Jackson S.D., Hargreaves J.S.J. Metal Oxide Catalysis

Подождите немного. Документ загружается.

244 6 Photoelectron Spectroscopy of Catalytic Oxide Materials

critically review a couple of areas of metal oxide catalysis in relation to the applica-

tion of PES.

6.2

XPS as a Surface - Sensitive Technique

6.2.1

Basic Principles

(X - ray) PES is based on the observation of the photoelectric effect, in which a

photon reaching a sample can liberate an electron, which subsequently escapes

into the vacuum. As a result of energy conservation, such a transition occurs only

if the energy of the impinging photon ( h ν ) is higher than the energy required to

remove the electron from the system. This latter is expressed as a sum of two

terms: ﬁ rstly, the binding energy ( E

) describing the energy needed to remove an

electron from a certain orbital, and secondly the work function ( φ ), which is the

energy required to take an electron from inside the solid to the vacuum level.

Therefore in the simplest view, the binding energy of an electron can be calculated

by measuring the energy - dependent photocurrent and using the equation

EhE

=−−νφ

(6.1)

where E

is the measured kinetic energy of the emitted electron.

The photoemission process creates a ﬁ nal state that has a positive hole, that is,

it is lacking one electron with respect to the initial state. Hence, in a more funda-

mental treatment, the binding energy of an orbital x will be the difference between

the total energies of the ﬁ nal hole state (with N − 1 electrons) and of the initial

state (with N electrons):

EENEN

Btot tot

()

=−

()

−

()

(6.2)

Since solving the Schr ö dinger equation for systems investigated in XPS is not

possible, several approximations have been applied to theoretically describe the

photoelectric process and to calculate binding energies. It is far beyond the scope

of this chapter to present these models; instead the reader is referred to the cor-

responding literature [1, 2] .

The fact that binding energies of a particular core level in the same atom but in

different chemical environments are different has made XPS a popular technique

for characterizing solid materials. The energy difference observed in binding ener-

gies is called the chemical shift. In a simpliﬁ ed model, chemical shift occurs owing

to different screening of the Coulomb interaction between the nucleus and the

probed orbital through valence electrons. Therefore, binding energies are readily

correlated with the partial charge on an atom. With increasing positive charge

(i.e. lower screening through valence electrons) on an atom/ion in a molecule, the

binding energy increases and vice versa. In fact in this very effective view, binding

energy is a function of initial state effects. Equation 6.2 indicates clearly, however,

the often neglected fact that the measured binding energy of the investigated

system is not only affected by its initial state but also by its ﬁ nal state. Such ﬁ nal

state effects, for example inter - or extra - atomic relaxations and core polarization,

are necessary to account for secondary structures observed in photoelectron

spectra.

6.2.2

Instrumentation

6.2.2.1 The Conventional Setup

XPS instruments have been commercially available for several decades. Obviously,

signiﬁ cant developments have been made to the instrumentation since the tech-

nique was introduced, and we will only brieﬂ y describe the currently available

standard equipment. Major manufacturers in this area include: JEOL, Kratos,

Omicron, PHI, Scienta, SPECS and Thermo VG.

Photons provided by the X - ray source illuminate the sample, giving rise to the

processes described in the previous section. Typically, two methods are used to

produce X - rays: laboratory experiments use the bombardment of a metallic target

with high - energy electrons (dual anode source: Al K α , Mg K α ), whereas setups

connected to a synchrotron beamline can make use of the highly intense synchro-

tron light. The electron lens system is placed between the sample and analyzer

and transmits electrons into the analyzer. Electrons passing through the lenses

are focused onto the input slit of the analyzer and are retarded to the pass energy

