interpretation of the results will be provided. After getting an introduction to the
experiment and the equipments by the advisor you will follow the four steps (A-D) listed
below.

A. Performance of the analyzer and characterization of the W sample

The advisor will show how to clean the W ribbon and you will repeat the operation twice.
Afterwards, the sample can be brought into the measurement position. Turn it in order to
have an electron emission angle of 45◦ (this also corresponds to a 45◦ incidence angle of the
photon beam relative to the surface normal). The advisor will explain the operations for
switching on the X-ray source (high voltage: 12 kV) and the HSA analyzer, and you will
learn how to run measurements via computer. Importantly, make sure that you have set
the correct value of the pass energy before each measurement. Acquire the following scans,
while simultaneously doing the data analysis to understand what you are measuring:

1. Take an overview scan with 100 eV ≤ Ekin ≤ 1500 eV (energy step: 1 eV). Since
you are not interested in the fine details of this wide scan, you can set Ep = 50 eV
(→ low energy resolution). Repeat the scan until a good statistics is obtained. If
needed, you can sum all scans using an appropriate computer software.

2. Try to assign all maxima in the previous spectrum to known core level and Auger
peaks: this may be easier if you plot the spectrum in the binding energy scale, using
the appropriate computer software (for the moment just use a reasonable value for
the analyzer work function).

3. Zoom-in into the W 4f core level region, using an energy step of 0.1 eV. Record a
sequence of spectra with different values of Ep (e.g., 7.5, 10, 15, 20, and 50 eV). In
order to have good statistics, at each value of the pass energy select appropriately
the integration time per energy point. Observe the effect of the pass energy on the
analyzer resolution. In which situation is the spin-orbit splitting clearer?

4. Analyzing the previous data with the computer software, estimate the FWHM of
the W 4f7/2 component and plot it as a function of Ep . Which trend do you observe
(e.g., parabolic, linear etc.)? Does it appear reasonable to you?

5. Consider the W 4f spectrum at Ep = 10 eV and look for the tabulated binding energy
of the W 4f7/2 component in polycrystalline tungsten. Can you now determine the
work function of the analyzer?

6. Zoom-in into the C 1s and O1s core level regions at Ep =50 eV. Is the surface
completely free of contaminants?

7. Take a wide scan at 200 eV ≤ Ekin ≤ 530 eV (Ep = 50 eV, ∆Estep ∼ 0.3 eV), with
the X-ray source operated at a reduced anode voltage of 1.4 kV and an emission
current of 5-6 mA (consult the advisor to change the operation conditions). Now
that the Al Kα emission is suppressed, what is the origin of the spectral features
observed? Can you draw any conclusions regarding the cleanliness of the Al target
of the anode?

18
B. Deposition of the Mg metal film
With the help of the advisor you will now deposit a Mg film onto the W ribbon, with the
latter held at room temperature. You can monitor the change of pressure in the chamber
during the deposition. Proceed according to the following instructions:

8. Record an overview scan from 50 to 1500 eV (θe = 45◦ , Ep = 50 eV, ∆Estep = 1

eV). Assign all the features of this spectrum and assess qualitatively the cleanliness
and thickness of the film. Can you see any signal from the W substrate? Taking
into account the inelastic mean free path of electrons and the collection geometry
give a lower limit for the thickness.

9. Record the Mg KLL Auger spectrum in the kinetic energy range from 1060 eV
to 1200 eV (Ep = 20 eV, ∆Estep ∼ 0.2 − 0.3 eV) for:
(1) nearly normal electron exit (θe = 10◦ );
(2) nearly grazing electron exit (θe = 80◦ ).
Keeping in mind the different probing depth of the two geometries, can you conclude
that the (electronic) structure of the metallic film is uniform across its thickness?

10. Take a closer look at the previous spectra and with the help of the literature (e.g.,
Refs. [6, 13]) identify the features related to plasmon excitations. Determine the
plasmon energy.

11. Zoom-in into the Mg 1s region (θe = 45◦ , Ep = 10 eV, ∆Estep = 0.1 eV). Write
down the kinetic energy of the main line. Record a photoemission spectrum (Ep =
10 eV) from the O 1s region. Is the metal film oxygen-free? If you like, you can also
do a zoom-in into the valence band region and check whether you can identify the
Fermi edge.

C. Oxidation of the Mg film at room temperature

You will help the advisor to oxidize the Mg overlayer at room temperature, letting O2 into
the chamber (p = 2 × 10−7 mbar) for about 5 minutes. Check that the pressure remains
constant during this time and estimate the total exposure in Langmuir.

