Drake G.W.F. (editor) Handbook of Atomic, Molecular, and Optical Physics

Подождите немного. Документ загружается.

1343

Surface Physic

89. Surface Physics

This chapter describes various applications of

atomic and molecular physics to phenomena that

occur at surfaces. Particular attention is placed

on the application of electron- and photon-atom

scattering processes to obtain surface speciﬁc

structural and spectroscopic information.

The study of surfaces and interfaces touches

on many ﬁelds of pure and applied science.

In particular there are applications in the

ﬁelds of semiconductor processing, thin ﬁlm

growth, catalysis, corrosion and fundamen-

tal physics in two-dimensions. A number of

recent texts cover surface physics in gen-

eral [89.1–4] as well as speciﬁc areas such as

experimental techniques [89.5], surface electron

spectroscopies [89.6, 7], and the application of

synchrotron radiation to surface science [89.8].

Also a number of book series that deal with ar-

eas of particular interest are published at regular

intervals, such as surface chemistry [89.9], sur-

face vibrations [89.10] and stimulated desorption

processes [89.11].

89.1 Low Energy Electrons and Surface Science1343

89.2 Electron–Atom Interactions ..................1344

89.2.1 Elastic Scattering:

Low Energy Electron

Diffraction (LEED).......................1344

89.2.2 Inelastic Scattering:

Electron Energy Loss

Spectroscopy ............................1345

89.2.3 Auger Electron Spectroscopy .......1345

89.3 Photon–Atom Interactions ...................1346

89.3.1 Ultraviolet Photoelectron

Spectroscopy (UPS).....................1346

89.3.2 Inverse Photoemission

Spectroscopy (IPES) ....................1347

89.3.3 X-Ray Photoelectron

Spectroscopy (XPS) .....................1348

89.3.4 X-Ray Absorption Methods .........1348

89.4 Atom–Surface Interactions ...................1351

89.4.1 Physisorption ...........................1351

89.4.2 Chemisorption ..........................1352

89.5 Recent Developments...........................1352

References ..................................................1353

89.1 Low Energy Electrons and Surface Science

To obtain information speciﬁc to the ﬁrst few layers of

atoms at the surface of a solid, techniques must be de-

vised that discriminate between signals from the surface

and from the bulk of the material. If a solid has dimen-

sions of ≈ 1cm

, then only approximately one atom in

is located at the surface. This makes the application

of bulk techniques to the study of surfaces (e.g. X-ray

diffraction [89.12]) problematic.

There are several experimental approaches to achiev-

ing surface sensitivity. If the probe used is an atom or

molecule that can be scattered from or desorbed from

the surface, then these species can be analyzed by tech-

niques such as mass spectrometry or resonant ionization.

A more recent innovation has been the use of scanning

probe microscopies [89.13] (e.g., the scanning tunnel-

ing microscope and its variants) which exploit a surface

sensitivity such as the surface valence electron density.

However the most common surface analytical tech-

niques use the intrinsic surface sensitivity of low energy

(10–2000 eV) electrons.

The surface speciﬁcity of low energy electrons arises

from the very short inelastic mean free path (MFP)for

these electrons in a solid. This property can be seen in

Fig. 89.1 which plots measured values of the inelastic

MFP for a number of materials and electron energies

between 1 and 2000 eV. The curve is called “univer-

sal” because the same general trend of short inelastic

MFP is observed for nearly all materials [89.14]. The

dominant energy loss mechanisms are valence band

excitations (plasmons and electronic excitations), and

since most materials have similar valence electron den-

sities, the resultant inelastic scattering MFP is to a good

approximation, material independent.

The very short inelastic MFPs for low energy elec-

trons has the result that any that escape from the solid

without having undergone inelastic scattering can only

Part G 89

1344 Part G Aplications

100

10 100 1000

3456 2 3456 2 3456 2

Mean free path λ(Å)

Electron energy (eV)

Fig. 89.1 The variation of the inelastic mean free path

with electron kinetic energy (“universal curve”). Based on

Briggs and Seah [89.14]

have originated very close to the surface, usually

within a few atomic layer spacings. Many of the sur-

face analysis techniques described in this chapter use

this surface sensitivity to obtain surface structural and

spectroscopic information. The same techniques are

easily applicable to conducting and semiconducting

materials, and can also be applied to insulating ma-

terials, if the charging effects are dealt with in some

manner.

89.2 Electron–Atom Interactions

89.2.1 Elastic Scattering:

Low Energy Electron Diffraction

(LEED)

The elastic scattering of low energy (20–500 eV) elec-

trons at surfaces is historically important in physics as

the experiments of Davisson and Germer provided early

experimental evidence of the wave nature of electrons.

After these initial experiments, the technique was largely

unused until the advent of cleaner ultra-high vacuumsys-

tems and surface preparation techniques in the 1960’s.