( E

pass

) for subsequent energy analysis. The lens systems can usually be operated

in several different modes for spatial and angularly resolved studies and for dif-

ferent spot sizes to fulﬁ ll the requirements of different users. Commercial XPS

instruments use electrostatic analyzers, of which the hemispherical analyzer

( HSA ) is the most commonly used. Electrons entering the HSA through the

entrance slit are deﬂ ected into elliptical trajectories by the radial electrical ﬁ eld

between the inner and outer hemispheres. Only electrons with kinetic energies in

a certain energy interval are able to pass through the full deﬂ ection angle from

the entrance to the exit slit. Electrons selected by the analyzer are multiplied and

counted by the electron detector. These main components of an XPS setup are

contained within an ultra - high vacuum ( UHV ) system to (i) maintain low pressure

for the electron detector and the X - ray source, (ii) avoid discharge in the lens

system, (iii) increase spectrum intensity (signal/noise ratio), and (iv) maintain

surface cleanliness if required. UHV conditions are achieved with turbo pumps,

ion pumps or sublimation pumps. All of these components use electronics as their

power supply and control unit.

6.2.2.2 In situ (High - Pressure) Setup

The conventional (UHV) XPS technique is a powerful tool to study surface com-

position and electronic states. However, these properties are strongly dependent

6.2 XPS as a Surface-Sensitive Technique 245

246 6 Photoelectron Spectroscopy of Catalytic Oxide Materials

on the experimental conditions, such as the temperature and the gas phase above

the investigated sample. It is increasingly realized that catalysts are dynamic mate-

rials whose active centers are formed or transformed under reaction conditions.

Therefore direct in situ (i.e. not “ pre - treat and transfer ” ) experimentation is highly

beneﬁ cial in many ﬁ elds of material science, particularly in catalysis. Since any

photon source and detector require UHV conditions, the minimum requirement

for in situ operation is the separation of these units from the gas atmosphere.

A schematic of a high - pressure XPS setup is shown in Figure 6.1 . Independently

of the source applied, a thin X - ray window separates the X - ray source from

the experimental cell. As with conventional XPS instrumentation, X - rays from the

source hit the sample and induce the photoelectric effect. After traveling in the

sample cell, some of the photoelectrons reach the entrance aperture of the differ-

ential pumping stages and pass through to the electron energy analyzer. Applica-

tion of differential pumping thus allows minimization of the travel path of

photoelectrons in the gas phase. A major advantage of the system is that a gas

atmosphere can be introduced into the sample cell instead of the UHV conditions

that are obligatory for conventional XPS. The pressure of gas that can be applied

in the sample cell is limited by scattering of photoelectrons by gas - phase mole-

cules, which leads to a decrease in photoelectron signal. The maximum pressure

depends on several factors, such as the distance between the sample and the ﬁ rst

aperture, the aperture diameter, the intensity of the X - ray source, photoelectron

collection and detection efﬁ ciencies, the kinetic energy of photoelectrons and the

type of gas. Electrostatic lenses placed in the differential pumping stages can sig-

niﬁ cantly increase the collection of photoelectrons.

Differential pumping in XPS was ﬁ rst used by Siegbahn and coworkers to study

free molecules [3] . Ten years later, by using one differential pumping stage around

a high - pressure cell, the ﬁ rst experiments were carried out on solids in a gaseous

atmosphere [4] . Later designs applied additional pumping stages and/or electro-

Figure 6.1 Schematic representation of an in situ XPS setup.

static lenses accepting electrons from a higher solid angle [5, 6] . By using a

molecular beam directed towards a surface, relatively high local pressures (up to

− 2

mbar) can be achieved in front of a sample, while the base pressure of the

analyzer chamber is still in the 10

− 6

– 10

− 7

mbar range. Such experiments are carried

out, for example, in Steinr ü ck ’ s group [7] .

