Surface Analysis Methods in Materials Science
Surface Analysis Methods in Materials Science
Surface Analysis Methods in Materials Science
~ ONLINE LIBRARY
•
PhysICS and Astronomy LJ
Springer-Verlag Berlin Heidelberg GmbH http://www.springer.de/phys/
SPRINGER SERIES IN SURFACE SCIENCES
This series covers the whole spectrum of surface sciences, including structure and dynamics
of clean and adsorbate-covered surfaces, thin films, basic surface effects, analytical methods
and also the physics and chemistry of interfaces. Written by leading researchers in the field,
the books are intended primarily for researchers in academia and industry and for graduate
students.
Surface Analysis
Methods
in Materials Science
Second Edition
With 272 Figures
Springer
Associate Professor D. [ohn O'Connor Dr. Brett A. Sexton
School of Mathematical and Physical Sciences CSIRO Manufacturing Science and Technology
University Newcastle Private Bag 33
Callaghan, NSW 2308 Clayton South, VIC 3169
Australia Australia
Series Editors:
Professor Dr, Gerhard Ertl
Fritz-Haber-Institute der Max-Planck-Cesellschaft, Faradayweg 4-6,
14195 Berlin, Germany
ISSN 0931-5195
This work is subject to copyright. AII right s ar e reserved, whether the whole or part of the mat erial is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,
reproduction on microfilm or in any other way, and storagc in data banks. Duplication of this publication or
parts thereof is perrnitted only under the provisions of the German Copyright Law of September 9, 1965,in its
current version, and permission for use must always be obtained from Springer- Verlag. Violations are liable
for prosecution under the German Copyright Law.
springeronlinc.com
The use of general descriptive narnes, registered names, trademarks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the relevant protective laws and
regulations and therefore frec for general use.
Typesetling: camera ready copy by the authors
Cover concept: eStudio Calamar Steinen
Cover production: design & production GmbH, Heidelberg
Printed on acid-free paper 57/3111/mf - 5 4 321 SPIN 10977446
Preface
The success of the first edition of this broad appeal book prompted the prepa-
ration of an updated and expanded second edition. The field of surface anal-
ysis is constantly changing as it answers the need to provide more specific
and more detailed information about surface composition and structure in
advanced materials science applications. The content of the second edition
meets that need by including new techniques and expanded applications.
The idea for this book stemmed from a remark by Philip Jennings of Mur-
doch University in a discussion session following a regular meeting of the
Australian Surface Science group. He observed that a text on surface anal-
ysis and applications to materials suitable for final year undergraduate and
postgraduate science students was not currently available. Furthermore, the
members of the Australian Surface Science group had the research experi-
ence and range of coverage of surface analytical techniques and applications
to provide a text for this purpose. A list of techniques and applications to be
included was agreed at that meeting. The intended readership of the book
has been broadened since the early discussions, particularly to encompass
industrial users, but there has been no significant alteration in content. The
editors, in consultation with the contributors, have agreed that the book
should be prepared for four major groups of readers:
The contributors mostly come from Australia, with the notable exception
of Ray Browning from Stanford University. Australia is a very large coun-
try with a relatively small population, so it is inevitable that the Australian
surface science community is spread rather thinly across the country. One
aim in producing this book has therefore been to bring together the breadth
of expertise within this Australian scientific community. All of the authors
have made significant contributions to the techniques and their applications,
in many cases over a period of more than 20 years. A second aim is to em-
phasise by example the very wide spectrum of information that can now be
obtained from the use of a variety of surface analytical techniques applied to
VIII Preface
the same material. Here, the intention has been to encourage people involved
in research, development and process control to become aware of the increas-
ing usefulness of surface analysis in their own fields of materials science.
Finally, we believe that the approach adopted here - namely descriptions
of the basic techniques, their limitations and their applications - will be
accessible and beneficial to people in any of the four groups of potential
readers in any country of the world. The authors are all closely involved
with the international scientific community in their own areas of research
and this is reflected in their selection of examples. We hope that this book
will fulfil a need by extending the range of techniques from those covered by
more specialised texts confined to one or two techniques, and by providing
examples of real applications to research, development and problem solving
in materials science.
The strategy and structure of the book is as follows: the book is de-
signed for use by all those interested in the surface characterisation of ma-
terials. This group is expected to include scientists (e.g. chemists, physicists,
biochemists, biologists, geologists, geochemists), technologists, metallurgists,
engineers and workers in the various biomedical and microelectronics appli-
cations areas. It is also intended for those requiring an introductory overview
of surfaces and techniques for their analysis, in particular final year under-
graduate and graduate students in any of the specialties listed above. It is
important to state clearly that it is not intended for experienced practitioners
in particular techniques of surface analysis. It does not attempt to critically
review all techniques or all variations and restrictions on the use of a particu-
lar technique. For instance, we are well aware that there are many difficulties
in procedures for quantification, chemical mapping and depth validation for
particular types of sample that are not explained in the short presentations
on each technique given in this book. They will need to be learnt when un-
dertaking a specific investigation of a material using one of the techniques
described here and will probably require further reading in more advanced,
single-technique books.
Part III describes some major applications. Again, some judgement has been
exercised in the choice of these applications, with emphasis placed on major
areas in materials science and technology. Options for addition and deletion
also exist in this section.
The book is best approached by initially reading the whole of Part I and
selected sections from Parts II and III. The reader may wish to focus on
particular materials of direct interest to his or her research, development
or technological application. This itself will suggest particular techniques as
being most appropriate initially. It may also require more detailed reading of
single-technique descriptions in other books: the references in Parts I and II
give specific guidance to books most suitable for this purpose.
Part I Introduction
2 UHV Basics
C. Klauber. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 71
2.1 The Need for Ultrahigh Vacuum. . . . . . . . . . . . . . . . . . . . . . . . . . . .. 71
2.2 Achieving UHV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 74
2.3 Specimen Handling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 76
2.4 Specimen Handling: ASTM Standards. . . . . . . . . . . . . . . . . . . . . . .. 78
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 81
Part II Techniques
15 EXAFS
R.F. Garrett, G.J. Foran ......................................... 347
15.1 Introduction ............................................... 347
15.2 Experimental Details ....................................... 348
15.2.1 Synchrotron Radiation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
15.2.2 Synchrotron Beamlines for EXAFS ..................... 352
15.2.3 Detectors ........................................... 354
15.2.4 The Sample ......................................... 355
15.2.5 Acquiring EXAFS Data ............................... 357
15.3 Theory of X-ray Absorption ................................. 359
15.3.1 EXAFS ............................................. 359
15.3.2 XANES ............................................. 361
15.4 EXAFS Analysis ........................................... 362
15.4.1 Data Reduction ...................................... 362
15.4.2 Conversion to k-space ................................ 363
15.4.3 Background Subtraction .............................. 363
15.4.4 Fourier Transformation ............................... 364
15.4.5 Fourier Filtering and Back Transformation .............. 364
15.4.6 Modelling and Least Squares Fitting
to the EXAFS Equation .............................. 365
15.5 Case Studies ............................................... 366
15.5.1 Surface EXAFS of Titanium Nanostructure Thin Films ... 366
15.5.2 Ion-Implantation Induced Amorphisation of Germanium .. 370
References ..................................................... 371
XVIII Contents
20 Coated Steel
R. Payling ..................................................... 473
20.1 Applications ............................................... 474
20.1.1 Grain Boundaries in Steel ............................. 474
20.1.2 Steel Surface ........................................ 475
20.1.3 Alloy Region ........................................ 478
20.1.4 Metallic Coatings .................................... 478
20.1.5 Treated Metallic Coating Surface ...................... 483
20.1.6 Metal-Polymer Interface .............................. 483
20.1.7 Polymer Surface ..................................... 484
20.2 Conclusion ................................................ 486
References ..................................................... 486
Part IV Appendix
Introduction
1 Solid Surfaces, Their Structure
and Composition
OWe use the term "selvedge" to emphasize that the surface is often a region (with
depth) rather than a two-dimensional layer.
to four times that of the original reduced metal. Such enormous expansion
accentuates the material's failure. Whilst such massive corrosion is controlled
by mass and charge transport in the solid state, its initiation is certainly
a surface process. The corrosion need not be visibly dramatic in order to
eventually affect the materials performance. Steels are typically protected
from corrosion by painting, yet these coatings will also eventually oxidize,
crack and fail to perform their intended task.
Not only chemical stability, but mechanical stability is of engineering im-
portance. Wear can be minimized not only be lubrication, but also by surface
hardening. This might be achieved by methods of nitridization or carburiza-
1 Solid Surfaces, Their Structure and Composition 7
Having reminded ourselves of the variety of forms in which solid materials can
exist, it is a useful exercise to conceptually "create" a surface from one of these
materials. Taking the simplest case of a crystalline solid, we might imagine an
ideal case of a sudden termination of the crystalline periodicity at the solid-
vacuum interface as in Fig. 1.1a. In layer, chain or sheet structures, where
relatively weak dispersion forces hold the constituents of the solid together,
such an "ideal" surface case might be envisaged. In strongly bonded solids
such as metallic, covalent or ionic systems, the termination of the periodicity
means that valence electrons will spill out into a continuum with no positive
cores, freed covalent bonds will be "dangling" into space and the Madelung
constant, evaluated on the basis of 3-dimensional symmetry, will no longer
apply. Invariably this breakdown of previously balanced bonding forces leads
to a rearrangement of the outer layer or layers and to a periodicity, and
possibly unit cell structure, different from that previously existing in the bulk.
8 C. Klauber and R. St. C. Smart
vacuum
---8000000solid
d=a
-----8000000 0000000000 0 0 0
0000000 0000000
0000000 (a) 0000000 (el
000
~OOOOOO
~000000 0000008°8<d8<do
0000000 0000000
0000000 (bl o 0 0 0 0 0 0 (d)
Fig. 1.1. Schematic of (a) a solid surface created by terminating the bulk of a
crystalline solid. Bonding imbalances cause the outer layer to move (b) and even
undergo reconstruction (c). When then exposed to a reactive medium, foreign atom
adsorption can occur (d), possibly leading to surface compound formation
The surface in this sense becomes the ultimate defect that a solid material
can have.
The simplest rearrangement is that of outer layer relaxation or contraction
(Fig. LIb), the layer either moving away from or towards the inner layers.
Dynamical low energy electron diffraction (LEED) calculations indicate that
for metals this movement may be 0-25% of the normal inter layer spacing [6].
As no bond breaking is required this rearrangement is spontaneous. A more
complex alteration occurs when the structure parallel to the surface under-
goes what is known as reconstruction, shown schematically in Fig. 1.1c. This
is particularly common in covalently bonded semiconductor materials. The
surface atoms rearrange themselves so as to minimize the dangling bonds
which arise from the original surface creation. Reconstruction can be spon-
taneous or may require heating to be induced. Surface reconstructions are
dealt with further in Sect. 1.4 of this chapter and Part II of the book. A large
amount of research work carried out in the discipline of surface science is con-
cerned with the in situ creation of such virginal surfaces (atomically clean
and ordered) and their subsequent controlled reaction with various gases.
These can be created in situ by a number of methods:
• cleaving or fracturing in the vacuum system; useful for single crystal work,
grain boundaries and fracture mechanics, but very limited for materials
analysis
• evaporating of clean films (which can be substrate epitaxial); has appli-
cation in nucleation, catalysis, adsorption and adhesion studies but is
generally only useful for model systems
1 Solid Surfaces, Their Structure and Composition 9
o substrate atoms
o oxide anions
adventitious carbon
MONATOMIC
~ STEP
PROTUSION
(2 UNIT CELL)
PIT
(4 ATOM VACANCY)
Fig. 1.2. Two representations of part of a solid surface depicting a variety of dif-
ferent surface sites. These sites are distinguishable by their number of nearest or
coordinating neighbours. Low coordination sites (surface defects) are the preferred
locations for adatom adsorption and can thus be the initiation point for a variety
of surface processes. Also depicted are an oxide facet, a cubic etch pit and a region
of extensive adventitious carbon contamination
1 Solid Surfaces, Their Structure and Composition 11
regions and bonded interfaces. All of these examples relate particular surface
structures to particular forms of reactivity. They are a few grains of sand on
the seashore of surface structural features.
To instrumentally probe a medium , in this case a solid surface, one of six ba-
sic probes may be applied to the surface: electrons, ions, neutrals, photons,
heat or a field. The analysis consists of measuring the surface's response, also
evident in one of these six ways. Combining the six probes and six responses
gives 36 fundamental classes of experimental technique by which we may
analyse a surface. By altering the energy and/or mass and/or character of
1 Solid Surfaces, Their Structure and Composition 13
these probes the variety of possible experimental methods increases even fur-
ther. Obviously not all options will provide an experimentally viable pathway
to gain useful surface information i.e. the surface structure (physical and elec-
tronic) and the surface composition (chemical). Whilst all possible variants
have not yet been explored to the fullest extent, the number of surface analyt-
ical techniques currently in use is nonetheless quite impressive. Appendix A.l
lists over 130 technique acronyms; some of these are admittedly restricted to
esoteric research endeavours, but many have gained widespread use in hun-
dreds of laboratories throughout the world. As it is beyond the scope of this
book to cover all of these, we will concentrate only on the more common and
versatile, twenty techniques in all. These are outlined in Table 1.3 within a
subset of 12 of the 36 fundamental classes. There is also a variety of more
classical surface-specific techniques such as basic measurements of contact
angle, zeta potential, particle size distributions, surface areas and porosities
and also electrochemical methods such as cyclic voltammetry which are not
included. They are generally covered in a number of good texts on colloid
and surface chemistry (e.g. refer to the bibliography at the end of the book).
Table 1.3. Summary of the various techniques considered in this book for ana-
lyzing the structure and composition of the surface selvedge layer. The 12 basic
categories, each of which may encompass a range of techniques, represent a subset
of a possible 36
tThe AFM technique essentially analyses force due to electric and magnetic fields at surfaces
1 Solid Surfaces, Their Structure and Composition 15
Fig. 1.4. Comparison of (a) secondary electron (b) backscattered electron (BSE)
and (c) topographic images from scanning electron micrographs of the same area
of a polished ceramic surface showing grain pull-out and porosity. The white bar
indicates magnification of 10 run. Note that the BSE image, in addition to porosity,
indicates regions of low (dark) and high (white) average atomic number
1 Solid Surfaces, Their Structure and Composition 17
Q '1,um J
Q ' 05,um
a b c
conditions, and capable of atomic resolution, are now older techniques (with
a large literature) to some extent superceded by STM. They apply only to
refractory metals and require special sample preparation not normally appli-
cable to materials science. They have greatly increased our understanding of
surface structure but must be regarded as essentially fundamental research
tools.
Low energy electron reflection microscopy (LEERM) is a relatively new
technique [26], used in combination with LEED, giving images from elas-
tically reflected electrons in the 0- 300 eV range. Its practical resolution is
about 50 nm but this can allow a combination of image and diffraction pat-
terns from differentiated surface features , a result of considerable importance
for surface phase formation in catalysis, reactivity, leaching, corrosion and
surface segregation. It is not yet sufficiently accessible for most materials
scientists to warrant a full discussion but may well be a technique for the
future.
20 C. Klauber and R. St. C. Smart
Provided that low energies are used, the scattering of electrons, atoms and
ions can be restricted to the top few atomic layers giving surface sensitive
probes for atomic positions (interatomic spacings), periodic structures (via
diffraction of the scattered beams), faceting, defect sites and disorder. Low
energy electron diffraction (LEED) uses electrons in the 10 to 500 eV range
with escape depths of 5-10 A (Chap. 13). Atomic beam scattering (ABS)
uses atomic beams of low-mass, unreactive gases (e.g. helium) monoenergetic
in the range 30 to 300 meV, which scatter and diffract from literally the top
atomic layer. Ions with energies less than 5 ke V i.e. low energy ion scattering
(LEIS) also scatter predominantly from the top layer due to the very large
cross sections for ion-atom elastic scattering (Chap. 11). They can be used
for studies of shadowing (of one atom by another) and atomic displacement
(i.e. interplanar spacing and defect) effects. Scattering of ions with higher
energies, i.e. medium energy MEIS, (20-200 keY) and high energy HEIS, or
Rutherford backscattering RBS (200 keV-2 MeV), can give information on
atomic positions and defects (e.g. interstitials) in the bulk of the solid. Pene-
tration is deeper and blocking effects due to displaced atoms are more evident
(Chap. 9). Similarly, medium (MEED, 500 eV to 5 keY) and RHEED reflec-
tion high energy electron diffraction (RHEED, 5-200 keV) , using the same
principles as LEED, give information particularly on surface topography and
faceting but with less surface sensitivity (up to 10 nm depth) (Chap. 12). All
of these techniques give primary data which can only be transformed to the
surface lattice (LEED), surface atomic positions (ABS, LEIS) or interlayer
spacings (MEIS, HEIS) by calculation using a model for the dynamical pro-
cesses occurring at the surface. Programs for routine analysis in LEIS, MEIS
and HEIS are readily available and reliable to a high level of accuracy. In
the other cases, the final structural information is only as good as the as-
sumptions the model allows and particularly for ABS and LEED, relatively
sophisticated models are required for reliable results.
The electrons elastically scattered from surface atoms (i.e. rv 1-10% of
the primary beam) in LEED studies of crystal surfaces will diffract sharply
only from well ordered domains of 100 nm 2 or greater as determined by
the coherence length over which the electrons will remain in phase. Disorder
in the surface can thus give diffuse spots and loss of intensity, providing
data for analysis of this disorder. Faceting appears as extra spots in the
diffraction pattern and steps on the surface can produce multiple splitting
of the primary spots in the pattern. The height of the steps can be inferred
from the "appearance voltage" of the splitting. The accuracy of interatomic
spacings determined from LEED calculations is about 0.1 A if the surface is
well-ordered and a dynamical model is used.
One of the major applications of LEED to surface structural elucidation
has been in the determination of the nature and magnitude of surface relax-
1 Solid Surfaces, Their Structure and Composition 21
Incident probe
Particle Energy Energy Current Beam
resolution or flux diameter
0.5meV
SIMS ion 0.5-30keV NA 1pA-100[-lA 500A-2mm
static 1pA-lOnA
dynamic > l[-lA
ISS ion 1-3keV 5-lOeV lOpA-1[-lA 200 [-lm-2mm
STM electron O.leV NA 0.5-1 nA 2A
EM electron 1O-50keV eV 0.1 [-lA-lOpA 2nm-1[-lm
UPS photon 10-50eV 3-20meV 10 11 _1012 S-l 1-3mm
2170
Am = -
E2- + 0 .72 a 1. 5 EO. 5 for inorganic compounds (1.2)
1 Solid Surfaces, Their Structure and Composition 25
100
10
A (nm)
0.1
10 100 1000
energy (eV)
Fig. 1.8. Compilation of electron inelastic mean free path measurements (IMFP,
A) for elements from Seah and Dench [36]. Solid curve of best fit represents the
"universal" IMFP applicable to all elements
A
a= ( - -
)1/3 x 1Q8 nm (1.3)
pnNA
where A is the molecular weight, p the bulk density in kg m- 3 , n the number
of atoms in the molecule and NA is Avogadro's number.
Note that the Seah and Dench approach is essentially structureless as sin-
gle crystal, polycrystalline and amorphous examples of the same substance
are only differentiated on the basis of their bulk densities. More recent com-
ments on the Seah and Dench formulae have been made by Ballard [37] and
Tanuma et al. [38].
The electron IMFP is particularly useful since, by assuming a process
of homogeneous attenuation, a Beer-Lambert type expression results for the
description of electron flux reduction:
1 = 10 exp ( -z
A cos
e) (1.4)
where 10 and 1 are the incident and emergent intensities and z/ cos is the e
e
path length for electrons travelling off-normal through a material of depth
z along the normal. Although, as Ballard [37] points out, such an expression
strictly only applies to electromagnetic radiation, in practical terms 1/10 plots
are found to be sufficiently linear to be useful for approximate estimates. The
e
total electron intensity reaching the surface at an angle from above the depth
z, as a fraction of intensity reaching the surface from all depths, can then be
established from a simple integration to be:
Fz,e = 1 - exp -z e)
(A cos (1.5)
26 C. Klauber and R. St. C. Smart
Table 1.5. Data set for the three-phase model of a passivated aluminium surface
e
Thus for = 0° , 63% of all electron intensity reaching the surface comes from
within one IMFP of the surface, 86% from within two IMFPs and 95% from
within three. For the electron spectroscopies IMFP is not only relevant to
surface sensitivity, but also to quantification of the analytical method .
Table 1.6. Relative Al2p and 0 Is photoelectron intensities for two escape angles
from the passivated surface. No correction has been made for relative cross-sections
and values are normalized for unit Al2p emission at () = 0°
Phase () = 0° () = 80°
Al2p photoelectrons AIO(OH) 0.059 0.048
from: a-Ab03 0.827 0.105
Al 0.113 < 10- 6
o Is photoelectrons AIO(OH) 0.117 0.090
from: a-Ab03 1.002 0.107
production modified by its transport to the detector. Since the X-ray attenu-
ation with depth can be considered negligible compared with the effects of A,
photoelectron production will be proportional to the relative atom density d;
and the layer thickness z, with subsequent escape dependent upon A and z.
Equation (1.4) is a statement of the probability of an electron on trajectory
Breaching z = 0 from a depth z. Hence, by integrating 1/10 from z = 0 to 00
we derive a maximum electron escape factor of A cos B. The fraction for a film
of z < 00 is given by (1.5). Combining both with d; yields an expression for
relative photoelectron production and escape from a selvedge of thickness z:
(1.6)
It follows that electrons from layer 2 passsing through layer 1 will emerge
with a relative intensity of:
exp (AI ~:~ B) d;,2 A2 cos B[1 - exp (A2 ~:: e) ] (1.7)
does serve to illustrate the influence of A upon surface analysis. Note that
one consequence of A is that isotropic electron production within a solid is
transformed to an anisotropic cos () dependence with transport to the sur-
face. Further general consideration to analyzing surface composition is given
in the rest of Sect. 1.5, although formal treatment is left to the individual
techniques in Part II.
Whether or not a component is detected thus depends initially upon its
depth z from the outermost layer (assuming that it is within the lateral region
being probed) and the probability of its responding with sufficient signal to
be detected. If the concentration within the volume element falls below its
sensitivity limit then that component will be invisible. The questions of varied
elemental sensitivities, practical limits to detection, spatial resolution and the
definition of chemical state information all affect how well the surface selvedge
composition is determined. Before considering any examples of composition
determination it is worth considering some of these basic constraints.
For the unknown surface confronting the researcher, the first problem is that
of detection of the elements present. Not all elements will be detected with
equal ease and the variation of detect ability changes between the various sur-
face analytical techniques. Figures 1.10-1.12 illustrate the relative elemental
sensitivities across most of the periodic table for three of the most common
techniques i.e. AES, XPS and SIMS. The two electron spectroscopies, AES
and XPS, both relying on ionization from particular energy levels and electron
detection, have comparable relative sensitivity with variations of less than two
orders of magnitude. However, things are quite different for ion spectrome-
try. In secondary ion mass spectrometry (SIMS), described in Chap. 5, an
accelerated ion beam (e.g. Ar+, Cs+, ot, 500 eV-5 keY) is focussed (down to
< 40 nm) on the surface and the secondary ions, both positive and negative,
sputtered from the surface are mass-analysed in a quadrupole or time-of-flight
mass spectrometer. The beam can be rastered to produce chemical mapping,
as in SAM, with very low (ppm) detection limits for most elements. So-called
dynamic SIMS uses high current densities and sputters many monolayers per
second whereas static SIMS uses low current densities and removes a sin-
gle surface layer over periods up to an hour. The ease of ion detection and
discrimination provides SIMS with an enormous dynamic range, easily en-
compassing five orders of magnitude in terms of relative sensitivity. Whilst
that provides SIMS with its greatest strength, because atom removal is in-
volved (rather than an intra-atomic excitation), these relative sensitivities
are dramatically matrix dependent. For example a SIMS spectrum of GaAs
based on positive ion detection would, on inspection, suggest a sample of pure
Ga and a 0.1 atom % trace impurity of As. Conversely, the negative ion spec-
trum would suggest a sample of pure As [41]. Such widely varying matrix
1 Solid Surfaces, Their Structure and Composition 29
LO
'<t
Z 3 keV electrons
LO 10 C')
C\J
'<t ...J
Z C')
Ru
•
LO C\J
0 ...J
:::2: 1.0 ~
Ol
« I)"
1.0
LL
.8 00
u .8
~
.:;: 0.1 Be
0
0 Ta
• ~
.:;:
:;:::;
:;:::;
'iii
'iii
c 8>~~~ 0.1 c
• Hg•
Q)
Q)
en en
Er Q)
Q)
> 0.01 >
.~
.~
KLL MNN 0.01 Q)
~ .....
0 20 40 60 80 100
atomic number
Fig. 1.10. Experimental elemental sensitivities relative to Ag M5N45N45 and F
K L 23 L 23 across most of the periodic table for Auger electron spectroscopy (AES).
Values are for an incident beam energy of 3 keV and include cylindrical mirror anal-
yser transmission function. Not all transitions have been included. Those selected
are the most intense, conveniently located transitions, e.g. intense, low-energy va-
lence transitions are difficult to utilize quantitatively. [Compiled from L.E. Davis,
N.C. McDonald, P.W. Palmberg, G.E. Riath, and R.E. Weber, Handbook of Auger
Electron Spectroscopy, 2nd ed. (Perkin Elmer Corporation, Eden Prarie USA,
1976)]
~
c
a
Q)
en
100
10
Gd
100
-
"0
Q)
u
Q)
.....
.....
au
0.0001
-
'E
-
10 0.001 c
en a
-
en c
a..... 1.0
a U
u '00 Q)
1.0
en 0.01
Q)
c 'E "0
a en ....
0.1 Q)
~ c >-
.!::! ctl
!:;
ctl
c 0.1 0.1 "0
a 0.01 ..... c
'0 en
Q)
a
(5 >- E
..c:: nl
a. 0.001 0.01 c
0 20 40 60 80 100 co
atomic number
Fig. 1.11. Elemental sensitivities relative to F Is across most of the periodic ta-
ble for X-ray photoelectron spectroscopy (XPS). Values are based upon theoretical
photoionization cross-sections with Mg K" radiation for the most intense levels (-).
Thf: second data set (0) illustrates the influence of electron spectrometer transmis-
sion on the relative sensitivities. The transmission is for a concentric hemispherical
analyser run at constant analyser energy (constant absolute resolution). [Compiled
from J.H. Scofield, J. Electron Spec., Rel. Phen. 8, 129 (1976); and A.E. Hughes
and e.C. Phillips, Surf. Interface Anal. 4, 220 (1982)]
30 C. Klauber and R. St. C. Smart
• •
Ca Ga •In
+ pure element
• compound
M+ yield from 13.5 keV 0-
101 ~.~~__~~~__~-L__~~~
o 20 40 60 80 100
atomic number
Fig. 1.12. Elemental sensitivities relative to F across most of the periodic table for
secondary ion mass spectrometry (SIMS). Note that as these are for ion production
they differ from the simple sputter yields shown in Fig. 1.13. [Compiled from H.A.
Storms, K.F. Brown, and J.D. Stein, Anal. Chern. 49, 2023 (1977)]
atom % to 0.7 with phosphate at 5.53 ± 0.24 atom %. Moreover RBS could
determine the anion concentrations (as atomic % of the characteristic anions
and AI) to an accuracy of a few percent.
The total quantity of material being probed can vary from a single atom
as in STM STM to up to 10 16 atoms in a broad area spectroscopy such as
XPS. In order to detect a single element, XPS would need the presence of
up to 10 13 _10 14 atoms so that STM would win in the adjusted sensitivity
competition. The advantage in XPS is the wealth of chemical information
provided. By comparison a destructive bulk analytical method such as atomic
absorption spectroscopy may need as few as 1010 atoms for detection [45].
Despite their many advantages neither XPS nor AES display sufficiently high
sensitivities for use in areas such as semiconductor doping. Here the highest
impurity levels which affect performance can be below their detection limits.
SIMS is generally used for this reason. In an alternative area, XPS has been
applied to trace analysis by Hercules et al. [46]. Utilizing chelating glass
surfaces it was found that heavy metals in solution could be easily analyzed
down to 10 ppb, without any optimization of the method.
use beam energies higher than 30 ke V for more precise focussing provides
diminishing returns because of the reduced cross-sections for Auger electron
production, especially for the lighter elements. Most electron excited Auger
work at high spatial resolution is carried out in the 5-10 keY range. As noted
in Chap. 6, the lateral resolution limit in SAM has not been reached and
30 A, i.e. the IMFP limit, is not considered impossible. Currently, in terms
of practical surface analysis in materials science, the "limit" is about 500 A,
see e.g. [48].
As with electrons, incident ion beams can also be focussed into fine beams.
The popular technique of static SIMS has a scanning variation akin to SAM's
relationship to conventional AES. A variety of means exist for the production
of ion beams, but for microfocussing the most effective is the liquid metal field
emission ion source. A liquid metal film, such as Ga, wets a solid needle and
is drawn out into a fine tip at the needle's end. By the application of a high
field, positive metal ions are extracted by field emission [49]. Such beams can
be focussed down to 500 A.
X-rays, in contrast to charged particles, are not easily focussed into small
spots. A Fresnel phase plate lens has been constructed which can focus Al
Ka radiation with 4000 A spatial resolution [50]. However, this is yet to be
applied in surface analysis. Less successful methods of localizing X-rays, such
as the use of crystal optics to focus the beam down to 0.15 mm or the prox-
imity method (generating the X-rays at a fine point, adjacent to a thin film
to be analysed) down to 20 !lm, have found some use to date [51]. Other than
X-ray localization, several methods do exist to achieve selected area XPS
(SAXPS), or the more general effect of photoelectron microscopy (PEM).
These are considered by Drummond et al. [51]. The approach which is easiest
to implement, because of its compatability with existing spectrometers, is
the use of a transfer lens system to magnify the surface area projected onto
the spectrometer entrance aperture [52]. Selection of a small entrance aper-
ture then means the spectrometer is only viewing part of the surface. This
can be further restricted by an additional aperture prior to the transfer lens.
Selected areas down to 150 !lm in diameter are available commercially [53].
Whilst relatively easy to implement, this approach does suffer from reduced
sensitivity with the photoelectron count rate decreasing in proportion to the
area being sampled. In practice this usually necessitates operating at reduced
resolution and scan widths in order to obtain information in a viable time
interval. By an extension of the detection electron optics it is feasible to pro-
duce an energy filtered photoelectron image of the surface with a resolution
of lO!lm [54]. Note that, unlike SAM or imaging SIMS, the input beam is
not rastered over the surface, and only a single selected area is analysed at a
given time. A comparison of the possible different methods of selected area
surface compositional analysis is given in Table 1.7.
Although in principle the depth, i1z, offers the highest spatial resolution
(i.e. layer-by-Iayer) of the volume element dimensions, the achievable reso-
eN
Table 1.7. Comparison of the possible different methods of selected area surface compositional analysis "'"
Method Information available Spatial resolution Advantages and disadvantages
Q
Surface elemental composition 20~150 [Am Chemical shift information obtainable with a
Selected area XPS ~
and chemical state analysis depending upon minimum of radiation damage. Physical i'O
(SAXPS) movement of specimen may be required to .:::
method selected 0-
(see text) obtain an image. Generally suffers from ..,ro
inferior signal to noise. §
0..
X-ray photoelectron Surface elemental composition 10 [Am Chemical shift information obtainable with a ?J
UJ
microscope and chemical state analysis minimum of radiation damage. Image M-
>Q) 8 • Cd
12 >Q)
0 0
0 K 0
0
6 8
~
LO
"0 Te "0
CD CD
'>' 4
OJ
c::
N~ Mo 50 y 4
'>'
OJ
V
·c 2 c::
Q)
L 0 ·c
::::: Si V
Ge
Mo W
Q)
:::::
:::J
c.. 0 C 0 :::J
c..
(/)
(/)
0 20 40 60 80 100
atomic number
Fig. 1.13. Elemental sputtering yield across most of the periodic table calculated
for Ar ions of energies 500 and 1000 eV [41]
lution with depth will depend upon the profiling approach. The commonly-
employed method of sputter profiling can be fraught with pitfalls as out-
lined in Chap. 4. Of most significance is the variation of sputtering yields
e.g. Fig. 1.13, leading to preferential sputtering which can dramatically alter
relative selvedge compositions. Two alternative approaches exist to extract
compositional information with depth. The first, principally applicable to the
electron spectroscopies, simply relies upon altering the surface sensitivity by
rotating the specimen such that the analyser (of narrow acceptance angle)
e
receives electrons from increasing angles off-normal. As the IMFP does not
e
alter, the analysed depth alters by cos and the technique thus becomes more
surface sensitive at high e. Angle-resolved XPS is explained in more detail in
Chaps. 7 and 16. A typical example of this is illustrated in Fig. 1.14. It is the
marked alteration in surface sensitivity which enables the detection of the
variety of altered (oxidised) surface groups on plasma-treated polystyrene.
The second approach, which is especially useful for deeper interfaces, in-
volves a mechanical lapping through the selvedge layer at some angle to the
original surface i.e. bevelling. Analysis is then achieved by scanning across
the lapped face, either mechanically or by beam raster. Depth resolution is
controlled by the angle of the lap and the lateral resolution of the probe.
Tarng and Fischer [56] point out that an ultimate depth resolution of ~ 35 A
would be possible with AES using a 0.20 lap and a beam and manipula-
tor of 0.5!lm resolution. A variation on straight lapping is ball cratering
[57], in which a shallow spherical pit is ground into the surface and the spa-
tially resolved probe is moved across the crater and thus to greater depths
(Fig. 1.1.5). Craters are more rapidly produced than an angled lap, but both
methods have a drawback in that the surfaces still have to be ion etched prior
36 C. Klauber and R. St. C. Smart
polystyrene
C 1S
5 min H20 plasma treatment
o o
II II
c c- o
/\
o 0 C= O
o
II
c-o
Fig. 1.14. Method of enhancing the surface sensitivity of the electron spectro-
scopies by increasing the take-off angle at which the analyser views the surface.
Here the surface functionalities created by plasma treatment of a polystyrene sur-
face are more readily detected. Adapted from Evans et al. [55]
r
/
( /
"l. / sample surface
, . . -_ _' ___ d2--/.....---~ T
r------------~
~~~d~l~~ ~
d 22 - d/
layer thickness t = Sr
de
depth of crater is - where de is crater diameter
Sr
Fig. 1.15. Schematic cross-section of a ball cratered surface selvedge. Rastering the
incident probe across the crater enables a profile analysis over a considerable depth
..•
·.
CH 3CH 2 CH 2 COO.Na
CO:C H = 1 : 0
".....; .
• •
...., .. '" -...
.·
o CH 3CH 2 COO.Na
.-., , •
.... '.
CO:C H = 1 :0
OJ
~ ..../ ........
c 0
...• •.
·.
:::J
o CH3COO.Na
U
•.., • \ CO:C H = 1 :0
.~,
........ T
25 cis
1
o
HCOO.Na Fig. 1.16. C Is photoelectron
..:-...
peaks for the sodium salts
....
of the first four fatty acids.