LEED is one of the most important and widely used

surface characterization techniques due to its surface

sensitivity and wide utility [89.5].

Diffraction from a two-dimensional net of scatterers

results in a two-dimensional array of reciprocal lattice

“rods” oriented normal to the surface. All kinematically

allowed rods intersect the Ewald sphere at all energies

— unlike the case in bulk X-ray diffraction. The short

inelastic MFP for scattering low energy electrons yields

the surface sensitivity of LEED for a crystal surface. The

kinematic theory used in X-ray diffraction is not directly

applicable to LEED since the low energy electrons can

undergo several elastic collisions in the surface region.

The elastic MFPs for low energy electrons are compara-

ble in magnitude to the inelastic mean free paths shown

in Fig. 89.1. This “multiple scattering” does not affect

the positions of the diffraction beams, but does alter their

intensities due to interference effects.

Kinematic analysis is sufﬁcient to determine the

diffraction beam positions for a proposed structure, and

this is the most common use for LEED. A diffraction

pattern is often sufﬁcient to determine the surface pe-

riodicity and unit mesh size. However determining the

surface crystal basis from the intensities of the diffracted

LEED beams requires moderately sophisticated calcu-

lations to be performed for proposed structures.

Quantitative LEED generally compares measured

“I(V) curves” (the intensity I of a particular diffrac-

tion beam as a function of electron energy measured

in volts) with a calculated I(V) proﬁle for a proposed

surface crystal structure. These calculations determine

the propagating electron wave functions Ψ

LEED

that

take into account electron-ion core scattering cross

sections and phase shifts, available multiple scatter-

ing pathways, inelastic scattering cross sections and

Debye-Waller effects. The structural parameters are var-

ied systematically until agreement can be reached with

the experimental I(V) curves [89.15, 16]. These calcu-

lations can yield surface atom positions to better than

0.1 Å vertically and 0.2 Å horizontally. Over 1000 sur-

face structures have been determined using LEED and

associated techniques [89.17].

Clean surfaces often have a different crystallography

than simple termination of the bulk crystal structure. Due

to the absence of neighbors above the surface, the surface

atoms can undergo both relaxation (change in interlayer

spacing) and reconstruction (changes in periodicity and

bonding) [89.2]. While relaxation and reconstruction do

occur on all types of surfaces, the reconstructions found

on semiconductors are most striking. The strong co-

valent bonds that are broken when a semiconductor is

cleaved give rise to high energy “dangling bonds”. The

surface energy is minimized by having the surface re-

construct to reduce the total number of these bonds. The

resultant structure is formed through a balance between

eliminating as many dangling bonds as possible and the

resultant stress caused in other bonds due to the dis-

Part G 89.2

Surface Physics 89.2 Electron–Atom Interactions 1345

placements of the surface atoms. One example of this

is the clean Si(111) surface, in which 49 surface atoms

form a new periodic arrangement to make up the recon-

structed Si(111)–(7×7) unit cell [89.2]. The number of

dangling bonds is thereby reduced from 49 to 19.

89.2.2 Inelastic Scattering:

Electron Energy Loss Spectroscopy

Low energy surface excitations such as phonons, plas-

mons and electron-hole pair excitations can be studied

using inelastic electron scattering. High resolution elec-

tron energy loss spectroscopy (HREELS) [89.18]uses

incident electron beam energies of 1–300 eV with

energy resolutions of 1–10 meV. Inelastic scattering

phenomenon are divided into three types: dipole, impact

and resonant scattering. Surface vibrations are also stud-

ied using infra-red absorption spectroscopy and inelastic

atomic beam scattering techniques [89.19].

Dipole Scattering

In dipole scattering HREELS, the incident electron

(1 < E

< 20 eV) undergoes a long range Coulomb in-

teraction with dipole ﬁelds associated with a dynamic

dipole moment. The main characteristic of dipole scat-

tering is that the inelastically scattered electrons have

an angular distribution that is strongly peaked (width

∆θ ≈

ω/E

) close to the specular direction (θ

= θ

In this limit there is no momentum exchange with the

surface vibrational mode (i. e., k



= 0 in the bandstruc-

ture), so dipole scattering HREELS is similar to infrared

spectroscopy in that only ir active vibrational modes at

the zone center can be studied. At the surface of a con-

ductor, dielectric screening guarantees that only ir active

modes having a component perpendicular to the surface

are visible — modes having dipole moments parallel to

the surface are screened by the surface and so cannot

be excited by dipole scattering. This “quasi-selection

rule” allows the site symmetry for some systems to be

decided [89.18]. Dipole scattering is most commonly

used in the study of vibrations of molecules on sur-

faces, which often have dipole-active vibrational modes

that are both intrinsic and caused by adsorption (i. e.,

frustrated translations and rotations).