Our survey of the recent literature on high - pressure XPS in the (0.1 – 5) mbar

range indicates that mainly two groups (Salmeron at Berkeley and that of the

authors) contribute to this ﬁ eld. These instruments were developed as a result of

close cooperation between the two groups, and their operation principle is nearly

identical. Both setups operate at their local synchrotron source. The samples are

usually transferred from a pre - chamber and can be mounted inside the reaction

cell, 1 – 2 mm away from an aperture, which is the entrance to the differentially

pumped stages of the lens system of the hemispherical analyzer. The sample can

be heated from behind using an infrared laser system. Gas ﬂ ows into the reaction

cell are regulated by leak valves and mass - ﬂ ow controllers. The gas - phase composi-

tion, to monitor ambient cleanness or catalytic activity, is recorded on - line by mass

spectrometers located either in the ﬁ rst differentially pumped lens stage or con-

nected directly to the reaction chamber. The surface of the sample can optionally

be cleaned by Ar

sputtering.

The high - pressure setup can be easily adapted for use in near - edge X - ray Absorp-

tion Spectroscopy ( XAS ). In this process the incoming excitation energy is scanned

across a range (dependent on the synchrotron light available and the absorption

edge of interest). All electrons emitted by the sample are collected and ampliﬁ ed

to produce a current. This form of XAS is called Total Electron Yield ( TEY ). (The

reader is referred to Stoehr [8] for an in - depth discussion of XAS.) The possibilities

of this technique will be brieﬂ y highlighted in Section 6.3.2 .

6.2.3

Quantiﬁ cation of XPS Spectra

The intensity of an XPS peak ( I

) is a strong function of (i) the incoming photon

ﬂ ux, (ii) the concentration of the given element, (iii) its photoionization cross -

section (which is excitation - energy dependent), (iv) the mean free path of the

emitted photoelectron, and (v) further instrumental parameters (such as photo-

electron collection and detection efﬁ ciency). By deﬁ ning atomic sensitivity factors

( S , as an overall factor summing up the effects of iii – v), the atom fraction of any

element in a sample can be calculated as:

XIS IS

iiAAA

()()

∑

(6.3)

assuming constant photon ﬂ ux, as in the typical XPS setup using Mg or Al K α

excitation, and homogeneous distribution of the constituent elements. Such sen-

sitivity factors are known and commonly used; however, the accuracy of the cal-

culation relies on the rare situation of an atomically homogeneous distribution of

elements in the sampling depth. For example, overlayer - type contamination has a

6.2 XPS as a Surface-Sensitive Technique 247

248 6 Photoelectron Spectroscopy of Catalytic Oxide Materials

strong effect of decreasing the intensity of low E

(high E

) peaks, affecting the

calculation. If a synchrotron source is used (with a known photon ﬂ ux versus

energy function), this effect can be circumvented by tuning the photon energy of

each investigated core level to record continuously electrons with similar kinetic

energy. As an additional beneﬁ t, the sampling depth for all elements will be

similar. This approach opens up the possibility to perform non - destructive depth

proﬁ ling experiments via intentional variation of photoelectron E

Based on the Lambert – Beer - type law of exponential attenuation of peak intensity

( I ) with excitation depth ( z ),

Iz I

()

−

()

λθcos

(6.4)

where λ is the attenuation length and θ is the angle of emission with respect to

the surface normal, known depth distributions of constituents can be modeled

and the thickness of layers can be calculated [9, 10] .

6.2.4

The XPS Spectrum (a Qualitative Picture)

An X - ray photoelectron spectrum is an electron - intensity (e.g. counts per

second) plot as a function of binding energy. It is either a wide scan showing

different types of transitions (core levels, Auger lines, valence levels, secondary

structures) or plotted in a narrow E

range displaying only one particular feature.

The baseline going through an elastic transition increases with increasing binding

energy, owing to electrons that have suffered some energy loss during their

escape.