12. Zoom-in into the Mg 1s and O 1s regions, as done previously. Do you observe any
change in the position of the Mg 1s line upon oxidation? Explain the results in the
light of what you have learnt about the chemical shift.

13. Monitor the changes in the Auger spectra at θe = 10◦ and 80◦ . Which conclusions
can be drawn based on the different surface sensitivity of the two collection geome-
tries? Try to understand whether the film is uniformly oxidized and structurally
homogeneous across its thickness, or instead the surface region is different from the
inner part of the film.

D. An experiment at your choice

This last part of the experiment is intended to stimulate your own initiative. Depending
on your interest, you can either choose to study an additional ex-situ prepared sample
(option a) or to learn further details on the sample you have studied so far (option b).
Any suggestions and ideas from your side are - of course - very welcome!
Follow one of the two steps below:

14a. The ex-situ prepared sample has been already mounted in the load-lock chamber by
the advisor. For example, an organic SAM on gold. You will get a brief explanation

19
 of the structure of this molecular film and of the relevance for technological appli-
cations. After checking the pressure in the load-lock system, open the gate valve
to the main chamber and bring the sample into the measurement position in front
of the X-ray tube. You will learn how to optimize the position with respect to the
analyzer entrance slit. Take all XPS spectra (e.g., overview spectra and finer scans
on single core levels, such as C 1s, O 1s, N 1s, S 2p) which you deem are necessary
to fully characterize the elemental composition of the sample and the presence of
chemically inequivalent species. Using tables for the photoionization cross-sections
you can determine semi-quantitatively elemental ratios and, with some inputs from
the advisor, even make a guess on the chemical formula of the adsorbed molecules.
14b. If you prefer, you can get more insight into the oxidation of the Mg film, performing
the oxidation - under the supervision of the advisor - with the sample kept at a mod-
erate temperature (50-100 ◦ C). By recording relevant core level and Auger spectra,
you will be able to check whether the oxidation proceeds further, deeper and more
efficiently under these conditions.

7 Drafting your report

Write a short but exhaustive report of your experiment. After a concise introduction on
the XPS technique and instrumentation, discuss the experimental results addressing the
issues listed below. Note that all photoemission spectra should be plotted in the binding
energy scale, whereas the Auger spectra can be kept in the kinetic energy scale.
A1. Assess the cleanliness of the W ribbon and assign all features of the overview scan.
Focus your attention on the W 4f doublet, give an estimate of the analyzer work
function and discuss the effect of the pass energy on the resolution, plotting the
experimentally obtained FWHMs for the 4f7/2 component as a function of Ep . Ex-
plain the spin-orbit splitting and for the case Ep =10 eV compare the intensity ratio
of the two spin-orbit components with the expected branching ratio. Finally, from
the experimental FHWM determine a lower limit for the lifetime of the 4f7/2 core
hole by making use of Eqs. (4) and (5).
A2. Referring to the XPS spectrum at reduced anode voltage discuss the presence of
contaminants on the Al anode, which can lead to additional X-ray emission lines.
B1. Assign all features in the overview spectrum of the pristine Mg film. Consider the
plasmon loss progression in the Auger KLL spectra. Compare the value you have
determined for the bulk plasmon energy (in eV) with the predicted one in the free
electron model: ~ωp = [n~2 e2 /me 0 ]1/2 , with me electron mass and n conduction
electron density. From the mass density and atomic weight of Mg and assuming two
conduction electrons per atom, justify that nMg ' 8.6 × 1028 electrons/m3 .
B2. By consulting available literature (e.g., Refs.[6, 13]), identify the main transitions in
the Auger spectra (KL2,3 L2,3 , KL1 L1 etc.). You may also check the reliability of the
work function value determined above by comparing the tabulated and experimental
values for the intense KL2 L3 (1 D2 ) line5 .
B3. By comparing the Auger spectra in different collection geometries comment on the
thickness of the metallic Mg film and make conclusions on its uniformity.
C1. Compare the Mg 1s and O 1s spectra before and after oxidation, showing each
region on the same plot. Describe the effect of the oxidation (structural and chemical
evolution) by commenting on binding energy values and chemical shift.
5
Note that tabulated values are often given with the kinetic energy measured relative to the Fermi level.

20
 C2. Is the Mg film uniformly oxidized through its thickness or can you infer differences
between the surface layer(s) and the region underneath?

Da1 Draw conclusions on the sample composition in comparison with your expectations;
discuss the possible presence and origin of contaminants; do some semi-quantitative
analysis. In the SAM case make a guess on the chemical formula of the thiol
molecules, and comparing the binding energy of the S 2p3/2 core level to literature
values show that the S-H bond of the molecule is cleaved upon SAM formation.