..,.- ~ Adapted from the original
work on ESCA by Siegbahn et
o
al. [61]. The oxygen-attached
295 290 285 280 275 carboxyl carbon appears at
binding energy (eV) the higher binding energy
~
'c ~
:::l 'c
g
>- :::l
~
:0 ~
~ :0
~
UJ
~ [jJ
g z
Z
"0
Fig.1.IT. SiL 23 VV Auger peaks [in differential dN(E)/dE and integral N(E)
modes] from elemental silicon (Si-Si), chemisorbed hydrogen on Si(lOO) (Si-H),
silicon nitride film (Si-N) and silicon dioxide film (Si-O) [64]
of the past have now been largely overcome with the improved sensitivity of
FTIR over earlier dispersive instruments. In the attenuated total reflectance
(ATR) mode it is particularly valuable for in situ solution studies of the solid-
liquid interface. (This, in particular, will be considered in further detail in
Chap. 8). This contrasts with most modern surface analytical methods where
the studies are generally carried out ex situ.
Fig. 1.18. Transmission electron micrographs of NiO crystals annealed in air for 4 h
at (a) 700°C, (b) noo°c, and (c) 1450°C (2 regions). The transition from defective,
equiaxed crystals with rounded, relatively smooth surfaces to relatively perfect
crystals with basically flat steps and ledges can be seen. [Reprinted with permission
from Figs. Ib, c, C.F. Jones, R.L. Segall, R.St.C. Smart, and P.S. Turner,. J. Chern.
Soc. Faraday Trans. I 73, 1710 (1977)]
Fig. 1.19. (A) Structure of several high-index stepped surfaces with different ter-
race widths and step orientations. (B) LEED patterns of the (a) Pt(755), (b)
Pt(679), (c) Pt(544), and (d) Pt(533) stepped surfaces. [Reprinted from Gabor A.
Somorjai: Chemistry in Two Dimensions: Surfaces. Copyright (c) 1981 by Cornell
University. Used by permission of the publisher, Cornell University Press]
46 C. Klauber and R. St. C. Smart
b
Friedel
Result: oscillatio7 electronic charge
density, n(x)
c
-2ft. o 2ft. x
Fig. 1.20. (a) Schematic illustration of the band model of metals (without electric
fields). The work function cP is defined relative to Eo. (b) The band model of a p-
type semiconductor surface involving both the space charge (CPs) and dipole moment
(CPdip) components. [Reprinted with permission from Figs. 7.1, 7.3, J. Nowotny and
M. Sloma in "Surface and Near-Surface Chemistry of Oxide Materials", ed. by L.-
C. Dufour and J. Nowotny, Ch.7 (Elsevier, Amsterdam 1988) p.281). (c) Charge
density oscillations and redistribution at a metal-vacuum interface. [Reprinted from
Gabor A. Somorjai Chemistry in Two Dimensions: Surfaces. Copyright (c) 1981 by
Cornell University. Used with permission of the publisher, Cornell University Press)
(1.8)
Very often in materials science, it is not the top atomic or molecular surface
layer that we are concerned with but a near-surface region from 2-100 nm
thick. Some examples of surface phenomena in this category are:
The range of surface concentrations in these examples may vary from major
matrix elements to ppm (as in n-type or p-type semiconductor dopants).
52 C. Klauber and R. St. C. Smart
A 8 1~----150A------'.1
o
---
----150A-1
...--.-~" .....•.
o
...... ....··JO·....-..···.Cr .....~"--~
Si
-
:,0 /"
..,........... .. "..,
'
: ....
. ..-.-----:-----..,
Cr '•
,____y-'S~
'.'
o 5 10 15 20 25 30 35 o 5 10 15 20 25 30 35 40
ETCHING TIME, MINUTES ETCHING TIME, MINUTES
100
C
l
z
0
~ 0 o
cr: Ta
z>- so
w
()
Si
z
0
()
()
~
0
t<
o Si (ox)
27 54
SPUTIER TIME (min.)
o
~pt+Si 106 _pt_l_ _I- PtSi 51-'
/ ..............\
"' 10 5
....... . _________ _
(\
~o 1 4
0 ..............
! ,"'\......... ________________ _
28Si++ ~ ~_~~s~:
I ,
Q..
~ 103 ," \'0, - 194 pt+ I ,
'00
c ," ,,---' \,
~ 10 2
,
:
I 10'
~~
100~~~~~~--~--_r--~~~~~~ 100+---r---~-,~~--~--~~~~-r-
o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
f..·................·......····\
106 ~PtSi_I-Si_
10 5 @
1~ ,~ _____ _
10'
100+---~--~--~--~--~~~~~--~- 100+---~--~--~--~--~uw~~-r---r---
o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Depth beneath surface (~m) Depth beneath surface (~m)
1 Solid Surfaces, Their Structure and Composition 53
Fig. 1.21. (previous page) (A) Auger (AES) depth profiles of 15 nm thick nichrome
film deposited on a silicon substrate recorded immediately after deposition and
(B) after heating in air at 450° C for 30 s. The heat treatment has produced both
oxidation and migration of the Cr and Ni. (C) XPS depth profiles of a tantalum
silicate film on Si0 2 substrate. The Si( ox) signal used was the Si(2p), 103 e V binding
energy emission; the elemental Si is the Si(2p) intensity at 99 e V binding energy. (D)
SIMS depth profiles showing redistribution of B implanted in Si during platinum
silicate growth; (a) as-deposited; (b) after annealing at 400°C for 5min; (c) after
annealing at 400°C for 30 min; (d) after 600°C annealing for 30 min. [Reprinted
with permission from Figs. 7, 8,12, K. Bomben and W.F. Stickle, Surface Analysis
Characterisation of Thin Films in "Microelectronic Manufacturing and Testing"
(Lake Publishing, Libertyville, IL, USA 1987)]
There are still a few surface structural features as yet not fully explored by
the preceding sections. They tend to be specific or localised features on or in
the surface and are largely associated with ceramic materials although, in all
cases, they can be found in other materials as well. We will look at techniques
for their examination.
Ceramics, rocks, ores, and composites exhibit these features. They are of
major importance in determining the surface properties of the material.
Large grain (i.e. > l/lm) multiphase materials can be imaged and chem-
ically mapped using SEM/EDS and backscattered electron images (see Sec-
tion 1.4.1). The analysis depth is relatively large (i.e. 0.1-1 /lm) in these
modes. For more surface-sensitive structural differentiation (i.e. 1-10 nm)
SAM can map features down to 50 nm. An example is shown in Fig. 1.22.
If the structural differentiation is even finer than this, TEM or STEM, with
EDS or EELS for analyses, must be used. These techniques can distinguish
grain structures, phase distributions and inclusions down to unit cell level
(i.e. 1-2 A) but do require electron-transparent samples. STM, with atomic
resolution, would rarely be justified for this application unless there is a need
to image and analyse (via STS) sub-structure within one of these regions.
All of these techniques can be used on the majority of materials as listed
in the beginning of this book. Extensive use of these techniques can be found
in the literature of metals, alloys, catalysts, minerals (i.e. in rock-forming)
and ceramics. A particularly valuable application is in the definition of re-
gions of different structure in blended polymers. These can be: elastomeric
inclusions; phase-separated regions; segregated layers or regions (e.g. silicones
in polyester matrices); unreacted monomer inclusions; and regions containing
oxidised products or unsaturated bonding (e.g. Fig. 1.23). Similar applica-
tions can be found for composites and many natural materials.
Materials generally fracture along regions of least resistance, i.e. those across
which bonding at the boundary is weakest. In single-phase materials, this
may occur predominantly along grain boundaries (although trans-granular
fracture also occurs under high-energy deposition). Surface analysis of the
56 C. Klauber and R. St. C. Smart
- 2 ~m
Fig. 1.22. SAM images from a precipitate in an Fe/Mn metal alloy (a) secondary
electron image showing substructure in the precipitate; (b) Fe distribution showing
absence of iron in the precipitate region; (c) S distribution with high concentration
except in the substructure and; (d) Mn distribution showing separation of Mn into
substructure of the precipitates. [Reprinted with permission from Applications Note
No. 7903 (Perkin Elmer Physical Electronics Division, MN, USA)]
presented grain boundaries, using SAM, XPS and static SIMS, has revealed
segregation of large impurity atoms into the atomic-width boundaries. The
detailed structure of displaced atoms can be seen using TEM, STM, LEIS
and ABS.
In multiphase materials, particularly multi phase ceramics, rocks and min-
erals, the region between grains of different phase is usually wider than a
single atomic layer and has a composition different from that of either of the
two adjacent phases. This integranular region can vary in width from 2 nm,
as in the ceramic Synroc [101], to several micrometres, as in alumina-based
ceramics. The structure of these intergranular films is often amorphous. Frac-
ture in these materials is predominantly intergranular with the layer tending
to "peel" off with one grain or the other leaving only a monolayer or two on
the other surface. The material in these regions, the bonding and structure,
can dominate such properties as mechanical strength, heat transfer, physical
and chemical durability and mineral separation (i.e. processing).
1 Solid Surfaces, Their Structure and Composition 57
C 15 5 C 15
Polyethylene
C 15 (a) (b)
0 15
Mylar 0 Mylar
_.
C-()
w
W
z3
Figure 1.24a illustrates the use of SAM to study fracture faces and inter-
granular films from a ceramic material. A surface showing enhancement of
alkali metal cations [e.g. (Na, Cs), Al and 0 is found, removable by etching
with an ion beam to a depth of > 2 nm. These surface enhancements are
confirmed by XPS and static SIMS as in Fig. 1.24b and c. TEM can be used
to image the width of the film (Fig. 1.24d)]. A considerable body of work now
exists on the study of inter granular films in ceramics using these techniques
(e.g. [102]).
Fig.1.24. (A) Secondary electron image (a) of a pore in a fracture face of the
ceramic Synroc C with Cs maps of the same area (b) before and (c) after removal
of ca. 15 nm by ion etching. Similar images (d, e, fJ of a Cs-rich area (e) before and
after (fJ removal of ca. 375 nm. The fracture face exposes closed porosity and thin
« 2 nm) remnants of intergranular films in the ceramic. (B) XPS depth profiles
from a Synroc C fracture face before (full line ) and after (dashed line ) immersion in
doubly distilled water for ca. 30 s. An enhancement of Cs, Al and, possibly, Mo in
the fresh fracture face is reduced by the water. The numbers on the right indicate
averaged atomic % of each element in the bulk of the material. (e) Sta tic SIMS
depth profile of a fresh fr acture face of Synroc C demonstrating enchancement of
Cs and Na, but not Ca and Ti, in the intergranular region. CD) High resolution
TEM of an integranular film at the interface between grains of perovskite and
magnetoplumbite. Cleavage has been initiated, with the crack (arrowed) running
between grains, the intergranular film adhering to the pervoskite grain . [R eprinted
with permission from Figs. 1, 4, 5, 6; J .A. Cooper, D .R Cousens, J.A. Hanna, RA .
Lewis, S. Myhra, R.L. Segall, RSt.C. Smart, P.S. Turner , and T.J. White, J. Amer.
Ceram. Soc. 69(4), 347- 352 (1986)J
1 Solid Surfaces, Their Structure and Composition 59
B DEPTH , NM DEPTH. NM
c SPUTTER DEPTH, NM o
0.4 0,8 1,2 1.6 2.0
5r-------------------------------~
i:~:.::::-----: : : : : :.~;~:
20 40 60
SPUTTER T IME, MIN
80
60 C. Klauber and R. St. C. Smart
Fig. 1.25. TEM of an ion beam -- thinned section of Synroc B (Synroc C with-
out addition of simulated nuclear waste) after hydrothermal attack for 1 day. The
perovskite grains have completely dissolved, recrystallising in-situ as brookite (B)
and anatase (A). Hollandite (H) grains are slightly attacked with a few small Ti02
crystallites on their surfaces. Zirconolite (Z) appears unaltered by this treatment.
[Reprinted with permission from T. Kastrissios, M. Stephenson, P.S. Turner, and
T.J. White, J. Amer. Ceram. Soc. 70, 144-146 (1987)]
Table 1.8. Accumulated threshold charge densities Dc for detectable electron radi-
ation damage to occur for a variety of materials and electron energies. The threshold
levels relate to maximum exposure times and optimal spatial resolutions possible
without damage occurring
References
1. D. Tabor: Surf. Sci. 89, 1 (1979)
2. P.H. Abelson: Science 234, 257 (1986)
3. W.A. Grant, R.P.M. Procter, J.L. Whitton (eds.): Surface Modification of
Metals by Ion Beams (Elsevier Sequoia, Lausanne 1987)
4. J.W. Rabalais, S. Kasi: Science 239, 623 (1988)
5. P. Danielson: In Kirk-Othmer Encyclopedia of Chemical Technology, Vol. 20
(Wiley, New York 1982) pp. 811, 812
6. F. Jona: J. Phys. C 11, 4271 (1978)
7. J. Verhoeven: J. Environ. Sci. 22, 24 (1979)
8. R.G. Musket, W. McLean, C.A. Colmenares, D.M. Makowiecki, W.J.
Siekhaus: App!. Surf. Sci. 10, 143 (1982)
9. G.A. Somorjai, M.A. Van Hove: Structure and Bonding 38, 1 (1979)
10. G.A. Somorjai: Chemistry in Two Dimensions: Surfaces (Cornell University
Press, Ithaca 1981)
11. V.M. Bermudez: Electronic structure of point defects on insulator surfaces.
Prog. Surf. Sci. 11, 1 (1981)
12. E.A. Colbourn, W.e. Mackrodt: Theoretical aspects of H2 and CO chemic-
sorption on MgO surfaces. Surf. Sci. 117, 571 (1982)
13. C.F. Jones, R.A. Reeve, R. Rigg, R.L. Segall, R.St.C. Smart, P.S. Turner: Sur-
face area and the mechanism of hydroxylation of ionic oxide surface. J. Chem.
Soc. Faraday I, 80, 2609 (1984)
14. E.A. Colbourn, W.C. Mackrodt: Irregularities at the (001) surface of MgO:
topography and other aspects. Solid State Ionics 8, 221 (1983)
15. P.W. Tasker, E.A. Colbourn, W.C. Mackrodt: The segregation of isovalent
impurity cations at the surfaces of MgO and CaO. J. Am. Ceram. Soc. 68,
74 (1985)
16. C. Ocal, B. Basurco, S. Ferrer: Surf. Sci. 157, 233 (1985)
17. J.E. Crowell, J.G. Chen, J.T. Yates Jr.: Surf. Sci. 165, 37 (1986)
18. C.F. McConville, D.L. Seymour, D.F. Woodruff, S. Bao: Surf. Sci. 188, 1
(1987)
19. LP. Batra: J. Electron. Spectrosc. Relat. Phen. 33, 175 (1984)
20. W.A. Anderson, W.E. Haupin: In Kirk-Othmer Encyclopedia of Chemical
Technology, 3rd ed., ed. by M. Grayson, D. Eckroth (Wiley, New York 1978)
Vol. 2, pp. 181-183
66 C. Klauber and R St. C. Smart
38. S. Tanuma, C.J. Powell, D.R. Penn: Surf. Sci. 192, L849 (1987)
39. C.D. Wagner, W.M. Riggs, L.E. Davis, J.F. Moulder, G.E. Muilenberg (eds.):
Handbook of X-ray Photoelectron Spectroscopy (Perkin-Elmer, Eden Prairie,
Minnesota 1978) pp. 42, 50
40. M.F. Ebel: J. Electron Spectrosc. Relat. Rhenom. 22, 157 (1981)
41. C.W. Magee: Nucl. Instrum. Meth. 191, 297 (1981)
42. D.P. Leta, G.H. Morrison: Anal. Chern. 52, 514 (1980)
43. Riber MIQ 256 SIMS/Ion microprobe
44. P. Skeldon, K. Shimizu, G.E. Thompson, G.C. Wood: Thin Solid Films 123,
127 (1985)
45. Based on an absorption sensitivity for iron of 0.15 f.lg/ml/1% Abs (p. 825)
and a sample volume of 5 f.ll (p. 355), In Instrumental Methods of Analysis,
H.H. Willard, L.L. Merritt Jr., J.A. Dean (eds.) (Van Nostrand, New York
1974)
46. D.M. Hercules, L.E. Cox, S. Osnisick, G.D. Nichols, J.C. Carver: Anal. Chern.
45, 1973 (1973)
47. R.C. Weast, M.J. Astle (eds.): CRC Handbook of Chemistry and Physics, 63rd
edn. (CRC, Boca Raton 1982) F-161
48. R. Browning: J. Vac. Sci. Technol. A 2, 1453 (1984)
49. P.D. Prewett, D.K. Jeffries: J. Phys. D 13, 1747 (1980)
50. N.M. Ceglio, A.M. Hawryluk, M. Schattenburg: J. Vac. Sci. Technol. B 1,
1285 (1983)
51. 1.W. Drummond, T.A. Cooper, F.J. Street: Spectrochimica Acta 40B, 801
(1985)
52. K. Yates, R.H. West: Surf. Interface Anal. 5, 217 (1983)
53. For example, VG Scientific ESCALAB 20-X
54. For example, VG Scientific ESCASCOPE
55. J.F. Evans, J.H. Gibson, J.F. Moulder, J.S. Hammond. The PHI Interface 7,
1 (1984) (Perkin Elmer, Eden Prairie, USA)
56. M.L. Tarng, D.G. Fischer: J. Vac. Sci. Technol. 15, 50 (1978)
57. V. Thompson, H.E. Hintermann, L. Chollet: Surf. Technol. 8, 421 (1979)
58. D.R. Clark, L.L. Hench: An overview of the physical characterisation of
leached surfaces, Nuel. Chern. Waste Manag. 2, 93 (1981)
59. J.W. Strojek, J. Mielczarski: Spectroscopic investigations of the solid-liquid
interface by the ATR technique, Adv. Colloid Interf. Sci. 19, 309 (1983)
60. R.F. Willis (ed.): Vibrational Spectroscopy of Adsorbates (Springer, Berlin,
Heidelberg 1980)
61. K. Siegbahn, C. Nordling, A. Fahlman, R. Nordberg, K. Hamrin, J. Hedman,
G. Johansson, T. Bergmark, S.-E. Karlsson, 1. Lindgren, B. Lindberg: ESCA,
Atomic, Molecular and Solid State Structure Studied by Means of Electron
Spectroscopy (Almqvist and Wiksells, Uppsala 1967) p. 79
62. R. Opila, R. Gomer: Surf. Sci. 105, 41 (1981)
63. J. Hulse, J. Kiippers, K. Wandelt, G. Ertl: Appl. Surf. Sci. 6, 453 (1980)
64. H.H. Madden: J. Vac. Sci. Technol. 18, 677 (1981)
65. M. Barber, R.S. Bordoli, G.J. Elliot, R.D. Sedgwick, A.N. Tyler: Anal. Chern.
54, 645A (1982)
66. C.J. Powell, N.E. Erickson, T.E. Madey: J. Electron Spectrosc. Relat. Phen.
17, 361 (1979)
67. C.J. Powell, N.E. Erickson, T.E. Madey: J. Electron Spectrosc. Relat. Phen.
25, 87 (1982)
68 C. Klauber and R. St. C. Smart
68. V.E. Henrich: Electronic and geometric structure of defects on oxides and
their role in chemisorption, In Surface and Near Surface Chemistry of Ox-
ide Materials, ed. by J. Nowotny, L.-C. Dufour (Elsevier, Amsterdam 1988)
pp.23-60
69. A.B. Kunz: Theoretical study of defects and chemisorption by oxide sur-
faces, In External and Internal Surfaces in Metal Oxides, ed. by L.-C. Dufour,
J. Nowotny, Materials Science Forum (Trans. Tech. Publications, Claustal-
Zellerfeld 1988) pp. 1-30
70. R.L. Segall, R.St.C. Smart, P.S. Turner: Oxide surfaces in solution, In Surface
and Near-Surface Chemistry of Oxide Materials, ed. by J. Nowotny, L.-C.
Dufour (Elsevier, Amsterdam 1988) pp. 527-576
71. V.E. Henrich: Ultraviolet photoemission studies of molecular adsorption on
oxide surfaces, Prog. Surf. Sci. 9, 143 (1979)
72. V.E. Henrich, G. Dresselhaus, H.J. ZeIger: Chemisorbed phases of H20 on
Ti02 and SrTi0 3 , Solid State Commun. 24, 623 (1977)
73. M.W. Roberts: Metal oxide overlayers and oxygen-induced chemical reac-
tivity studied by photoelectron spectroscopy, In Surface and Near-Surface
Chemistry of Oxide Materials, ed. by J. Nowotny, L.-C. Dufour (Elsevier,
Amsterdam 1988) pp. 219-246
74. G.A. Somorjai: Adv. Catal. 26, 1 (1979)
75. G. Heiland, H. Liith: In The Chemical Physics of Solid Surfaces and Het-
erogeneous Catalysis, Vol. 3 , ed. by D.A. King, D.P. Woodruff (Elsevier,
Amsterdam 1984) p. 147
76. S.C. Chang, P. Mark: Surf. Sci. 45, 721 (1974); ibid. 46, 293 (1974)
77. C.C. Schubert, C.L. Page, B. Ralph: Electrochim. Acta 18, 33 (1973)
78. J. Marien: Phys. Status Solidi A 38, 339, 513 (1976)
79. J. Nowotny, M. Sloma: Work function of oxide ceramic materials, In Surface
and Near-Surface Chemistry of Oxide Materials, ed. by J. Nowotny, L.-C.
Dufour (Elsevier, Amsterdam 1988) pp. 281-344
80. S.R. Morrison: The Chemical Physics of Surfaces (Plenum, New York 1977)
81. H. Wagner: Physical and Chemical Propert'ies of Stepped Surfaces, Springer
Tracts Mod. Phys. (Springer, Berlin, Heidelberg 1978)
82. H.E. Clark, R.D. Young: Surf. Sci. 12, 385 (1968)
83. H.D. Hagstrum: Science 178, 275 (1972)
84. H.D. Hagstrum: Phys. Rev. 150, 495 (1966)
85. G. Heiland, H. Liith: Adsorption on oxides, in The Chemical Physics of Solid
Surfaces and Heterogeneous Catalysis, ed. by D.A. King, D.P. Woodruff (El-
sevier, Amsterdam 1984) p. 156
86. N.S. Huck, R.St.C. Smart, S.M. Thurgate: Surface photovoltage and XPS
studies of electronic structure in defective nickel oxide powders, Surf. Sci.
169, L245 (1986)
87. E. Garrone, A. Zecchina, F.S. Stone: Philos. Mag. B 42, 683 (1980)
88. W. Gopel: Prog. Surf. Sci. 20, 9 (1985)
89. J. Cunningham: Photoeffects on metal oxide powders, in Surface and Near-
Surface Chemistry of Oxide materials, ed. by J. Nowotny, L.-C. Dufour (El-
sevier, Amsterdam 1988) pp. 345-412
90. M.W. Roberts, R.St.C. Smart: XPS determination of band bending in defec-
tive semiconducting oxide surfaces, Surf. Sci. 151, 1 (1985)
91. A.W. Adamson: Physical Chemistry of Surfaces (Wiley, New York 1986)
1 Solid Surfaces, Their Structure and Composition 69
C. Klauber
Modern surface analytical methods over the last three decades have been
dominated by those requiring ultrahigh vacuum (UHV) chambers in which
to carry out the analyses. This is not a universal requirement for surface
analysis and several of the techniques such as Fourier transform infrared
(FTIR), scanning tunnelling microscopy (STM) and ellipsometry do not have
a mandatory vacuum requirement. Vacuum is of course required by those
techniques utilizing beams of particles and higher energy radiation so that
the beams may be generated and travel undisturbed until intercepting the
surface. The requirement for UHV or vacua of S; 10- 10 mbar (10- 8 Pa) is
fundamental to surface analysis when those beams are employed. This arises
due to the flux of residual gas molecules striking the surface i.e. the number
of molecules per unit area per unit time, which is responsible for the pressure
that those gas molecules exert upon the surface. By knowing the pressure the
flux can be evaluated. From the kinetic theory of gases [1] the molecular flux
Z is given by the Herz-Knudsen equation:
Nc
z=-
4V'
(2.1)
where N IV is the number of molecules per unit volume and c is the average
speed of the molecules. For a gas of molecular weight M at temperature T
[1]:
c = J8RT (2.2)
7fM
where R is the gas constant. Combining the above with the ideal gas equation
PV = nRT and N = nNA where NA is Avogadro's number, gives:
Z = nNAPJ(8RTI7fM)
4nRT '
and therefore
Z= NAP (2.3)
V27fMRT
72 C. Klauber
10 8
1 year
- 10 7
1 month
- 106
1 week
- 105 1 day
U
Q)
5 hours
.!E.-
Q) 1 hour
E
- 103 :;:::;
10 minutes
Fig. 2.1. Relationship between gas pressure, surface contamination times and mean
free path lengths. The earth's atmospheric altitude variation is included for com-
parison. trn is the time to form a monolayer of nitrogen with sticking probabilities
s = 0.1 and 1.0, ts is the average time before a surface atom is struck by a gas
phase molecule. Ae is the mean free path of a 100 eV electron in N 2 , 02 or CO, Ai
is the mean free path for that collision to be ionizing, Ax is the mean free path of
a molecule of X in X [2]
Fig. 2.2. Front elevation of a typical multitechnique surface analysis system (top)
and a cross sectional view from above of specimen transporter mechanisms for
such a system (bottom). Key to the diagrams: (1) Fast entry specimen insertion
lock (stainless steel). (2) UHV specimen preparation chamber (stainless steel). (3)
UHV experimental/analysis chamber (mu-metal). (4) Vi ton sealed gate valve. (5)
Rotary drive to specimen transfer mechanism. (6) Titanium sublimation pump
vessel (stainless steel). (7) Viewport. (8) Autocarousel motor drive. (9) High pre-
e
cision specimen translator (X, Y, Z translation and tilt). (10) Twin anode X-
ray source (Al/Mg) . (11) UV discharge source (UPS) . (12) Monochromated X-ray
source (Al/Mg). (13) 2000 A electron source (AES, SEM, SAM). (14) Scanning ion
source. (15) Electron energy analyzer vessel. (16) Detector (single or multichan-
nel). (17) Specimen fracture stage. (18) Static, broad beam ion source. (19) High
pressure gas reaction/catalysis cell. (20) Specimen transfer fork (wobble stick). (21)
6-specimen mobile carousel. (22) lO-specimen autocarousel. (23) Preparation vessel
specimen transfer "railway". (24) Binocular microscope. (25) Alternative fast en-
try lock position or optional extension chamber port. (26) Port for monochromated
electron source. (27) Port for specimen heating/cooling stage. Reproduced courtesy
of VG Scientific Ltd, UK
76 C. Klauber
A variety of vacuum pumps exist which can pump down to the URV
regime, each type has its own virtues. The most popular main chamber
pumps are diffusion, turbomolecular or ion pumps. Diffusion pumps using
modern fluids such as polyphenyl ethers are excellent performers but liquid
nitrogen cooled traps are mandatory. Turbomolecular pumps can be used
without traps, although trapping improves their ultimate performance. Ti-
tanium sublimation pumps, utilizing chemical gettering, are useful for addi-
tional pumping. Cryogenic based sorption pumps are particularly useful for
rapidly evacuating large chambers. A comprehensive description of modern
vacuum technology can be found in Weissler and Carlson [5].
Aspects of the various probes such as electron and ion guns, X-ray sources,
mass analysers and electron spectrometers are dealt with in the subsequent
chapters pertaining to the individual techniques. In commercially produced
instruments these are normally produced by the one manufacturer. One-off
research instruments often consist of a custom central chamber with a variety
of accessories, often in-house constructed.
Means by which specimens of interest are introduced into vacuum can vary.
Dedicated research instruments may need to be brought entirely up to atmo-
sphere in order to change samples via unbolting flanges. The instrumental
cost saving by eliminating a separate introduction chamber is impractical for
analytical purposes as specimens need to be changed frequently. In any case,
the subsequent baking required exposes the specimen to high pressures of
residuals whilst at temperature, thereby forming a considerable contamina-
tion selvedge. The alternative fast entry systems are of two general types,
one involves a specimen probe or rod onto which the sample is mounted, the
rod then being pushed into the analytical region via a series of differentially
pumped seals. The other, illustrated in Fig. 2.3, involves moving the speci-
men, mounted on a small stub, through one or two isolatable locks via rack
and pinion or pulley specimen transporters. Stubs are a more convenient ap-
proach, but they can be less versatile with respect to heating and cooling
of the specimen or making specific electrical connections e.g. as in attach-
ment of a thermocouple. It is feasible to introduce up to a half-dozen stubs
at once via a carousel system. With the specimen probe system electrical
connections are made ex situ and the wiring travels with the specimen. The
specimen translator for dedicated research instruments are usually superior
at heating, cooling and can have additional degrees of mechanical freedom,
such as azimuthal adjustments for angle-resolved studies. Figure 2.3 shows
an internal view inside a vacuum system utilizing the stub method of sample
transport.
The method of specimen mounting is largely governed by the nature of
the material being examined. For most spectroscopies, good earthing of con-
2 URV Basics 77
Fig. 2.3. View inside a surface analysis vacuum system incorporating an auto-
carousel and utilizing the stub method of sample transport. Reproduced courtesy
of VG Scientific Ltd, UK
tape, however, is quite gaseous under X-ray irradiation and many workers
avoid its use.
Viable specimen dimensions and masses are dependent on the vacuum
system and technique(s) being employed. For instance, spot electron-excited
Auger analyses can be performed quite easily on particles microns across,
whereas signal-to-noise requirements for XPS mean that dimensions of 5-
10 mm are routine. Large specimens that cannot be cut usually pose the
greatest difficulty. Some systems will accept wafers up to 50 mm in diameter,
but beyond a few mm in thickness they soon become unwieldy to handle. Of
particular concern is a specimen's prehistory, especially the nature of fluids
with which it has been in contact, as not all materials are DRV compatible.
With, say, high vapour pressure lubricants, any attempts at solvent cleaning
may disturb the surface that is really of interest and the information required
may be lost.
Specimen storage and transfer. Storage is a problem even in very clean labo-
ratories. Hence the best approach is to analyse specimens as soon as possible.
Glove boxes, vacuum chambers and desiccators are generally the best means
of storage as they can minimize oxidation/hydrolysis effects. These can also
be adapted to directly connect to entry locks to avoid any atmospheric expo-
sure of the specimen surface. The container itself should not be a source of
contamination and materials with volatile components should be kept sepa-
rate to avoid cross contamination. During shipping most factors can be easily
controlled except temperature, which can be quite detrimental.
80 C. Klauber
Specimen mounting. This has been briefly touched upon in Sect. 2.3. The
most serious difficulty which might arise concerns specimens which are poor
electrical conductors. In addition to the approaches pointed out in Sect. 1.11.2
to minimize charging it may also be feasible to place a conductive mask, wrap
or coating about the specimen. This mask e.g. metal foil or colloidal graphite,
2 URV Basics 81
References
Techniques
3 Electron Microscope Techniques
for Surface Characterization
Electron microscopy in its various forms has developed over the past fifty
years into one of the major techniques of materials science. Surface analytical
techniques are more recent additions to the materials scientists' range of
experimental methods for learning about materials properties. Microscopy
and spectroscopy are complementary, and the use of one alone can result in
an inadequate characterization of a material. In this chapter we will consider
the various modes of electron microscopy and attempt to demonstrate the
critical importance of applying electron optical imaging together with surface
analytical techniques in studies of surfaces.
The textbook picture of a specimen to be examined using surface ana-
lytical techniques such as XPS or SIMS is typically a perfectly fiat surface
of a laterally homogeneous material (Fig.3.la). Of course, as the chapters
of this book make clear, no real sample of any significant interest is like
that, and techniques such as scanning Auger microscopy can provide in-
formation about variations in composition across a surface, with resolution
to < 100 nm. Nevertheless, there is a tendency to interpret surface analyt-
ical data in terms of a postulated model of the surface. Only by looking
at the surface, with whatever resolution is required in order to see the rel-
evant detail, can one be confident of the interpretation of the surface an-
alytical data. Thus microscopy - optical, electron, scanned probe, etc. --
must be employed in conjunction with the surface spectroscopies. For ex-
0.1 mm
~
,a-MO
~
ample, the two samples illustrated in Fig. 3.1 b,c could give identical XPS
spectra and depth profiles, whereas one consists of a continuous amorphous
oxide film and the other of a fine dispersion of oxide crystallites across the
surface.
The aim of this chapter is to introduce the basic concepts of electron mi-
croscopy of surfaces to readers who are not familiar with electron microscopy.
We will consider the ways in which scanning and transmission electron micro-
scopes can provide information about surfaces, and the nature of that infor-
mation. Two other surface imaging systems - scanned probe microscopy, and
scanning Auger microscopy - are covered elsewhere in this book (in Chaps. 10
and 6 respectively) and will not be considered here. An excellent introduc-
tion to more advanced aspects of the subject has been presented by Venables
et al. [1]. The subject was covered in the proceedings of two NATO Advanced
Study Institutes [2,3], and at the 1989 Wickenburg Workshop [4].
In this short article, we cannot cover all relevant aspects of the instru-
ments and techniques in detail. There are many excellent texts on electron
microscopy, some of which are listed at the end of this chapter as sources for
further reading [5-11]. In particular, more detailed descriptions of the major
components of electron microscopes, and more comprehensive explanations
of their operation and performance, should be studied by anyone proposing
to learn to use electron microscopy for the characterization of surfaces. A
good starting point is the short book by Goodhew [5].
In the following sections, we will consider first the nature of the informa-
tion we may require from electron microscopy, then the major characteristics
of electron optical systems, and the factors which determine the nature and
quality of electron images. The roles of the scanning electron microscope and
the transmission electron microscope in surface studies will be described and
illustrated. In the final section we will look briefly at other developments in
the imaging of surfaces.
• whether there are minor phases present in the nominally single phase
material;
• the orientation of grains;
• whether chemical attack results in the growth of precipitates;
• whether attack occurs preferentially at boundaries, or on certain grains;
• the topography of fractured surfaces, and whether they reveal intergran-
ular or transgranular fracture;
• whether the sample is porous.
There are two ways in which images can be obtained using electron beams.
The traditional transmission electron microscope (TEM) is a close analogue
of the light optical microscope, and involves the illumination of a transparent
object by a beam of electrons, and the formation of a magnified image of the
electron waves emerging from the object using an objective lens and two or
more projector lenses. The scanning electron microscope (SEM) uses lenses to
form a demagnified image of the electron source. This fine probe of electrons
is scanned across the sample, and one of a number of possible signals arising
from the electron-specimen interaction is detected, amplified, and used to
modulate the intensity of a TV-like image tube, which is scanned at the
88 P.S. Turner, C.E. Nockolds, and S. Bulcock
same rate as the electron probe. Clearly there are fundamental differences
between these methods, but both use electron sources, electron lenses and
electron detectors, and therefore both are limited by the performance of such
devices.
The TEM and SEM techniques converge in the STEM (scanning TEM)
in which a sub-nanometre diameter probe is scanned across a thin sample
and a transmitted electron signal - often also filtered to remove inelastically
scattered electrons .. used to form the image. Both TEM and STEM can
be operated in reflection mode, in which the electron beam is incident at a
glancing angle and reflected and/or diffracted from a flat sample (REM and
SREM).
In forming an image, two factors govern the information which may be
obtained about the object - the resolving power of the imaging instrument
(the smallest resolvable separation of two distinct points in the object) and
the contrast in the image (the difference in intensity which allows us to dis-
tinguish resolved detail from a background intensity level).