6.2.4.1 X - ray Transitions

Core - level excitation from a closed sub - shell results in the most intense and typi-

cally narrow photoelectron lines. The width of a core level measured at half its

maximum height, full width at half maximum ( FWHM ), is usually deﬁ ned as the

convolution of the width of the photon source, the natural line - width and the

analyzer resolution. All transitions (except from s levels) give rise to doublets due

to spin – orbit coupling, in which the spin of the unpaired electron left in the orbital

can couple in a parallel or an anti - parallel manner with its orbital angular momen-

tum. As all the relevant elements (except H) show at least one transition detectable

by this technique, and the core levels of different elements are usually readily dis-

tinguishable, XPS is widely used for elemental analysis in the sampling depth of

the topmost atomic layers.

Following the formation of an initial core level vacancy ( E

), an electron from

a higher lying level ( E

) drops down to ﬁ ll this vacancy. If the energy difference

between these two orbitals is high enough to remove a third electron from its

orbital ( E

), then this electron can leave the sample with kinetic energy

EEEE

KBBB

123

()

≈−−

(6.5)

This process is called Auger transition and as a result Auger lines will show up

in the XPS spectra. They appear as a group of lines, as several Auger transitions

can occur. Because Auger electrons have kinetic energies independent of the initial

photon radiation, they appear in the binding energy plot at different energy posi-

tions when different photon excitation is applied.

The energy levels approximately 5 – 20 eV below the vacuum level (i.e. low E

)

correspond to bonding or delocalized orbitals. In a typical solid sample these

closely spaced orbitals overlap and form a valence (electronic) band. The Fermi

level, E

, designates the energy separation between ﬁ lled and empty states at zero

Kelvin and represents a reference position of zero on the binding - energy scale of

solid samples. The structure of the valence (and conduction) band determines

optical and electrical transport properties, chemical reactivity and catalytic proper-

ties. As, however, all the components of the investigated sample contribute to the

valence band spectrum, its interpretation generally requires theoretical (density of

states) calculations with high predictive ability.

Core polarization will give rise to multiplet splitting of the investigated core level

if the system has an unpaired open valence conﬁ guration. The coupling of the

remaining core electron with the unpaired valence electrons can create several

ﬁ nal states with different energies.

After core hole formation, relaxation in the valence orbitals can give rise to pro-

motion of valence electrons into unoccupied levels. If this reorganization is fast,

and the energy required for this transition is not available to the primary photo-

electron, shake - up satellites can show up on the low kinetic energy (high E

) side

of the main peak. Further loss lines can be created if the photoelectron passing

the solid excites group oscillation of the conduction electrons (plasmon loss).

Additional secondary features, X - ray satellites, can appear in the spectra if a

non - monochromatic source is used; moreover X - ray ghost lines show up when the

exciting photon originates from impurity elements in the X - ray source.

6.2.4.2 Binding Energy in Practice; An Issue?

A substantial part of the information obtainable from a photoemission experiment

concerns the qualitative data derived from line positions and line proﬁ les for

unresolved contributions to a given excitation line. This information, dealing with

binding energies of core and to a lesser extent of valence lines, is used to deduce

qualitative chemical information about the bonding and reactivity of species under

consideration. The acronym “ ESCA ” ( electron spectroscopy for chemical analysis )

expresses this application of PES and was historically the major driver for the

popularization of the method before it became apparent that it is a surface - sensi-

tive technique requiring detailed information about surface structure and purity

of a sample in order to derive meaningful average information about the bulk

chemical bonding.

For the purpose of catalyst analysis, the weaknesses of ESCA turn into strengths

as it is the chemical bonding of the outer surface of a solid that is of interest and

only to a lesser extent its bulk chemical structure. Most of our theoretical under-

standing of chemical bonding refers, however, to the bulk state (crystal structures)

6.2 XPS as a Surface-Sensitive Technique 249

250 6 Photoelectron Spectroscopy of Catalytic Oxide Materials

and cannot directly be applied for the interpretation of PES result. As outlined

elsewhere in the chapter, the surface sensitivity of PES is a function of its excita-

tion energy and is thus to a certain degree variable. Under no conditions can pure

surface sensitivity be realized; the information must always be divided into a bulk

and a surface contribution.