Impact Scattering

In impact scattering, the incident electron samples the

short-range interatomic potential, and is not restricted

to dipole-active vibrational modes [89.18]. Higher elec-

tron energies are used (30 < E

< 300 eV) since the cross

section for impact scattering generally increases with en-

ergy. The inelastically scattered electrons are distributed

at all angles, and so there can be exchange of parallel mo-

mentum (k



) with the surface, allowing the mapping of

surface excitation bandstructure

ω(k



). The cross sec-

tions for impact scattering HREELS can be calculated

using an approach similar to LEED, but now consider-

ing the normal displacements of atoms from equilibrium

positions. The scattering potential V(r, [R]) (where [R]

is the set of position vectors of the N atoms in the sur-

face region) can be expanded in terms of the normal

displacements µ

from equilibrium according to



r, [R]



= V



r, [R

]





i=1



∇



r, [R]







]

· µ

+··· .

(89.1)

The ﬁrst term is responsible for elastic scattering

(LEED) and the second term is the dominant term for

impact scattering. The inelastic cross section for mode i

is then [89.20]

dσ

dΩ

=|Ψ

∗

LEED

∂V

∂µ

|Ψ

LEED

)|

(89.2)

using electron wave functions calculated using the same

formalism as LEED. The symmetry properties of this

matrix element can be used to determine the polariza-

tion direction of surface vibrational modes by searching

for systematic absences in the inelastic intensity in par-

ticular high symmetry direction of the surface [89.20].

Impact scattering studies are most commonly made in

the study of surface phonon bandstructure and the vi-

brational modes of adsorbed molecules that are not

dipole-active.

Resonant Scattering

Resonant electron scattering at surfaces [89.21, 22]is

usually applied to the study of adsorbed molecules, and

has much in common with resonant scattering from gas-

phase atoms and molecules (see Chapt. 47). In resonant

scattering, the incident electron combines with a target

molecule to form a short-lived molecular ion, which

subsequently decays and can leave the molecule vi-

brationally or electronically excited. The cross sections

for this process have a typical proﬁle. For example, in

a shape resonance the formation of the temporary nega-

tive ion intermediate corresponds to adding the incident

electron to a particular unoccupied orbital of the tar-

get molecule. The study of the angular dependence of

resonance scattering cross sections [89.22] can yield in-

formation on the orientation of molecules on surfaces,

Part G 89.2

1346 Part G Aplications

since the electron capture and emission cross sections

are ﬁxed by the molecular orientation and the resonance

symmetry.

89.2.3 Auger Electron Spectroscopy

Auger electron spectroscopy (AES) is one of the most

widely used surface science techniques due to its chem-

ical and surface sensitivity [89.5,14,23]. Core holes are

created in near surface atoms using a high energy elec-

tron beam (2–5 keV, 1–100 µA) or less commonly, an

X-ray source. For these low binding energy core holes,

the Auger decay mode is highly probable (Sect. 61.2).

Although the Auger energy and lineshape can give spec-

troscopic information, in surface science this is seldom

used – rather it is the chemical ﬁngerprint of the atoms

which is of interest.

Auger electron spectroscopy is a surface sensi-

tive technique by virtue of the low kinetic energy

(50–1000 eV) of the emitted Auger electrons. Auger

electrons from atoms more than a few Ångstroms below

the surface are inelastically scattered and so not detected

by the energy selective detector. The kinetic energy of

the Auger electrons is

= E

− E

− U (89.3)

where E

is the binding energy of the initial core elec-

tron and E

, E

are the binding energies of the other

electrons (one or both are valence levels) involved in

the Auger process. Energy shifts and relaxation are

accounted in the term U which includes hole-hole inter-

actions and atomic and solid state (dielectric) screening

of the holes, and hence can be sensitive to the local

chemical environment.

Auger spectra are typically obtained in derivative

mode [i. e., N



dN(E)/ dE] to separate the Auger tran-

sitions from the secondary electron background, and

comparison is made to reference spectra [89.24]. The

raw chemical sensitivity of AES is very high, and sur-

face concentrations of ≈1% of many common chemical

species can be detected. Semiquantitative measurements

may be made by comparison to these reference spectra,

but such comparisons only give atomic concentrations

within a factor of 2 (or worse) since the measured AES

signal can be modiﬁed by a number of factors. More

precise quantitative measurements can be made by cali-

bration of the AES intensity for the atomic constituents

at a surface [89.5]. The AES sensitivity to atoms A on

a clean substrate (atoms B) can be determined if the ab-

solute quantity of A can be established by some other

means.

The chemical sensitivity of AES can also be ex-

ploited as a form of chemical microscopy (scanning

Auger microscopy) since the exciting electron beam that

is used may be focused to a very small size. The Auger

signal from the small target volume can be analyzed for

speciﬁc chemical components. By rastering the incident

electron beam, a chemical map of the surface can be

made.