The best resolution attainable with a given instrument is determined by
the quality of the objective lens in the TEM or the probe-forming lens in the
SEM. The wave nature of the electrons leads to a fundamental diffraction-
limited resolution, but the high aberrations of magnetic electron lenses de-
termine the actual resolution. :For a given lens, the TEM resolution and the
smallest probe diameter which that lens could produce are essentially iden-
tical (e.g., rv 0.3 nm at 100 ke V), but in practice a typical commercial SEM
has a best resolution significantly worse than a standard TEM, due to lower
electron energies and larger focal lengths (e.g., 3nm at 15keV). With the
recent widespread availability of field emission sources for SEM, and probe
sizes down to 1 nm, greatly improved resolution of surface is possible, but is
often not achieved due to limitations in contrast.
The contrast of an image is limited by the random nature of the emission
of electrons from the source. If we consider a particular element in the image
(a picture element or pixel) with an intensity corresponding to the detection
of N electrons in that pixel, then the standard deviation in the intensity will
be proportional to .JR. The difference /',.N between signals from nearby pixels
should be at least 3.JR if contrast between these pixels is to be detected in
the image [8,9]. In the TEM, all image points are detected simultaneously
(parallel detection), and in practice it is usually easy to ensure that the noise
level (.JR / N) is very low. But in the SEM each pixel is detected in sequence
as the probe scans across the sample (serial detection); the noise level in
scanned images is a critical factor in determining the image contrast and
detectable resolution.
3 Electron Microscope Techniques for Surface Characterization 89
A critical factor in electron imaging is the current density of the electron beam
at the object. The electron optical brightness, (3 of the system, is defined by
(3.1 )
(3.2)
Thus for scanned imaging systems, with a given beam current, small probe
sizes require high convergence angles, and vice versa.
There are three types of electron source used in contemporary EMs, dif-
fering in maximum brightness and in maximum beam current. The tradi-
tional electron gun uses a tungsten wire, heated to about 2700 K, from which
electrons are emitted and accelerated to the required energy. Under optimal
adjustment, a brightness (3 = 5 x 105Acm- 2 sc 1 is obtained at 100keV,
with a total current of up to 100 f.tA from an effective source diameter of
about 30 f.tm. A more recent development is the lanthanum hexaboride (LaB6)
source, which, operating at lower temperatures, gives about ten times higher
brightness, but requires better vacuum conditions. The field emission source,
a pointed tungsten tip operated in URV at room temperature, can provide
brightness values as high as 109 Acm- 2 sr- 1 at 100keV, but at much lower
total beam currents. For details of the construction and operation of elec-
tron guns using these sources, the reader is referred to texts on electron
microscopy, e.g. [6,9,10]. The standard guns for TEMs and SEMs, using W
or LaB 6, are very flexible, and easily adjusted to provide for a range of beam
currents at or close to the corresponding maximum brightness values.
These electron guns are almost monoenergetic, with a narrow range of
electron energies around the mean, due to the thermal spread in energy of
the emitted electrons. In the thermionic guns, this is about 1.5 eV, whereas
for the field emission gun it is about 0.25 eV.
Electric and magnetic fields having cylindrical symmetry act as lenses for
charged particles. Electrons emerging from a point in an object are brought
to a focus in the corresponding image plane. In comparison with light optical
90 P.S. Turner, C.E. Nockolds, and S. Bulcock
The standard SEM is the most useful electron optical system for surface imag-
ing, and should be applied routinely to characteri;l,e all surfaces which will
be, or have been analyzed by XPS, SIMS, etc. (It will usually be preferable
to image after surface analysis, to avoid the effects of contamination during
SEM examination.)
The essential elements of an SEM are shown schematically in Fig. 3.2. The
electron gun, fitted with a W, LaB 6 or Field Emission (FE) gun operates
typically over the range 0.1-30 k V accelerating voltage. A condenser lens
produces a demagnified image of the source, which in turn is imaged by the
Condenser
Specimen
Fig. 3.2. Basic features of the SEM, showing electron gun, condenser and objective
(probe-forming) lenses, scanning system, detectors and image display CRT. The
diameter of the aperture and the working distance from lens to specimen determine
the convergence angle of the probe. The probe at the specimen is the demagnified
image of the source, broadened by lens aberrations
92 P.S. Turner, C.E. Nockolds, and S. Bulcock
probe forming lens (often called the objective lens) onto the specimen. The
electron path and sample chamber are evacuated. Scanning coils deflect the
probe over a rectangular raster, the size of which, relative to the display
screen, determines the magnification. Detectors collect the emitted electron
signals, which after suitable amplification can be used to modulate the inten-
sity of the beam of the display video screen, which is rastered in synchronism
with the probe.
resolution of the whole BSE signal can be improved by reducing the beam
energy but this is offset by the fact that the detectors become less efficient
for the lower energy electrons. The current generation of SEMs have the
capability of operating at low kV with very small beam diameters and high
beam currents and this opens up new possibilities for the study of surface
composition using BSE.
do = J4Ip/7f2{3la . (3.4)
80
E'70
c:
~60
Qj
0;50
E
~40
,, --:a::
o 5 10 15 20
Aperture (mrad)
Fig. 3.3. Variation of SEM probe diameter with aperture angle, in accordance
with (3.4). The contributions due to diffraction (dd: -- + -- ); spherical aberration
(ds: - x - ); chromatic aberration (de: --. -- ) are shown for the case of 15 kV,
2eV energy spread, C s = 7.5mm and C c = 10mm. Curves for the geometrical
probe diameter do (-0-) and total probe diameter d p ( - - - ) are shown for two
probe current values and gun brightness 3 x 10 4 A cm -2 sr- 1 . The optimum probe
diameters of 12 and 20 nm are obtained for apertures of 6 and 10 mrad respectively
(a,rrows)
100jJm
Fig. 3.4. SEl micrograph of a crushed single phase mineral sample, pressed onto a
conducting adhesive surface, as used for some XPS characterizations
3 Electron Microscope Techniques for Surface Characterization 95
Fig. 3.5. (a) SE and (b) BSE SEM images of a polished sample of Synroc, showing
the strong edge contrast in the SE image at pores; in both images, the contrast
differences between grains is due to backscattered electrons
a b c 10 pm
a V'\~\~'1It..-flY"~{\(VV{vt/'"rvtfVy.'ViY(''''\~
b
~lyl/l~liljbMVth~j~wv~~~\~~{W'i
c
Fig. 3.6. SEM images and line traces from a fracture face of Synroc, for (a) SE
(b) BSE compositional and (c) BSE topographical signals. The much greater noise
levels in the BSE signals, are evident in the line traces; the topographical BSE
image is formed by subtracting signals from the two halves of the BSE detector,
thus removing the compositional contrast, but losing detail to enhanced noise levels
at low kV. One of the major advantages of operating at low beam energies
is the reduction in the interaction volume. Monte Carlo calculations [26]
show that at 10 ke V the maximum depth of penetration of the beam in Si is
approximately 1 ~m, while at 1 keY it is close to 0.03 ~m. This means that the
secondary electron image at 1 kV gives a better representation of the surface
detail. Another advantage of low kV operation is that the secondary electron
yield increases very strongly, approximately as a function of the inverse of
the beam energy.
Low voltage SEM has other advantages, especially for non-conducting
materials for which a conducting coating (usually about 10nm of gold) is
necessary for conventional SEM imaging. There is an electron energy, usu-
ally around 1 kV, at which no charging occurs. Thus no metal coating of
such samples is required, and fine surface detail is not obscured. In addition,
3 Electron Microscope Techniques for Surface Characterization 97
Fig. 3.7. FESEM images of scale structure of a Merino sheep wool fibre. At 1 kV on
wool, no significant charging of the fibre occurs, allowing imaging without the need
for a conductive coating. Fine details of the surface are revealed. At 5 kV, electrons
penetrate further into the fibre, and the contrast of surface detail is greatly reduced.
(This sample was coated with a thin film of evaporated carbon, to avoid charging at
5 kV.) Images courtesy of Mr David Watson, CSIRO Textile and Fibre Technology.
The TEM is usually used to study the internal microstructure and crystal
structure of samples which are thin enough to transmit electrons with rela-
tively little loss of energy. This requires thicknesses in the range 20-300nm,
depending on the average atomic number of the material, using the typical
200 ke V TEM. Clearly the bulk samples used in surface analysis cannot be
examined directly this way. The transmission microscope is nevertheless an
important adjunct to the SEM in studies of surfaces, because of its greater
resolving power (down to crystal lattice dimensions), its ability to provide
surface sensitive diffraction data and images, and associated analytical capa-
bilities through X-ray and electron energy loss spectroscopies. In addition,
images and diffraction patterns may be formed in the reflection mode, off
smooth surfaces or facets. The challenge is to ensure that the surfaces of
samples examined in the TEM are equivalent to those studied by surface
analysis. The most satisfactory approach would be to combine the surface
analytical and microscopy techniques in the one instrument; some progress
has been made in this direction (Sect. 3.5). However, we will consider here
the more typical situation of separate instrumentation, and the consequent
requirements on sample preparation to achieve our aim.
The TEM is very similar to the conventional light optical microscope in terms
of optical principles. The electron source is followed by two condenser lenses
to provide a uniform illumination of the specimen over the area of interest,
adjustable as the magnification is changed. The sample is mounted on a
stage to provide suitable movement. The primary image is formed by the
objective lens, which determines the resolution obtainable. The final image
is projected onto a viewing screen through two or more projection lenses,
and can be recorded on film placed below the screen, or using CCD array
detectors.
100 P.S. Turner, C.E. Nockolds, and S. Bulcock
Gun
Condenser 1
lens
Condenser 2
C. aperture
Specimen
Objective lens
O.ap.
SA ap.
Intermediate
lens
Projector lens
Final
image 1...4E-------' Screen
Film
Fig. 3.8. The TEM column. The double condenser lens system forms a focussed
beam at the specimen, which is immersed within the magnetic field of the objective
lens. The objective aperture, placed at the back focal plane, controls the image
contrast. The projector lenses magnify the image onto the screen
by selecting some of the diffracted electrons and preventing the rest from
contributing to the image; and to select specific areas of images from which
to obtain diffraction information. Magnetic deflection coils are used not only
to align the electron beam, but also to control it to provide specific image
modes, and to correct for astigmatism in the lenses.
The wavelength of 100 keV electrons is 37 pm, which is about 50 times smaller
than the typical nearest-neighbour separation in many crystals. Electrons
incident on a thin sample are diffracted though angles of order 10- 2 radian;
the resulting electron diffraction pattern can be observed by imaging the back
focal plane of the objective lens. Using an aperture at the level of the first
image (the "selected area aperture"), diffraction data from areas down to
about 300 nm can be obtained; smaller areas may be selected for diffraction
analysis if the TEM has the facility to form fine illumination probes (down
to 2nm).
An amorphous film shows broad rings in its diffraction pattern, with peaks
corresponding to the most probable interatomic spacings. Crystalline mate-
rials show the sharp spots corresponding to Bragg diffraction by the periodic
lattice. Thus information about the degree of crystallinity, the orientation
and crystal structure of individual grains, etc., can be obtained from the
electron diffraction patterns. Although the dominant features of the patterns
will arise from the bulk crystal, weak diffraction spots arising from surface
structures can often be detected, and may be used to form dark field images
of the corresponding surface detail.
those of the lattice planes of the Bragg diffracted beams included in the im-
age. In general such images are very complex, but under certain conditions
(very thin crystals, electron beam orientated down a crystal symmetry axis,
a specific defocus of the objective lens) "structure images" can be recorded
which relate directly to the crystal structure with a resolution corresponding
to the spacing of the planes giving rise to the Bragg reflections passing the
aperture.
The resolution achieved in the images is determined by a balance between
the diffraction limit (rv )..j a) and the effects of spherical aberration (rv C s ( 3 )
and chromatic aberration (rv Cea AE / E). The simplest estimates of resolu-
tion may be made using (3.3). More accurate estimates of resolution involve
consideration of the effects on the phases of the electron waves of defocus,
spherical and chromatic aberration, and the objective aperture which limits
the angular range of diffracted waves contributing to the image (see [10,11]
for detailed discussions of these topics).
Fig. 3.9. Phase contrast image of an MgO crystal, after exposure to water vapour
and electron beam irradiation; recorded at 100 keVin a TEM. Phase contrast is
achieved by defocussing the image, revealing a pattern of small (1-2 nm) rectangular
depressions and protrusions across the initially smooth {100} surfaces
References
1. J.A. Venables, D.J. Smith, J.M. Cowley: HREM, STEM, REM SEM and STM,
Surf. Sci. 181,235 (1987), see also J.A. Venables: Ultramicroscopy 7,81 (1981)
2. A. Howie, U. Valdre (eds.): Surface and Interface Characterization by Electron
Optical Methods; Proc. NATO ASI, Erice 1987 (Plenum, New York 1988)
3. P.K. Larson, P.J. Dobson (eds.): Reflection High-Energy Electron Diffrac-
tion and Reflection Imaging of Surfaces, Proc. NATO ASI, Veldhoven 1987
(Plenum, New York 1988)
4. J.A. Venables, D.J. Smith (eds.): Proceedings of Workshop on Surfaces and
Surface Reactions, Wickenberg Inn, Arizona, 1989: Ultramicroscopy 31 (1989)
5. P.J. Goodhew: Electron Microscopy and Analysis (Wykeham, London 1975);
P.J. Goodhew, F.J. Humphreys: Electron Microscopy and Analysis, 2nd edn.
(Taylor and Francis, London 1988)
6. P.W. Hawkes: Electron Optics and Electron Microscopy (Taylor and Francis,
London 1972)
7. I.M. Watt: The Principles and Practice of Electron Microscopy (Cambridge
University Press, Cambridge 1985; Second edition 1997)
8. J.I. Goldstein, D.E. Newbury, P. Echlin, D.C. Joy, C. Fiori, E. Lifshin: Scan-
ning Electron Microscopy and X-ray Microanalysis (Plenum, New York 1981)
9. L. Reimer: Scanning Electron Microscopy, Springer Ser. Opt. Sci. Vol. 45
(Springer Berlin, Heidelberg 1985; Second completely revised and updated
edition 1998)
10. L. Reimer: Transmission Electron Microscopy, Springer Ser. Opt. Sci., Vol. 36
(Springer Berlin, Heidelberg 1984; fourth edition 1997)
3 Electron Microscope Techniques for Surface Characterization 105
11. J.M. Cowley: Diffraction Physics, 2nd edn. (North Holland, Amsterdam 1981)
12. A.E. Ringwood: Safe Disposal of High Level Nuclear Reactor Wastes (ANU,
Canberra 1978); J.A. Cooper, D.R Cousens, RA. Lewis, S. Myhra, R.L. Segall,
RStC. Smart, P.S. Turner: J. Am. Ceram. Soc. 68, 64 (1985)
13. A.F. Moodie, C.E. Warble: J. Cryst. Growth 10, 26 (1971)
14. D. Cherns: Phi!. Mag. 30, 549 (1974)
15. K. Kambe, G. Lehmpfuhl: Optik 42, 187 (1975); G. Lehmpfuhl, Y. Uchida:
Ultramicroscopy 4, 275 (1979)
16. M. Klaua, H. Bethge: Ultramicroscopy 11, 125 (1983)
17. K. Takayanagi, Y. Tanashiro, M. Takahashi, S. Takahashi: J. Vac. Sci. Techno!.
A3, 1502 (1985)
18. L.D. Marks, D.J. Smith: Nature 303,316 (1983); D.J. Smith: Surf. Sci. 178,
462 (1986)
19. L.D. Marks: Surf. Sci. 139, 281 (1984)
20. For reviews of REM, see K. Vagi: J. AppL Cryst. 20, 147 (1987); K. Vagi:
Electron microscopy of surface structure, Adv. Opt. Elec. Microsc. 11, 57
(1989)
21. N. Osakabe, Y. Tanashiro, K. Vagi, G. Honjo: Japan J. AppL Phys. 19, L309
(1980); Surf. Sci. 109, 353 (1981)
22. Y. Tanishiro, K. Takayanagi: Ultramicroscopy 31, 20 (1989)
23. T. Nagatani, S. Saito: Proc. 11th Int. Conf. Elec. Microsc., Kyoto, Japan (1986)
2101; K. Ogura, M. Kersker: Proc. 46th EMSA (1988) p.204
24. E. Bauer: Ultramicroscopy 17, 51 (1985); E. Bauer, M. Mundschau, W. Swiech,
W. Telieps: Ultramicroscopy 31, 49 (1989)
25. G.G. Hembree, P.A. Crozier, J.S. Drucker, M. Krishnamurthy, J.A. Venables,
J.M. Cowley: Ultramicroscopy 31, 111 (1989)
26. D.C. Joy: Monte Carlo Modelling for E.M. and Microanalysis (Oxford Univer-
sity Press, Oxford, 1995)
27. V.N.E. Robinson: J. Microscopy, 103, 71 (1975)
28. G.D. Danilatos: Micron and Microscopica Acta, 14, 307 (1983)
29. B.J. Griffin: Scanning, 22, 234 (2000)
30. J.L. Pouchou, F Pichoir: Scanning, 12, 212 (1990)
4 Sputter Depth Profiling
B.V. King
The task for the experimentalist is to deduce the structure of the target from
the measured depth profile. For example, Fig. 4.1a shows a SIMS depth pro-
file of a 16 period SiGe superlattice structure deposited onto a Si substate
(Fig. 4.1b). Analysis of this depth profile requires the establishment of ac-
curate concentration and depth scales as well as an appreciation of whether
the features in the depth profile correspond to the structures in the target
or to artifacts of the sputter profiling process. This chapter will provide a
summary of possible artifacts and their effect on depth resolution.
(a) (b)
- iH}_ lnm Ge
1500
.~
til
C
Q)
C
:;; 1000
Q)
.!!1
(;j
E
za 500
o ~~~~~~~~~~~~~
40 50 60 70 80 90
Apparent Depth (nm)
Z = Y JMt/1000epNAn (4.1 )
4 Sputter Depth Profiling 109
(4.2)
where ~ = Ed Eit and Eit (in keY) and Kit [9] are scaling constants depending
on the atomic numbers and masses of the ion and target atoms
and Uo is the surface binding energy in eV. Uo is usually taken as the sub-
limation energy [10]. The reduced nuclear stopping cross-section, Sn(~)' has
been estimated [11] by
(4.6)
110 B.V. King
for Bi < 80° with n ~ 1. For angles of incidence less than 45°, n ~ 1 and
J ex cos B so that the erosion rate does not vary significantly with angle. The
results of these formulae are accurate to within a factor of two for typical
sputtering energies and species but break down for very low ion energies (as
used in SNMS), for light ions, when the energy transferred is insufficient to
initiate a collision cascade or for high energy heavy ions when interactions
occur between moving atoms in the cascade. They also underestimate the
sputter yield of insulators when energy deposited into electronic excitation
and ionization may lead to atom displacement and sputtering [12]. For keY
energies, the yields for Ar+ bombardment of most materials range from 0.5
to about 2 and are approximately linearly dependent on energy.
Sputter yields, energy deposition and ion ranges may also be calculated
by computer simulations. The best known of these is TRIM, a Monte Carlo
simulation for amorphous targets [13,14] as well as to calculate sputter yields
of rough surfaces [15]
The depth scale across an interface between two pure elements is, to a first
approximation, given by the separate depth scales in the individual layers,
where the interface is determined by the point at which the concentration
falls to 50%. This analysis can, however, be difficult for thin layers or layers
where the concentrations cannot be accurately determined. In Fig. 4.2a the
~ 80
-;!2.
~
z (j)
0 60 I-
Z
~
a::
=>
o
I- 40 ()
Z
W
()
z 20
0
()
5 keV Ar+
/5 keV e- Si
Si Si Si Si Si ~
Ti Ti Ti Ti a 20A
Ti
40A
Ti
120A
Ti
240A
Ti b
Fig. 4.2. Sputter depth profiles of the TijSi multilayer shown using (a) AES and
(b) SIMS
4 Sputter Depth Profiling 111
a Ti
b
Ti
CJ) CJ)
t- t-
Z
::J
Z
::J
o
o() o()
(4.7)
For more detail on many of the above aspects, the reader is directed to an
extensive review of quantitative sputtering [21].
The signal measured in anyone of the surface techniques used with sputter
profiling is related to target elemental concentrations by factors which have
been discussed in other chapters. In summary the counts, ni, measured in
time t for species i are given as
(4.8)
where J is the sputter ion current density (cm- 2 ), Y is the sputter yield,
N is the atomic density (cm -3), X.Ion is the cross-section for secondary ion
production, d is the Auger or XPS information depth, I is the primary elec-
tron or photon flux (S-l) and Xel is the cross-section for Auger electron or
photoelectron production.
In sputter depth profiling, quantification is often difficult since the sen-
sitivity factors, Y, Xion ' d and Xci' generally change with sputter time. For
example, in Fig. 4.2, depth profiling using AES shows four peaks in the Ti+
concentration whereas a SIMS profile shows 5 Ti peaks which do not corre-
spond to dips in the Si+ signal. The two peaked structure from the deepest
Ti layer is characteristic of SIMS profiles of concentrated layers and is due
to the dependence of secondary ion yields on the matrix. In the above case,
the Ti rich layer has been found to contain TiSi 2 by Auger lineshape anal-
ysis. The double peaked structure then results from the difference in yield
of Ti+ sputtered from a TiSi 2 matrix or from a predominately Si matrix.
The presence of peaks in Si+ signal at the position of the shallower Ti layers
4 Sputter Depth Profiling 113
is probably due to oxygen enhancement of the Si+ yield. Oxygen ion bom-
bardment or oxygen flooding during inert gas bombardment would reduce
these SIMS matrix effects by maintaining a relatively constant oxygen con-
centration throughout the depth profile. For example, in Fig. 4.3, the Ti+
profile shows a dip then peak at the Ti-Si interface when profiled by Ar+.
When oxygen flooding is used the Ti+ is shown to vary smoothly across the
deposited film as expected. The Si+ profile also corresponds better to the
known distribution when flooding is used.
Quantification of the concentration is not only related to the matrix ef-
fects and interferences referred to above and in other chapters but is also
determined by factors related to ion sputtering. The most important are ion
beam induced compositional changes, sample contamination and preferential
sputtering. For example, ion beam induced loss or gain of elements at the
surface or in the bulk of the target can markedly affect bulk quantification.
For example, ion irradiation causes loss of 0 from Ta205 to form a mean
composition of Ta20 [22]. On the other hand, elements may be gained from
contamination of the sample from i) impurities in the ion beam, ii) adsorp-
tion from the residual gas in the analysis chamber and iii) redeposition of
previously sputtered material onto the target during profiling.
Implantation into the target of impurities in the ion beam may cause
erroneous depth profiles. These impurities may be ions which pass magnetic
mass separation in the ion beam line, e.g. ArH+ during depth profiling using
Ar ion beams, or neutrals formed in the beam line and transported with the
ion beam. To avoid
(4.12)
at equilibrium. As a result the surface composition of A would change from
the bulk value to the value in (4.12) as sputtering proceeded. The result of
preferential sputtering is that transients occur in measured profiles of surfaces
and interfaces. In Fig. 4.4 YAg > YTa so that Ag is depleted at the surface
with sputtering as Ta is enriched. These transients may however also be
caused by other processes - preferential recoil implantation of light elements
or radiation enhanced surface segregation. Preferential recoil implantation
is seen for alloys with large mass differences between the components, i.e.
mA/mB > 5. The depth over which this effect is evident is of the order of
twice Rp. Surface segregation will be further discussed in a later section.
The above artifacts can be mostly overcome by the use of standards for
calibration of the concentration and depth scales. For SIMS depth profiles
implanted standards are used. If the implant is laterally homogeneous, both
in implant dose and depth distribution agreement to within 10% can be
obtained for analysis of B in Si by different SIMS instruments [24]. Thin film
standards for calibration of depth scales in AES have also been developed,
e.g. by National Physical Laboratory [25] and National Bureau of Standards
[26,27].
104
Cu Cu
4 keV Ag Si
+ Ta 50
Ne
35
103
U)
I-
Z
::J
A B
Z 1·0,--"",,""-~----,
z
°0·84 ------, o
~
II:
:, ~
c::
!2:
w
0·5 :
'
~~
/y \1
/\
" I-
r5
B
I A II c I
:1:; I
o : ~J
5°·16 ------:-- -
o °
a
~'
.L...._ _ --1:""""';""'::'_-1
b °
~ ~::l
SPUTTER DEPTH Z C SPUTTER DEPTH Z
LAYER 1
LAYER 2
LAYER 3
Fig. 4.6. Summary of the mechanisms which can lead to a loss of depth resolution
in sputter profiling. (A) ejection of particles from the surface of a smooth crater; (B)
subsurface ejection due to a larger escape depth; (C) analysis of material sputtered
from the crater walls; (D) redeposition from crater walls into the analyzed crater;
(E) analysis of material sputtered by neutrals in the ion beam; (F) ejection of
particles from a rough crater; (G) atomic mixing within the target
25.----------------------------,
4·5
3
2
Z 15
o
~
310
o(f)
Fig. 4.7. The depth resolution I'1z
~ 5
• • as a function of sputtered depth z
for (e) 1,2,3 and 4.5keV Ar+ ir-
O~---.-----r-----.----._--~
radiated NiCr mulitlayers [32] (_)
o 100 200 300 400 500
2.5, 5.5 and 14.5 keY Cs+ irradi-
SPUTIERED DEPTH (nm) ated GaAs [33] (~) Si0 2 /Si [34]
view of the analysis optics [38]. The field of view then is generally minimised
by the use of lenses in the analysis optics. The region around the crater is
also likely to contain higher oxygen concentrations than inside the crater so
the ionisation probability and hence SIMS yield would increase. Wittmaack
[39] has estimated that to achieve a dynamic range of 106 in depth profiling,
a pressure lower than about 10- 7 Pa is required for profiling with a high
dynamic range in most analysis systems. Figure 4.8 shows an amalgamation
of different experimental results for the influence of instrumental effects on
SIMS depth profiles of B implanted into Si.
1028~------------------------~
/··\:-ATIC BEAM
... RASTER ONLY
~~ ..~
, 0
Fig. 4.8. Combination of var-
P gun (10-4 Pal
ious depth profiles of B im-
\RASTER & GATE & LENS planted into Si showing the
.... decrease in the background
••• MINICHIP
signal caused by (i) ion raster-
ing, electronic gating and ex-
traction lenses [40] (ii) resid-
ual gas pressure [37] and (iii)
0·20 0·40 0·60 0·80 1·00 minichip sample preparation
DEPTH (pm) _ _ [36]
The other beam effect, induced charging, are most important for insulat-
ing samples. Ion beam induced sample charging causes dramatic changes in
the energies of secondary ions and so in apparent yield of secondary or scat-
tered ions and electrons. It can also cause field assisted migration of mobile
species like Na within the sample. Surface charging is usually overcome by
flooding the target with low energy electrons.
One of the other instrumental contributors to a loss in depth resolution,
sample contamination, has been previously discussed. The final factor, instru-
ment geometry and stability, is small for properly designed instruments. For
4 Sputter Depth Profiling 119
example, only when the misalignment between the ion and electron beams in
an AES instrument is greater than 0.8 times the variance of a static gaussian
ion beam does the loss in depth resolution become significant in comparison
to other factors [41].
of the original depth distribution and a function g(z, x) which depends on the
amount of ion mixing. The effect of ion mixing is to smooth the profile where
concentrations change rapidly and to bring the peaks toward the surface with
respect to their original positions (Fig. 4.5c). The redistribution of atoms
during sputter profiling leads to a broadening of a delta function impurity
distribution into a gaussian impurity distribution at a certain depth x (if
sputtering and matrix relocation is ignored) with variance [50]
(J"2 = 2D(x)t = 0.608¢Fd(X)R~/NEd (4.16)
where Fd (x) (e V I A) is the energy deposited into collisional processes at depth
x, N is the atomic density (atoms/A3), (R~/Ec) is a factor which only de-
pends on the target (A 2 leV), ¢ is the ion fl uence (ions I A 2 ) and D (x) is the
diffusion constant (A2 Is) over an irradiation time t at depth x. The diffusion
formalism allows a parameter (Dtl¢Fd) to be used as a measure of mixing
efficiency - the rate of mixing per unit of deposited energy. Experimental
values for (Dtl¢Fd) typically are in the range 1-200A 5 /eV. Low values are
found for targets of high cohesive energy in which the cascade energy density
is low [51]. The lower limit also corresponds to results from analytic theories
of ballistic mixing [50,52]. The depth resolution Llz due to ion mixing may be
found from (Dtl ¢Fd) by solution of a diffusion equation together with asso-
ciated boundary conditions [53]. It is found that Llz depends on the degree
of preferential sputtering as well as (Dt I ¢Fd) but that the depth resolution
is not a sensitive function of (Dt/¢Fd). For the intermixing of Ge and Si,
Kirschner and Etzhorn [54] also estimate Llz to be about 1.5 Rp. As a guide
then, if ion mixing is dominant, Llz is about 1-2 Rp.
Since the interaction between the inert gas and the target atoms is small,
it is reasonable to disregard the influence of inert gas implantation on mea-
sured depth profiles. Chemical reactions between different target elements
4 Sputter Depth Profiling 121
and between implanted ions and target atoms will however alter depth pro-
files by either aiding or negating diffusional mixing. In general impurities with
the high heats of reaction will preferentially segregate. Hues and Williams
[55] in a study of the effect of an O 2 jet on ion mixing, found that impurities,
like Ca, with higher oxygen affinities than Si, segregated to the surface whilst
Ag with lower oxygen affinity than Si were displaced further into the bulk.
C(O, z) = J
C(x, O)g(z ~ x) dx . (4.17)
100
E
c: 10
~
10 100 10
Z, nm Z, Ilm
Fig. 4.9. Typical values of !!.z for different targets as a function of the depth, z,
eroded by keY Ar+ [56]
35
E
.s
z
0
t=
=>
...J 25
0
en
w
II: DEPTH
w (nm)
0
~
II:
CD 50
w 15 ® 113
I-
~ @ 163
@) 226
5
0° 30° 60° 90°
ION BEAM ANGLE
Fig. 4.10. Variation of!!.z with ion beam angle to the target normal after sputtering
to the indicated depths using 4keV Ar+ [32]
4 Sputter Depth Profiling 123
model is however only valid in the low concentration limit and alternatives
exist, Fourier transform and maximum entropy methods, for deconvolution
[59].
4.3 Conclusion
The theoretical developments associated with the study of effects due to the
ion beam probe and also effects due to ion irradiation in general suggest that
a reasonable understanding of the processes is emerging. Ion irradiation with
the deliberate aim of inducing mixing and recoil implantation is likely to
be a very significant, practical form of surface modification technology. As a
consequence, there is a need to develop the available models in greater detail,
for subsequent use in systems of practical importance.
References
1. H. Oechsner: In Thin Film and Depth Profile Analysis, ed. by H. Oechsner,
Topics Curr. Phys. Vo!' 37 (Springer, Berlin, Heidelberg 1984) p.63
2. J. Kirschner, H.W. Etzhorn: App!. Surf. Sci. 3, 251 (1979)
3. M.P. Seah: J. Vac. Sci. Techno!. A 3, 1330 (1979)
4. H.H. Andersen, H.L. Bay: In Sputtering by Particle Bombardment I, ed. by
R. Behrisch, Topics App!. Phys. Vo!' 47 (Springer, Berlin, Heidelberg 1981)
p.145
5. Y. Yamamura, H. Tawara: Atomic Data and Nuclear Data Tables 62,149-253
(1996)
6. G. P. Chambers and.J. Fine: In Practical Surface Analysis, 2nd ed., Vo!' 2, (ed.
D. Briggs and M. P. Seah, Wiley 1992)
7. P. Sigmund: Phys. Rev. 184, 383 (1969)
8. P. Sigmund: Phys. Rev. 187, 768 (1969)
9. P.C. Zalm: J. App!. Phys. 54, 2660 (1983)
10. KA. Gschneidner, Jr.: Solid State Phys. 16, 344 (1964)
11. W.D. Wilson, L.G. Haggmark, J.P. Biersack: Phys. Rev. B 15,2458 (1977)
12. J. Schou: Nuc!. Instr. Meth. B 27, 188 (1987)
13. J.P. Biersack, W. Eckstein: App!. Phys. A34, (1984) 73
14. J.F. Ziegler, J.P. Biersack, U. Littmark: In The Stopping and Range of Ions
in Solids (Pergamon, Oxford 1985)
15. M. Kustner, W. Eckstein, V. Dose, J. Roth: Nuc!. Instrum. Meth. Phys. Res.
B145, 320 (1998)
16. R. Kelly, N.Q. Lam: Radiat. Eff. 19,39 (1973)
17. C. Tian, W. Vandervorst: J. Vac. Sci. Techno!. A 15, 452-459 (1997)
18. K Wittmaack: J. Vac. Sci. Techno!. B. 16, 2776-2785 (1998)
19. C.W. Magee, G.R. Mount, S.P. Smith, B. Herner, H.J. Gossmann: J. Vac. Sci.
Techno!. B16, 3099-3104 (1998).
20. G. Betz, G.K Wehner: Sputtering by Particle Bombardment II, ed. by
R. Behrisch, Topics App!. Phys. Vo!' 52 (Springer, Berlin, Heidelberg 1983)
p.11
124 B.V. King
lillu ·
MASS
SCAN
h\
b
lLL DEPTH
PROFILE
c
IMAGE
SECONDARY
IONS
VACANCIES
IMPlANTED ATOMS
Fig. 5.2. Schematic diagram of the processes which take place after ion impact onto
a surface. SIMS is concerned with analysis of the sputtered ions. The impact of one
primary ion can cause many target atom displacements which affect the surface
seen by subsequent ion impacts
Pel. - +0.00 V
()
III
(.I)
"(.I) 4
+-'
C
:l
o
U
m
.3 2
o so 100
Mass [amu ]
Fig. 5.3. Positive ion mass spectrum for Cu bombarded with 12 keY Ar+. Promi-
nent peaks are seen for Na, K, Ca, and the two isotopes of Cu
130 R.J. MacDonald and B.V. King
8t-------~----~--~----~--~--~----~--~--__t
0 16
+
7 o
AL 27
U X ZR 90
ILl
OAU 197
"-!!'" 5~ _ __
C
::l 4
o
U
01 3
o
..J 2
o lOOO 2000
Tlme [sec]
Fig. 5.4. SIMS depth profile of 0, AI, Zr and Au as a function of primary ion
bombardment time for a thin film sample comprising two layers of mixed Zr, Au
The yield of secondary ions of a given element, ejected from a solid surface,
will depend on a number of parameters. We could write
(5.1 )
a) The Incident Ion Beam Density. This is a function of the ion source
and the experimental measurement to be made. So-called static SIMS requires
a very low primary beam density, perhaps 10 9 particles cm- 2 S-l deposited
over a few mm 2 to give a total dose in any experiment below 10 13 cm- 2 ,
so that each incoming primary ion "sees" an undamaged surface If each ion
impact affects an area of 10 nm 2 then only 10 12 impacts cm -2 will cause
10% of the top monolayer to be damaged. For a particle beam density of
1nAcm- 2 , this damage threshold would be reached after 160s.