The fact that PES strongly perturbs the electronic structure of an atom [11 – 13]

(also the preceding section) renders it difﬁ cult to compare the binding energy of

an energy level obtained from the position of an experimental spectrum to a cal-

culated value of the inverse dissociation energy of the respective state obtained by

theoretical treatment. Moreover, there are serious experimental issues with deﬁ n-

ing the energy scale of the method in an absolute way. Today, it is customary to

use the following conventions for an experimental binding - energy scale and to

discuss the positional information in PES in terms of “ chemical shift ” , which is

given for each element and each energy level with the convention that the elemen-

tal state of the atom serves as zero for the chemical - shift scale. For a few elements

of relevance in the present context this deﬁ nition is not easy, as the elemental state

is not solid, which causes additional problems with the shift scales. In these cases

(notably for oxygen) experimental binding - energy positions frequently observed

are used as zero for the shift scale (e.g. 530.0 eV for oxygen 1s).

The quantities contributing to the position of a photoelectron ( PE ) line in solids

can be summarized by the following equation:

EE EE E E

B h K rel exp ch

=−−−−

(6.6)

Detailed treatments of this issue can be found in the literature [3, 14, 15] . Here

we only brieﬂ y review the contributions under practical considerations and without

discussing the deep quantum chemical implications involved.

The binding energy ( E

) should be given as the position of the center of gravity

of a ﬁ t of the experimental line proﬁ le obtained after satellite subtraction (for labo-

ratory experiments) and non - linear background correction [16] . The proﬁ le func-

tion used to approximate the lines must be calibrated for a given analyzer and

must not be changed during the ﬁ tting of different spectra.

The excitation energy ( E

h ν

) should be as monochromatic as possible to maximize

the resolution of the experiment. The kinetic energy ( E

) is the direct experimental

result of the PE experiment. It is used as energy scale in Auger spectroscopy but

today no longer in PE spectra. Note that this is not the case in older literature

where often the kinetic energy was plotted.

The kinetic energy contains the original chemical information of the PE experi-

ment. If the kinetic energy of the emission from the same energy level is compared

for samples with a different chemical bonding (different ground state electronic

structure) then one observes a shift between the positions of the emission maxima

that is considered as being proportional to the difference in ground state electron

density in the valence band. This effect is the origin of chemical shift scales that

exist as empirical correlations for many elements. As a rule of thumb, it is assumed

that a difference of one formal oxidation state unit will lead to a ground state

chemical shift of about 1 eV for d - block elements. In main group elements this

shift increment can be larger. It needs to be pointed out that these rules are not

supported by quantum chemical considerations and that many exceptions exist.

Nevertheless, they are very popular and give rise to frequent and extensive debates

in the literature about the nature of chemical bonding in a particular system. In

light of the lack of theoretical justiﬁ cation for such shift scales, it is not recom-

mended to use such shift arguments in attempts to derive a description of the

chemical bonding. Other properties of the experimental spectrum, such as the

satellite features [11, 14, 17, 18] or the valence band, should be used instead.

Despite these caveats, the use of these correlations is frequent in catalysis science

and several examples of their application in vanadium - based systems will be dis-

cussed below. It is, however, no surprise that substantial differences and debates

exist in the literature about the nature of the chemical bonding as derived from

such empirical correlations. It is only considered useful to consider empirical shift

arguments as ordering criteria for systems with different chemical bonding, being

related, for example, to functional performance. The jump to the conclusion that

ground state electronic differences may be responsible for observed functional

differences (activities or selectivity in catalysis) is unjustiﬁ ed on the basis of chemi-

cal shift arguments. A prominent example of this debate is the role of putative

pentavalent vanadium compounds in selective oxidation catalysis, which will be

elaborated in detail below. An exception would be the comparison of spectra of

different samples of the same general chemical compound of a given element, for

example the analysis of a series of defect species of an oxide of a given transition

metal. Also in a series of compounds with large electrostatic contributions to the

chemical bonding (salts) [19 – 22] , the assumption of a large contribution of ground

state electron density to the chemical shift may be valid.