89.3 Photon–Atom Interactions

89.3.1 Ultraviolet Photoelectron

Spectroscopy (UPS)

Photoelectron spectroscopy (PES) (Sect. 61.1) has been

historically divided into UV photoelectron spectroscopy

for low photon energies (generally the study of valence

electron states) and X-ray photoelectron spectroscopy

(study of core electron levels). The use of low en-

ergy photons (5 < hν<50 eV) for PES of solids has

the advantage that the photons have a negligible mo-

mentum (k ≈ 0). This allows straightforward band

mapping since the transitions are vertical in momentum

space:

hν = E

(k+ G) − E

(k) (89.4)

where k is the electron state wavevector and G is a re-

ciprocal lattice vector.

Most UPS studies are done using angle-resolved

photoelectron detection (also called angle resolved pho-

toelectron spectroscopy or ARPES). The kinetic energy

and emission angle θ of the photoelectrons are

measured, allowing the initial state binding energy and

momentum parallel to the surface





to be determined

from





2mT

sin θ. (89.5)

The mean free path of the photoelectrons from va-

lence levels allows both surface and bulk electron states

to be studied. Separation of the surface from bulk bands

can be accomplished in several ways [89.5]. Since a crys-

tal surface has two-dimensional symmetry, only the k



component of momentum is a good quantum number

for the surface states. Also, in transporting the surface or

Part G 89.3

Surface Physics 89.3 Photon–Atom Interactions 1347

bulk state photoelectron of momentum kthrough the sur-

face to the detector, only the k



component is conserved.

A bulk state disperses in three-dimensions, requiring

knowledge of k= k



+ k

⊥

. The magnitude of k

⊥

cannot

be determined unless the inner potential (change in po-

tential normal to the surface) is known by some other

means. This property allows bulk and surface states to

be distinguished since surface states remain ﬁxed in en-

ergy for a constant k



, while bulk states disperse if k

⊥

changed while k



is kept constant. This is done by mea-

suring the photoelectron spectrum for a range of UV

photon energies at a ﬁxed value of k



(often at θ = 0so



= 0). Bulk states disperse since k

⊥

varies as the pho-

ton energy is changed, but the surface states remain at

a ﬁxed binding energy in the photoelectron spectrum.

The photoelectron transition matrix element be-

tween initial and ﬁnal states |i and | f  due to the

incident photon vector potential A is

I ∝







A· p+ p· A







≈







2A· p







. (89.6)

The spatial variation of A near the surface (the surface

photoeffect)can be neglected, although this is not strictly

valid at these low photon energies [89.26].

Due to the low photoelectron kinetic energies in

most UPS work, precise calculation of photoelectron

spectrum intensities is rather difﬁcult due to multiple

scattering and phase shifts sensitive to valence electrons.

However, the initial electron state symmetries can be de-

termined using the symmetry properties of the matrix

element (89.6). Selection rules allow the state symme-

tries to be determined by measuring the photoemission

spectrum along high symmetry directions of the surface

using polarized UV radiation [89.5].

The valence electron states of adsorbed molecules

can also be studied using UPS. Peaks in the photo-

electron spectrum due to valence levels of adsorbed

molecules tend to have larger linewidths (≈ 1 eV typ-

ically) than in the gas phase due to solid state and

instrumental effects, so vibrational structure is seldom

resolved. The positions of the molecular valence states

are shifted in energy due to the surface work function

and solid state relaxation (dielectric screening) effects.

In addition to these rigid shifts, chemical shifts are also

observed due to bonding (chemisorption) interactions

between the molecule and surface. The molecular char-

acter of the valence orbitals is often retained from the

gas phase, so the shifted levels can be identiﬁed by using

the symmetry properties of the photoemission matrix el-

ement. For example, photoemission from gas phase CO

shows three valence states: 5σ ,1π and 4σ in order of

increasing binding energy. A series of photoemission

spectra [89.25] for gas phase CO, solid CO and CO

chemisorbed on several transition metal surfaces (in or-

der of increasing CO–surface binding energy) is shown

in Fig. 89.2. For solid CO and weakly chemisorbed

CO/Ag(111) all three valence orbitals of molecular CO

are well resolved, though the adsorbed CO spectrum is

shifted rigidly in energy by relaxation effects. Since CO

is chemisorbed more strongly on Cu(111) and Pd(111),

the 1π and 5σ valence states shift relative to one another.

The strongly overlapping 1π and 5σ states observed for

chemisorbed CO can be individually resolved by using

the photoemission selection rules and linearly polarized

UV radiation [89.25]. The origin of the chemical shift

between the 5σ and 1π levels has been ascribed to bond

formation involving the CO 5σ level and σ -symmetry

d-electron states on the transition metal surfaces [89.2].