In dynamic SIMS, beam densities ofO.1-10mAcm- 2 are typical although
the exact value used depends on the time required to profile the complete
structure. For example, to profile Si to a depth of 1 flm in 20 min requires a
current density of about 1 rnA cm -2. Typically ot, Cs+, or Ar+ beams are
used with beam diameters of 1-10 flm. The best depth resolution however
is obtained when the detected ions all come from the same depth i.e. when
the eroded crater is perfectly flat. Typically this is achieved by scanning the
primary ion beam over areas of about 250 flm x 250 flm on the sample. The
requirement of rapid depth profiling places constraints on the size of the iOll
132 R.J. MacDonald and B.V. King
beam used. For example, if the ion beam used in the above Si depth profile
was scanned in this way, a beam current of 0.6 flA would be required to
achieve the desired current density of 1 mA cm -2. Such a beam current could
only be typically achieved for Ar+, ot or Cs+ ions by using a beam of at
least 50 flm diameter. More details on dynamic SIMS are found in the chapter
on depth profiling.
Imaging SIMS may be performed using an ion microscope or imaging
microprobe. In microscope mode, the spatial distribution of ions emitted
from the surface is retained through the secondary ion transport optics and
measured with about 1 flm resolution by a position sensitive detector or on
a fluorescent screen. In microprobe mode, a well focussed primary ion beam
is rastered over the surface, as in the dynamic SIMS mode discussed above,
and the secondary ion emission from the small bombarded area measured as a
function of the beam position. This mode is inherently inefficient compared to
the microscope mode since only a small part of the surface is being detected
at anyone time.
In most circumstances encountered, the density of the incident ion beam
is such that the collision cascades initiated are not overlapping so that the
sputtered secondary ion signal is linear with incident ion beam density. If,
however, primary beams of molecular ions like ot are used, the molecule
splits upon impact. The energetic atoms produced then initiate collision cas-
cades in the target which can overlap in space and time. This will lead to
nonlinear relationships between the incident ion beam density and the yield
of secondary ions.
U = JASA (5.2)
NA '
where J A is the ion current density and N A is the elemental atomic density.
The sputter yield of a target depends on the ion energy and species as well
as the target atomic number, surface crystallinity and topography. However,
theoretical sputter yields calculated for smooth amorphous targets generally
give good agreement with values found from experimental depth profiles of
single element targets. Therefore theoretical sputter yields can be profitably
used to calculate erosion rates in dynamic SIMS measurements. This then
allows the SIMS depth profile which is given in terms of sputter time (Fig. 5.4)
to be recast in terms of sputtered depth.
A theory of the sputtering of atoms from the smooth surface of an amor-
phous solid consisting of a single atomic species has been well developed
by Sigmund [7]. In this case, the sputter yield of atoms, S, from a target
bombarded by ions of mass Ml and energy E is proportional to the nuclear
5 SIMS - Secondary Ion Mass Spectrometry 133
stopping power, Fd(E) (i.e. the energy spent in elastic collisions between
atoms.)
VACUUM LEVEL
Fig. 5.5. Schematic energy diagram of an atom leaving a metal surface. EF is the
Fermi level and ¢ is the work function. Initially the atomic level Ei is broad and
may lie above EF before the atom is sputtered. The variation of the image potential
causes Ea to lower with separation until Ea = EF at the crossing point. Electrons in
the metal can tunnel out to fill the atomic level once Ea < EF beyond the crossing
point
5 SIMS - Secondary Ion Mass Spectrometry 135
-q;t--
...
--1
(5.5)
(5.6)
where d(zc) is the width of the ionisation level at the distance Zc at which
the level Ea crosses E F . V.L is the ejected atom or ion velocity component
136 R.J. MacDonald and B.V. King
of
1
V(r)
2~ d
C
e
r::F+o excited
states
4• a
M+ +0-
b r
IM+O
Ground slale
Fig. 5.7. Schematic diagram of the M-O system during its dissociation with in-
creasing distance, r, from the surface. In this model of ionisation of sputtered atoms,
energy is transferred from the collision cascade to a quasi-molecule comprising a
substrate metal atom with an adsorbed oxygen atom (inset 1). As the metal and
oxygen atoms move together (inset 2) the system moves to the left up the repulsive
side of the potential energy curve. If the energy imparted in the collision is great
enough the system may cross onto curves c, d or e at points C3, C4 or C5 respectively,
and excited metal atoms or M+ + 0 will be sputtered. For the more probable low
energy collisions, the system will return to the right along curves a or b and so
metal and oxygen ions or atoms will be sputtered (inset 4)
logical suggestion is that, for a surface coverage which is small so that there
is essentially M+O- bonding, the number of M+ ions ought to equal the
number of 0- ions.
It is obvious from the above that the general picture of secondary ion for-
mation is known, but that an accurate analytical model allowing calculation
of a cross-section for ionisation in (5.1) is not available. This remains the
major barrier to quantification of SIMS.
(5.8)
The neutralisation parameter will favour slower ions, hence it is likely that the
energy spectrum will have its peak shifted to higher energies relative to the
138 R.J. MacDonald and B.V. King
The main elements of a SIMS analyser are (i) a primary ion source (ii) optics
to transfer these ions to the target surface and to collect secondary ions from
the surface and direct them to (iii) a mass spectrometer.
All SIMS machines must have a source of ions available to initiate the
sputtering process. In general, the ions will have energy in the range 1-
20keV, but other characteristics of the incident beam will depend on the
type of application. Many add-on SIMS analysers will use a reasonably simple
ion beam, with a beam diameter of order 0.1-1 mm and with the facility to
raster the beam across larger areas. Dedicated SIMS analysers on the other
hand often have sophisticated ion beam systems, particularly if the analyser
is likely to be used in microscope mode.
All ion guns comprise an ion source and a method of transferring the
ions to the sample. This is normally done by floating the ion source at the
accelerating voltage and leaving the target at ground potential. In some in-
struments, e.g. the Cameca IMS6F, the target may also be floated at a high
potential. There are four common types of ion sources used, namely electron
bombardment ion sources with and without plasma formation (duoplasma-
tron, hollow cathode), surface ionisation sources and field ionisation sources
(liquid metal ion sources).
In duoplasmatron sources, ions are extracted in a strong external elec-
trostatic field from a high density plasma which is magnetically confined
5 SIMS - Secondary Ion Mass Spectrometry 139
len~
quadrupole
Fig. 5.B. Schematic diagram of a PHI duoplasmatron source and primary ion optics
between an anode and intermediate electrode. These sources are bright (10 6 -
10 7 A m- 2 sr- I ) with an energy spread of about 10 eV, and so allow current
densities of about 10 rnA cm -2 into beams of diameter 1-50 !lm. Both inert
and active gas beams can be produced by this source. Figure 5.8 shows a
schematic diagram of the PHI duoplasmatron source and ion transport op-
tics.
In a surface ionisation source, alkali metals like caesium are varporised
from a porous metal surface which has a high work function. When the tem-
perature of the surface is high so that the caesium surface concentration is
low, then the caesium is almost all desorbed as ions. The brightness of the
source is about 106 Am- 2 sr- 1 but the source has a very low energy spread
« 1 e V) and so current densities and spot sizes comparable to duoplasma-
trons can be achieved.
In liquid metal ion sources a low melting point metal (Ga, In, Cs) is
supplied as a liquid over a metal tip which is subject to a high electric field.
Electrons tunnel from surface atoms to the tip, leaving the metal ions to
be extracted. These sources have high brightness (1010 Am- 2 sr- I ) [14] and
allow nA currents to be delivered into spots of diameter less than 50 nm or
pA to be delivered into a minimum spot size of 10 nm. Figure 5.9 shows a
schematic diagram of the layout of the FEI single lens liquid metal ion source.
Duoplasmatron sources are mainly used for production of ot
and Ar+ for
depth profiling and mass scans. Surface ionisation sources allow depth profil-
ing using Cs+ with the attendant advantages of increased negative ion yields
and decreased matrix effects in positive ion depth profiling. Liquid metal ion
sources are exclusively used for imaging SIMS since currents are normally too
small for rapid depth profiling. However, for high spatial resolution imaging,
140 R.J. MacDonald and B.V. King
the gun must be matched with a high transmission spectrometer since, for
example, a volume of 20 nm x 20 nm x 5 nm contains only 50000 atoms. For a
dilute impurity, e.g. at a concentration of 20 ppm, there will only be one atom
in the volume. The impurity can then only be detected if the spectrometer
transmission is high.
Primary ion optics, for transferring ions onto the sample consists of usu-
ally two electrostatic lenses for beam focussing, xy deflection for beam po-
sitioning, rastering and neutral rejection, a stigmator for focus correction
independently in x and y directions when small spot sizes are used, and a
mass filter to remove beam contamination, either multiply charged ions, clus-
ters or impurity ions. The most commonly used mass filters are a Wien filter
or a magnetic sector. In a Wien filter crossed magnetic and electric fields pro-
duce equal and opposite forces on ions of a certain M / Z ratio which then pass
through the filter undeflected. Ions of other M / Z ratios do not pass through
the filter. The electric fields may be shaped by shims to produce additional
focussing. In a magnetic sector, ions are bent in a uniform magnetic field.
Ions of energy e V and a specific M / Z ratio are bent in the horizontal plane
with a radius of curvature
R =
J2V OM
eB2 Z'
(5.9)
5 SIMS - Secondary Ion Mass Spectrometry 141
which, if it matches the radius of curvature of the instrument, allows the ions
to be transmitted through the magnet. Focussing in the vertical plane can
be achieved by rotating the magnetic boundaries with respect to the particle
trajectory.
The dynamic range and depth resolution achievable in depth profiling
is dependent on the crater flatness and the background signal reaching the
detector. Flat craters are achieved by scanning the ion beam typically over
five beam diameters. Electronic gating of the detected ions then ensures that
only the ions emitted from the centre of the crater are counted. For this
procedure to work correctly, neutrals arriving with the ion beam must be
eliminated since they will generate secondary ions independent of any raster
gating. A bend of at least 1 0 followed by apertures must be included in the
section of the ion optical path to reject neutrals from the beam striking the
sample.
SIMS analyses of insulating samples, common in static SIMS, creates
problems due to surface charging. The injection of positive ions and ejection
of secondary electrons (with an efficiency of almost unity at SIMS energies)
can cause surface potentials to vary with time to a few hundred volts which
can completely stop secondary negative ion emission and alter the energy
spectrum of emitted positive ions making optimisation of ion optics difficult.
Charging may be reduced by using thin samples, low ion current densities,
electron flooding or fast atom bombardment. Electrons with energies of a
few e V to a few hundred e V can be generated from just a filament or a
simple gun. If the electron current is matched to the ion current, then the
surface charging can be reduced. This process is self-stabilising since a positive
surface potential will reduce the loss of secondary electrons which will in turn
decrease the surface potential.
In the application of SIMS to analysis, the information on surface compo-
sition is contained in a range of secondary ions of variable mass, distributed
over a broad angle and energy range. In order for the equipment to have a
reasonable mass resolution, the energy range of particles accepted into the
mass analysis section of the spectrometer must be narrow. This energy range
is given by I1E which is the spread in the ion energy E seen by the mass
spectrometer. The energy E is equal to qV + Eo, where q and Eo are the
charge and energy, respectively, of the secondary ion and V the extraction
potential. A spread in E then results from the spread in the initial energy of
the secondary ions and from any drift in V. This energy resolution must be
smaller if a larger mass resolution M / 11M is desired. A SIMS mass spectrome-
ter must therefore consist of an energy analyser preceding the mass analyser.
The secondary ions may be extracted into this energy analyser by a high
(> 1 kV mm- 1 ), or low « 10 V mm- 1 ) electric field. Higher transmission re-
sults from the use of a high extraction field since more ions are collected into
the analyser and the relative energy spread of the secondary ions (I1E / qV)
is less. Typical electrostatic energy analysers have high transmission over a
142 R.J. MacDonald and B.V. King
a c
tbr·..·
0.3
2~~ . ,,
unstable
a \
0.2
0.1
Fig. 5.11. Schematic diagram of the ion optics of a Cameca IMS-4F ion microscope.
The electrostatic sector selects the ion energies passed to the double focussing mag-
netic sector. The ions may then be displayed on a fluorescent screen in microscope
mode or collected in a Faraday cup or electron multiplier for quantitative static or
dynamic analysis. The elements of the analyser are (1,2) - ion sources, (3) - pri-
mary beam mass filter, (4) - immersion lens, (5) - specimen, (6,17,20) - deflectors,
(7,11,15) - lenses, (8,10,14) - slits, (9) - electrostatic sector, (12) - spectrometer,
(13) - electromagnet, (16) - projection display and detection, (18) - channel plate,
(19) - fluorescent screen, (21) - Faraday cup, (22) - electron multiplier
2000 are typical for rf quadrupoles. They are useful for adding SIMS analysis
onto other analysis systems or where space is critical. Further discussion on
rf quadrupole systems is given by Dawson [18].
Magnetic sector spectrometers send ions through a perpendicular mag-
netic field. The radius of curvature of the ions then depends on (M / Z) 1/2 as
found previously. The mass resolution of the magnetic sector is proportional
to the radius of curvature of the path but is sensitive to angular and energy
spread in the incoming secondary ions. The loss in resolution may be alle-
viated by energy analysis in an electrostatic sector. The transmission of the
instruments is high (> 10%) and the mass resolution is high (M / /',.M > 10 4 )
over a large mass range. The high mass resolution is important for many
types of analysis. For example, a typical problem in semiconductors is to an-
alyze P in Si. The mass of p 31 is 30.9859 amu whereas the mass of Si 3 0H is
30.9736 amu. If the analyser has a mass resolution M/ /',.M less than 4000, the
two peaks will overlap and unambiguous determination of low concentrations
of P in Si will be made impossible.
Figure 5.11 shows a schematic diagram of a common magnetic sector in-
strument, the CAMECA IMS4F system. This is a SIMS microprobe which
has been available for a number of years and has progressed through several
generations of different analysers. It is a microscope system capable of pro-
144 R.J. MacDonald and B.V. King
ducing images of about 1 flm spatial resolution. It has a high mass resolution
(up to 10 4 in spectrometer mode) which is achieved by the use of a high
extraction potential (about 5 kV), secondary ion energy and mass selection
in 90° electrostatic and magnetic sectors. The combination of transfer lenses
and sectors forms a stigmatic image of up to a 400 flm x 400 flm area of the
target onto the fluorescent screen. Alternatively, the secondary ions can be
directed into an electron multiplier for static and dynamic SIMS analysis.
Time of flight instruments rely on the measurement of the drift time of
secondary ions along a flight tube. If the energy of the ions is known, eVa,
by accelerating them through a high potential difference, then the mass to
charge ratio of the ion is related to the time taken to drift a distance d by
(5.10)
Typical flight times are 10 flS-l ms depending on the ion mass and the flight
path. The secondary ions must, however, all start at the same time. This
requires that the formation of the secondary ions must occur in short pulses,
typically sub-nanosecond in duration for mass resolutions up to 10000. The
time between pulses is set by the maximum flight time plus the data pro-
cessing time. Typically the primary ion beam is pulsed by rapid deflection
past slits in the primary ion column. Alternatively, secondary ion formation
above the surface from sputtered neutrals could be pulsed, for example, by
laser ionisation. The transmission of time of flight instruments is very high
( > 50%) but is dependent on the energy spread of the secondary ions. This
may be circumvented by incorporating an energy compensation section or
an energy filtering stage in the flight tube so that higher energy ions travel
further than low energy ions of the same mass to charge ratio and ions of
different energy arrive at the detector simultaneously [19]. For example, a
270° spherical sector energy filter has been used in the TFS surface analyser
from Charles Evans & Associates (Fig. 5.12).
A detector system will then follow the mass filter, with the type of detector
depending on the likely signal strength, i.e. it will consist of an ion counting
system for low count rates or a low current amplifier if the count rate is above
about 10 7 counts per second. Postacceleration of the secondary ions into the
detector may also be used to improve detection limits.
There have been major advances in instrument design in the last few years,
embodied in the new machines from the major manufacturers (e.g. ADEPT-
1010 from PHI, IMS 6f, IMS1270 and NANOSIMS 50 from Cameca, Ion-TOF
from the group of Prof. Benninghoven at the University of Muenster, SIMS
4500 from Atomika and Ionoptikas FLIG Floating Low Energy Ion Guns).
The trends in instrumentation are to reduce the primary ion energies to
sub ke V for depth profiling, and to have increasing automation, especially
for wafer-sized samples, in sample analysis and data reduction.
5 SIMS - Secondary Ion Mass Spectrometry 145
,
Cross Sectional View of the
TFS Surface Analyser
\-
0.'
..~ TRIFfTM Spectrometer
Fig. 5.12. Schematic diagram of a direct imaging time-of-flight SIMS analyser. The
270 0 analyser filters the energy of the secondary ions, allowing a mass resolution
of more than 3000. The analyser can be used in microscope mode, with a lateral
resolution of 1--2 ~m or in microprobe mode (using a liquid metal ion source) with
a 0.1 ~m resolution
It was indicated in Sect. 5.1.3d that the magnitude of the SIMS signal for a
given ion was significantly affected by the presence of active gas adsorbates
on the surface of the target. This is because the cross-section for ionisation
is very dependent on the chemical state of the surface. The signal enhance-
ment factor due to the presence of an active gas layer can be more than
100 times that of the ion yield from a clean surface. For example, the yield
of positive secondary ions under 8 ke V Ar+ bombardment [20] increases by
up to three orders of magnitude when the target surface is saturated with
oxygen (Fig. 5.13). This change is not uniform even between cluster and mul-
tiply charged ions from the same element since, as the oxygen coverage of
a silicon surface is changed from zero to a saturation coverage, the yield of
Si+ increases by 2.3 orders of magnitude, but SiO H , Si 2+ and Sit remain
the same or decrease [21] (Fig. 5.14). Negative ion yields can be similarly in-
creased by introducing an electropositive element like Cs+ onto the surface,
146 R.J. MacDonald and B.V. King
y+ Cr
Mg
1
Si Ti
Mo
Fe
Sr
-1
10
Ni
Ba w
Ge
-2
10
AI 1
-3
10
V
Mn Nb
Cu
-4
10
Ta
i) the enhanced yield is still a widely varying number from element to ele-
ment as shown in Fig. 5.15. Systematic studies of the secondary ion yield from
a wide range of elements have not been undertaken, in part because of the
sensitivity of such yields to experimental parameters such as the cleanliness
of the surface, the background gas concentration, the transmission character-
istics of the SIMS analyser, etc.
ii) when an active gas is used to enhance the ion yield and hence the sensitiv-
ity, the measured ion yield is a function of the active gas concentration at the
surface. This concentration of active gas may be established by irradiating
the surface in a background environment of the active gas or alternatively,
the incident ion beam may consist of ions of the active gas itself.
The use of primary Cs+ beams and the subsequent detection of MCs+
species or, for elements M with high electron affinity, MesH [23] has pro-
vided a powerful alternative to ot
beams for reduction of variations of the
5 SIMS - Secondary Ion Mass Spectrometry 147
1E+08
1E+07
(i)
Q.
.£
~ lE+06
·in
<:
III
.E lE+05
<:
.2
e:-
m
."
c: lE+04
0
U
III
(J)
lE+03
1E+02
2 3 4 5
Fig. 5.14. Ion yields of silicon secondary ions for clean Si, oxygen-saturated silicon
or silicon dioxide surfaces
The energy of the secondary ions which generate the SIMS signal is important
for SIMS analysis in two ways. First, it allows molecular mass interferences to
be overcome by tuning the acceptance energy in quadrupole systems. Second,
it may be used to overcome matrix effects in the so-called infinite velocity
method [27].
Many SIMS experiments do not require mass resolution equivalent to iso-
tope separation, but it is often necessary to detect the signals due to atomic
and molecular ions. This can be done on the basis of the energy spectrum
of the secondary ions. The energy spectrum of molecular ions, e.g. M;:t or
Mn ot,
maximise yield at a lower energy than that of the atomic species
148 R.J. MacDonald and B.V. King
a lE+08 r--------------------------------------.
lE+07
i!:" lE+06
'iii
c:
2
-=c: lE+05
.2
2:-
CII
U
c: lE+04
o
u
(I)
CI)
lE+03
lE+02+-~--_:~----~~~~==~~~--~
o 50 100 150 200 250
Ion Mass
b lE+07~----------------------------------_.
lE+06
i!:"
.~ lE+05
(I)
£
c:
.Q
2:- 1E+04
CII
u
c:
o
u
(I)
CI)
lE+03
lE+02t-----~~----~~----_,~----~~-----!
~ 20
Ion Mass
Fig. 5.15. (a) Positive ion yields for the indicated targets bombarded with
13.5keVO- at normal incidence. (b) Negative ion yields for the indicated targets
bombarded with 16.5 keY Cs+ at normal incidence
5 SIMS - Secondary Ion Mass Spectrometry 149
"As + Implant
InSI
5 J( 1O l '/em'
7Q~Y
10'
'iii'
~u.,.
•
~
.i'
i
:6'"
j
10'
(5.11)
where IE and IR are the secondary ion intensities for element E and reference
(or matrix) element Rand C E and C R are their respective concentrations.
For trace element analysis, C E is small so C R doesn't change with a change
in CEo In this case, we rename the reference (R) as the matrix (M) and can
combine the matrix concentration with the elemental RSF to give
(5.12)
Then
CE=RSF~ . (5.13)
EM
RSF values for positive impurity ions [3] from oxygen bombardment of silicon
range from less than 10 21 atoms cm- 3 for Na to more than 10 24 atoms cm- 3
for common dopants like P. For Cs bombardment, negative ion RSFs range
from less than 10 22 atoms cm- 3 for S to more than 10 27 atoms cm- 3 for Be.
Tables of RSFs can be found in [3]. RSFs measured on the same machine
vary by < 50% over time. Differences between different instruments are of
the same magnitude, at least for GaAs [5.30], and are primarily ascribed to
different primary ion impact angles.
Static SIMS is ideally used to obtain the chemical composition of the sur-
face without damaging the surface. A common SIMS problem is to identify
differences in trace element concentrations between two targets, e.g. from
contaminated and uncontaminated samples. In this case absolute concentra-
tion determinations are not required, thus avoiding SIMS ion yield uncer-
tainties. In the following example, an epitaxial CdTe film had been grown
on a CdTe(111) substrate by molecular beam epitaxy. The film was a sin-
gle crystal, but measurement of optical properties indicated the presence of
impurities in the film. Figures 5.17a, b shows SIMS mass spectra of a CdTe
substrate and the CdTe film, respectively, which were taken on a Riber MIQ
156 RF quadrupole-based analyser. The level of oxygen and hydrocarbons
(masses 12-16) in the CdTe substrate is much lower than in the CdTe film.
5 SIMS - Secondary Ion Mass Spectrometry 151
6
a
5
U
Q)
til
"
"-til 3
....,
c:
:J 2
0
U
01
0
-l
0
o SO 100 1'50
Mass [amu]
7t-----------------~------~--------~------------------4
b
6
U
III ~
til
"....,
til 3
c:
::l 2
o
U
en
o 0
-l
o 50 100 150
Mass [amu]
Fig. 5.17. (a) Positive ion spectrum from a clean CdTe substrate. (b) Positive ion
spectrum for a MBE-grown CdTe thin film
152 R.J. MacDonald and B.V. King
6000
5000
4000
3000
2000
II
1000
o 160 180
III I:
Fig. 5.18. SIMS mass spectra from a PET surface. The protonated monomer has a
mass of 193 Daltons. The many lower mass peaks are due to contaminant molecules
and molecular fragments.
FRACTURE
I I PLANE
200 fl m Fig. 5.19. SIMS composition
map of an adhesive bond
fracture on 2024 aluminium
,,/ " ~ treated with silane, showing
Adhesive o the locus of fracture based on
Q.
4) the distribution of Si and AI.
I The dark grey areas in the
X· Si· O· Si· map contain no Al or Si and so
I I correspond to adhesive at the
o 0 surface. The white areas con-
I I tain both Al and Si so corre-
Adhesive Bond Details spond to the surface fractur-
2024 Aluminium Alloy ing at the silane-oxide inter-
abraded + silane coupling agent
bonded with FM- 73 epoxy adhesive face
5 SIMS - Secondary Ion Mass Spectrometry 153
There is also increased Fe (mass 56) in the deposited film. These are probably
the impurities responsible for the altered optical properties.
A more complicated static SIMS spectrum, from a polymer surface, is
shown in Fig. 5.18 [32]. In this case many of the peaks may be due to fragment
ions. To make sense of specific spectra may well require an understanding
of molecular fragmentation and attachment processes. However, libraries of
static SIMS spectra [33] will give a first guide to the interpretation of results.
Principal factor analysis has also proved useful for static SIMS analysis [34].
References
23. Y. Gao, J.W. Erickson and R.A. Hockett: In Secondary Ion Mass Spectrometry
X eds A. Benninghoven, B. Hagenhoff, H.W. Werner (John Wiley, Chichester
1997) p.339
24. K. Wittmaack: In Secondary Ion Mass Spectrometry X eds A. Benninghoven,
B. Hagenhoff, H.W. Werner (John Wiley, Chichester 1997) p.39
25. H. Gnaser: J. Vac. Sci. Techno!. A 12, 452 (1994)
26. Y. Homma, Y. Higashi, T. Maruo, C. Maekawa, S. Ochiai: In Secondary Ion
Mass Spectrometry IX eds A. Benninghoven, B. Hagenhoff, H.W. Werner
(John Wiley, Chichester 1995) p.398
27. Secondary Ion Mass Spectrometry X eds A. Benninghoven, B. Hagenhoff,
H.W. Werner (John Wiley, Chichester 1997) p.131
28. Reference [3] page 1.8.5
29. http://ois.nist.gov /srmcatalog/
30. http://www.npl.co.uk/npl/cmmt/sis/refmat.html#Tantalum
31. Secondary Ion Mass Spectrometry X eds A. Benninghoven, B. Hagenhoff,
H.W. Werner (John Wiley, Chichester 1997) p.135
32. S. Bryan, F. Reich, B.W. Schueler, G. Marsh: In Secondary Ion Mass Spec-
trometry X eds A. Benninghoven, B. Hagenhoff, H.W. Werner (John Wiley,
Chichester 1997) p.939
33. http://www.surfaceSpectra.com/
34. Secondary Ion Mass Spectrometry X eds A. Benninghoven, B. Hagenhoff,
H.W. Werner (John Wiley, Chichester 1997) pp.313, 887
6 Auger Electron Spectroscopy
and Microscopy - Techniques and Applications
P.C. Dastoor
6.1 Introduction
Auger electron spectroscopy (AES) is one of the most commonly used surface
analytical techniques available to the materials scientist. It has the ability
to measure the chemical composition of the first few monolayers of a given
surface with a sensitivity of the order of 0.1 atomic % and a spatial resolution
of the order of 10 nm [1]. Its ease of interpretation means that AES is often
the analysis technique of choice and has been used to study a wide range of
different materials.
This chapter outlines the basic principles of Auger electron spectroscopy,
the equipment used and the advantages and disadvantages of the technique.
The various methods of performing an Auger analysis are described, with a
view to providing the reader with expert insight into how AES may used to
tackle a variety of surface materials analysis problems. The diverse capability
of Auger electron spectroscopy is illustrated by a number of examples of its
application in a wide range of fields. Finally, the possible future trends in
the development of the technique are discussed, foreshadowing the future
capabilities of the technique.
6.2 Fundamentals
The Auger effect was first independently discovered by Meitner [2] and
Auger [3] who found that if an atom is ionised, the resulting electronic re-
organisation can cause the ejection of an electron with an energy that is
characteristic of the originating atom. The Auger process is essentially a
three-electron process that is typically initiated by the ejection of a core
electron by an incident high-energy electron. The core hole thus created is
rapidly filled by an electron from a higher energy shell leaving the atom in
an energetically excited state (Fig. 6.3). The excited atom can lose energy in
one of two ways. Firstly, an X-ray photon of the appropriate energy can be
ejected. The measurement of the energy of the energy of this X-ray, which is
characteristic of the originating atom, forms the basis of the analytical tech-
nique known as X-ray fluorescence (XRF). The second way that the excited
atom may lose its excess energy is through the ejection of another electron
156 P.C. Dastoor
with a certain kinetic energy. The energy of this so-called Auger electron is
also characteristic of the source atom and is the measurement of the energy
distribution of these Auger electrons that forms the basis of Auger electron
spectroscopy. This relaxation involves both the electrons in the atom itself
[5] and the surrounding material [6].
So, what is it that determines whether an atom de-excites through the
emission of an X-ray photon or through the ejection of an Auger electron?
It turns out that the probability, or cross-section, of these two processes is
highly dependent upon the atomic number (Z) of the excited atom. For low
Z elements (Z < 40), the de-excitation is dominated by the Auger process
and thus for many of the more abundant elements a strong Auger signal is
observed.
If the Auger electron energy (EAuger) is written in terms of experimental
photoelectron energies (referenced to the Fermi level) then the energy can be
written as:
(6.1)
where Ex, Ey and Ez are the binding energies of the three participating
electrons and Ueff is the extra energy needed to remove an electron from a
doubly ionised atom [7].
The characteristic energy distribution of electrons emitted from a solid
surface upon excitation by a primary electron beam is shown in Fig. 6.2.
This spectrum can be divided into main three regions. The large peak at low
energies (~ 50 eV) in region 1 is due to the so-called "true secondary cascade" .
In region 2 the presence of Auger peaks can be observed superimposed upon a
background of red iffused primary and backscattered electrons. In region 3, the
elastically scattered electrons can be observed in the form of the elastic peak.
Indeed, it is these electrons that are used in low energy electron diffraction
6 Auger Electron Spectroscopy and Microscopy 157
Primary Electrons
'True' Secondary Electrons
N(E)
Loss Peaks
__ - - -Rediffused Primaries
Electron Energy
Fig. 6.2. Schematic diagram of the characteristic secondary electron spectrum from
a solid surface. The peak at high energy corresponds to the elastically scattered elec-
trons and thus occurs at the energy of the primary beam. Electron energy loss peaks
are observed on the low energy tail of the elastic peak and correspond to electrons
that have undergone inelastic interactions with the material (such as generating
surface and bulk plasmons). The peak at low energy corresponds to the so-called
"true secondaries", which have undergone multiple inelastic collisions within the
solid surface, and is typically located at an energy of 50-100eV. The Auger peaks
(shown exaggerated in the figure) are superimposed upon the long exponential tail
from the low energy peak and these electrons initiate further contributions to the
background
e-beam
Ar+beam a
Fig. 6.3. Schematic diagram of the common electron analysers used for Auger
electron spectroscopy. (a) A single pass cylindrical mirror analyser (CMA). (b) A
concentric hemispherical analyser (CRA)
158 P.C. Dastoor
(LEED) studies. On the low energy side of this peak there are a number of
features associated with electron loss processes in the solid, including losses
due to both bulk and surface plasmons. These electrons are the ones that are
probed during electron energy loss spectroscopy (EELS).
The surface sensitivity of the technique arises from the energy range of the
Auger electrons that are ejected. When an energetic electron passes though
a solid, it continually has inelastic collisions with the electrons in the solid.
These collisions ultimately result in the electron becoming incorporated in
the electronic structure of the solid itself. The number of collisions that occur,
and hence the range of the electron in the solid, is known as the inelastic mean
free path (imfp) and is a strong function of the electron's kinetic energy. Early
studies suggested that the imfp of most materials lay on a universal curve [4].
Indeed, for the usual kinetic energies encountered in Auger spectroscopy (20-
2000 eV), the imfp of the Auger electrons is very low; typically between 5 and
20 A. Therefore, only electrons that originate within the first 5-20 A( or 3 to
10 atomic layers) of the surface will be ejected from the surface and hence
detected. However, as will be discussed later in this chapter, recent work has
shown the concept of a "universal curve" for imfp is inappropriate [8].
6.3 Instrumentation
Although Auger spectra can be stimulated using incident x-rays or ion beams,
most dedicated Auger systems employ an electron beam to irradiate the sur~
face. Historically, the incident electrons typically have a kinetic energy of
either 3 ke V or 5 ke V, although increasingly beam energies of 10 ke V are be-
ing employed. The choice of beam energy actually influences the yield of
secondary electrons emitted from a solid surface. Typically, the yield of sec-
ondary electrons of a given energy, E, is maximised when the beam energy
is 3E. The use of an electron beam provides a number of further advantages
when it comes to using Auger spectroscopy to provide chemical images of sur-
faces. In this application, the incident electron beam can be readily focussed
to produce a small spot size upon the surface and subsequently rastered over
the surface. Many of the early Auger systems used electron sources that were
based around a thermionic electron gun. Although relatively low cost, the in-
herent electron-optical aberrations associated with these devices meant that
they were only capable of minimum beam spot sizes of greater than 0.2 !tm.
For applications requiring a high brightness, small spot size electron emitter,
single crystal sources (such as lanthanum hexaboride, LaB 6 ) or field emission
sources are used. The spot size of the electron beam governs the ultimate
resolution of the Auger imaging system and current systems are capable of
spot sizes of the order of 10nm [9].
The second main component in any Auger system is the electron en-
ergy analyser that allows the energy distribution of secondary electrons to
be collected. There are three main designs of electron energy analyser used
6 Auger Electron Spectroscopy and Microscopy 159
6.4 Quantification
The functional form of the true secondary electron cascade (region 1 in
Fig. 6.2) is important for quantitative analysis in Auger spectroscopy since
most of the Auger peaks are superimposed upon this background. This slowly
decaying electron signal arises from inelastically scattered Auger electrons,
and backscattered primary electrons, that in turn generate secondary elec-
trons in the solid. Using a diffusion-based argument, Sickafus showed that
this cascade of electrons has the general form:
where N(E) is the energy distribution of electrons and A and m are constants
related to the material properties of the solid [13-15]. This exponential form
for the background has been shown to be suitable for many materials in
the region E < ~ Eprimary (where the contribution of losses from primary
electrons can be neglected). In an improvement to this approach, Matthew
160 P.C. Dastoor
et al. suggested that a more appropriate form for the background function
was:
N(E) = A(E + J)-m (6.3)
where J is the mean ionisation energy of the surface atoms [16].
One of the major requirements of any surface analytical technique is the
quantitative determination of the chemical composition of a surface. Indeed,
the application of AES to quantitative surface analysis has a long history, with
a number ofreviews published over the last 20 years [17-19]. As discussed by
Seah [19], there are three main approaches to quantitative analysis:
1. Calculate the relevant terms from first principles. Unfortunately, this ap-
proach is not practical since the origin of the secondary electron spectrum
is still not sufficiently well understood.
2. Calculate the surface concentrations from published databases. While
more practical, this approach does not take into account the effect of the
measuring instrument.
3. Calibrate the concentration using locally produced standards and data-
bases. This is the most desirable approach since it compensates for the
transmission function of each individual apparatus.
It should be noted that there are a number of different approaches that can
be used for quantifying Auger spectra in different circumstances, such as di-
lute alloys, thin films or multi-component systems [20]. The most common
method for quantification of surface concentration involves the use of sensi-
tivity factors from pure elemental standards [21,22]. These sensitivity factors
will differ from one apparatus to another and thus each operator needs to
generate their own set of standard spectra. Using bulk elemental standards,
the surface concentration, Gi , of element i, is given by:
G- Ii/If' (6.4)
"- Lj FijIj/If'
where j is the number of elements on the surface, Ii is the Auger signal of
element i, is the intensity of the pure element standard and Fij is the matrix
correction factor. The matrix correction factor includes all of the effects as-
sociated with the interaction of the primary beam with the target, such as;
changes in backscattering [23], atomic density and attenuation length between
the pure elemental standard and the unknown sample, microtopography [24]
and primary electron diffraction [25,26].