The relaxation energy ( E

rel

) is a sum of contributions describing the response of

the electronic structure to the creation of the core hole on the femtosecond tim-

escale. This response is detected by the photoelectron leaving as an exit wave from

the excited atom and modiﬁ es its energy to a considerable extent. The effect is

larger the deeper the core hole in energy, and can reach values of about 10% of

the total energy. For deep core holes the relaxation is only weakly dependent on

the chemical nature of the sample, but this is not the case [15, 23, 24] for high -

lying core levels. The relaxation phenomena are caused by the partial or complete

screening of the initial core hole by the surrounding electrons. The effectiveness

of the screening depends on the extent of overlap between the wave functions of

the parent state and the state that contributes to the screening. The screening is

largely dominated quantitatively by weakly localized valence states, for example

“ free electrons ” from metals are very effective. Also valence electrons from sur-

rounding atoms can effectively contribute to the core hole relaxation (extra - atomic

contribution). These two contributions mean that the chemical structure can have

a substantial effect on the extent of core hole relaxation when comparing chemical

compounds with insulating properties and high oxidation states to matrices with

metallic properties and low oxidation states. As the total contribution of relaxation

to the experimental binding energy is large (easily a few tens of electron volts) it

6.2 XPS as a Surface-Sensitive Technique 251

252 6 Photoelectron Spectroscopy of Catalytic Oxide Materials

is quite likely that small changes in the relaxation pathways created by a difference

in chemical bonding of an atom will cause substantial changes in the binding -

energy scale (a few electron volts) that are not caused by the difference in ground

state valence electron density. A common example of such a relaxation shift is the

change in PES spectra of nanoparticles as a function of their particle size [23, 25,

26] . Nanostructures change lattice constant or long - range structural ordering at

much larger sizes than the transition that occurs from collective to atomic elec-

tronic structure (at cluster sizes below 100 atoms). The subtle changes in long -

range ordering exert a sizeable effect on relaxation (up to about 1 eV) whereas the

local ground state electronic structure stays largely unperturbed (e.g. metallic). At

the limit of nanoparticles PES can become, by the action of relaxation phenomena,

a structure - sensitive method, although the basic physical information is massively

dominated by the inner - atomic contributions [27, 28] .

The term E

exp

summarizes all contributions arising from the experimental setup.

In PES it is necessary that electrons ﬁ rst leave the sample where they have to

overcome, besides the local binding energy, the collective attraction of the work

function. At the surface of the electron detector, the photoelectron has to penetrate

back into a solid where the inverse of the work function becomes liberated as

energy gain. As these two quantities, with absolute values between 3 and 6 eV, are

largely unknown and vary strongly with geometric surface state and with the pres-

ence of adsorbate in the spectrometer and at the sample, it became customary to

lump these contributions into a correction voltage that is applied to the analyzer

and set by calibration using a series of polycrystalline clean metal samples [29]

(copper, silver and gold) that can also be used to determine the linearity of the

binding - energy scale. The term thus does not occur in evaluations of correctly

calibrated instruments as long as solids are being measured. In the present context

of in situ catalytic studies, the data are often compared to gas - phase spectra. Such

samples do not exhibit a work function and thus the work function correction

applied to the spectrometer gives rise to an incorrect calibration of the binding -

energy scale for gases. The convention for gases is to use the lone pair emission

lines for Ar and di - nitrogen as calibration standards [3], and these show different

values in solid - state spectrometers and in dedicated gas - phase instruments. Care

must thus be taken when comparing binding energies between adsorbates and

free molecules, where not only strong relaxation effects but also substantial experi-

mental peculiarities of the instruments used may lead to hard - to - interpret “ chemi-

cal shifts. ” In unfortunate cases the effects may cancel out, as relaxation shift and

unintentionally applied work function correction are of the same order of magni-

tude (5 eV).