89.3.2 Inverse Photoemission

Spectroscopy (IPES)

Inverse photoemission spectroscopy is properly classi-

ﬁed as an incident electron technique but is included here

(eV)

20 15 10 5

0 =

vac

CO/Pd(111)

= 142 kJ/mol

CO/Cu(111)

= 47 kJ/mol

CO/Ag(111)

= 19 kJ/mol

SOLID

COND

= 7 kJ/mol

GAS

4σ 5σ1π

s.u.

Fig. 89.2 AsetofUPS spectra for CO adsorbed on various

surfaces, as well as gas phase and solid CO. The peak

labeled “su” is a shake-up satellite peak. After [89.25]

Part G 89.3

1348 Part G Aplications

as a natural companion to UPS. IPES utilizes low energy

electrons (5–50 eV) incident on a surface. Transitions

from a high-lying initial electron state in the contin-

uum to a lower unoccupied state cause a UV photon

to be emitted. By detecting the emitted photon inten-

sity as a function of energy, the joint density of states is

measured [89.7]. IPES is used to study the unoccupied

portion of surface and bulk bandstructure, particularly

the region between the Fermi level and the vacuum level,

which is difﬁcult to access by other means. This allows

the study of unoccupied states of adsorbed molecules

(e.g., antibonding molecular levels) as well as intrinsic

surface states such as the Rydberg-like states of electrons

trapped in the image potential at the surface [89.7].

89.3.3 X-Ray Photoelectron

Spectroscopy (XPS)

X-ray photoelectron spectroscopy allows the study of

the energy of atomic core levels via the Einstein photo-

electric equation T

= hν − E

,whereT

is the kinetic

energy of the photoemitted electron and E

is the

binding energy of the core level. Atoms bound in dif-

ferent chemical environments (e.g., at particular sites

on a surface) experience different chemical shifts, and

so are measured at slightly different binding energies.

This sensitivity to chemical environment allows XPS to

characterize the different types of binding sites for sim-

ilar atoms, and measure their abundance by measuring

the relative intensities of XPS emission from different

species.

Shifts in T

arise from two sources: intra-atomic

and inter-atomic relaxation shifts. The intra-atomic re-

laxation E

is due to screening of the core hole by other

electrons in the emitter atom. The inter-atomic relax-

ation E

is important in solids (particularly for metals)

and is due to screening of the core hole by the dielectric

response of the surrounding medium. These relaxation

shifts (on the order of a few eV) tend to increase the ki-

netic energy of the emitted photoelectron, so the kinetic

energy in the adiabatic limit is

= hν − E

+ E

. (89.7)

Since photoemission is a rapid process, nonadiabatic

processes lead to shakeup and shakeoff features in which

other electrons are excited to higher energy levels, caus-

ing the photoelectron to have lower kinetic energy [89.5]

(see Sect. 62.4.4). It is also possible to excite discrete

excitations of the solid, such as plasmons and a contin-

uum of low energy electron-hole excitations, causing the

XPS peak to have an asymmetric lineshape. The over-

all XPS distributions of T

thus contain contributions

from the adiabatic channel, and lower energy photoelec-

trons in a series of discrete peaks or a continuum from

nonadiabatic processes.

The chemical abundance of a species can be deter-

mined from the sum rule that the total XPS cross section

is proportional to the sum of the adiabatic peak and

all the shake-up and shake-off components. However it

is usually only convenient to measure the intensity of

the adiabatic peak. If comparisons are made between

chemical species in different chemical environments,

the shake-up and shake-off intensities may differ. Hence

the adiabatic channel intensities might not reﬂect the

true abundance. Very often this problem can be mini-

mized by careful calibration, and chemical analyses can

be made to an accuracy of a few percent.

The surface sensitivity of XPS is due to the short in-

elastic MFP for the photoelectrons, which can be made

to have energies in the range 10–1000 eV by appropriate

choice of the X-ray wavelength. The XPS signal is pro-

portional to exp[−z/(λ cos θ)],wherez is the depth of

the emitter, λ is the inelastic mean free path of the pho-

toelectron and θ is the angle measured from the surface

normal.

The popularity of XPS as a surface analysis tech-

nique is due to the availability of convenient and

sufﬁciently intense monochromatic X-rays from lab

sources (usually Al and Mg K

X-ray lines at 1486.6

and 1253.6 eV). The increasing availability of con-

tinuously tunable X-rays from synchrotron radiation

sources allows improved measurements due to the abil-

ity to tune the X-rays to slightly above threshold,

where the XPS cross-sections are maximized, the in-

elastic MFP is shorter (hence more surface speciﬁc)

and electron monochromators can operate with higher

resolution [89.5].