Db1. Describe the results of the oxidation of magnesium at moderate temperature.

8 Useful Exercises
One of these exercises, at your choice, might be asked during the final colloquium:

1. Determine after which time interval an originally clean metal surface (e.g., Mg) will be
covered with one monolayer of O atoms at a pressure p = 2 × 10−7 mbar = 2 × 10−5 Pa
of molecular oxygen O2 .

[Hint: make use of the Hertz-Knudsen equation (see §3.3) to determine the impingement rate
of O2 onto the surface; assume s = 1 for the sticking coefficient; give an estimate of the surface
density of Mg atoms, e.g. on a Mg(0001) surface, for which the size and area of the primitive cell
are known; remember that the adsorption of O2 is dissociative.]

2. Assume a hemisherical analyzer with radii R1 (inner) and R2 (outer). What is the
potential difference ∆V that has to be applied to the hemispheres in order to bend the
trajectory of electrons of kinetic energy Ep into a circular orbit of radius R0 = (R1 +R2 )/2?
Numerical values: R1 = 9.5 cm, R2 = 10.5 and Ep = 5 eV.

[Hint: the centripetal force bending the electron’s trajectory must equate the force exerted by
the electric field; make use of the Gauss’s law.]

9 Useful tables
• Electron binding energies for all elements:
http://xdb.lbl.gov/Section1/Table 1-1.pdf

• Energies of the X-ray emission lines for all elements:

http://xdb.lbl.gov/Section1/Table 1-2.pdf

• Subshell photoionization cross-sections:

http://xdb.lbl.gov/Section1/Sec 1-5.pdf

References
[1] F. Reinert and S. Hüfner, Photoemission Spectroscopy − from Early Days to Recent
Applications, New Journal of Physics 7, 97 (2005); doi:10.1088/1367-2630/7/1/097.

[2] S. Hüfner, Photelectron Spectroscopy — Principles and Applications, Springer-Verlag,

Berlin, 2003.

[3] P. Willmott, An Introduction to Synchrotron Radiation — Techniques and Applica-

tions, J. Wiley & Sons Ltd., 2011.

21
 [4] D. P. Woodruff and T. A. Delchar, Modern Techniques of Surface Science, Cambridge
University Press, 1994.

[5] M. C. Biesinger et al., Resolving Surface Chemical States in XPS Analysis of First
Row Transition Metals, Oxides and Hydroxides: Sc, Ti, V, Cu and Zn, Applied
Surface Science 257, 887 (2010); doi:10.1016/j.apsusc.2010.07.086.

[6] J. C. Fuggle et al., X-ray Excited Auger and Photoelectron Spectra of Magnesium,
Some Alloys of Magnesium and its Oxide, Journal of Physics F: Metal Physics 5, 375
(1975); doi:10.1088/0305-4608/5/2/020.

[7] J. J. Sakurai, Modern Quantum Mechanics, Addison-Wesley Publishing Company,

Reading - Massachusetts, 1995. See pp. 341-345.

[8] B. D. Ratner and D. G. Castner, Electron Spectroscopy for Chemical Analysis (Chap-
ter 4), in: Surface Analysis — The Principal Techniques, edited by J. C. Vickerman,
J. Wiley & Sons Ltd., 1997.

[9] M. Henzler and W. Göpel, Oberflächenphysik des Festkörpers, Teubner Studienbücher,

Stuttgart, 1994.

[10] K. Oura, V. G. Lifshits, A. A. Saranin, A. V. Zotov and M. Katayama, Surface

Science — An Introduction, Springer-Verlag, Berlin, 2010. Chapter 3.

[11] J. C. Rivière, Instrumentation (Chapter 1), in: PRACTICAL SURFACE ANALYSIS,

Vol. 1: Auger and X-ray Photoelectron Spectroscopy, edited by D. Briggs and M. P.
Seah, J. Wiley & Sons Ltd., 1990.

[12] I. W. Drummond, XPS: Instrumentation and Performance (Chapter 5), in: SUR-
FACE ANALYSIS by Auger and X-ray Photoelectron Spectroscopy, edited by D.
Briggs and J. T. Grant, IM Publications and SurfaceSpectra Limited, 2003.

[13] P. Steiner et al., The KLL Auger Spectra of Na and Mg Metal and Their Plasmon
Structure, Physica Status Solidi (b) 90, 45 (1978); doi:10.1002/pssb.2220900104.

22