6.5 Techniques
Generally speaking there are four main modes of AES operation: point analy-
sis, line scan, profiling and scanning. The four different modes originate from
the typical surface regions that are analysed.
6 Auger Electron Spectroscopy and Microscopy 161
In this mode, the incident electron beam spot remains stationary on the
surface and the Auger electron spectrum is measured. Figure 6.4 shows the
direct form of a typical survey spectrum for a copper surface, with the num-
ber of detected electrons N(E)/E plotted as a function of electron kinetic
energy. The characteristic copper peaks between 700 and 1000 eV are clearly
observed, together with peaks associated with surface contamination. Figure
6.4 also shows the differentiated form of the same survey scan, illustrating
how this method tends to emphasise the fine structure in the direct spec-
trum [19]. The point analysis mode is the most common analysis mode for
Auger electron spectroscopy since it allows for the rapid determination of the
elemental composition of a small region of the material surface. Although
Cu
'-v-'
Cu
o 200 400 600 800 1000
Electron energy, eV
Fig. 6.4. Auger surface obtained from a contaminated copper surface. (a) The
direct spectrum EN(E). (b) The differential spectrum, d'k{EN(E)}, obtained by
sinusoidal energy modulation
162 P.C. Dastoor
Beam
diameter
Electron beam
Sample surface
Backscattered electrons
X- rays
Fig. 6.5. Schematic diagram showing the interaction volume produced by an in-
cident high-energy electron beam on a solid surface. The incident electron beam
induces a pear-shaped interaction volume and generates both electrons and X-rays
to be ejected from the surface. The short inelastic mean free path of the relatively
low energy Auger electrons means that AES information is obtained only from the
uppermost atomic layers. As a consequence the lateral resolution of AES is close to
the diameter of the primary beam. The secondary electron escape depth is larger
than that of the Auger electrons whereas the backscattered electrons , whose energy
is close to that of the primary beam , escape from depths of 100- 1000 nm. The X-
ray escape depth and lateral resolution correspond to the dimensions of the entire
interaction volume
the interaction volume of the high energy incident electrons is of the order
of a micron, the short imfp of the escaping Auger electrons means that the
lateral extent of Auger emission is determined by the minimum spot size of
the electron gun (Fig. 6.5). For modern electron guns, the size of this spot
(and hence the lateral resolution of the technique) is typically of the order of
lOnm.
In this mode the electron beam is rastered laterally over the surface in one
dimension while simultaneously collecting the ejected Auger electrons. The
main application for this analysis mode is actually in determining the depth
variation of a material's chemical composition. For materials that consist of
many layers, or perhaps possess a relatively thick surface coating, the long ion
6 Auger Electron Spectroscopy and Microscopy 163
100r---r---~--~--~--'---'---'-~
SAMPLE 4, (d)
z80
o
~
g:60
z 8.8 SEC
w
()
z 18 W/cm 2
840
MAX TEMP: 6700 C
2 4 6 8 12 14
SPUTTER TIME (MINUTES)
Fig. 6.6. Auger depth profile from a study of the formation of tungsten silicide on
silicon. A titanium film has been interposed between the tungsten surface film and
the silicon substrate. The study showed that upon annealing, the formation of tung-
sten silicide was enhanced by the presence of the titanium film with a corresponding
decrease in surface roughness and improvement in adhesion
sputtering times required for in-situ depth profiling may well be impractical.
However, the variation of chemical composition with depth can be obtained
for such materials by collecting a line scan over their cross-section. This
cross section may be obtained by ball-cratering or by fracturing the sample
and analysing the cross-section of the fracture cross-section [27-30]. Indeed, a
number of surface analysis systems that are designed for industrial use possess
facilities to allow samples to be fractured in-vacuo. The relatively long times
required for a complete survey scan mean that typically a small number of
Auger peaks are collected during each line scan. Each monitored element is
determined by selecting an energy window that covers the appropriate Auger
electron energy. The intensity of the Auger electron peak that corresponds to
the element of interest can thus be automatically measured as a function of
lateral distance along the cross-section, which in turn corresponds to a depth.
Fig. 6.7. Micrographs of SiC sintered in a nitrog(m atmosphere. (a) Optical mi-
crograph of polished and etched microstructure. (b) 15 keY backscatter electron
image. (c) Carbon X-ray map by wavelength dispersive analysis. (d) Monochrome
reproduction of chemical phase image derived from Auger imaging. The phase im-
age is of one particle at the triple point between SiC grains. Note the different scales
between this and the X-ray image in (c)
6 Auger Electron Spectroscopy and Microscopy 165
The first instruments were analogue systems that involved the combination
of a URV electron microscope with a CRA [33]. The images were stored on
magnetic tape and the system was relat.ively slow and cumbersome. There
followed the development of a digital instrument [34] and shortly afterwards
the development of commercial instruments.
The spatial resolution of state-of-the-art commercial instruments [36] is
currently of the order of 5-10 nm, with the electron beam generated using a
Schottky field emission electron gun [37]. This resolution is far superior to
that commonly available with other surface microprobe techniques, such as
imaging X-ray photoelectron spectroscopy with a typical working resolution
of the order of a few microns. The versatility of the SAM technique means
that it has found applications in studying a wide variety of material surfaces
including geological specimens [38-43], semiconductors [44], superconducting
oxides [45,46] and engineering ceramics [47,48].
As an example, Fig. 6.7 shows a comparison of the optical, backscattered,
X-ray wavelength dispersive and Auger microprobe images of a SiC surface
sintered in a nitrogen atmosphere [49] .
............-.......
.....
o
----------_.---------- 84%
..
·0
o
00
16% --------------------.-
....-..
00
...................................
23 25 27 29 31 33 35 37 39 41
Depth Eroded (nm)
Fig.6.S. Schematic diagram of the Auger depth profile obtained from a 30 nm
Ta20s/Ta standard. The resolution of the interface (~z) is taken as the depth
between the 84% and 16% levels of the 0 KLL Auger signal. The 50% point is
taken as the position of the interface
166 P.C. Dastoor
NiCrAI
~
(a)
""
c
N; s
~
;
I Ni15Cr13AI + 0.5 Zr
1
NiCrAIZr
i
" 1 dN(E)
"::t)1:;;,AIN; c
/1
~-----.L>.(.:' __ i z, _ (b) -
t~ V/O
""1 -......v_-
~--I_-----2.
450 550 650 750 850 I 750 850 750
<5xl0-9Torr 10-6 Torr 02
ANNEAL TEMPERATURE, °C
NiCrAIY
c Cr a Cr
(e)-
/'
o 100 200 300 400 500 600
ELECTRON ENERG~ eV
It is now possible to obtain the most up to date measurements for the imfps
for a variety of materials using the National Institute of Standards and Tech-
nology (NIST) Electron Inelastic-Mean-Free-Path Database [65].
6.6 Applications
is to determine the two factors Rand C that express the measured spectra
(D) in the form:
(6.6)
where R is a column vector containing representative spectra for the chemical
components in the profile and C is a row vector containing the concentrations
of these spectra for each sputter depth. In order to accurately decompose the
acquired depth profile into the factors Rand C, the number of significant
spectral components (or eigenvectors) needs to be determined. Typically this
involves a decision by the spectroscopist as to which elements are present
on the surface, each species then has a characteristic Auger spectrum corre-
sponding to a significant eigenvector for the purposes of the analysis.
Once Rand C have been determined it is possible, in principle, to split the
original data matrix into terms corresponding to different spectral features
of the original matrix, such that:
(6.7)
6.7 Future
Auger electron spectroscopy as a technique is continually being developed
and improved, both from the point of view of experimental practice and in-
strumentation. Increasingly, there are international efforts to standardise the
practice of AES and the quantification of spectra. Initial efforts were focussed
around the Versailles Project on Advanced Materials and Standards (VA-
MAS), with the establishment of a working party and the determination of
a standard reference spectrum for copper [96]. Since then a committee of the
International Standards Organisation (ISO) has been established (ISO jTC
201 for Surface Chemical Analysis) with the purpose of establishing interna-
tionally recognised standards for Surface Chemical Analysis. Indeed, a num-
ber of standards are being developed that are directly related to AES. For
example, ISO 17973 - Surface chemical analysis - Medium resolution Auger
electron spectrometers - Calibration of energy scales for elemental analysis
specifies a method for calibrating the kinetic energy scale of Auger electron
spectrometers with an uncertainty of 3 eV for general analytical use for iden-
tifying elements at surfaces and specifies a method to establish a calibration
schedule [97,98].
In terms of instrumentation, one of the major trends in Auger spec-
troscopy has been the continual advance in the ability to produce Auger mi-
croscopic images of surfaces. The study of multichannel image collection and
processing techniques have resulted in the development of the multi-imaging
scanning analytical microscope (MULSAM) [35,99]. The development of the
hyperbolic field analyser (HFA) offers the possibility of collecting an entire
energy spectrum in parallel, thus allowing the very rapid acquisition of en-
ergy spectra [100]. Indeed, a new instrument, the so-called spectrum imaging
scanning analytical microscope (SISAM), based on the implementation of the
HFA [101]. These, and other developments, mean that the study and appli-
cation of Auger electron spectroscopy is likely to continue for many years to
come.
References
1. H. Todokoro, Y. Sakitani, S. Fukuhara, Y. Okajima: J. Electron Microsc. 30,
107 (1981)
2. L. Meitner: Z Phys. 17, 54 (1923)
3. P. Auger: Comptes Rendus 177, 169 (1923)
4. M.P. Seah, W.A. Dench: Surf. Int. Anal. 1, 2 (1979)
5. D.A. Shirley: Phys. Rev. A. 7, 1520 (1973)
6. R. Hoogewijs, L. Fiermans, J. Vennik: Surf. Sci. 69, 273 (1977)
7. R. Weissmann, K. Muller: Surf. Sci. Rep. 105, 251 (1981)
8. S. Tanuma, C.J. Powell, D.R. Penn: Surf. Interface Anal. 25, 25 (1997)
9. C.-O. A. Olsson, S.E. Hrnstrm, S. Hogmark: In Surface Characterisation: a
user's sourcebook, ed. by D. Brune, R. Hellborg, H.J. Whitlow, O. Hunderei,
(Wiley, New York 1997)
172 P.C. Dastoor
50. BCR reference material no. CRM 261 available from Institute for Reference
Materials and Measurements (IRMM), Retieseweg, B-2440, Geel, Belgium.
(www.irmm.jrc.be)
51. C.P. Hunt, H.J. Mathieu and M.P. Seah: Surf. Sci. 139, 549 (1984)
52. C.S. Wei, M. Setton, J. Van der Spiegel, J. Santiago: J. Appl. Phys 61, 1429
(1987)
53. P.H. Holloway: Surf. Sci. 66, 479 (1977)
54. R. Li, L. Tu, Y. Sun: Surf. Sci. 163, 67 (1985)
55. T. Sekine, A. Mogarni, J.D. Geller, SEM, SEM Inc, AMF O'Hare, Chicago
(1981), p.245
56. H.J. Mathieu: In Surface Analysis - Principal Techniques, ed. by J.C. Vick-
erman (Wiley, New York 1997), Chap. 4
57. A. Jablonski, C.J. Powell: J. Electron Spectrosc. Relat. Phenom. 100, 137
(1999)
58. C.J. Powell, A. Jablonski, LS. Tilinin, S. Tanuma, D.R. Penn: J. Electron
Spectrosc. Relat. Phenom. 98-99, 1 (1999)
59. C.J. Powell, A. Jablonski: Surf. Interface Anal. 29, 108 (2000)
60. S. Tanuma, C.J. Powell, D.R. Penn: Surf. Interface Anal. 11, 577 (1988)
61. S. Tanurna, C.J. Powell, D.R. Penn: Surf. Interface Anal. 17, 911 (1991)
62. S. Tanurna, C.J. Powell, D.R. Penn: Surf. Interface Anal. 17, 929 (1991)
63. S. Tanurna, C.J. Powell, D.R. Penn: Surf. Interface Anal. 20, 77 (1988)
64. S. Tanuma, C.J. Powell, D.R. Penn: Surf. Interface Anal. 21, 165 (1994)
65. C.J. Powell, A. Jablonski: NIST Electron Inelastic-Mean-Free-Path Database
(Version 1.0), National Institute of Standards and Technology: Gaithersburg,
MD,1999
66. P. Morgan, B. Jorgensen: Surf. Sci. 208, 306 (1989)
67. P.J. Lurie, J.M. Wilson: Surf. Sci. 65, 476 (1977)
68. T.W. Hass, J.T. Grant, G.J. Dooley III: J. Appl. Phys. 43, 1853 (1972)
69. Y. Mizokawa, T. Miyasato, S. Nakamura, K.M. Geib, C.W. Wilmsen: Surf.
Sci. 182, 431 (1987)
70. H.H. Madden: Surf. Sci. 126, 80 (1983)
71. C.D. Wagner: Anal. Chern. 47, 1201 (1975)
72. L.P. Erickson, B.F. Phillips: J. Vac. Sci. Technol. B 1, 158 (1983)
73. H.H. Busta, C.H. Tang: J. Electrochem. Soc. 133, 1195 (1986)
74. F. Marchetti, M. Dapor, S. Girardi, F. Giacomozzi, A. Cavalieri: Mater. Sci.
Eng. A 115, 217 (1989)
75. R. Pantel, F.A. D'Avitaya: Thin Solid Films 140, 177 (1986)
76. S.J. Pearton, A.B. Emerson, U.K. Chakrabarti, E. Lane, K.S. Jones,
K.T. Short, A.E. White, T.R. Fullowan: J. Appl. Phys 66, 3839 (1989)
77. T.T. Huang, B. Peterson, D.A. Shores, E. Pender: Corrosion Science, 24, 167
(1984)
78. S. Mathieu, La Revue de Metallurgie, Jan 1989, p. 73
79. E.R. Malinowski, Factor Analysis in Chemistry, 2nd ed. (Wiley, New York
1991)
80. W. Palmer: Appl. Phys. A 42, 219 (1987)
81. B.M. Clemens: J. Non Cryst. Solids 61-62, 817 (1984)
82. H.J. Grabke: ISIJ 29, 529 (1989)
83. M. Militzer, J. Wieting: Acta Metall. 37, 2585 (1989)
84. D.C. Peacock: Appl. Surf. Sci. 26, 306 (1986)
174 P.C. Dastoor
M.H. KibeI
7.1.1 Theory
Presented schematically in Fig. 7.1 are the related processes that are involved
in the ejection of a photo- or Auger electron. XPS involves the removal of a
single core electron, while AES is a two-electron process subsequent to the
removal of the core electron, with the Auger electron ejected following reor-
ganisation within the atom. Auger electrons are produced in XPS along with
photoelectrons, and these can complicate the interpretation of the subsequent
spectra as will be discussed later in this chapter.
The photoemission process is shown on an energy level diagram in Fig. 7.2.
The sample is irradiated with X-rays of known energy, hv, and electrons of
binding energy (BE) Eb are ejected, where Eb < hv. These electrons have a
kinetic energy (KE) Ek which can be measured in the spectrometer, and is
given by
(7.1)
where cJ>sp is the spectrometer work function, and is the combination of the
sample work function, cJ>s, and the work function induced by the analyser.
176 M.H. KibeI
(a) XPS
• • • • •• L'.3 OR 2p
•• L, OR 28
~ //////,"oromc,"o,
o. KOR18
or
(b) AES
...
... AUGER ELECTRON
~L,OR28
~ .
,/
/'
L',3 OR 2p
I
I
I
-~.'-".I---- K OR 18
Fig. 7.1. Representation of the processes involved in XPS and AES. (a) XPS in-
volves the removal of a single core electron while (b) AES is a two-electron process
following the removal of the core electron
Since we can compensate for the work function term electronically, it can be
eliminated, leaving
(7.2)
or
(7.3)
---------------------1--
hv
r----
(1)5
Fig. 7.2. Schematic diagram of the
photoemision process. The sample is
irradiated with X-rays of known en-
ergy, hv , and electrons of binding en-
EB Spectrometer ergy Eb are ejected. These electrons
X-ray have a kinetic energy, E k , which can
photon
_L. be measured in the spectrometer, and
is given by (7.1). P s is the work func-
tion of the sample, and P sp is the
Sample spectrometer work function
Te 3<\
Te 3dlt2
(dIHNNI
Te(HNN)
(d 3d~
(d3~ I Hg 4fSQ
Te 4d
(d3~ ( ls
MgKa
Fig. 7.3. MgKo: XP spectrum of Hg o.6 Cdo.4 Te grown via Metal Organic Chemi-
cal Vapour Deposition (MOCVD). Inset : (a) Te 3d region of full spectrum before
cleaning, showing the presence of oxide. (b) Te 3d region after ion etching, which
has removed the oxide
178 M.H. Kibei
A = Am cose (7.4)
e
where is the angle of emission to the surface normal. Thus electrons emitted
perpendicular to the surface (e = 0°) will arise from the maximum escape
depth, and hence will carry information which may be indicative of the bulk
material, whereas electrons emitted nearly parallel to the surface (e '" 90°)
will be purely from the outermost surface layers. Thus XPS is an extremely
surface sensitive technique, and as we will see from Sect. 7.5.3, varying the
angle of detection during an experiment, as described above, can enhance
this surface sensitivity.
7 X-Ray Photoelectron Spectroscopy 179
7.2 Instrumentation
As with most techniques, certain components are essential, while others, al-
though desirable, can be optional. In this section we discuss the components
used in modern XPS systems, and consider different alternatives available for
many of these components.
Computer
X-ray line. From Table 7.1, the only lines which are sufficiently energetic and
narrow are the Na K a, Mg K a and Al K a lines. However it is very difficult to
design a suitable, stable sodium anode, thus the Mg and Al sources are most
commonly used. Many manufacturers now provide X-ray sources with dual
Mgj Al anodes as standard. Other dual-anode combinations are also available,
such as MgjZr and AljZr, used since the Zr La line has a particularly high
sensivity to elements such as aluminium and silicon [11].
The sources described above do not, however, produce single X-ray lines,
but a series of lines superimposed on the Bremsstrahlung continuum. One
method of removing the unwanted components, and eliminating the contin-
uum, is to monochromatise the radiation. This is most conveniently achieved
by using a diffraction grating, with the line spacings for selecting the correct
component determined from the Bragg relation [12]. Of course the result-
ing monochromatised radiation will be greatly reduced in intensity, due to
dispersion.
x x
y y
---1
I
I
I
I
I
I
I
,I
I
IVocwm
e::..m~~ ___ _______ ,I
Fig. 7.5. Schematic diagram showing the operating principles of a concentric hemi-
spherical analyser (eRA) (see text for explanation), together with the electronics
used to produce XP spectra
and
(7.6)
where the entrance and exit slits are separated by an angle of 180 0 • This
results in focussing in two dimensions, i.e. at a point.
The resolution of an analyser can be defined in two ways. One is the
absolute resolution I1E measured as the full width at half-maximum height
(FWHM) of a peak. The other is the relative resolution, e, of a peak at KE
Eo
e = E o/ I1E. (7.7)
Thus the absolute resolution is independent of peak position, but the relative
resolution must be referred to the kinetic energy of the peak.
182 M.H. KibeI
The current actually reaching the analyser exit slit following photoionisa-
tion is typically in the region of 10- 16 _10- 14 A, which is well below conven-
tional current-measuring techniques. Thus pulse counting is required, and an
electron multiplier is used as the detector. There are two types of electron
multiplier currently in use in modern spectrometers: the traditional discrete-
dynode type and the more common channel electron multiplier.
The dynode multiplier functions by allowing the electrons to strike the
first electrode (or dynode), and then achieving current amplification via sec-
ondary electron multiplication of many stages (typically 10-20). A resistor
chain is used to establish the separate dynode potentials. Multipliers of this
type can achieve amplification of lO c 10 7 , depending on their design.
A channel electron multiplier consists of a small curved glass tube, the
inside wall of which is coated with high resistance (~ 109 fl full length)
material. When a potential (~ 3 k V) is applied between the ends of the tube
the resistive surface becomes a continuous dynode. An electron entering the
low potential (typically +500V) end of the multiplier generates secondary
electrons on collision with the wall of the tube. These are accelerated until
they strike the wall again, giving an avalanche effect. The gain of these devices
is of the order of 107-10 8 .
Apart from single multipliers, substantial increase in performance is pos-
sible with multichannel acquisition via position-sensitive detection [32]. This
involves use of multiple arrays of discrete dynodes or channeltrons, providing
vastly increased signal-to-noise (S/N), thus reducing acquisition times and
thus less sample exposure to X-rays.
As seen in Fig. 7.5, the multiplier output is normally taken through a
pre-amplifier, an amplifier, a discriminator and a rate-meter system, with
the spectrum being displayed on an X-Y recorder. More commonly, a com-
puter interface is situated between the discriminator and rate-meter, with the
spectra subsequently digitised and stored for future reference. Also shown
in Fig.7.5 is a schematic of the electronics required to operate the anal-
yser / detector system.
is also extremely valuable for residual gas analysis and, if the correct geome-
try is chosen, for thermal desorption spectroscopy. Many of these options are
discussed elsewhere in this book, and in review articles [18,19].
(a) W
(b) W0 2 (brown)
494 1066
(j
492 1064 ..:
Q)
+-'
:;-
Q)
E
~ 490 1062 co
"-
co
I!) a...
-.:t
z 488 1060
"-
Q)
I!)
-.:t 0>
«
:::J
Z
-.:t
~ 486 "0
Q)
~
"0
484 0
~
482~-.~--r-~~-r-'--~~-r-,--~~-4
where Ek(A) = Auger electron KE, and EdP) = photoelectron KE. One of
the main advantages of a is that it is independent of static charging, and is
characteristic of a particular chemical state. Unfortunately, for some systems
a can have negative values, however (7.12) can be modified by using (7.2) as
follows:
(7.11)
or, the "modified" Auger parameter,
(7.12)
Thus a graph Ek(A) vs Eb(P) becomes independent of photon energy.
Such two-dimensional plots have been generated for many elements, one of
which is tellurium, Te, shown in Fig. 7.7 [8]. Many other examples of the use
of the Auger parameter can also be found in the literature [35-41].
Ql,2 Q3 Q4 Q5 Q6 {J
CuO
CuO
Fig. 7.8. The copper 2pl/2 and 2P3/2 XP spectra from (a) CuO, showing very
prominent shake-up satellites, and (b) Cu 2 0, where such satellites are absent. (Re-
produced with permission from [42])
Ghost lines are small peaks appearing in an XP spectrum which result from
X-radiation from foreign material. For example, common ghost lines arise
from Mg impurity in the Al source, or vice vcr-sa in a dual anode source.
7 X-Ray Photoelectron Spectroscopy 189
Other sources can be eu from the anode base structure or X-ray photons
arising from the foil window. The positions of these ghost lines can be easily
calculated, and Table 7.3 shows where such lines are expected to occur.
Anode material
Contaminant Radiation Mg Al
(7.13)
For all elements the fundamental, or first, bulk plasmon will always be ob-
servable, with multiple plasmon loss peaks also visible in some cases. Depend-
ing on surface conditions, and the energy of the corresponding photoelectron
peak, surface plasmons may also be seen, especially in the case of clean metal
surfaces. An example of this can be seen in Fig. 7.9, for aluminium, showing
plasmon loss peaks, both bulk and surface, for the 28 photoelectrons [44].
One of the major advantages of XPS is the ease with which quantitative data
can be routinely obtained. This is usually performed by determining the area
under the peaks in question and applying previously-determined sensitivity
factors. For a homogeneous sample, the number of photoelectrons per second
190 M.H. KibeI
in a given peak, assuming constant photon flux and fixed geometry, is given
by [45]:
1= KNa)\AT, (7.14)
Sx = KOAAT, (7.15)
then
Ix = NxSx (7.16)
or
Nx = Ix/Sx . (7.17)
AI 28
(a)
~
·iii
c
.l!l
E
I-- Ilwb
(b)
Fig. 7.9. XP spectrum of the A12s region, showing a succession of sequential bulk
plasmon losses (11Mb = L5.2 eV). In spectrum (a), for clean AI, a surface plasmon
loss can also be seen (nws = 10.7 e V), which disappears on oxidation, as in spectrum
(b). (After [44])
7 X-Ray Photoelectron Spectroscopy 191
(7.18)
This approach will provide semi-quantitative results for most situations, ex-
cept where heterogeneous samples are involved, or where serious contami-
nation layers obscure the underlying elements. Also, any peak interference
(e.g. overlapping Auger lines or another XPS peak) must be avoided. Ide-
ally elemental sensitivity factors should be determined specifically for each
instrument.
A simple application of this type of quantitative analysis has been pro-
vided by Leech et al. in a study of the effect of etchants on II-VI semiconduc-
tor materials [46,47]. Their work used XPS to determine the relative amounts
of Hg, Cd and Te in the samples before and after etching, and hence were
able to follow the changes which the etchants induced.
As mentioned in Sect. 7.2.1(a), many X-ray sources are available for use in
XPS, with the most common being a dual Mg/ Al source. What is the value
of the availability of two different X-ray lines? Firstly, the Mg K a line is
narrower than the Al K a, thus allowing a better resolution to be achieved.
Secondly, some core levels may not be accessible with the Mg source, and
require the slightly higher (by 233 eV) energy Al line. Thirdly, often in an
XP spectrum Auger lines and XP peaks may overlap, and changing to another
source can alleviate this problem, since the Auger electrons always have the
same KE, whereas the photoelectron KE depends on the ionising radiation
as the BE is constant. Thus the Auger lines appear to move relative to the
XP peaks as the X-ray energy is varied. An example of this application is
shown in Fig. 7.10, where spectra have been recorded for galvanised iron. In
the Al Ka spectrum (b), the Zn(LMM) lines overlap with the 0 Is and
Cr 2p orbitals, while the Cr(LM M) lines interfere with the Zn 2p peaks.
Changing to Mg K a (a) solves this problem, with the Cr( LM M) lines clear
of any other features, and the Zn(LM M) lines now overlapping with the C
Is peak.
This topic has been broached previously in this book, both generally and
with specific application to AES (Chap. 6). Depth profiling has applications
in XPS also, although it has been less utilised than in AES. The process
192 M.H. KibeI
5.10
x 104 MgKa
4.59
4.08
"0
C 3.57
0
c.>
Q)
(/) 3.06
Q; 2.55
0-
(/) Zn (LMM)
C
2.04
,------,
::l
0 1.53
() C15
1.02
0.51 a
0.00
x 103
Cr LMM
AIKa
0.99
0.88 Zn (LMM)
"0
C
,------,
0
c.> 0.77
Q) 015
(/)
0.66
Q;
0-
0.55
(/)
C 0.44
::l
0
()
0.33
0.22 Zn3s
0.11
b
0.00
1000.00 800.00 600.00 400.00 200.00
Binding Energy (eV)
Fig. 7.10. XP spectra of galvanised steel obtained using (a) Mg Ka and (b) Al
K a X-ray sources. Note the relative positions of the Auger and Photoelectron lines
again involves etching, or sputtering, away the material using a rare-gas ion
beam (usually Ar) and recording spectra as a function of depth. However for
an XPS depth profile a relatively large surface area must be etched. This is
usually necessary since the X-ray beam cannot easily be focussed to a small
spot in most instruments (as discussed in Sect. 7.2.4).
Although ion sputtering can change the chemical states in a material,
much information is still available from an XPS depth profile. An excellent
example is the variation of the oxidation state of an element in a material.
Figure 7.11 shows a simple depth profile for a CdTe film on a nickel substrate,
where the tellurium has oxidised [37]. The first thing to notice is that Cd
seems to be preferentially concentrated at the surface, whereas the Te which
appears near the surface occurs largely as the oxide Te02, which disappears
quickly on sputtering. The sequence of XP spectra recorded as a function of
7 X-Ray Photoelectron Spectroscopy 193
80
~60~~_ _
~ ~---~---------
~40 / .......Te
I
I
20
. Te* N'I
............
O~--~a=~==~~~~~~
'-'-'-'-
o 57 104 171 228 285
Eteh Time (Sees.)
Fig. 7.11. XPS depth profile for a CdTe electrodeposited film. "Te*" represents Te
arising from Te02
'0
C
o
(.)
a.>
en
.l!l 20
c
:J
o
o
While discussing surface specificity in Sect. 7.1.3, it was pointed out that the
depth from which the photoelectrons emanated depended on the angle of
detection, i.e. the angle of emission to the surface normal, B. This is shown
diagrammatically in Fig. 7.13, where it can be seen that detection close to the
normal enhances the signal from the bulk relative to the surface, while detec-
tion close to the surface plane enhances the signal from the surface relative to
the bulk. Thus varying the angle of detection can yield non-destructive depth
information, an example of which is presented in Fig. 7.14, showing data for
a thin film of Si0 2 on Si [48]. For low values of B the main contribution to the
spectrum is from the bulk Si, while at larger values of B the contribution from
the oxide layer becomes substantial. This approach is obviously preferable to
the destructive ion etching, but is limited to very thin layers.
8 = take-off angle
Layer
Bulk
When insulating materials are irradiated by X-rays they quite often develop
a static charge due to their inability to replace the photo emitted electrons.
This charging can be substantial, say 2-5 e V, but is always positive and quite
small in comparison to electron-induced AES where charging can be of the
order of several hundred eV. By using reference elements, such as a small
amount of gold or silver (or, in some cases, adventitious carbon) the BE shifts
can easily be calculated. In some instances, partial charging is possible, e.g.
insulating domains on a conducting substrate (see Fig. 17.6 in Chapter 17).
The most common approach to this problem is to use a neutral ising source,
such as a low energy electron flood gun, to compensate for the charge. Using
a reference peak to observe the amount of BE shift, the flood gun can be
tuned to provide just the right amount of current to shift the peaks back to
their "uncharged" binding energies [32].
7 X-Ray Photoelectron Spectroscopy 195
.--.
:!:::
C
:J
-
~
C1l
"-
:"-
0
~
~
·W
-c
Q)
C
c 8 =66°
e
t5
-
Q)
Q5
o
o
..c
D....
a.
C\J
U5
Fig. 7.14. Study by XPS of the in-
terface between silicon and a thin
(0.91 nm) film of Si02 on its surface. ()
varies from 21 ° to 81 ° away from the
normal to the surface. (Reproduced
Binding (eV) with permission from [48])
Obviously no one technique can ever solve every surface problem, and for
most studies two or more techniques are combined. Compromises must be
made in the use of a particular method, between, for example, resolution or
sensitivity, or lateral resolution or chemical information. In order to place
XPS in perspective with respect to other surface-science tools, we will com-
pare its characteristic properties with those of two other frequently used
techniques, viz. AES and secondary ion mass spectrometry (SIMS). (Refer to
Tables 1.3,1.4 for a list of the specific properties of these surface analytical
techniques. )
All three spectroscopies have similar element detection capabilities, al-
though only SIMS can detect hydrogen. XPS can detect all other elements,
while AES can detect all but hydrogen and helium. Thus, except for special
applications, the range of elemental detection is fairly comparable. Quan-
titative data can be obtained from all the methods, although standards are
196 M.H. KibeI
needed, however AES and XPS can provide semi-quantitative information di-
rectly from peak heights. In some situations SIMS can detect elements down
to concentrations of less than 0.01 ppm (10- 6 Atomic %), whereas XPS and
AES are virtually identical in their detection ability of only 0.1 Atomic %. All
three techniques are comparable when it comes to depth resolution, however
AES and SIMS can achieve high lateral resolution unlike XPS. This arises
because it is much easier to focus an electron or ion beam than an X-ray flux,
7 X-Ray Photoelectron Spectroscopy 197
although as discussed in Sect. 7.2.4, the ability to image with X-rays, and
the resultant lateral resolution, has improved recently.
One of the major advantages of XPS is its ability to provide consistent
chemical information. SIMS also provides chemical information, although
more difficult to interpret, and AES chemical information is both difficult
to obtain routinely and to interpret. It is difficult to study organic materials
or many adsorbate systems, due mainly to electron-beam induced damage,
although this is minor compared to the destructive nature of SIMS. XPS is
by far the least destructive of the three techniques considered here, and X-
ray damage is not often a problem. Finally, it is much simpler to eliminate
static charging during XPS than when using AES or SIMS, since the latter
methods rely on beams of charged particles.
In summary, the main advantages of XPS (in comparison to AES and
SIMS) are its ability to provide vital chemical information, simple quanti-
tative information, low sample damage and the fact that it can be used to
study insulating materials. The only major disadvantage of XPS is its rela-
tively poor lateral resolution which has been improved in recent times.
7.7 Applications
While there are a plethora of applications in which XPS has played a major
role, the list is far too exhaustive to consider in any detail here. Instead the
main categories in which XPS has been used are tabulated in Table 7.4 along
with the information yielded by the technique in each case, and references to
some of the original, and more recent, work.
7.8 Conclusion
X-ray photoelectron spectroscopy has been used extensively over the past
twenty years in many areas of surface and materials analysis. The disad-
vantages of the technique, such as the need for URV and the poor spatial
resolution of the X-ray source, are more than adequately outweighed by the
advantages, viz. the ease of interpretation of spectra and the ability to derive
chemical-state and simple quantitative information. Over the past few years
the technical aspects of XPS have improved substantially, and this trend
will continue into the next decade. The number of areas of application for
XPS has also grown considerably, making this technique almost mandatory
in modern analytical laboratories.