As an example of the chemical sensitivity of XPS,

photoelectron spectra of the 4f

7/2

core levels of W(111)

and Ta(111) are shown in Fig. 89.3. The spectra show

discrete peaks from both bulk and surface atoms. The

binding energies of the surface atoms are affected by the

adsorption of hydrogen.

XPS can also be used to give local structural in-

formation due to elastic scattering of the photoemitted

electron in the region near the emitter [89.5]. One form

of this is the use of forward scattering from a buried

emitter. A high energy XPS photoelectron is focused in

the direction of nearby atoms due to a lower effective

potential close to the atomic core. Angular scanning of

the XPS detector will then detect a more intense signal

along the bond axis. This method has proved useful in

Part G 89.3

Surface Physics 89.3 Photon–Atom Interactions 1349

32.0 31.5 31.0

Binding energy (eV)

23.0 22.5 22.0 21.5

Binding energy (eV)

23.0 22.5 22.0 21.5

Binding energy (eV)

Intensity (arb. units)

W(111) 4f

7/2

hv = 70 eV

Intensity (arb. units)

W(111) + 11LH

Intensity (arb. units)

Ta(111) 4f

7/2

hv = 66 eV

Intensity (arb. units)

Ta(111) + 500LH

∇

Fig. 89.3 XPS spectra of the 4f

7/2

states of W(111) and Ta(111). Contributions from the bulk (b) and surface (S1,S2)

atoms can be distinguished, and the surface atom peaks are shifted by the adsorption of hydrogen. After [89.28]

studying the orientation of adsorbed molecules and the

structure of hetero-epitaxial thin ﬁlms [89.27].

89.3.4 X-Ray Absorption Methods

X-ray absorption methods [89.8] measure the decay of

the core hole rather than the intensity of emitted pho-

toelectrons, since that XPS process is complicated by

a number of possible ﬁnal state processes, as discussed

above. To obtain surface sensitivity, the X-ray absorp-

tion is measured using a low energy electron emitting

channel such as an Auger electron emission or the to-

tal electron yield, which are proportional to the overall

X-ray absorption. These methods require an intense and

tunable source of monochromatic x-radiation near the

excitation edge for the core level of a particular atom,

and so are usually performed using synchrotron radia-

tion. Synchrotron sources have the additional beneﬁt of

linearly polarized light, which is crucial for NEXAFS

and useful for SEXAFS discussed next.

SEXAFS: Measurement of Bond Lengths

The surface extended X-ray absorption ﬁne structure

(SEXAFS) technique is most commonly used to mea-

sure the bond lengths for atoms adsorbed on a surface.

SEXAFS utilizes the elastic backscattering of the emit-

ted XPS photoelectron from nearby atoms that surround

the emitter. Elastically scattered waves arrive back at the

emitter and add coherently (constructively and destruc-

tively) to the outgoing wave, thus modifying the matrix

element for the transition to the ﬁnal state [89.29]. Exper-

imentally, the cross section σ(hν) for X-ray absorption

above the threshold photon energy is modiﬁed by an

oscillatory structure. The ‘atomic’ contribution to the

absorption cross section can be removed using

χ =

σ − σ

, (89.8)

where σ and σ

are the measured surface and free atom

X-ray absorption cross sections. If only single scattering

events are involved in backscattering to the emitter then

the ﬁne structure function χ is

χ(k) =



i=1

(k) sin



2kR

+ φ

(k)



, (89.9)

where k is the magnitude of the photoelectron wavevec-

tor and the sum is done for N ‘shells’ of atoms

surrounding the emitter. The distance R

is the radius

of the ith shell and 2kR

is the associated phase factor

Part G 89.3

1350 Part G Aplications

for the backscattered photoelectron, with an amplitude

(k). The phase shift φ

(k) is required due to the

backscattering path through the potential surrounding

the emitter and scattering from the atom at R

.Ifthe

could be ignored, then a simple Fourier transform of

χ(k) would reveal the radial distribution function and

the bond lengths R

[89.29]. In practice, the phase shifts

cannot be neglected, but very often these can be found by

studying chemically similar systems in which the bond

lengths are known. The phase shifts for photoelectrons

well above the absorption edge (having kinetic ener-

gies greater than ≈50 eV) are dominated by the atomic

ion cores, and so are not sensitive to the valence elec-

tronic structure. The phase shifts can also be calculated

in a straightforward way since this problem is essen-

tially the same as done in LEED multiple scattering

calculations.