The term E

stands for charging contributions. This term is absent in spectra

of samples exhibiting a ﬁ nite density of states at the Fermi level. For all practical

purposes this is correct for all true metals and for many semiconductors with

intrinsic states near zero binding energy. Many systems relevant in catalysis do,

however, not fulﬁ ll this condition (all non - black samples, glasses, porous materi-

als, supports) or even worse, are composites of metallic and non - metallic systems,

giving rise to mixed metallic – insulating behavior of their surface under PES. Such

systems are mostly supported catalysts [30] but also working catalysts with a metal-

lic bulk covered with non - metallic deposits or with a water ﬁ lm. Partly reduced

semiconducting solids, or strongly defective semiconductors exhibiting extrinsic

states from defects within their band gap, also belong to the class of systems which

exhibit no deﬁ ned zero point of binding energy. Three phenomena will result from

this situation, all affecting individually or in concert the binding - energy scale.

In the simplest case of extrinsic semiconductors, the temperature and all chemi-

cal processes modifying the abundance and nature of the defect states will modify

the zero of the binding - energy scale and so cause “ shifts ” that are solely due to

calibration and not at all to local electronic structure modiﬁ cation (e.g. most oxida-

tion states in an oxide are not affected if a fraction of the extrinsic defects are

created or healed by diffusion of oxygen species, yet shifts up to 2 eV can result

from this process).

The second most frequent state is an apparent shift of the binding - energy scale

due to insufﬁ cient ﬂ ow of electrons between sample and spectrometer (insulation)

causing a stable electric ﬁ eld gradient [31, 32] to occur between sample surface

and spectrometer. This ﬁ eld can be very substantial, causing charging shifts that

range between a few electron volts in laboratory instruments to a few hundreds

of electron volts in synchrotron - based experiments, as the magnitude of the ﬁ eld

depends on the photoelectron ﬂ ux. Typical measures for avoiding the adverse

consequences are (i) ” calibration ” with an assumed position of a photoelectron line

(carbon at 285 eV or oxygen at 530 eV are frequent examples) (ii) charge injection

into the surface by shining electrons or low - energy photons on it with the strong

risk of modifying the chemical state of the native surface (chemical reduction) and

(iii) increasing the electron carrier density (electrons or protons) by heating the

sample or by performing a chemical reaction in situ .

The most difﬁ cult case of a poorly deﬁ ned binding - energy scale is the occurrence

of differential charging. This very frequent case occurs with chemically inhomo-

geneous systems (supported catalysts, metals with hydrocarbon deposits, oxide

patches on metals, partly reduced oxides). In these cases, a distribution of binding -

energy scales exist giving rise to a distribution of emission line positions. Neither

common referencing nor the application of charge neutralization will remove this

distribution. In fortunate cases, a series of discernible peak structures will occur.

In many cases, however, an unresolved distribution will cause line broadening of

unknown shape. A clear indication of the presence of differential charging is the

modiﬁ cation of emission lines in shape and/or position with a variation in X - ray

excitation conditions (intensity or angle of incidence). If this occurs the data are

basically not interpretable and the experiment cannot be continued in any mean-

ingful way.

For the purpose of the present chapter, a comment on the interpretation of

oxygen 1s binding energies in oxides seems appropriate. The contribution of the

surface terminating layer and of a possible adsorbate to the total O 1s spectrum

of an oxide catalyst is low (an estimated 15% for laboratory XPS) as in oxides the

depth of information [33] is commonly larger than estimated from the universal

relation between kinetic energy and escape depth, which is valid for metals only.

6.2 XPS as a Surface-Sensitive Technique 253