References
30. H.W. Ade, J. Kirz, S.L. Hulbert, E.D. JohnHOn, E. Anderson, D. Kern: Appl.
Phys. Lett 56, 1841 (1990)
31. U. Gelius, B. Wannberg, P. Baltzer, H. Fellner-Felegg, G. Carlsson, C.-G. Jo-
hansson, J. Larsson, P. Munger, G. Vegefors: J. Electron Spectrosc. 52, 747
(1990)
32. P.J.C. Chappell, M.H. Kibei, B.G. Baker: J. Catalysis 110, 139 (1988)
33. C.D. Wagner: Anal. Chem. 47, 1201 (1975)
34. C.D. Wagner, L.H. Gale, R.H. Raymond: Anal. Chem. 51, 466 (1979)
35. R.H. West, .I.E. Castle: Surf. Interface Anal. 4, 68 (1982)
36. J.C. Riviere, J.A. Crossley, B.A. Sexton: J. Appl. Phys. 64, 4585 (1986)
37. M.H. Kibei, C.G. Kelly: Materials Aust. 19 (10), 15 (1987)
38. C.D. Wagner, D.E. Passoja, H.F. Hillery, T.G. Kinisky, H.J.A. Six,
W.T. Jansen, J.A. Taylor: J. Vac. Sci. Technol. 21, 933 (1982)
39. A. D'Agnano, M. Gargano, C. Malitesta, N. Ravasio, L. Sabbatini: J. Electron
Spectrosc. 53, 213 (1991)
40. R. Alfonsetti, G. De Simone, L. Lozzi, M. Passacantando, P. Picozzi, S. San-
tucci: Surf. Interface Anal. 22, 89 (1994)
41. M.H. Kibei, P.W. Leech: Surf. Interface Anal. 24, 605 (1996)
42. M. Scrocco: Chem. Phys. Lett. 63, 52 (1979)
43. J.M. Ritchie: Phys. Rev. 106, 874 (1957)
44. A. Barrie: Chem. Phys. Lett. 19, 109 (1973)
45. B.A. Sexton: Materials Forum 10, 134 (1987)
46. P.W. Leech, P.J. Gwynn, M.H. Kibei: Appl. Surf. Sci. 37, 291 (1989)
47. P.W. Leech, M.H. Kibei, P.J. Gwynn: J. Electrochem. Soc. 137, 705 (1990)
48. A. Ishizaka, S. Iwata: Appl. Phys. Lett. 36, 71 (1980)
49. B.L. MaRchhoff, K.R. Zavadil, K.W. Nebesny, N.R. Armstrong: J. Vac. Sci.
Technol. A 6, 907 (1988)
50. D.E. Fowler, J. Rogozik: J. Vac. Sci. Technol. A 6, 928 (1988)
51. T. Czeppe, P. Nowak: Materials Sci. Forum 25/26, 525 (1988)
52. T.M.H. Saber, A.A. El Warraky: J. Mater Sci. 23, 1496 (1988)
53. E. Paparazzo: Surf. Interface Anal. 12, 115 (1988)
54. E.W.A. Young, J.C. Riviere, L.S. Welch: Appl. Surf. Sci. 28, 71 (1987)
55. N.S. McIntyre: In Practical Surface Analysis by Auger and Photoelectron Spec-
troscopy, ed. by D. Briggs, M.P. Seah (Wiley, New York 1983) Chap. 10, p. 397
and references therein
56. D. Landolt: Surf. Interface Anal. 15, 395 (1990)
57. P.M.A. Sherwood: Anal. Chim. Acta 283, 52 (1993)
58. M.F. Guimon, G. Pfister-Guillouzo, M. Bremont, W. Brockmann, C. Quet,
J.Y. Chenard: Appl. Surf. Sci. 108, 149 (1997)
59. M. Ben-Haim, U. Atzmony, N. Shamir: Corrosion Sci. 44, 461 (1988)
60. K.L. Rhodes, P.C. Stair: J. Vac. Sci. Technol. A 6, 971 (1988)
61. J.C. Langevoort, I. Sutherland, L.J. Hanekamp, P.J. Gellings: Appl. Surf. Sci.
28, 167 (1987)
62. F.-M. Pan, P.C. Stair: J. Vac. Sci. Technol. A 5, 1036 (1987)
63. R.J. Bird: Metal Sci. J. 7, 109 (1973)
64. H.J. Brabke: Surf. Interface Anal. 14, 686 (1989)
65. J. Jedlinski, A. Glazkov, M. Konopka, G. Borchardt, E. Tscherkasova,
M. Bronfin, M. Nocun: Appl. Surf. Sci. 103, 205 (1996)
66. I. Milosev, H.-H. Strehblow, B. Navinsek: Surf. Interface Anal. 26, 242 (1998)
200 M.H. Kibei
67. S. Akhter, X.-L. Zhon, J.M. White: App!. Surf. Sci. 37, 201 (1989)
68. J.G. Clabes, M.J. Goldberg, A. Viehbeck, C.A. Kovac: J. Vac. Sci. Techno!.
A 6, 985 (1988)
69. H.M Meyer III, S.G. Anderson, Lj. Atanasoska, J.H. Weaver: J. Vac. Sci.
Techno!. A 6, 1002 (1988)
70. J.L. Jordan, C.A. Kovac, J.F. Morar, R.A. Pollack: Phys. Rev. B 36, 1369
(1987)
71. H. Ohno, T. Ichikawa, N. Shiokawa, S. Ino, H. Iwasaki: J. Mater Sci. 16, 1381
(1981)
72. C. Argile, G.E. Rhead: Surface Sci. Rep. 10, 277 (1989)
73. W.W. Zhao, F.J. Boerio: Surf. Interface Ana!. 26, 316 (1998)
74. A.J. Hartmann, R.N. Lamb: Current Opinion in Solid State & Mater. Sci. 2,
511 (1997)
75. N. Bertrand, B. Drevillon, A. Gheorghiu, C. Senemaud, L. Martinu,
J.E. Klemberg-Sapieha: J. Vac. Sci. Techno!. A 16, 6 (1998)
76. G. Moretti, G. Fierro, M. Lo Jacono, P. Porta: Surf. Interface Ana!. 6, 188
(1980)
77. T.L. Barr: In Practical Surface Analysis by Auger and X-Ray Photoelectron
Spectroscopy, ed. by D. Briggs, M.P. Seah (Wiley, New York 1983) Chap. 8
p. 283 and refences therein
78. J.S. Brinen: In Applied Surface Analysis ed. by T.L. Barr, L.E. Davis (ASTM,
Philadelphia 1978) p.24
79. J. Grimblot, L. Gengembre, A. D'Huysser: J. Electron Spectrosc. 52, 485
(1990)
80. B.G. Baker, Chapter 17, and references therein
81. F. Gaillard, P. Artizzu, Y. Brulle, M. Primet: Surf. Interface Ana!. 26, 242
(1998)
82. J.H. Thomas III, G. Kaganowicaz, J.W. Robinson: J. Electrochem. Soc. 135,
1201 (1988)
83. T. Wada: App!. Phys. Lett. 52, 1056 (1988)
84. R.N.S. Sodhi, W.M. Lan, S.LJ. Ingrey: Surf. Interface Ana!. 12, 321 (1988)
85. C.F. Yu, M.T. Schmidt, D.V. Podlesnik, R.M. Osgood, Jr.: J. Vac. Sci. Tech-
no!. B 5, 1087 (1987)
86. M.D. Biedenbender, V.J. Kapoor, W.D. Williams: J. Vac. Sci. Techno!. A 5,
1437 (1987)
87. H.J. Kim, R.F. Davis, X.P. Cox, R.W. Linton: J. Electrochem. Soc. 134,2269
(1987)
88. T. Hattori: J. Vac. Sci. Techno!. B 11, 1528 (1993)
89. W. Palmer: Surf. Interface Ana!. 22, 331 (1994)
90. L.-Y. Chen, G.W. Hunter, P.G. Neudeck, D. Knight: J. Vac. Sci. Techno!. A
16, 2990 (1998)
91. A. Nelson, S. Glenin, A. Frank: J. Vac. Sci. Techno!. A 6, 954 (1988)
92. B. Mutel, O. Dessaux, P. Goudmand, J. Grimblot, A. Carpentier, A. Szarzba-
jnski: Rev. Phys. App!. 23, 1253 (1988)
93. D.T. Clark, D.R. Hutton: J. Polym. Sci. A 25, 2643 (1987)
94. T.J. Hook, J.A. Cardella Jr., L. Salvati, Jr.: J. Mater Res. 2, 117 (1987)
95. G. Gilberg: J. Adhes. 21, 129 (1987)
96. D. Briggs: In Practical Surface Analysis by Auger and Photoelectron Spec-
troscopy ed. by D. Briggs, M.P. Seah (Wiley, New York 1983) Chap. 9, p.359
and references therein
7 X-Ray Photoelectron Spectroscopy 201
There are many forms of vibrational spectroscopies applied to surfaces but the
most accessible techniques are Fourier transform infrared spectroscopy and
Raman spectroscopy. They will be discussed in this chapter. Other vibrational
spectroscopies (e.g. PAS, EELS) are listed in the Appendix and Bibliography.
8.1 Introduction
Infrared spectroscopy has been a widely used technique in industry for the
structural and compositional analysis of organic, inorganic and polymeric
samples and for quality control of raw materials and commercial products.
With the advent of Fourier transform infrared spectroscopy (FTIR) the range
of applications and the materials amenable to study has increased enor-
mously, owing to its increased sensitivity, speed, wavenumber accuracy and
stability. Oils, coal, shale, polymers, paints, catalysts, pharmaceuticals and
indu~trial gases have been successfully analyzed [1]. More recently FTIR has
been developed for the quality control of minerals and in exploration [2]. Cel-
lulosic materials, wood, paper and cotton have also proved amenable [3]. It is
also now possible to study electrode processes using FTIR [4,5]. Biochemical
processes once excluded from the field of infrared spectroscopy owing to the
presence of water now form one of the fastest growing areas of research in
FTIR. Specific examples are discussed in the text.
Conventional infrared spectroscopy relies on the dispersion of an infrared
beam via a grating into its monochromatic components, and slowly scanning
through the entire spectral region of interest. When a sample is placed in the
beam, various wavelengths of infrared radiation are absorbed by the sample
as the beam is scanned, and the result recorded as the infrared spectrum of
the sample. To maintain a resolution of 4 wavenumbers, this typically requires
around 2-3 minutes for a single scan of the spectrum.
Fourier transform infrared (FTIR) spectroscopy relies on a totally dif-
ferent principle to record the same information - that of interferometry. A
Michelson or Genzel interferometer forms the basis of the FTIR spectrome-
ter. Figure 8.1 shows a comparison of the optics for dispersive and Michelson
interferometric instruments. The FTIR instrument consists of a standard in-
frared source, collimation mirrors, a beamsplitter, a fixed mirror and moving
mirror. 50% of the beam is passed through the beamsplitter and 50% is re-
flected both on the initial pass and from reflections from the two mirrors. This
204 R.L. Frost and N.K. Roberts
DISPERSIVE OPTICS
Exit
slit
Collimator
Diagonal mirror
I mirror
Entrance
slit
INTERFEROMETRIC OPTICS
Fixed mirror
...- ..
~
Movable
L1.- -_~I'
.* +..* mirror
Source .. +.... +----+
mirror motion
~~ L2
Beam splitter • •
.-
Fig. 8.1. Comparison of optics for dispersive and Michelson interferometric instru-
ments
The main techniques for studying surfaces with FTIR spectroscopy are,
Of these techniques (a), (b) and (c) have been used successfully with dis-
persive IR instruments on solid samples. FTIR has extended infrared spec-
troscopy to aqueous solutions using (c) and enabled (d) to be carried out
in the IR region; previously PAS was only possible in the UV-visible region.
Both ATR and PAS allow depth profiling of the surface. (e) IES enables the
FTIR spectra of surfaces and thin films to be obtained at elevated tempera-
tures and in situ.
i) the surface of solid samples, provided good contact can be achieved be-
tween the sample and the ATR element. Depth profiling is also possible
by varying the angle of incidence provided it does not exceed the criti-
cal angle. This technique is ideal for solid samples that can make good
contact with the ATR element, in this class fall rubber, polymer films
fabrics, coated and painted surfaces.
ii) ATR has been used with excellent results to investigate the solid water in-
terface. The application of FTIR-ATR Spectroscopy to aqueous systems
has opened up a whole new area of study, from the clotting mechanism
of blood on foreign surfaces to the mode of action of flotation collectors
on mineral surfaces.
8 Vibrational Spectroscopy of Surfaces 207
k
(8.1)
s
208 R.L. Frost and N.K. Roberts
R
I
~
R'oo (sample)
-=-:-=:-'----::----':-:- (8.2)
00 ~ R'oo (standard) .
R'oo (sample) represents the single-beam reflectance spectrum of the sample
and R'oo (standard) is the single-beam reflectance spectrum of a selected non-
absorbing standard exhibiting high diffuse reflectance throughout the wave-
length regions being studied. Potassium chloride or potassium bromide with
a particle size of less than 10 f!m is often used as a standard. The depth
of the KCI or KBr matrix required to yield samples of "infinite thickness",
i.e where further increasing the thickness causes no change in the spectrum,
is less than 5 mm. The Kubelka-Munk theory predicts a linear relationship
between the molar absorption coefficient, k, and the peak value of f(Roo)
for each band, provided s remains constant. Since s is dependent on particle
size and range, these parameters should be made as consistent as possible if
quantitative data are needed.
Diffuse reflectance can be used with dispersive IR instruments, however
the low signal level and the poor sensitivity limits the range of applications.
The advent of FTIR spectroscopy has made the technique more accessible.
The samples must be highly scattering, that is, the optical depth, f!P, must be
larger than the geometrically average path length, f!D, through the sample.
For scattering samples f!D is a function of particle size. Light impinging on
an opaque, nonscattering sample is completely absorbed, except for a small
reflected component (rv 7% at normal incidence). If the surface is rough, a
diffuse reflectance accessory (rv 20% efficient) is required to collect this small
amount of light. The great advantage of DRIFT is the short preparation time
and the ability to obtain a strong spectrum with less than 100 f!g of sample
in a suitable matrix. Diffuse Reflectance has been successfully applied to
finely powdered samples, however it is not suitable for samples in the form of
coarse powders, granules or intractable lumps. High pressure and tempera-
ture DRIFT cells are now available for studying in situ catalytic reactions at
surfaces. Samples may also be used without being mixed with a matrix pow-
der, but considerable distortions in relative peak heights within the spectrum
may occur if the diffuse reflectance accessory does not completely eliminate
the specularly reflected radiation.
A successful application of this technique is ex-situ investigation of the
adsorption of the flotation collector, amylxanthate, on the mineral pyrite [8].
For the more interesting and relevant in-situ investigations this method can-
not be used and the ATR method is the method of choice (see next section).
Figure 8.2 shows the DRIFT spectra of pyrite after wet grinding at natural
pH (5.1) and adsorption of the collector amylxanthate. The sample, of which
50% was less than 7 f!m, was dispersed in KBf. Above monolayer coverage
bands appear at 1026cm- 1 (CS stretching mode) 1259-1261cm- 1 (C-O-C
8 Vibrational Spectroscopy of Surfaces 209
0.2 2.51O- 3 M
5.851O- 5 M
3.60
9 51O- 3 M
1.191O-3 M
5.90
0.0
h 1 1O-2M
4.161O-3 M
7.12
WAVENUMBERS
Fig. 8.2. Diffuse reflectance spetra of pyrite after wet grinding at natural pH (5.1)
and adsorption of amylxanthate with a conditioning time of 10 min. For each spec-
trum are given the initial and final concentration, and the statistical surface cover-
age
stretching in dixanthogen) and bands due the hydrocarbon chain at 1347 and
1465 cm -1. The bands at 1746 and 1716 cm -1 indicate the presence in the
adsorbed layer of the dimer of amylmonothiocarbonate from an oxydegrada-
tion reaction. The bands between 600 and 900cm- 1 are due to the oxidation
products on pyrite surface after grinding and before contact with the xanthate
solutions.
dp
A . 2 e-
= -(sm n~1)1/2 (8.3)
27f
where d p = depth at which the intensity is reduced to 1/e of its original
value. e is the angle of incidence and n21 the ratio of the refractive indices of
210 R.L. Frost and N.K. Roberts
"Y&al (0,)
sample (n 1 )
~-L t
Fig. 8.3. Schematic of the atten-
uated total internal reflection pro-
dp cess
the sample (n2) and the internal reflection element (nl). Consequently the
penetration is greater at longer wavelengths. Over the mid-IR range (2.5-
25 !lm) there is a factor of 10 difference in sampling depth. Furthermore
e
as gets larger, dp gets smaller and approaches 0.1 A at grazing incidence
(e = 90°) for high index media. Theoretically d p is infinitely large at the
critical angle. It is not possible to vary the angle continuously and several
elements of fixed angle, usually 30°,45° and 60°, are employed, provided they
do not exceed the critical angles. The sensitivity of the ATR technique can be
markedly increased by using multiple internal reflection elements. A typical
crystal 50 mm long with its end faces cut at the crystal angle can produce
between 20 and 30 reflections. If l is the crystal length and d its thickness,
the number of reflections is given by
N = lldcote. (8.4)
Table 8.1. Penetration depths (d p ) in [lm for polyethylene for various ATR ele-
ments
Wavenumber [em-I]
ATR element Angle 2000 1700 1000 500
KRS-5 45 1.00 1.18 2.00 4.00
KRS-5 60 0.55 0.65 1.11 2.22
Ge 30 0.60 0.71 1.20 2.40
Ge 45 0.33 0.39 0.66 1.32
Ge 60 0.25 0.30 0.51 1.02
8 Vibrational Spectroscopy of Surfaces 211
ATR spectroscopy can therefore study the surfaces of samples in the range
of a fraction of a !-lm to several !-lm provided intimate and uniform contact is
maintained with the ATR element. It is particularly important to maintain
uniformity of contact if further data processing such as spectral subtraction is
to be made. A torque wrench to press the sample against the ATR crystal ele-
ment should be used to obtain reproducible contact. The surfaces of the ATR
element should be clean and highly polished. For the best results the sample
should be pliable and have a smooth surface. In practice, a spectrum is run of
the ATR element alone then in contact with the sample. The latter spectrum
is then ratioed to the former, since FTIR instruments are single beam in-
struments, producing the spectrum of the surface of the sample alone. Depth
profiling of surfaces is possible by varying the crystal material and the angle
of incidence. An interesting comparison of ATR and PAS depth profiling of
some biocompatible polymer surfaces showed that PAS was more sensitive
to surface impurities and segregation [11]. The outstanding advantage of the
FTIR-ATR spectroscopy is the ability to study the solid/aqueous solution in-
terface, previously inaccessible to dispersive IR spectrometers because water
is a strong IR absorber. Two examples will suffice to show its capability.
The first example is protein adsorption on foreign surfaces from whole
blood. Interactions between flowing blood and a foreign surface are of great
practical significance when medical polymers such as those used in heart
valves, indwelling catheters, dialysis membranes, vessel grafts, and other ar-
tificial organs are implanted in the body. Most typical polymers in contact
with blood give rise to a number of complex reactions culminating in a blood
clot (thrombus). FTIR-ATR spectroscopy has contributed significantly to our
understanding of clot formation [12,13]. Since FTIR spectra can be collected
in fractions of a second, the kinetics of the process can be readily studied.
Figure 8.4 contains the spectrum of flowing blood, a reference saline blank,
Flowing blood
Saline
Subtraction
Additional
expansion
and the spectrum obtained by subtracting the latter from the former. Figure
8.5 is a continuation of Fig. 8.4, showing additional expansion of the sub-
traction spectrum. These spectra were obtained in 0.8 s. The presence of the
amide I (1650cm- 1 ) and amide II (1550 cm- 1 ) peaks due to proteins so close
to the strong OH binding vibration of water (1640cm- 1 ) shows that suc-
cessful spectral subtraction can be obtained in less than 1 s close to strong
interfering peaks. Figure 8.6 shows the total amount of protein adsorbed, as
measured by the amide II peak (1550cm- 1 ), as a function of time. Careful
analysis of the spectrum below 1550cm- 1 allows a more detailed eludica-
tion of the proteins adsorbed. The ATR crystal may be bare, as in this case,
or coated with a very thin film of polymer (1 OO-lOOOA) for a study of its
biocompatibility.
The second application of FTIR-ATR spectroscopy is the adsorption of
flotation collectors on mineral surfaces. FTIR-ATR spectroscopy is partic-
ularly suitable for studying the in situ adsorption of flotation collectors on
mineral surfaces. Previously only ex situ methods were available for studying
the process, i.e. the mineral with adsorbed collector was removed from the
aqueous environment, dried and the spectrum recorded. However, virtually
nothing is known about the effect of removing the water from the sample.
FTIR-ATR spectroscopy allows ready subtraction of the water and mineral
spectra to leave only the spectrum of the adsorbate in situ [14]. The exper-
imental procedure is as follows. Firstly, the mineral is vacuum evaporated
onto an ATR element to produce a thin film (rv 0.05 !lm). A word of caution
is required here. It is important to be sure that the mineral is unchanged
during evaporation. The term "thin film" is a misnomer, as it is known that
the "thin film" consists of very small crystallites. Evaporation of tin (IV)
oxide produces a variety of oxides. Fortunately in this case the lower oxides
8 Vibrational Spectroscopy of Surfaces 213
SHEEP
:::i
120.0
---
<i
::: 100.0
~
~
~ 80.0
w
f-
~
§1 60.0
;'Ii
40.0
20.0
0.0 +1- - r - - - - - , - - - , - - , - - - - - r - - - , - - r
0.0 7.0 14.0 21.0 28.0 35.0 42.0 49.0
TIME IN MINUTES
Fig.8.6. Plot of amide II (1550cm- 1 ) band intensity versus time of blood flow
showing total amount of protein adsorbed in the time period indicated. Spectra
collected every 5 s
may be converted to tin IV) oxide by annealing in oxygen. Fluorite and other
minerals have been successfully produced as "thin films". If thin films cannot
be obtained it may be possible to obtain spectra on suspensions of the min-
eral, e.g. Goethite, or use a transmission technique if the mineral is infrared
transparent in the appropriate region, for example sphalerite. Secondly, a
spectrum is run of the ATR element coated with the mineral film in the pres-
ence of water. The experimental set-up is shown in Fig. 8.7. The water is then
i
collector in solution
'0 '0 b adsorbed collector
? cassiterite coating
·nfrared beam
ATR element
'l. '0
b
Fig. 8.7. Schematic representation of the use of ATR method to study collector
adsorption
214 R.L. Frost and N.K. Roberts
2.3189
DIGILAB FTS-IMX
2.1165
1.9142
1.7110
CfJ
f-
Z
~
>-
a: 1.5095
«
a:
f-
Ol
a:
«
1.3071
1.1040
0.9025
0.7001
1600 1200 850
WAVENUMBERS
Fig. 8.8. FTIR-ATR spectrum of 300 ppm SPA adsorbed on a thin film (rv 0.05 [.lm)
of tin (IV) oxide at pH 4.5 (100 scans)
pumped out and replaced with the collector solution. Spectra may then be
run at appropriate intervals and difference spectra obtained showing the ad-
sorbate on the mineral surface. Figure 8.8 shows the result of the adsorption
of styrene phosphoric acid (SPA) on cassiterite. Figure 8.9 is the spectrum of
styrene phosphoric acid. Comparison with the corresponding SPA complexes
shows a close resemblance between the adsorbate species and the correspond-
ing tin (IV) and titanium (IV) complexes. On adsorption of the collector the
bands due to the v(P=O), vas(P-O) and vs(P-O) vibrations of the phos-
phonic acid group are replaced by a complex broad band due to the resonance
stabilized phospho nate group. The kinetics of the adsorption process may be
followed in situ along with the effect of such variables as pH, temperature
and ionic composition. At the same time spectra of the solution species may
be obtained for comparison with the adsorbed species.
8 Vibrational Spectroscopy of Surfaces 215
34+---______+_-----------4-----------r
DIGILAB FTS-IMX
29
24
19
o
o
w
~14
[;3
cc
oC/)
tIl
« 9
-1
o
Ho11./
Sn
I R-~-O, 1 ./
I Sn
HO: 1 ' "
OH HO /1 '"
monoanion cassiterite
of SPA surface !fast
R-P~
ft . . O, Sn1 ./
"'0"'1 '"
bidentate surface
complex of SPA
216 R.L. Frost and N.K. Roberts
•j ~
IR SOURCE
DATA
SYSTEM
electric;;:'
signal RECORDER
IR beam
_1:[electrical
signal
[~~ i.oo="k41Wl
INTERFEROMETER PAS SAMPLE CELL
KBrWINDOW
GAS
~
IMODULATED
IR BEAM
TO DATA SYSTEM
I
D
SAMPLE PRE-
AMPLIFIER
PHOTOACOUSTIC SIGNAL
POWER SUPPLY
W
...J
<{
()
rJJ
>-
a:
<{
a:
f-
a;
a:
~
~
~ 0
w
f-
~
E 7.0
=>.
I 6.0
f-
(!)
zw 5.0
...J
Z
0 4.0
iii
=>
LL
LL 3.0
0
...J
«
::;;
2.0
a:
w 1.0
I
f-
Fig. 8.12. Thermal diffusion length
~oo ~m ~~ ~oo ~oo 1~
of cellulose at 1730 cm -1 versus the
OPTICAL VELOCITY (em 5- 1 ) optical velocity of the interferometer
properties. Figure 8.11 shows the PAS-FTIR spectra of cotton yarn sized
with a polyurethane at different interferometer velocities [15].
The intensity of the peak at 1246 cm -1, due to the polyurethane sizing
agent, increases dramatically as the optical velocity changes from 0.235 cm S-l
to 1.119 cm s-1, indicating that the agent is concentrated at the surface. Fig-
ure 8.12 shows the thermal diffusion length of cellulose at 1730 cm S-l for
different optical velocities. Hence it may be concluded that the polyurethane
sizing agent is more concentrated in the top 2.8!tm and gradually decreases
to a depth of 6.8 !tm.
8 Vibrational Spectroscopy of Surfaces 219
Applications. IES has many and varied applications. To illustrate the use
of IES, two examples from firstly the synthesis of a molecular sieve (a pillared
montmorillonite) and secondly an example of a mineral surface of industrial
importance are used. A typical set of lES data is shown in Figure 8.13 which
displays the infrared spectra of an Al 13 pillared montmorillonite from 200 to
750°C at 50°C intervals. To show the advantages of the technique, the letters
220 R.L. Frost and N.K. Roberts
750°C
show the types of changes that are observed upon thermal treatment. Spectral
changes at [A] report the dehydroxylation of the montmorillonite, at [B] the
combustion of organics in the sample and at [C] the changes in the structure
as the clay is pillared. A second example is the thermal treatment of natural
gibbsite (Figure 8.14). Clearly the processes of dehydration, dehydroxylation,
combustion of organics and the phase change of gibbsite to an alumina phase
may be observed.
and provided ready access to the extensive data handling facilities that are
available with a commercial FT-IR spectrometer.
FTRS spectra may be obtained using several instruments: e.g. the Perkin-
Elmer 2000 series Fourier Transform spectrometer fitted with a Raman ac-
cessory or the Biorad series 2 FTIR spectrometer. The Perkin-Elmer 2000
series FTIR spectrometer equipped with a Raman accessory comprises a
Spectron Laser Systems SL301 Nd-YAG laser operating at a wavelength of
1064nm, and a Raman sampling compartment incorporating 180 0 optics. The
FT-IR spectrometer contains a quartz beam splitter capable of covering the
spectral range 15,000-4000cm- l . The Raman detector is a highly sensitive
indium-gallium-arsenide detector and is operated at room temperature. Un-
der these conditions Raman shifts would be observed in the spectral range
4000-150 cm -1. Spectra are corrected for instrumental function and detector
response. Raman spectra may also be obtained using a Biorad series 2 FTIR
spectrometer equipped with a Raman accessory comprising a Spectraphysics
T10-1064C Nd-YAG diode laser operating at 1064nm. Measurement times
of between 0.5 and 2 hours are used to collect the Raman spectra with a
signal to noise ratio of better than 100/1 at a resolution of 2 cm -1. A laser
power of 100 m W is used. This power is low enough to prevent damage to
most surfaces, but was found to be sufficient to produce quality spectra in
a reasonable time. No significant heating, as may be evidenced by the lack
of thermoluminescent background, is observed. Raman spectra are collected
as single beam spectra and are corrected for instrumental effects. Spectral
bands are ratioed to the instrument profile function determined by recording
the spectrum of a calibrated grey body, calibrated at the National Standards
Laboratory at C.S.I.R.O., Linfield, Sydney, Australia.
(e)
(d)
(e)
(b)
(a)
3550 3570 3590 3610 3630 3650 3670 3690 3710 3730 3750
Wavenumber/em- 1
Fig. 8.15. Raman spectra of the hydroxyl stretching region of (a) chrysotile, (b)
montmorillonite, (c) vermiculite, (d) tremolite, (e) pyrophllite
Fig. 8.16. Raman spectra of the low frequency region of (a) chrysotile, (b) ortho-
clase, (c) pyrophllite, (d) montmorillonite, (e) muscovite, (f) tremolite, (g) vermi-
culite
224 R.L. Frost and N.K. Roberts
~
(jj
c
Q)
E
c
co
E
co
a:
Fig. 8.17. Raman spectra ofthe hydroxyl stretching region of (a) low defect kaolin-
ite, (b) high defect kaolinite, (c) DMSO intercalated low defect kaolinite, (d) DMSO
intercalated high defect kaolinite
,::-
-en
c
.$
E
c
'E"
'"
a:
Fig. 8.18. Raman spectra of the hydroxyl stretching region of (a) low defect kaoli-
nite (ldk), (b) ldk intercalated with KAc at 1 bar and 25°C (c) at 1 bar and 100°C
(d) at 2 bars and 200°C
may be measured. This has particular application to the study of sols and gels.
Because water is a strong infrared absorber, infrared spectra from components
of aqueous systems are very difficult to obtain. Further silicate glasses have
weak Raman spectra and the Raman spectra of aqueous sols may be measured
in situ in a glass sample container [26,27]. The spatial resolution of a Raman
microprobe is about 1 [lm. This may be compared with infrared microscopy
where because of the wavelength of light used, the spatial resolution at best
is 20 [lm. The full potential application of Raman spectroscopy to the study
of surfaces is yet to be realised.
References
S.H. Sie
9.1 Introduction
charge build-up, which can affect the result. Typically a thin layer of carbon
is applied; Au is often used as well. The choice is determined to minimize
interference with the analyzed material. In RBS, the element identification
is poor for high Z due to limits in the detector resolution, and the sensi-
tivity for light element detection is also limited, especially in the presence
of high Z matrix. NRA is limited to only those elements exhibiting strong
nuclear reactions, occurring mainly in light nuclei. Interpretation of data can
be ambiguous when the sample is not homogeneous, for instance when pre-
cipitation, segregation or clustering of impurities in the matrix occur. The
surface texture of the sample must be smooth for accurate measurements.
In RBS surface roughness would affect the result, in that in effect the target
presents micro-surfaces at various angles to the incident beam, thus smear-
ing the results. This can in fact be used to measure surface roughness. If
microbeams are available, surfaces with rough texture can still be analyzed if
the undulation is larger than the microbeam size. In NRA, surface roughness
will degrade the depth resolution at the surface, and non-flatness can result
in error due to angular distribution effects.
The measurements are usually carried out in vacuum. While high vac-
uum (pressure in the 10- 5 to 10- 6 mbar range) is quite adequate for typical
measurements, lasting a few minutes, an ultra high vacuum may be required
for prolonged measurement on the same spot to prevent carbon build-up on
the target due to break-up of residual hydrocarbons in the chamber under
beam bombardment. This would affect NRA in shifting the beam energy, and
may cause interference in RBS when measurements of light elements (e.g. 0)
are involved. External beam measurements are possible, where the beam is
brought out of the accelerator through a thin window or through differen-
tially pumped aperture. In such case the RBS detector is usually operated in
an inert gas atmosphere to minimize background problems. At any rate the
resolution of the detector suffers, and this kind of measurements are usually
conducted only for special cases, e.g. in-situ measurements of large objects.
Table 9.1 summarizes the salient features of the methods to be used as a
rough guide. Special cases can extend the general limits shown. The usefulness
of the methods is enhanced with microbeams, which for energetic beams are
typically limited to a few microns in diameter for viable beam intensities.
In the following, basic principles of the methods are presented and ex-
amples of applications to both simple and complex samples are given as
illustrations. The scope is limited to applications in stoichiometric analysis,
and thus the channelling method will not be discussed in detail. For more
detailed treatment of the subject, and all other IBA methods, the reader
can refer to a number of excellent handbooks, the latest being 'Handbook
of Modern Ion Beam Materials Analysis' edited by Tesmer and Nastasi [1].
Proceedings of conferences based on the theme of ion beam analysis are also
excellent resource material and usually contain the repertoire of the applica-
tions in various disciplines. One of the most well known series is the biennial
9 Rutherford Backscattering Spectrometry 231
RBS NRA
International Ion Beam Analysis conference, now approaching its 20th year.
The latest proceedings published as an issue of Nuclear Instruments and
Methods is given in [2].
9.2 Principles
Both RBS and NRA involve measurements of energy and flux of the radia-
tions resulting from the interaction of the ions and the atoms of the material
under analysis. In RBS, they are the beam projectiles themselves after being
scattered by the target; and in NRA they could be gamma-rays or fragments
of the reacting nuclei. Unless stated otherwise the general nomenclature is as
follow:
Zl, m2 - ion beam projectile atomic number and mass (in amu)
Z2, m2 - target atomic number and mass
E - incident ion beam energy (MeV)
q - charge state of the incident ions
e - electronic charge (1.6 x 10- 19 Coulomb)
I - ion beam current (Amperes)
F - ion beam flux = J I dt/eq
N - target molecular density (molecules/cm 3 )
s - solid angle of the detector (steradians) subtended
at the target.
dO"
dY=N-bE / -dE (9.1)
dw dx
where:
where:
To obtain the yield for analysis of the data, this basic formula can be inte-
grated.
Energetic ions lose energy mainly through electronic excitation of the stop-
ping medium atoms, which does not involve change in direction, referred to
as electronic stopping. At low velocities « cj137), nuclear collisions become
more frequent, giving rise to an additional energy loss mechanism known
as nuclear stopping, usually involving many changes of directions. But this
occurs towards the end of the ion trajectory, and thus will not affect or be
observed in the data obtained. For instance for a Sn target, the nuclear stop-
ping contribution (estimated theoretically [3]) to the total is only about 1%
for 1 MeV proton rising to ",15% at 0.5 MeV. Nuclear stopping correction
can be significant and must be included in determining the range R for low
energy particles. The corrections to R range from rv 1O~4 for protons in Be
to ",0.2 for the heaviest element in heavy element medium.
Extensive measurements over the past two decades have established the
values for the stopping power of protons and alpha particles in various ele-
mental material. The data are not as extensive for heavy ions, but these can
be related to the stopping power of protons by a scaling factor based on the
9 Rutherford Backscattering Spectrometry 233
9.2.2 Straggling
The statistical nature of the energy loss at low velocities involving many
binary collisions gives rise to fluctuations in the energy loss, resulting in the
straggling effect. This effect progressively broadens the energy distribution
of the incident particle as it loses energy, degrading the depth resolution in
RBS and decreasing the sensitivity of NRA. A theoretical estimate of this
effect was first given by Bohr [8]:
(9.4)
where Q B is the width of the energy distribution at one standard deviation
(corresponding to FWHM/2.355), and dR is the target thickness. Another
theoretical estimate [9] incidated a smaller effect than Bohr's theory, partic-
ularly in the low energy region and for heavier elements. It is instructive to
use Bohr's estimate to determine whether straggling should be considered
when information at depths of a few microns is required. For instance for Sn,
at 1mg/cm2 depth (1.74[lm) the Bohr estimate gives 7.5keV FWHM. The
total effect on backscattered particles is the sum in quadrature of this and
another for the outgoing path, and the detector resolution. For a detector
resolution of 16 ke V, the resulting effect is a broadening of the resolution to
19keV.
234 S.H. Sie
In a typical RBS analysis the scattered beam particles are usually detected at
the "back angles" (> 90° with respect to the beam direction) using a surface
barrier detector. The energy distribution and the yield contain respectively
the identity and of the concentration of the target nuclei. In this method, h
and 12 in (9.2) are unity, as is the detection efficiency e f. However the energy
of the scattered particle is attenuated by the target matrix. The particles
scattered from a layer thus have an energy uniquely determined by the depth
of the layer, and the species of the scattering atom. The yield is determined
by the Rutherford scattering cross section:
(9.6)
E ml +m2
The angular dependence of K is strongest near 90° and the contrast for
different m2 is higher for larger ml, as can be seen in Fig. 9.1, calculated for
Hand 4He.
Consider now a scattering from a thick target set at an angle Bl with
respect to the beam, and detection at angle B2 (Fig. 9.2). At a depth dt,
the incident energy Eo is attenuated by dE = (dEjdx)Eo dtj cosB l , and the
scattered energy is K' E" = K' (Eo - dE) where K' is the kinematic factor
at E". The scattered particle energy is further attenuated on the way out by
dE' = (dEjdx)EII dtj COS(B2). This particle is separated in energy from that
scattered from the surface by bE. For infinitesimal dt a quantity S can be
defined:
bE = Sdt (9.7)
where
The quantity S reflects the total stopping power for incident and scattered
particles, and can be used to define the depth scale. This general expression
9 Rutherford Backscattering Spectrometry 235
Cu Au
1·0 \ ~u Sn~
o·g
, " ~" , '~----~ __
!\ \- --~~Sn
\ ,,---- Si
0·8 \
\
'"
Cu
\ '
0·7 \ "
\ " Be
\ "
0·6 \
\
K \
0·5 \
\
0-4
\
,
""
0·3
--H
"" , ,
0·2 _ _ _ 4He
--~
0·1
O~----~30~----~6~O~----~9~O------1~2~O----~1~50~----1~80
e (degrees)
Fig. 9.1. Kinematic factors for Hand 4He beams on selected targets as a function
of angle
E"
E'
is the basis for manual analysis of the spectrum, which is practical for thin
targets.