The amplitudes A

(k) due to scattering from shell i

at distance R

from a point source emitter are [89.29]

(k) =

∗

| f

(π, k)| exp



− 2







×exp



− 2R

/λ



, (89.10)

4500 4600 4700 4800

02 684

2468

Photon energy (eV)

Intensity (arb. units)

S = Cu(111)(√3 × √3) R30°–I

B = Bulk CuI

R(Å)

Intensity (arb. units)

I–Cu

χ(k)

k(Å

–1

)

k(Å

–1

)

hω

Fig. 89.4 (a) X-ray absorption data taken near the iodine edge for bulk CuI and Cu(111)–I. (b) The extracted ﬁne structure

function χ(k), shown multiplied by k

to enhance the high k structure. (c) The Fourier transform of χ(k) showing the

location of the ﬁrst shell of Cu atoms from the I emitters. (d) Back-transformed k

χ(k) after applying a ﬁlter to extract

the nearest neighbor data. After [89.30]

where N

∗

is an effective number of atoms and





is the

mean-square displacement of the atoms in shell i.The

backscattering amplitude | f

(π, k)| has been separated

from its phase factor φ

(k) in (89.9). The inelastic mean

free path for the photoelectrons reduces the contribution

from successive shells by a factor exp(−2R

/λ),andit

is this term that allows the kinematic single-scattering

approach to be used. Multiple scattering paths involve

longer trajectories and so are more strongly attenuated

by this exponential factor. If the near-edge energy region

of the absorption cross section is not included in the

analysis, then the scheme outlined in (89.9)and(89.10)

is reasonable. The near-edge region (within ≈ 50 eV of

the absorption edge) is troublesome, not only because

of multiple scattering, but also because the phase shifts

(k) for low energy photoelectrons are more sensitive to

valence electron distributions, and so are more sensitive

to the details of the local chemical environment.

The SEXAFS method is illustrated with the data

of Fig. 89.4 for iodine adsorbed on Cu(111). Panel (a)

shows the measured X-ray absorption σ(hν) for both

the surface system Cu(111)–I and bulk CuI. The ex-

tracted χ(k) and their Fourier transforms are shown in (b)

Part G 89.3

Surface Physics 89.4 Atom–Surface Interactions 1351

and (c). The Cu–I nearest neighbor bondlength is clearly

shownby the peak in (c), and is found to be 0.07 Å longer

than in bulk CuI. SEXAFS is most commonly applied to

atomic adsorption systems since molecular systems are

difﬁcult to analyze, as they can contain several similar

bond lengths, and the shells containing different atoms

are difﬁcult to model.

NEXAFS: Molecular Orientation at Surfaces

In SEXAFS, the X-ray absorption close to the excita-

tion edge is avoided due to the problems of multiple

scattering, and phase shifts that are chemically sensi-

tive to valence electrons. The study of near-edge X-ray

absorption ﬁne structure (NEXAFS) can avoid these

difﬁculties by using only the symmetry properties of

the transition matrix element without concern for the

absolute amplitudes [89.32]. For isolated molecules,

unoccupied molecular states having σ or π symme-

try are very commonly found close to or just below

the threshold for photoemission. These ‘molecular res-

onances’ often remain for adsorbed molecules, and

the symmetry properties of the absorption intensity

I ∝|f | A· p|i|

can be used to determine the molecular

orientation.

For the overall matrix element to be nonzero, it

must be totally symmetric. Using linearly polarized syn-

chrotron radiation, the direction of polarization A can

be varied by rotating the crystal. For example, the K-

edge X-ray absorption of CO on the Ni(100) surface in

Fig. 89.5 shows ﬁnal state resonances A (π-symmetry)

and B (σ-symmetry). The intensity of these features de-

pends of the direction of polarization of the incident

X-rays. X-rays with a polarization vector parallel to

the surface (θ = 90

◦

) strongly excite the π-symmetry

absorption while the σ-symmetry absorption is absent.

For grazing incidence X-rays polarized normal to the

surface, the σ-resonance is prominent. From dipole se-

lection rules, this polarization dependence of the X-ray

absorption is evidence that the CO molecule is adsorbed

280 290 300 310

Photon energy (eV)

Carbon Auger yield (arb. units)

45°

10°

= 90°

CO on Ni(100)

C K-edge

X-rays

Fig. 89.5 Near-edge X-ray absorption data from the C K-

edge from CO adsorbed on Ni(100) as function of incident

photon angle θ. The molecular π (peak A) and σ (peak B)

resonances are observed. The polarization dependence of

the absorption allows the CO orientation to be determined.

After [89.31]

with its bond perpendicular to the plane of the Ni(100)

surface.

The σ-resonance ﬁnal state in NEXAFS corresponds

to multiple scattering of the photoelectron along the

bond axis. The overall phase shift for this ﬁnal state

is approximately





E − V(r) dr ≈



E − V(r) R = const. (89.11)

where R is the bond length. This sensitivity of the ﬁnal

state phase shift to bond lengths allows the σ-resonance

energy to be used as a measure the molecular bond

length [89.33]. A plot of the σ -resonance energy vs.

1/R

shows a linear relationship, as is found in the case

of simple hydrocarbons in the gas phase and adsorbed

on a Cu(100) surface [89.32].