For thick and complex targets numerical simulation of spectrum is nec-
essary in the analysis. While many laboratories tend to produce their own
RBS simulation programs, some are available from a number of authors,
e.g. [10,11]. In calculating the simulated spectrum the response of the sur-
face barrier detector must be taken into account. The response of the de-
tector to monoenergetic particles appears as a Gaussian distribution in an
energy dispersive pulse height analyzer with the width dominated mainly by
the electronic noise of the system. Incomplete charge collection and multiple
scattering effects in the detector produce a continuum extending from the
full energy peak to low energies, which could amount to as much as 10% of
the total counts.
9.3.2 Examples
Figure 9.3 shows a simulation spectrum for a target consisting of a thin
layer of Sn on a thick eu substrate, analyzed using a 2.0MeV 4He beam.
9 Rutherford Backscattering Spectrometry 237
,200
.
0; 0.2
c:
c: • 0.4
.t:. x 0.8
~ 600
:. 1.6
'E::l 3.2
0
U • 6.4
400 12.8
,. - , ). 10.'
:., , . ...
X ~.
200 )( ... _ x
. ....sG iti·.:
!II: .....
<
! • I ...
: :., ~
\."J \
"
500 ' 000 1500 2000 2500
The peak height would increase with target thickness until the maximum
height H is reached. Further increase results in the increase of the width of
the peak with a trapezoidal shape. The Sn peak is partially separated from
the spectrum from Cu through kinematic effect. The Cu spectrum shows a
high energy edge corresponding to the interface layer, and extends to the
low energy region showing the characteristic shape of the spectrum from a
thick target. The depth scale for the Sn peak can be determined from the
dispersion of the spectrum. If dE is the energy per channel of the spectrum,
then the height of the spectrum H at the edge, corresponding to the surface
is given by:
H = F (da)' sN dE
dw Eo So
238 S.H. Sie
The channelling effect: when the target is a single crystal, the backscatter
yield will be reduced when the incident beam is aligned with the major (low
index) axes [12]. This phenomenon can be used to probe the site of an im-
purity atom. For interstitial sites, the backscatter yield will increase again
towards the random direction value.
Resonant scattering: for a number of light nuclei, e.g. 160, 12C, a number
of resonances (marked by increase of yield) occur for certain energies. Appli-
cation of this is similar to that which will be discussed further in the NRA
section below.
9 Rutherford Backscattering Spectrometry 239
1400
(a)
Simulated RBS Spectra
1200 0.5 mg/cm2 target
E(a) = 2 MeV Q = 50 ~lC
1000
0 =160° Q =6msr ·· ""
PI
W
x eo
· co
0;
"" 800
r.
'"
.s::;
+ Au
~
·
a
c: 600 s-
U
"0 o Q,
·
Co
400 '1\
x Al
200
0
0 500 1000 1500 2000 2500
Backscallered Particle Energy (keV)
1600
(b)
Simulated RBS Spectra
1400 1 mglcrri' complex
target on Cu substrate
E(a ) = 2 MeV a
= 50 )JC
a=
1200
1600 Q =6 msr
0; 1000 . Pb
""CO
.s::; Cu
o SnPb3
• SnPb
800
~
c: > Sn3Pb
U
"0 800
Sn
400 ':-a
.~
200 "
f>
0
0 500 1000 1500 2000 2500
Backscattered Particle Energy (keV)
Fig.9.4. (a) Calculated RBS spectra from a series of targets of 0.5 mg/cm 2 layer
of various elemental material. The spectra are computed for the same conditions
as in Fig. 9.3. The width of the spectra generally increases with lower Z. This is
mainly due to kinematic and stopping power effect . Occasionally this trend does
not hold, e.g. in the case of Ba. (b) Calculated RBS spectra from a series of targets
consisting of a Cu substrate with a 1 mg/cm 2 surface layer of Sn, Pb and their
various compounds. Note the variation of the height and width of the surface peak
as a function of the material
240 S.H. Sie
Forward recoil: in this technique, under bombardment with heavy ions, the
recoiling nuclei are detected in the forward direction. This provides some
depth profile information but it is usually limited to a depth of around 1 !lm
due to straggling effects [13].
15
10
~--~11~0~0--~1~3700~~~~~17~0~0---1~970~0~
Eproton (keV)
Fig. 9.5. Resonances in the reaction between a proton beam and a fluorine target,
appearing as peaks in the reaction yield, e.g. gamma rays, as a function of the
incident beam energy
9 Rutherford Backscattering Spectrometry 241
They are known as sub-Coulomb transfer reactions. The most common reac-
tions are capture reactions which show resonances corresponding to discrete
high energy states in the product (compound) nucleus. In proton capture
reactions, alpha particles are sometimes emitted, and can be used to detect
the resonance. These reactions occur at specific energies for specific elements,
and this is the feature exploited in NRA. At the resonant energy, the reaction
occurs at the surface, sampling the surface region for which the energy loss
of the beam corresponds to the resonance width. For higher beam energies
the reaction will occur deeper in the target where the resonance energy is
reached. Hence this method can be used to determine the concentrations of
the particular atom involved in the resonance as a function of depth, giving
the depth profile of the specific element. With typical resonance widths in
the few ke V range depth resolutions of a few nm are obtained.
For example, the resonances in the reaction between a proton beam and
a fluorine target can be seen in Fig. 9.5 as peaks in the reaction yield against
the incident beam energy. The criteria for selection of the reactions for NRA
applications are that the cross section should be large and that the radiation
can be detected readily. Gamma rays are commonly detected in this reaction.
9.4.1 Formalism
In the resonant NRA method the yield is obtained only at the resonance
energy, and from a layer corresponding to the natural width of the resonance.
The correction factor h (9.2) depends on the nature of the radiation; it is 1
for particles as long as they are not stopped within the target. For gamma
rays, the usual attenuation must be computed. The intensity of a gamma ray
after traversing a medium with a linear absorption coefficient fl' is given by:
(9.8)
where:
A more convenient variable is the areal density (g/cm 2 ), and the correspond-
ing coefficient is referred to as mass attenuation coefficient fl (= fl' / (!, where
(! is the density). For a compound medium, Bragg's rule of additivity applies.
Thus for a compound AnlBn2 ... the mass attenuation is given by:
9.4.3 Examples
Figure 9.6 shows the hydrogen profile from a set of samples of hydrogenated
amorphous silicon, obtained using the resonance at 6.412 Me V fluorine beam
energy, corresponding to the O.340MeV resonance (Fig. 9.5) when protons
are used on fluorine target [18]. In this particular case, the inverse of the
resonant reaction induced by protons is used, where now one uses the 19F
as the beam to detect hydrogen. The 6.13 MeV gamma rays emitted by the
reaction are detected in NaI detectors. The surface peaks on two of the sam-
ples indicate a depth resolution of 44 nm, corresponding to the resonance
width of 3 keV. The range of applicability is usually limited to a few I-1m,
because of increasing diffuseness of the beam with depth due to multiple
scattering. In the present case, the onset of another resonance corresponding
Depth (nm) Si
o 100 200 300 400
(a)
10 %H
I
,
"II
~ I I
'c I I
::l
g>- 10,000 I
: \
I
(b)
:0 I 13·7 % H
~ I
I
I
I
'0 I I
(ii I ,
'>' I , ....•....• 10·5 % H
OJ I ~ •.
() I ... ,
C
(1) I e' , \
\
§ 5,000
I
I \
\
,; ....
,,\
\
8 %H
(J)
OJ I ,,; I
I ...... _.... I
a::
I I
I I
I I
I I
, I \
I
I - __
0.
O~--__-=~ ______~___________ I__-L______ -_-_~
__-_-_-_-_-~~~
6·5 7·0 7·5
Fig. 9.6. A hydrogen profile of a selection of a-Si:H samples, obtained using the
resonance at 6.412 MeV fluorine bombarding energy. Two of the samples show sur-
face peaks which may arise from adsorbed moisture. The width of the surface peak
indicates the depth resolution of 44 nm obtained with this resonance. Note the non-
linearity of the hydrogen content scale due to change in the stopping power with
composition
244 S.H. Sie
A
0'51l0> IS,) 8 01>---0 Fig. 9.7. Carbon profiles (A
C ,....- - ... and B) from a series of
o .······4 samples of amorphous hydro-
genated Si-C compound, ob-
tained using the alpha reso-
nance at 4.26 MeV. The corre-
sponding backscattered particle
spectra (C) show the onset of
the resonance as well as that
due to oxygen at 3.036 MeV.
The oxygen arises from the
f " IN.V) glass substrate
B
22 r-~----,,-----.-----'------r-----'
20 ( ..--..
F o--~
18 H ...----0,
0 ' -,,--
16
1 o- ..a-o
~lr.
~ 12
e '0
c
~ 8
8
,. V;,
f "IN.V)
C ,- . - . - --,. -
lei
Ea =3.99MeV
...
It" \ Ea = 2.3 MeV
\
I,
,
'." Si
Ea = 4.27 MeV
}lC e-
Channel number
9 Rutherford Backscattering Spectrometry 245
to the 0.484 MeV in Fig. 9.5, limits the applicability of the reaction above to
a depth of only 1 ~m. The sensitivity of this method is limited by background
to around 100 ppm. This can be improved when anti-coincidence techniques
are applied to reduce the cosmic ray background. A detection limit as low as
10ppm has been reported [19].
The resonance yield can be calibrated against a hydrogen-rich target of
known composition. Organic polymers (e.g. kapton, polyethylene, mylar) are
convenient but they are easily damaged by the beam mainly due to thermal
effects. Care must be taken in the calibration run to minimize such effects
by running very low level and diffuse beam on the target. The damage can
be observed by monitoring the yield as a function of time or beam dosage,
and the true yield can be obtained by extrapolation to zero time or beam
dosage. More stable standards can be obtained using implanted material [18]
or hydrogenated amorphous silicon.
Sub-Coulomb barrier resonances in scattering of protons and alpha par-
ticles on some light nuclei can be utilized in this method, where the detected
radiation is the scattered particle. The main advantage are the somewhat
longer range of applicability, and higher efficiency of detection afforded by
particle detectors [h = 1 in (9.2)]. As an example, Fig. 9.7 shows the depth
profiles of carbon obtained from several samples of thin layers of hydrogenated
amorphous Si-C compound using the 4.26 MeV resonance of alpha scattering
on carbon, and the corresponding spectra obtained at various bombarding
energies showing the onset of the resonance [21]. The figure also shows the
resonance of oxygen from the glass substrate, at 3.036 MeV. These particular
resonances are peaked at 180 0 with respect to the beam direction, hence the
ideal measurement geometry is consistent with the standard RBS require-
ments.
9.5 Summary
Two of the suite of ion beam analysis techniques, namely RBS and NRA,
and variations thereof, are versatile and powerful non-destructive methods for
characterizing the elemental composition and structure of the surface region,
extending from a few hundred atomic layers to several microns. In this respect
they complement other surface analytical methods which probe fewer atomic
layers of the surface, and their chemical states. The chemical insensitivity
of the forces governing the interaction between the probing beam and the
sample results in methods which are highly quantitative, being amenable to
straightforward calibration procedures.
References
1. J.R, Tesmer, M. Nastasi eds. Handbook of Modern Ion Beam Materials Analysis
Materials Research Society, Pittsburgh, Pennsylvania, 1995.
2. Nucl. Instr. and Meth. in Phys. Res. B136-138 (1998)
3. J. Lindhard, M. Scharff, H.E. Schiott: Kgl. Danske Videnskab. Selskab., Mat.
Fys. Medd. 33, 14 (1963)
4. L.C. Northcliffe, R.F. Schilling: Nuclear Data Tables A7, no. 3-4 (1970)
5. H.H. Andersen, J.F. Ziegler: Hydrogen Stopping Powers and Ranges in All Ele-
ments (Plenum, New York 1977) and related titles in the series "The Stopping
and Ranges oflons in Matter", ed. by J.F. Ziegler (Pergamon, New York 1980)
6. W.H. Bragg, R. Kleeman: Phil. Mag. 10, 318 (1905)
7. J.F. Ziegler, J.M. Manoyan: Nucl. Instr. Meth. in Phys. Res. B35, 215 (1988)
8. N. Bohr: Kgl. Danske Videnskab. Selskab., Mat. Fys. Medd. 18 no.8 (1948)
9. W.K. Chu: Phys. Rev. A13, 2057 (1976); also in [1, page 1]
10. P.A. Saunders, J.F. Ziegler: Nucl. Instr. Meth. in Phys. Res. 218, 67 (1983)
11. L.R. Doolittle: Nucl. Instr. Meth. B9, 344 (1985)
12. B.R. Appleton, G. Foti: In Ref. [9.1] p.67
13. B.L. Doyle, P.S. Peecy: Appl. Phys. Lett. 34, 811 (1979)
14. J.A. Davies, J.S. Forster, S.R. Walker, Nucl. Instr. and Meth. in Phys. Res.
B136-138 (1998) 594
15. W. Assmann, J.A. Davies, G. Dollinger, J.S. Forster, H. Huber, Th. Reichelt,
R. Siegele, Nucl. Instr. and Meth. in Phys. Res. Bll8 (1996) 242
16. J.H. Hubbell: Atomic Data 3 no. 3 (1971)
17. R. Theisen, D. Vollath: Tables of X-Ray Mass Attenuation Coefficients
(Stahleisen, Dusseldorf 1967)
18. S.H. Sie, D.R. MacKenzie, G.B. Smith, C.G. Ryan: Nucl. Instr. Meth. in Phys.
Res. B15, 525 (1986)
19. H. Damjantschitsch, M. Weiser, G. Heusser, S. Kalbitzer, H. Mannsperger:
Nucl. Instr. Meth. in Phys. Res. 218, 129 (1983)
20. J.F. Ziegler et al.: Nucl. Instr. Meth. 149, 19 (1978)
21. S.H. Sie, D.R. MacKenzie, G.B. Smith, C.G. Ryan: Nucl. Instr. Meth. in Phys.
Res. B15, 632 (1986)
10 Materials Characterization
by Scanned Probe Analysis
S. Myhra
10.1 Introduction
The STM (see list of acronyms at end of book) was invented by Binnig et al.
in 1982 [1]. The two main protagonists, G. Binnig and H. Rohrer, were sub-
sequently awarded the Nobel Prize for physics. Thus began the age of SPM.
Much of the early development and excitement generated by the unequivocal
demonstration of spatial resolution on the scale of the single atom and of
local spectroscopies have been described in the literature [2-4]. The ratio-
nale for including a chapter on SPM in a book on surface analysis can be
inferred from Fig. 10.1 (adapted from Rohrer [5]). The impact of SPM in the
broad field of surface and interface science and technology can be illustrated
by its prominence at a conference in Birmingham in September 1998, which
brought together a representative cross-section of the international surface
science community through the 14th International Vacuum Congress, 10th
International Conference on Solid Surfaces, 5th International Conference on
Nanometer-scale Science and Technology and 10th International Coference on
Quantitative Surface Analysis. Of some 1350 invited and contributed papers
approximately 20% were based substantially on SPM techniques and method-
ologies, while the corresponding indices for the 'traditional' techniques and
'other' were 34% and 46%, respectively.
The demonstration of the 7 x 7 reconstruction of the Si (111) surface
was arguably the result which put STM on the scientific map. Irrefutable
evidence for single-atom real-space resolution, or even more convincingly for
single missing atoms and atomically resolved I-V spectroscopy, were also
significant events [6].
The second member of the SPM family, the AFM, was demonstrated in
1986 [7]. Subsequently, other tip-to-surface interactions (e.g. magnetostatic,
electrostatic, thermal radiation, etc.) have given rise to additional members.
As well, most of the original techniques have had offsprings by way of vari-
ations on the underlying theme (e.g. contact, non-contact, intermittent con-
tact, lateral force and force-vs-distance modes for the AFM).
In hindsight it may be that the invention of the STM was not, in itself,
the most significant aspect of the work of the group at IBM. Rather it was
the thinking which led to the STM, and its technical implementation, which
paved the way for a plethora of scanned probe techniques. For purposes of
248 S. Myhra
Dimension
mm
macro- SUrface!
molecules Analysis
nm
Computation
atom
SPM
I
SFM
III
SNOM SKP STM STS SThM
I
ECM
I
MFM SCMlEFM
I
AFM
F-d LFM
I
Contact Non-cont. Intermittent
Contact
Fig. 10.2. The SPM family tree in an abbreviated form. The acronyms are defined
at the end of the chapter
10 Materials Characterization by Scanned Probe Analysis 249
Table 10.1. SPM and traditional techniques - complementary strengths and weak-
nesses (adapted from [8])
Requirement SPM Other Techniques
Surface nanostructure STM/Contact AFM LEED (reciprocal space)
real space FIM and ISS (real space)
Morphology /topography STM/AFM SAM/SEM/ion microscopy
xlOOOOOO xlO 000
good z-resolution poor z-resolution
Grain structure STM/AFM SAM/SEM/Imaging SIMS
Phase structure SAM/Imaging XPS/SIMS
Electronic structure
Valence/conduction states STS (local) UPS/IPES/EELS (non-local)
Core states XPS/AES
Composition XPS/AES/ ...
Mechanical properties AFM/F-d N ana-indentation
Frictional properties LFM/F-d
Optical properties SNOM Optical microscopy
x 100000 x 1000
good z-resolution poor z-resolution
Magnetic structure MFM ('local')
Thermal properties SThM Scanning calorimetry
Surface conductivity SCM (local) Differential charging (non-local)
'Buried' information Profiling XPS/ AES/SIMS/ ...
In situ capabilities Vac/gas/liquid Usually URV
In situ reaction dynamics Potentially very good Generally limited
V TIP
D INTERACTION
SURFACE
the intrinsic versatility and inherent richness of the SPM system. In prac-
tice, exploitation of the system is predicated on definitions and acceptance
of sufficiency of knowledge. The consequences of the limitations regarding
sufficiency will be explored below. Also, it should be noted that the ter-
minology of 'microscop(e/y), implies that in the overwhelming majority of
instances the 'surface' is taken to be the unknown. Thus most of the discus-
sion of the SPM system will concentrate on the conventional microscop( e/y)
aspects/applications, but some attention will be given to other configura-
tions. The schematic description in Fig. 10.3 also makes it abundantly clear
that a resultant image must represent a complicated convolution of the above
three elements.
The vast majority of present SPM facilities are based on two generic types
of instruments.
Type II: Is based on operation within a DHV envelope, and tends to be single
technique, usually STM. Some recent systems do also support AFM in con-
tact and non-contact modes. Some special purpose DHV instruments may
incorporate hot/cold stages, transfer systems, specimen preparation cham-
bers, and various combinations of complementary surface/interface analysis
techniques. Such instruments are expensive, not user-friendly and relatively
inflexible. Their main utility is for applications where atomic or near-atomic
resolution is required, and where cleanliness of reactive surfaces is manda-
tory. They are not suited for rapid turn-around of specimens or for routine
analysis.
Type III: The gap between Types I and II is gradually being filled. A typical
Type III instrument retains the multi-mode functionality and flexibility of
the air-ambient instruments, while at the same time providing a controlled
10 Materials Characterization by Scanned Probe Analysis 253
10.4.1 STMjSTS
Electron tunnelling is the elementary process which accounts for the opera-
tion of STMjSTS. The approach adopted by Tersoff and Hamann [9] remain
an intuitive and useful tool for interpretation of data. Additional material
can be found in the review literature [5,10,11]. The energy level scheme of
a tunnel barrier is shown in Fig. 10.4. The respective wavefunctions outside
and inside the barrier are sinusoidal and exponential; the latter is particularly
relevant to the problem at han(i. The tunnel current is given by a summa-
tion over elastic, hence (E2 - Ed, tunnelling channels linking occupied states
Energy
! !
I I
I I
I I
Sample (1) Gap Tip (2)
I I
I I
r----
I I
I I
I
Evac- - - - -
---r --
I
I
$1 Evac
EF (1) --t $2
eVT
L __ EF (2)
Fig. 10.4. The energy level scheme of a tunnel barrier. Sample and tip are denoted
(1) and (2), respectively. Work function is 1>, and tunnel voltage is VT
254 S. Myhra
(as described by the Fermi-Dirac function f) on one side of the barrier with
unoccupied states (1 - f) on the other side
The two distributions of states are displaced by eVT, where VT is the tunnel
voltage. Figure 10.4 illustrates the case of tunnelling being two metals from
occupied states below EF to unoccupied states above E F . The matrix element
M 1 ,2 is the transfer-Hamiltonian of Bardeen; and is a weakly varying function.
A simple one-dimensional case of two identical free-electron metals with
identical work functions, ¢, will provide useful insight. Applying boundary
conditions, the wavefunctions in the gap may be written as
(10.2)
where the width of the tunnel barrier is s; k is a real number (typically
lOnm- 1 ), and is given by k = (2m¢/h 2 )1/2. It is straightforward to show
that
I ex L ItP~12ItP~12e-2k8 (10.3)
1,2
The expression gives qualitative insight into the tunnelling process, irrespec-
tive of tip shape, if the wave function anchored in the tip can be approximated
by an s-wave. Furthermore, if the overlaps of the respective wavefunctions
within the barrier are small, and if the tails of the two functions are of similar
shape, then
(10.4)
The local density of states in the sample surface at the position of the tip
can be written as
(10.5)
o Empty States
k k
Fig. 10.5. STS illustrated schematically. Electron states in the tip are being swept
past bulk and surface states in the specimen surface. The probability of elastic
tunnelling at a particular tunnel voltage depends on availability of occupied and
empty states and manifests itself as a change in tunnel current
• The lateral resolution in the x-y plane is due, in part, to the exponential
dependence of I on the barrier width. As well, the lateral decay of the
s-state located in the apex atom of the tip will confine the interaction lat-
erally to a radius of ca. 0.1 nm. Thus there will be single atom resolution
due to the lateral spatial decay of the density of states function. Of equal
significance is the fact that the elastic tunnel process picks out particular
states in the sample and tip, as a consequence of energy conservation.
Thus the tunnel voltage, VT , sets an energy window, with a width of a
few kT. Consequently one may choose a tunnel voltage for maximum res-
olution, i.e., an energy for which the corresponding electron states have
the greatest spatial variation. These observations account for the relative
ease with which 'atomic' resolution can be obtained for covalent materi-
als (e.g. Si); this is a consequence of the strong spatial dependence of the
density of states across the unit cell at the extrema of bands. Conversely,
similar resolution is much more difficult to obtain for metals where the
spatial variations are more gentle .
• The ability to 'tune' the energy window of the tunnelling process by
changing e VT is the basis for STS. Successive maps of the surface at dif-
ferent values of VT will reveal the real-space locations of the corresponding
equi-energy contours of the density of states. Local spectroscopy can be
carried out by fixing the lateral and vertical position of the apex of the tip
with respect to the surface and recording an h-VT curve. The principle
of the method is illustrated in Fig. 10.5, adapted from [12]. The h-VT
256 S. Myhra
An SFM probes a surface by sensing a force, or its gradient, between the sur-
face and a tip; therein is the underlying principle of operation. The attention
in this section will be principally with the AFM and its various operational
modes.
The principles of the SFM cannot be described with the same degree of
generality and neatness as for STMjSTS. Neither can the image formation
process be understood at the same level of detail and physical insight. Con-
tinuum theories will provide only a rough phenomenological description of
the tip-to-surface system, which may be adequate for most purposes, but the
10 Materials Characterization by Scanned Probe Analysis 257
full power of the SFM will only be realised once detailed microscopic theories
can be deployed on a routine basis.
Atoms respond to short-range repulsive exchange interactions and longer
range attractive ones. The tip-to-surface system may be modelled crudely
by pair-wise interactions between atoms in the tip and in the surface. The
interactions combine to produce a potential well with a location of lowest
energy, the binding energy E b , and a point at which there is zero net force.
The shape of the well defines a force gradient; the magnitude of the force
gradient is relevant for defining the operational mode and for the design
of the force sensing element of an SPM instrument. The contact mode has
its operating point in the repulsive force regime, where the force gradient
is high, and the repulsion is due to the overlaps of the electron clouds of
atoms in close proximity. The simplest description of the interaction in the
contact mode is in terms of the Lennard-Jones potential function; it turns
out, however, that a proper description of the contact mode requires a full
molecular dynamics simulation [15]. Intuitive and qualitative insight may be
gained from describing the system by two coupled springs. At equilibrium,
when the tip is tracing out equal force contours, then
(10.6)
where keff and kN are the effective force constants for the tip-to-surface inter-
action and the normal spring constant of the lever, respectively. Excursions
from the bottom of the potential well and from the point at which the lever
is undefiected are Zeff and ZL, respectively. Since in general keff » kN, then
Zeff « ZL, and a contact mode image will be a very good approximation to
the actual surface topography. The simple description also underscores the
need to image a 'soft' material with a probe with a low spring constant.
The non-contact mode, on the other hand, has its operating point in the
attractive force regime, where the force gradient is low and the attraction is
due to dispersion forces of the van der Waals type and other interactions.
Modelling of forces acting on extended bodies in close proximity is a non-
trivial exercise, and is beyond the scope of the present Chapter. Most of
the current attempts to reconcile theory with practice are based on a seminal
monograph by Israelachvili [16]. An AC resonance technique is generally used
to control the feed-back loop in the non-contact mode. The sampling of the
effective potential well is then via a driven and damped oscillator where
both amplitude and phase will depend on the average force gradient. The
sensitivity is greatest when the operating point is at the steepest part of the
resonance curve for the probe. Thus a non-contact image will represent a map
of constant force-gradient.
eigenmodes, but the amplitude is increased until the tip makes intermittent
excursions during each oscillatory cycle into the part of the potential well
dominated by the short-range repulsive interactions (hence the alternative
designation of tapping mode). The feed-back loop will now lock in either on
a constant decrement in amplitude or on a constant change in phase, where
the amplitude decrement/phase change is substantially dependent on effects
of the short-range interactions.
c
o (a)
~
't
"0
Q;
>
o Lift-off
.3
Nett repulsion
A
o
Nett attraction
:E
Stage travel
o
(b)
~
"0
'u;
\
"3 \
a.
c !" \
:g z'Q)"
0 \
Q) I \1
't
"0
0
_~ ___ -'l_
Q; I
>
Q)
"0
~
ZRO I
=
-l <Il I
<Il
1[; _ _ _ _ _ _ _ _ _ FLO
_
Z
~-..:..t
Indentation 0 Non-contact
Fig. 10.6. Schematic F-d curves. Lever deflection is plotted along the vertical axis.
(a) Interaction between a 'hard' surface and an incompressible tip is shown, with
stage travel as the independent variable. The effect of the meniscus aqueous layer
is apparent, and gives rise to instabilities at the points of snap-on and lift-off_ (b)
An incompressible tip interacting with a compressible surface and being subjected
to adhesive interactions at the interface. The horizontal axis now shows explicitly
the indentation of the surface by the tip
260 S. Myhra
• The force constant of interaction, ki' is simply the slope of the curve,
kNzL/lzd - zLI, where Zd refers to stage travel.
• The forces at 'contact' on approach and retract, FA and F R , may be
defined at the inflection points where the force constants of interaction
are the greatest.
• The snap-on and lift-off forces, Fso and FLO, are measured at the points
of greatest net attraction. The latter is generally taken to be the force of
adhesion.
• The distances ZRO and ZIO are measures of the elastic recovery and the
plastic indentation, both at zero lever loading.
• The extent of hysteresis in the system is given by the area enclosed by
the curves.
The parameters defined above cannot readily be related to the familiar macro-
scopic definitions of mechanical properties, e.g. hardness, adhesion, Young's
modulus, tensile strength, flexural strength, etc, unless the system can be
specified further (e.g. tip shape, contact area, surface free energy of tip, etc)
and unless additional assumptions are made (homogeneity, isotropy, surface
topography, etc).
A consequence of the discussion above is the need to match the force
constant of the lever to that of the interaction being investigated in order to
extract maximum information. If there is mis-match, then either the lever is
the only compliant element, and no information is obtained about the surface,
or the surface is the only compliant element, and the deflection of the lever
is not measurable.
The essential feature of the LFM is that the feed-back loop maintains control
as in the conventional contact mode, but that torsional deformations of the
lever are monitored by sensors orthogonal to those that generate the signal
for the AFM control loop. The difference signal from the orthogonal lateral
sensors as a function of position is then used to generate a friction force
image. LFM is gaining in popularity as a method capable of producing image
contrast from lateral differentiation; it has also extended tribology into the
nano-range, and can offer insights into mechanisms of friction and wear on
mesoscopic and atomic scales. The LFM arrangement is shown schematically
in Fig. 10.7.
The tip-to-surface interaction, assuming that the tip is moving in a di-
rection perpendicular to the long axis of the lever, can be described (in the
10 Materials Characterization by Scanned Probe Analysis 261
,~
Fig. 10.7. The LFM system. The lever applies and senses tip-to-surface forces with
components FN = kN Z, FT = kTx and FL = kLY
An SFM system may have useful lateral resolution down to 0.1 nm, while
the field of view may be as large as 100 x 100 [lm 2 ; thus the linear dynamic
range may be a factor of 106 . In practice the range will be covered by two or
more stages, e.g. a 1 [lm stage for high resolution scanning, a 10-25 [lm general
purpose stage, and a 100 [lm one when a large field of view is required. Stages
will generally be supplied fully calibrated with a suitable set of parameters
recognisable by the relevant control software package. In many cases the as-
received calibration will suffice. However, non-ideal characteristics such as
nonlinearity, hysteresis, aging and creep affect the accuracy, resolution and
reproducibility of spatial mapping.
In general x-y plane calibration procedures require the use of standard
specimens. Standard specimens are usually supplied by the manufacturer, e.g.
HOPG (highly oriented pyrolytic graphite) and mica, useful for x-y calibra-
tion of the highest resolution stage, and micro-machined semiconductor grids
for calibration over larger fields of view. As well, one can prepare well-ordered
specimens from mono disperse latex spheres of known diameter [20-22]. The
spheres are available in the size range from 50 nm to several [lm, and can be
obtained with certification traceable to primary size standards. Perfect hexag-
onally close-packed mono- and multi-layer structures can readily be prepared,
thus allowing calibration of all relevant x-y plane parameters (linearity, or-
thogonality and magnification). Point defects will be found occasionally, and
can serve as useful markers for determination of resettability and drift. The
features of a typical specimen are shown in Fig. 10.8; it will be apparent that
contour lines taken along particular directions and 2-D fast Fourier transform
techniques, if available, are convenient tools.
Two of the most important attributes of SFM are, firstly, that the depth
of field is limited only by the dynamic range of the stage in the z-direction,
and, secondly, that the spatial resolution in the z-direction may be better
than 0.1 nm (absolutely) and 1 part in 10 4 (relatively).
The precision in the z-motion will be conditioned by the performance
of the piezoelectric actuator. Once again it will be convenient to resort to
use of standard specimens. Single d-spacing features on cleaved surfaces (e.g.
MgO, Si(l11) and NaCl) will be useful in some cases, but it is more common
to use VLSI standards for calibration along the z-direction. A flat surface
with a known tilt angle, will give rise to a z-excursion over the field of view
comparable to the z dynamic range, and can be used to explore linearity
and hysteresis [23]. Alternatively, the known slope of a pyramidal tip may be
used for the same purpose [24]. A direct and convenient laser interferometric
method has been described [25,26]. As in the case of the x-y plane calibration,
layers of monodisperse latex spheres are excellent for z-calibration [22,27].
Monodisperse Au particles have also been used [28].
264 S. Myhra
a)
230
b)
o
o 2300 nm
Fig. 10.8. (a) Contact mode image of a polystyrene bead layer (200 nm spheres)
at an edge. (b) A z-direction line profile along a close-packed axis, extending to
the edge, shows that the scanning stage is calibrated accurately along X-, y- and
z-axes [22J
Both beam and V-shaped levers are in common usage. The supplier will
generally provide a generic value for kN; the actual value for a particular
lever may differ by a factor of two, however. At the present time suppliers
do not specify kT, much less k L . The expressions for kN and kT for a 'long'
beam-shaped lever are given by [31].
Et 3 w
kN -
-
- -
4£3 (10.11)
Et 3 w
(10.12)
kT = 6(1 + v)Lh2
where E is Young's Modulus, t is the thickness of the lever, v is Poisson's
ratio, and other dimensions are defined in Fig. 10.7. For completeness, the
expression for kL is
(10.13)
The 'difficult' quantities are t and E, the former being difficult to determine
even with SEM imaging and the latter being dependent on crystallographic
orientation, dopant concentration and stoichiometry. However, given that kN
can be determined reliably, the two other spring constants can be calculated
relatively accurately from
2L2
kT = kN 3(1 + v)h 2 (10.14)
L2
kL = kN 3h 2 (10.15)
In the case of LFM it can be shown that the limits of detectability for JL is
given by
(10.16)
The outcome of an AFM experiment will depend on the shape of the tip. The
as-received shape will vary from tip to tip. Moreover, the shape is in general
a dynamic property likely to alter during the conduct of an experiment, by
exchange of material between tip and surface. Thus it is desirable to check
tip shape before and after, or even during, an experiment. A convenient
method for tip characterisation is 'reverse' imaging [34]. The essential aspect
of the method consists of scanning an 'unknown' tip over a surface feature
having a shape approximating that of a delta-function spike. The result is an
image of the tip. The outcome is due to the spike having much higher aspect
ratio, Ar = (height of tip) / (width at the base), and much smaller tip radius,
R Tip , than those of the 'unknown' tip. The supplier's nominal value for Ar
will be reliable, in general, while the nominal value of RTip is less reliable.
Imaging other objects with known geometries (e.g. polystyrene spheres, gold
particles, TMV, etc.) can produce values for subsets ofthose parameters [35-
38]. While reverse imaging can provide accurate information about tip shape
in the mesoscopic domain (> 5 nm), information at the atomic level of detail
cannot be obtained by any routine methodology.
A commercially available sharpened tip is often an adequate approxima-
tion to a delta-function in comparison with a standard pyramidal tip [19,31].
An up-turned tip can be mounted as a standard specimen, or even be incor-
porated into the actual experiment. The tip height can be determined from
a line profile with a precision of about 5%. Fig. 10.9 shows reverse images
of tips at high resolution. An apex angle of 70° and R Tip = 35 nm can be
deduced from Fig. 10.9a; the results are in good agreement with the nominal
values of 70.5° and 40 nm, respectively.
The practical importance of tip analysis is apparent from Fig. 10.9b. Re-
verse imaging over two BaC0 3 spikes show that the actual 'tip' consisted of
two blunt apexes (RTip ~ 125 nm), offset by some 500 nm laterally and 65 nm
vertically. The use of non-ideal tips will result in anomalous outcomes, even
though the spring constants and other relevant parameters may be identical
and/ or well-characterised.