89.4 Atom–Surface Interactions

89.4.1 Physisorption

The binding interactions between atoms and surfaces

can be classiﬁed as physisorption (long range attrac-

tive dispersion forces) and chemisorption, in which

chemical bonds are formed. The long-range dispersion

force between a polarizable atom and a conducting

surface give rise to the leading term of the van der

Waals potential V(z) ∝−1/z

. At smaller atom-surface

separations, the location of the reference image plane

needs to be included, resulting in an attractive potential

V(z) =−C

/|z − z

where z

is the distance from the

last atomic plane to the image plane, typically 2–3 Å.

In principle, the constant C

is calculable from the

dielectric properties of the substrate and the atomic po-

larizability, but experiments have found values of C

Part G 89.4

1352 Part G Aplications

40% less than expected [89.34]. The cause of this dis-

crepancy has not been clariﬁed, although contributions

from surface roughness have been suggested.

At larger atom-surface separations, retardation ef-

fects must be taken into account in the dispersion

interaction (the Casimir–Polder force, Sect. 79.2.4), and

here theory predicts a 1/z

interaction potential. This

form for the potential has been conﬁrmed experimen-

tally [89.35].

As the atom approaches the surface, wave function

overlap eventually causes repulsion. In the absence of

chemical bond formation, the repulsive potential is sim-

ply proportional to the surface electron density n(z),

resulting in the overall physisorption potential

V(z) = Kn(z) −

|z − z

. (89.12)

The surface charge density n(z) decreases exponentially

above the surface [89.2], and the constant K can be de-

termined to reasonable accuracy by an effective medium

theoretical approach [89.36]. This shallowphysisorption

potential well has been studied experimentally in atomic

beam (often He) scattering experiments. Under certain

scattering conditions, it is possible for an incident atom

to be ‘selectively adsorbed’ on the surface by making

transitions into and out of bound states of the potential

well of (89.12). The incident atom with wavevector k

is diffracted by a surface reciprocal lattice vector G

into a bound state E

of the potential well [89.2]. The

diffracted atomic beam intensities due to scattering via

selective adsorption show strong variations which can be

related to the quantum numbers (n, h, k), and give infor-

mation on the shape of the physisorption potential well.

89.4.2 Chemisorption

Many atoms and molecules will chemisorb when

brought sufﬁciently close to a surface, forming chem-

ical bonds (covalent or ionic) that are much stronger than

the physisorption bond [89.3]. In chemisorption, there

is charge transfer between the adsorbate and the sur-

face, modifying the electronic structure of both. Valence

electronic levels of the adsorbate are shifted in energy

and also broadened by resonant interactions with the

delocalized valence electrons at the surface (e.g., free

electron-like s–p states). For many transition metal sur-

faces, interactions with the more localized d–states are

also important.

The adsorbate-surface bond energies are most com-

monly studied by thermal desorption spectroscopy

(TDS) [89.37, 38]. In this method, the adsorbate cov-

ered surface is heated using a linear temperature ramp,

and the desorption rate of a particular species is meas-

ured using a mass spectrometer. By using a range of

heating rates β, not only can the desorption energy be

measured, but the kinetics governing the desorption pro-

cess can be uncovered. For example, in the simplest

case of a coverage-independent adsorption energy and

ﬁrst order kinetics, if the peak desorption rate occurs at

a temperature T

then [89.39]

exp





(89.13)

where E

is the desorption energy and ν is the ‘attempt

frequency’ for desorption. Since both ν and E

are un-

known, values for both can be found by measuring TDS

spectra using two different heating rates β or by estimat-

ing ν (often ν ≈ 10

−1

). More complex desorption

kinetics are studied by utilizing the full desorption pro-

ﬁle, and a range of heating rates and initial adsorbate

coverages. These kinetic data are also applicable to ad-

sorption if the adsorption and desorption are reversible

processes. For irreversible adsorption systems, it is pos-

sible to measure the adsorption energies directly [89.40]

by monitoring the ir radiation emitted from a surface

as a submonolayer quantity of atoms or molecules is

adsorbed.

89.5 Recent Developments

Surface Physics has both rapidly matured and ex-

panded its connections with other disciplines in the

last eight years. A number of notable review works

have been published recently [89.41–45]. The appli-

cation of surface science techniques has expanded

rapidly in a number of technological areas, such as

semiconductor devices, catalysis, and magnetic ma-

terials. The emerging ﬁeld of nanotechnology has

drawn heavily from the techniques of surface sci-

ence. Important advances have been made in a large

number of areas, including biosurfaces [89.46,47], clus-

ter science [89.48], carbon fullerene materials [89.49,

50], electrochemistry [89.51], and surface photochem-

istry [89.52].

Continuing advances have been made in the appli-

cation of light from synchrotron radiation sources — so

Part G 89.5