10 Materials Characterization by Scanned Probe Analysis 267
(a) (b)
1900 nm
1700 nm 2900 nm
Fig. 10.9. Reverse images of AFM tips. (a) Tip shape can be determined from a
line profile. (b) Shape of an altered tip
(a)
TopTe
BottomTe
BottomW.
TopW 0
a = 6.3 A
b: 3.5 A
a'= 2.2 A
~I
. dSI
(b)
c=0.61 A
a)
b)
Fig. 10.11. Atomically resolved images of WTe2, obtained at negative sample bias,
showing the general surface structure and point defects. The spacing between the
close-packed rows is 0.63 nm
10 Materials Characterization by Scanned Probe Analysis 269
Thin organic films are the essential elements for many emerging and tradi-
tional industries, e.g. [41], based on products and processes such as adhesives,
composites, semiconductor planar lithography, paints, coatings in the food
industry, magnetic information mass storage media and advanced lubricants
[42-47].
While 'traditional' surface and interface analysis of thin films has much
to offer, AFM methodologies and F-d mode, in particular, has relevance in
the present context [48]. The description below is based on accounts in the
literature [49,50].
Deployment of F -d analysis on thin films will result in a system consisting
of three interacting compressible elements (lever, film and substrate) which
can be modelled by three linear springs and two interface interactions. An
idealised generic F-d curve is shown in Fig. 10.12. For simplicity it may
be assumed that the tip has a rectangular shape and that the compressed
surface will adapt to the shape of the tip. It can then be shown [49] that
the spring constant of the compressible substrate is defined for large force
loadings, when the thin film is fully compressed, by
k _ kNSsub
sub - 1- S sub (10.17)
(10.18)
270 S. Myhra
Fa Net repulsion
o r-----'--t_ _ t
-f - - t
Net attraction
o
Fig. 10.12. Generic F-d curve for a system with three compressible elements in
series - substrate (ksub), thin layer/film (kl ayer ) and lever (kN). Two attractive in-
teractions are sensed by the tip, when it touches the outside surface of the layer /film
(Fa), and when the layer is compressed to negligible thickness and the tip senses
the layer-to-substrate interface (FJ)
k - k SlayerSsub
layer - N S S (10.19)
sub - layer
where Slayer is the slope of the F -d curve in the region where the film is
compliant, and zd(l) is the tip position at the point where the tip 'senses'
the film-to-substrate interface.
Fig. 10.13. AFM image show details of the surface texturing for wool. The spacings
and heights of surface texturing are shown by the line profile
roughness was 11.4 ± 1.5 and 8.1 ± 1.0nm for the VH8 and DVe tapes,
respectively.
The outcome of F-d analyses on scoured wool is illustrated in Fig. 10.14;
similar results were obtained from the video tapes. There is an excursion
during approach due to net repulsive interaction which is followed by an
adhesive 'snap-on', and then by another excursion into the repulsive force
regime. In the context of the present measurements the thin film can be
thought of as a linearly compressible element of finite thickness. The three-
element system will be compressed so that the slope of the curve will reflect
the effective stiffness of the combined system. When the lipid layer is near
its full compression, the tip senses the adhesive interaction at the film-to-
substrate interface and is pulled into contact with the substrate. From that
point onwards the film is passive, having effectively been compressed to a
272 S. Myhra
100
-.
E
---cc
0
..;::::;
()
Q)
;:;:::: 50
Q)
Cl
....
Q)
>
Q)
....J
-100 0 100
Stage Travel (nm)
Fig. 10.14. F-d data obtained under water on scoured woo!. Note the similarity
with the generic curve in Fig. 10.12
Specimen klayer (N 1m) ksub (N/m) Elayer (MPa) F~~~ (nN) T (nm)
rigid layer of negligible thickness, and the system reverts to having only two
compressible elements.
The inferred spring constants of the substrate and the film, and the thick-
ness of the film are given in Table 10.4. The spread in values for T is due to
surface roughness, uncertainty in tip shape, and uncertainty in choosing the
point of initial contact. It has been shown, e.g. [55], that when a paraboli-
cally shaped tip indents a compliant surface, then Young's Modulus, E, can
be inferred from
FI 4(Rpara)1/2 ~ Z1.5
load = 3 1 - l/2 (10.20)
The load is the sum of the restoring force of the compressed layer and the
adhesive interaction, Fload = klayerT + Fa, Z = T is the extent of inden-
tation/ compression, and Rpara = R Tip = 50 nm. The results are given in
Table 10.4; the spread in values for E is a direct reflection of uncertainties
in determination of T. The adhesive 'snap-on' force at the interface was also
determined.
10 Materials Characterization by Scanned Probe Analysis 273
(b)
, 90 nm
L
\
\
700 nm
nmL
Fig. 10.18. F- d analysis: (Top) Calibra-
tion curve (straight line with steepest
130 slope on right) and results of indenting
a soft (left curve) and a stiffer region
(middle curve) of the plasma membrane;
o nm 400 nm (Bottom) the hard substrate will provide
support for the cellular structures being
'- - - - -.--------=-::-::-=---.. . _-- compressed
The outcomes were straight lines having slopes of 2, thus justifying the as-
sumption of conical tip shape for z » R, in accord with the expression
i1z cone = [F(~) tan 0:(1 - v 2)/E]1 /2, where 0: is the opening half-angle of
the tip. The effective Young's Modulus, E, was found to be 4.6 and 17 kPa,
respectively. The former represented the effective modulus for a weakly sup-
ported membrane section, while the latter higher value was obtained from
a location supported by a stiffer cytoskeletal substructure. The lower graph
in Fig. 10.18 demonstrates additional stiffening at high force loading, when
cytoskeletal elements become trapped between the tip and substrate.
The improved imaging conditions due to use of an activated substrate
described above suggest that strong adhesive tip-to-surface interactions, re-
sulting in shear stress, are principally responsible for variable imaging con-
ditions of live cells. The treatment of the surface of the tissue culture dish
used in these experiments is proprietary, but the trade literature claims neg-
ative charging and hydrophilicity for the surface. AFM Si3N4 tips should in
principle be oxidised and be hydrophilic, however , in practice they will be
hydrophobic due to hydrocarbon contamination. Adhesive interactions be-
tween functionalised tips and surfaces have been studied by Frisbie et al.
[63], who found that the strengths of adhesive lift-off forces ranged from 8
(COOH/COOH) to 3 (CH 3/CH 3) to < 1 nN (COOH/CH 3). A phospholipid
monolayer may be expected to have hydrophobic patches, and one would
therefore expect strong adhesive interactions at both bilayer-to-substrate and
tip-to-bilayer interfaces. Hence the tip will bond to sites in the phospholipid
surface and the resulting shear forces will disassemble the membrane. Large
segments of membrane material and other debris will be carried along tran-
siently by the tip, and will be the cause of loss of spatial resolution and
instability of imaging conditions. The results suggest that the activated sub-
strate acts as a selective scavenger thus preventing build-up of debris on the
tip.
10 Materials Characterization by Scanned Probe Analysis 277
4606 nm
180 min
2303 nm
Onm
SO.06 ~(m
50.06 ~mm--------25:ro;;;;;-_______~=:
25.03~m
during the application of the MTT test, which is commonly applied to a cell
population in order to determine the viable fraction.
(a) Diamond-Like Carbon (DLC) Films. DLC films have many at-
tractive properties which can be tailored for a range of actual or potential
applications [64-67]. In particular, the tribological properties of DLC films
are beginning to attract interest. Since many applications are concerned with
macroscopic surfaces in sliding contact, the traditional macroscopic methods
(e.g. pin-on-disk or ball-on-disk) are often the most relevant indicators of
performance. However, the trend is toward smaller devices and point con-
tacts. Accordingly functionality must also be validated on the (sub )flm-scale,
requiring measurements which are spatially resolved on the mesoscale and
indeed into the nano-regime. A study was undertaken in order to measure,
on the mesoscopic scale, friction and surface adhesion for DLC coatings from
two fabrication routes, and to explore the influence of the two most likely
service environments, air and vacuum [68].
Materials from two fabrication routes, IBAD (ion-beam assisted deposi-
tion) and RF-CVD (radio-frequency assisted chemical vapour deposition),
were considered. The primary LFM data were obtained from the 'friction
loop', an example of which is shown in Fig. 10.20. The average lateral force
along the track is given by half the average peak-to-peak signal of the loop
converted into lateral tip deflection multiplied by the torsional spring con-
stant, i.e. (FL ) = CT(VL /2)k T . A subset of the processed experimental results
obtained with a blunt, R Tip > 100 nm, tip are shown in Fig. 10.21. The
results can be described by a multi-asperity model, which predicts a linear
.r ,. ~
to
at
SOnN
Lsonm
Fig. 10.20. Typical friction loop. The track length plotted along the horizontal axis
was ca. 200 nm. The lateral force on the tip is plotted along the vertical axis. The
slopes of the near-vertical sections represent static friction when the stage reverses
its direction of travel
10 Materials Characterization by Scanned Probe Analysis 279
1000
800
/ .....0
o ./
/
/
600 /
Z /
.s /
...J /~
LL /
400
o //0
/
/
/
FN (nN)
dependence irrespective of the gross shape of the tip. The sets of data were
therefore fitted by straight lines from which coefficients of friction were cal-
culated. The results are listed below in Table 10.6.
The results from analysing a CVD-route film with a sharp tip, RTip = 10-
20 nm, are shown in Fig. 10.22. A non-linear dependence is apparent. In the
single-asperity Hertzian regime it is predicted for low loads that FL = CF~/3,
with c being a constant. A curve was fitted to the data and exhibited good
agreement with the 2/3-dependence.
Table 10.6. Tribological properties for DLC films inferred from data base
Material/ designation Condition !-!
IB-assisted deposition
A1(1) air/vac 0.08 ± 0.03
A1(2) air/vac 0.075 ± 0.03
A1(3) air/vac 0.09 ± 0.03
A2 air 0.09 ± 0.03
A2 vac 0.045 ± 0.02
RF -CVD deposition
B1(DLC4) air 0.06 ± 0.02
B2(DLC10) air 0.06 ± 0.02
B2(DLC10) vac 0.125 ± 0.04
280 S. Myhra
300,--------------------------,
Z
.s200
~
o
u..
~ 100
til [;]
...J
[;]
IjJ
O~--r-_.--~--._~--_r--~~
Fig. 10.22. Friction data obtained with a sharp tip on the CVD-type surface
All linear curves fitted to the friction data for DLC specimens intersect
the FL axis at positive values (the zero load lateral force, FLO) and extrapo-
late to intercept at negative values on the FN axis (the dynamic lift-off force
required to obtain zero lateral force, FNO)' The observation shows that there
is an adhesive interaction between tip and surface. The values of FNO for DLC
specimens ranged from 500 to 2200 nN. These values are far too high to be
explained by simple Van der Waals interactions, or by other chemical effects.
The alternative possibilities are concerned with meniscus forces and/or tribo-
generated electrostatic interactions. The former should lead to a correlation
with static F -d lift-off forces, and a meniscus force should be substantially
dependent on the ambient (i.e., air versus vacuum). No such correlation or de-
pendence was observed. Detailed consideration of the data base showed that
the meniscus makes only a minor contribution to the adhesive interaction
for DLC surfaces. The electrostatic interaction, on the other hand, should be
largely absent in the F -d mode since the measurement is static, as was ob-
served. Also, one would expect the insulating RF -CVD specimens to exhibit
a greater electrostatically generated adhesive interaction than the conducting
IBAD specimens. The results were in accord with that expectation.
Fig. 10.23. FeSEM images showing layer structure and preferential cleavage planes
(top) , and platey wear and fracture debris (bottom)
70
60
Z 50
.s..J
IL
I
Q) 40
~
0
IL
30
E
~
...J 20
10
Fig. 10.24. Representative data for measurements of FL versus FN. The data points
refer to: (L» an abraded face analysed in vacuum; (.) a scraped face exposed to air
for 45 mins, then analysed in vacuum; and (D) a scraped face analysed in vacuum
References
62. V. Parpura, P.G. Haydon, E. Henderson: J. Cell Sci. 104, 427 (1993)
63. C.D. Frisbie, L.F. Rozsnyai, A. Noy, M.S. Wright on, C.M. Lieber: Science 265,
2071 (1994)
64. R.F. Davis: Physica B. 185, 1 (1993)
65. H. Tsai, D.B. Body: J. Vac. Sci. Techno!. A 5, 3287 (1987)
66. F.M. Kimock, B.J. Knapp: Surf. Coat. Techno!. 56, 273 (1993)
67. A. Matthews, S.S. Eskildsen: Diamond Relat. Mater. 3 902 (1994)
68. A. Crossley, C. Johnston, G.S. Watson, S. Myhra: J. Phys. D: App!. Phys. 31,
1 (1998)
69. M.W. Barsoum, T. El-Raghy: J. Am. Ceram. Soc. 79, 1953 (1996)
70. S. Myhra, J.W.B. Summers, KH. Kisi: 39, 6 (1999)
71. A. Crossley, KH. Kisi, J.W.B. Summers, S. Myhra: J. Phys. D: App!. Phys.
32, 632 (1999)
11 Low Energy Ion Scattering
D. J. O'Connor
Low energy ion scattering (LEIS) is the study of the structure and compo-
sition of a surface by the detection of low energy (100eV-lOkeV) ions (and
atoms) elastically scattered off the surface. This technique is a subset of ion
scattering spectrometry which involves the use of incident ions with energies
ranging from 100 eV to over 1 Me V. The range of measurements possible over
such a large range of energies extends from purely atomic layer resolution
at the low energy end to analysis to depths of the order of microns at the
high energy end. Some of the high energy effects (> 250 ke V) are covered in
Chap. 9, while the intermediate energy range (medium energy ion scattering)
has been successfully developed as a near surface structure probe mentioned
briefly in Chap. 1. The use of low energy ions to measure the surface struc-
ture of solids was established by Smith [1] in 1968. In that study the basic
elements of LEIS were established and these have been built on over the
past 20 years to develop into a powerful surface atomic layer structure and
composition probe [2-6]. It has been successfully applied to a wide range of
practical surface problems which include the surface composition analysis of:
• binary alloys
• catalysts
• cathode surfaces
• polymers
• surface segregation
• adsorbates
• surface structure
• adsorbate site identification
LEIS involves the bombardment of the surface with either inert gas or alkali
ions and measuring the energy distribution of the ions (or less commonly
288 D. J. O'Connor
the neutrals) scattered off the surface. As the energy loss to electronic pro-
cesses (termed inelastic energy loss) is relatively small at low energies it is
possible to identify the mass of the target atoms from the energy of the scat-
tered projectiles. By applying the principles of the conservation of energy and
momentum the scattered projectile energy, E 1 , is given by
~e
p
Ep = 2.0 keV
8 = 135 0
Au
~" . e x4.3
Ep = 5000 eV
8=135 0 x047 alloy'
Cu
x3.8 x3.8
J \ ) \ ~1--""--------If.....L.'-------""---j
'iii
Ar+ + He+ c
Q) Ar+ ~ Au, Cu
Ar+ + He+ ~ Pure Au, Cu C
~'
Ep = 5000 eV
Ep = 2.0 keV Au
. 8 8 = 135 0
8 = 135 0
Cu
x1.0 x2.5
\ )
1600 1800 2400 350 450 2400 2500
Scattered Ion Energy (eV) Scattered Ion Energy (eV)
Fig. 11.1. Comparison of the analysis of a Au-Cu(43%) alloy with He and Ar
projectiles demonstrating the increased mass resolution attainable with heavier
projectiles. From [7]
11 Low Energy Ion Scattering 289
(11.2)
(11.3)
where E2 is the energy of the particle recoiling at angle of ¢ to the incident di-
rection. The angle of recoil is limited to 90° so in many systems the detection
angle is set at greater than 90° in give the dual benefit of improving the mass
resolution and removing the potentially complicating features introduced by
recoiling projectiles.
In some cases the identification of the existence of an element on a surface
is sufficient to allow some conclusions to be reached. The simplest example
of this form of analysis was performed by Smith [1] who analyzed a clean Ni
surface onto which CO had been adsorbed. In the spectrum scattered ions
from Ni and from 0 were observed, but no ion yield was measured from
the C atoms. It was concluded from this that the CO molecule was bonded
perpendicularly to the Ni surface with the C atom forming the bond hence
the C atoms were shadowed from the incident ions by the 0 atoms and no
scattered ion yield was observed off C.
The existence of recoils can be exploited in some applications for the spe-
cialized tasks of detecting light elements and for the very sensitive detection
of electronegative adsorbates. The detection of light elements is extremely
difficult directly with ion scattering however the recoils are easily identified
and measured [4-12] and has been used to identify hydrogen on surfaces.
While it is possible to observe recoil peaks in a positive ion spectrum it is of-
ten difficult to separate this contribution from other ion scattering processes
290 D. J. O'Connor
104
...J
W 1200
Z
Z
« 2
i3 103
a:
w
0..
CJ) 800
f- 2
5102
0
0
10oL-~~~~--~----~~~~~~
2.0 93.6 185 277 368 460 93.6 185 277 368 460
(0) (600) (1200) (1800) (2400) (3000) (600) (1200) (1800) (2400) (3000)
CHANNEL CHANNEL
1000 C
NEGATIVE IONS
3keV Ne+ _ Ni
800
...J
W
Z
Z
«
i3 600
a:
w
0..
CJ)
f-
5 400
o
o 0-
(1) (2)
200
I I
Fig. 11.2. The measured scattered and recoiled spectra obtained for different
charge states when a Ni surface is bombarded by 3keV Ne+. From [8]
11 Low Energy Ion Scattering 291
with inert gas ion yields at the same energy. To overcome this limitation it
is possible to take advantage of the fact that some elements will escape the
surface in positive, negative and neutral charge states while in general the
inert gas projectiles only escape as positive ions or neutrals (Fig. 11.2). In the
case of electronegative elements (0, Cl, etc.) the negative charge fraction can
be as large or larger than the positive so by measuring the negative ion spec-
trum the contribution from the electronegative elements stands out without
interferences from the projectiles and as a result it has been estimated that
as little as 10~5 of a monolayer can be measured [12].
(11.4)
Z l Z2 e2
V(r) = ¢(r/a) . (11.5)
4JrEor
Here Zl and Z2 are the atomic numbers of the interacting particles, r is
the interatomic separation and a is the screening length. While the Moliere
approximation [13] (11.6) has been used extensively and successfully for many
292 D. J. O'Connor
years it has recently been surpassed in accuracy by the ZBL potential [14]
(11.8) which is based on the wave functions of a large range of elements.
¢M(X) = 0.35 exp( -0.3x) + 0.55 exp( -1.2x) + 0.10 exp( -6x) . (11.6)
One of the reasons for the success of the ZBL potential is the different ex-
pression used for the screening length, a, given by
ao
a = -=:-;;-;;---=-;'""""
Z?·23 + Zg·23 .
(11.9)
This potential has been shown [15] to be the best fit to the range of currently
available measurements of the interatomic potential and in absolute terms
it may be accurate to a precision of the order of 10%, however the ratio of
the cross sections of two atoms will be of higher precision and may be better
than 1%.
A more detailed description of the numerical determination of the scatter-
ing cross section has been given by Torrens [16], however as a general guide
the cross section increases with
a b
Fig. 11.3. The principal charge exchange processes responsible for the neutraliza-
tion of low energy ions near surfaces. Part (a) depicts the resonance process in
which there is a match between the energy level of an excited state in the projectile
and a filled state in the band structure of the solid. The Auger process (b) involves
an electron filling the ground state of the projectile and a second electron in the
band acquiring the excess energy
The transfer of an electron from the solid to the projectile can occur by
two principal processes. The simplest is the resonance process in which the
electron transfers directly from a level in the solid to a vacant level in the pro-
jectile at the same energy (Fig. l1.3a). The second major process is the Auger
process (discussed in Chap. 6) in which an electron from the solid transfers
to a lower energy state in the projectile and the excess energy is transferred
to another electron of the solid (Fig. l1.3b). A complete description of the
charge exchange process is much more complicated with the inclusion of the
possibility of surface electronic states, the changing energy levels of the pro-
jectile due to the interaction with its image charge and with the electrons
of the solid. As well there are currently two unresolved views of the charge
exchange process which favour either an interaction of the projectile with the
distributed electron distribution of the solid or a description which favours a
discrete interaction with each atom of the solid.
While a completely predictive theory for charge exchange is not yet avail-
able recent intense effort, both experimental and theoretical, has established
a basis for the description of charge exchange which is most appropriate to
inert gas ions. Hagstrum [17,18] developed a model for very low energy ions
(l-lOeV), and while this is for much lower energies than encountered in
LEIS and some approximations were not valid at these energies, it proved
to successfully describe the transition rates. Recent, more rigorous, theoret-
ical studies have yielded the same basic equations given by the Hagstrum
description. The Hagstrum model considers a transition rate which varies
exponentially with distance from the surface and can be integrated over the
294 D. J. O'Connor
trajectory of the projectile. The resulting expression for the ion fraction P+
is given by
(11.10)
20
Ar - Cu (polycr.)
5 Eo =BOOeV 10
iT = 1-1 0-7 Alcm 2
3
2
5 ISMS 0.1
3 +-----~~_r~~----~--------~
10 50 100 200 min
i i , ,
3 10 20 30 50 monolayers
Fig. 11.4. Demonstration of the relative sensitivity of LEIS and SIMS to the exis-
tence of contamination on a Cu surface. While the Cu ion yield in SIMS decays by
two orders of magnitude as the surface concentration of Cu increases during sputter
cleaning, the ISS Cu yield increases only 10%. From [22)
11 Low Energy Ion Scattering 295
where a is the screening distance and r is the distance between the projectile
and the target atom. This model would suggest that the ion fraction depends
on the path followed in a mixed surface and that it would be difficult to
estimate the charge exchange probability in practical applications. Further
work is in progress to establish which of the two models of charge exchange
best describe all observations.
Despite this uncertainty it is possible to demonstrate that the role of
charge exchange in LEIS is less significant than in SIMS where similar pro-
cesses occur and it is well known that the ion fraction of secondary ions is
very sensitive to low levels of contamination on the surface. The relative sen-
sitivities of LEIS and SIMS to surface contamination were demonstrated by
Grundner et al. [22] as a eu surface was cleaned by ion bombardment and
the SIMS signal for eu decreased by a factor of 100 while the LEIS signal
for Ar scattered off eu increased by approximately 10% in keeping with the
increased concentration of eu on the surface as contaminants were removed
(see Fig. 11.4). Thus LEIS, while not as sensitive to low levels of contamina-
tion as SIMS, will more accurately describe the changing composition of a
surface.
While the scattering cross section and the neutralization probability are
major uncertainties in the quantification of LEIS, it is nevertheless possible to
perform quantitative analysis despite them. The form of these measurements
fall into the following categories:
1. Relative measurements
2. Standards
6 6
IS: He - Ni (III) IS: Ne - Ni (III)
Eo= 1keV Eo= 1keV
8 = 90° 8 = 60°
5 ljJ = 45° 5 ljJ = 30°
QQ)
iT = 5·10-8 A iT = 1.3·10-BA
.!!? AES: Ee= 2keV AES: Ee = 2keV
ie = 50llA 4--:- 1.0 ie = 50llA
-E4 2 Vptp :J
2 Vptp
:J
0 ~.~ 0
u
ni ~ 0.8
'"~3 3 C Q)
0)-
.- C
c
(f) -
Q) ch
~ 0.6
~2 2~ «
"0
u
C ~ 0.4
..Q
~
Ne-Ni IS
oOx10lS x Ni AES
"Ox10 3AES 0.2
-Ni IS " Ne-S IS
• Ni AES
0 1 2 3 4 5 10 234 5 6 7
Oxygen exposure (a.u.) Sulphur coverage (a.u.)
Fig. 11.5. Demonstration of the linearity of LEIS yield response with adsorbed
material for 0 and S on a clean Ni surface. The enhanced surface sensitivity of
LEIS is evident in the case of the 0 adsorption as the LEIS yield from Ni decreases
with increasing dose while the AES yield from Ni increases marginally. From [25]
if it is assumed that all the adsorbate remains on the surface. Examples can
be found where a linear relationship between the LEIS and AES yields breaks
down [29], in which case the interpretation of this difference may be that the
adsorbate moves below the surface so that it continues to be detected by AES
and not by LEIS, or that there is a change to the electronic properties of the
surface which affects the neutralization rate of the ions used in LEIS.
In other studies LEIS has been applied to compare the surface compo-
sition of clean and contaminated catalysts to identify the contaminate. In
such studies only the existence of an unexpected element is sufficient to al-
low a conclusion without the need for quantitative compositional analysis.
In the following list of studies of surfaces it was not necessary to establish
an absolute scale of concentration but instead the identification of a par-
ticular element on the surface or a measurement of the relative change in
concentration of an element as some change or profiling is undertaken was
sufficient.
In these cases the assumption that the ion fraction is constant was sufficient
to establish a relative concentration scale.
11.3.5 Standards
The analysis of surfaces using standards has been performed with both ele-
mental and molecular standards. In the case of alloy analysis the most con-
venient reference materials are targets of the pure elements, thus to analyze
a compound or alloy [42] of two components A and B which have surface
concentrations of Na and Nb (in atoms cm- 2 ) respectively, it is first neces-
sary to establish the measured ion yield 1~ and 1~ off pure surfaces of A and
B which have surface concentrations of N~ and N~ respectively. From the
measured ion yields for the alloy surface, 1a and h respectively, the surface
concentrations can be determined.
(11.12)
and
(11.13)
If only A and B are present, then the relative concentrations can be deter-
mined from the total number of surface sites No (= Na + Nb) as
S1a<
Na/No = S1a + h (11.14)
where
S = 1~ N~ = 1~ (db)2 (11.15)
12 N~ 12 da
The sensitivity factor S has been measured [43] for Cr and Fe off clean and
oxide surfaces and to within experimental accuracy the same value (1.5±0.1)
was obtained. In the same study it was established that the relative sensitivity
factors for O/Fe was 0.11 and for O/Cr was 0.08, highlighting the difficulty in
detecting 0 with LEIS. There have been variations on this approach [44,45]
with similar findings which support the assumption that the ion fraction is
little affected by the nature of the surface or the environment of the scattering
elements.
The use of the standards approach is complicated for 0 (and similar ad-
sorbates) as a pure 0 target is not available. One approach to overcome this
298 D. J. O'Connor
0
~
0:
'" Co/Mo
.6
>-
f-
[']
[']
B~ c:::::::::::::§
iii
z
W .4
i
f-
~ ¢ ¢
00 [']
~ .2
Fig. 11.6. The evolution of the relative concentrations of principal elements in the
surface of a CoMoS/r -Ab03 catalyst measured by LEIS as a function of tempera-
ture. From [47]
Eo
(keV)
0.5
~
1.0 iii_
ZOO
Wf-
f--
ZZ
_:::J
z>-
00:
-«
00:
Wf-
2.0 ffiii'i
I=~
«-
()
00
4.0
o .2 .4.6 0.2.4.6 o .2 .4 .6 o .2 .4 .6
SCATIERED ION ENERGY RATIO (Es/Eo)
"0
Q)
.;;;'
c B
.2
E/EO
••• • •• • • • •••
• •• • •• • • • ••••
Fig. 11.8. At low energies both single and
double scattering sequences as illustrated
AZIMUTHAL ANGLE
Fig. 11.9. The ion yield from single (large peaks) and multiple scattering peaks
(small peaks) is shown as a function of the azimuthal angle (relative to the [100]
direction) of the target for 6 keY Kr incident at 25° to a W(110) surface. The peaks
in the multiple scattering at 55° and 125° are evidence of the [111] directions, while
the influence of multiple scattering at 0°-[100], the 90° -[110] and ±25° -[113] can be
observed
the [100] direction) of a W(110) surface. The peaks in the multiple scattering
at 55° and 125° represent the influence of the [111] directions on multiple
scattering yields, while the multiple scattering from the [100] (0°), the [110]
(90°) and [113] (±25°) surface directions are less pronounced as expected by
their interatomic spacing.
Perhaps the most direct use of shadowing and the most easily interpreted is
the ICISS developed by Aono [50]. To explain its use it is first necessary to
11 Low Energy Ion Scattering 301
\ / / /
He+ (Eo 1.0 keV)
-
r~··f~·;;········································=··~·
........... Jo~
I \
Fig. 11.10. The shape of a typical shadow cone for 1 ke V He incident upon a Ti
atom. The clear area behind the atom is the excluded region or shadow cone. From
[50]
302 D. J. O'Connor
300
He+ NiAI (111l a 300~------------------------'
-z-
(/) Ts= 150K III
C.
c. u
u
(/)
100
20· 40· 60· 80· 100· I/',n 20· 40· 60· 80· 100· Ij>m
Fig. 11.11. Results of He scattering yield from Ni (a) and Al (b) on a NiAI(111)
surface both demonstrating that there is no Al terminated surface layer and allowing
the determination of the relative positions of the surface atoms. From [59J
the shadow cone will intersect the neighbouring atom. The scattered ion yield
will be zero below the critical angle and beyond this angle it will peak (from
the flux enhancement) then fall to an intermediate value. The measurement
of the interatomic spacing comes from the critical angle determination and
a knowledge of the shape of the shadow cone. An early application of this
technique was the location of C under the surface layer of Ti in a TiC(l1l)
surface [58]. In this application the C sits asymmetrically between the surface
Ti atoms and hence there are two critical angles depending on from which
direction the ion beam is incident. The position of the C is 87 ± 8 pm below
the Ti surface layer.
In a more recent demonstration of the power of this form of analysis [59]
a variation of this technique was applied to the NiAI( 111) (Fig. 11.11) surface
in conjunction with LEED and STM to determine the termination structure
of this binary ordered alloy. In Fig. I1.11a the yield from the Ni scattered
projectiles yields information on the relative positions of surface Ni atoms
while that in Fig.I1.11b reveals that the Al is confined to the second layer.
~ ~?_N_B_~_~_____________~ }
8CATTE"'D---/ }'/
B""-~v
ELECTROSTATIC ENERGY
ANALYZER
The second contribution arises because the target atom is not stationary and
thus the projectile collides with a moving target. As a result the energy width
of the scattered ions is broadening with a width given by
(11.17)
where Eth is the mean thermal vibration energy. At room temperature Eth is
0.025eV so the thermal energy contribution for a lOOOeV projectile is 5eV
or 0.5%. As the natural widths of LEIS peaks are greater than those encoun-
tered in electron spectrometries the energy resolution (!!'E / E) need be no
better than 0.5% in most practical circumstances. In some instruments the
solid angle is increased to increase sensitivity but this results in a greater
kinematic broadening and diminished mass resolution. The kinematic broad-
ening increases with increasing projectile mass and with increasing collection
solid angle.
304 D. J. O'Connor
References
1. D.P. Smith: J. Appl. Phys. 38, 340 (1967)
2. H. Niehus, W. Heiland, E. Taglauer: Surface Science Reports 17, 213 (1993)
3. H.H. Brongersma, P.A.C. Groenen and J.-P. Jacobs: "Application of low en-
ergy ion scattering to oxidic surfaces", in: Science of Ceramic Interfaces 2, Ed.
J. Nowotny, Elsevier (1994) pp. 113-182.
4. J.W. Rabalais: Surf. Sci. 300, 219 (1994)
5. M.M Sung, V. Bukov, A. Albayati, C. Kim, S.S Todorov, J.W Rabalais: Scan-
ning Microscopy 9, 321 (1995)
6. C. Kim, C. Hoefner, V. Bykov, J.W. Rabalais: Nucl Instr Meths in Phys Res
B 125, 315 (1997)
7. H.J. Kang, R. Shimizu, T. Okutan: Surf. Sci. 116, L173 (1982)
8. P.J. Schneidner, W. Eckstein, H. Verbeek: Nucl. Instrum. Meth. 218, 713
(1983)
9. B.J.J. Koeleman, S.T. de Zwart, A.L. Boers, B. Poelsema, L.K. Verhey: Nucl.
Instrum. Meth. 218, 225 (1983)
10. J.A. Schultz, R. Kumar, J.W. Rabalais: Chern. Phys. Lett. 100, 214 (1983)
11. B.J.J. Koeleman, S.T. de Zwart, A.L. Boers, B. Poelsema, L.K. Verhey: Phys.
Rev. Lett. 56, 1152 (1986)
12. D.J. O'Connor: Surf. Sci. 173, 593 (1986)
13. G. Moliere: Z. Naturforsch. 2A, 133 (1947)
14. J.P. Biersack, J.F. Ziegler: In Ion Implantation Techniques, ed. by H. Ryssel,
H. Glawischnig, Springer Ser. Electrophys. Vol. 10 (Springer, Berlin, Heidel-
berg 1982) pp. 122-156
15. D.J. O'Connor, J.P. Biersack: Nucl. Instrum. Meth. B15, 14 (1986)
16. I.M. Torrens: Interatomic Potentials (Academic, New York 1972)
17. H.D. Hagstrum: Phys. Rev. 96, 336 (1954)
18. H.D. Hagstrum: In Inelastic Ion-Surface Collisions, ed. by N.H. Tolk,
J.C. Tully, W. Heiland, C.W. White (Academic, New York 1977)
19. R.J. MacDonald, D.J. O'Connor: Surf. Sci. 124, 423 (1983)
20. R.J. MacDonald, D.J. O'Connor, P.R. Higginbottom: Nucl. Instrum. Meth.
Phys. Res. B2, 418 (1984)
21. D.J. O'Connor, Y.G. Shen, J.M. Wilson, R.J. MacDonald: Surf. Sci. 197, 277
(1988)
22. M. Grundner, W. Heiland, E. Taglauer: Appl. Phys. 4, 243 (1974)
23. H. Verbeek: In Materials Characterisation Using Ion Beams, ed. by
J.P. Thomas, A. Cachard (Plenum, New York 1978)
24. E. Taglauer, W. Heiland: Appl. Phys. Lett. 24, 437 (1974)
25. E. Taglauer, W. Heiland: Surf. Sci. 47, 234 (1975)
26. P.J. Martin, C.M. Loxton, R.F. Garrett, R.J. MacDonald, W.O. Hofer: Nucl.
Instrum. Meth. 191, 275 (1981)
27. A. Sagara, K. Akaishi, K. Kamada, A. Miyahara: J. Nucl. Mater. 93/94, 847
(1980)
28. H.H. Brongersma, G.C.J. Van der Ligt, G. Rouweler: Philips J. Res. 36, 1
(1981)
29. H. Niehus, E. Bauer: Surf. Sci. 47, 222 (1975)
30. Y.T. Tan: Surf. Sci. 61, 1 (1976)
31. M. Wu, D.M. Hercules: J. Phys. Chern. 83, 2003 (1979)
11 Low Energy Ion Scattering 305
32. L.L. Tongson, A.S. Bhaila, I.E. Cross, B.E. Knox: Appl. Surf. Sci. 4, 263 (1980)
33. W.L. Baun: Appl. Surf. Sci. 4, 374 (1980)
34. C.RK. Marrian, A. Shih, G.A. Haas: Appl. Surf. Sci. 24, 372 (1985)
35. C. Creemers, H. Van Hove, A. Neyeens: Appl. Surf. Sci. 7, 402 (1981)
36. W.V. Lampert, W.L. Baun, B.C. Lamartine, T.W. Haas: Appl. Surf. Sci. 9,
165 (1981)
37. H. Jeziorowski, H. Knozinger, E. Taglauer, C. Vogdt: J. Catal. 80,286 (1983)
38. J.P. Jacobs, S. Reijne, R.J.M. Elfrink, S.N. Mikhailov, H.H. Brongersma: J.
Vac. Sci. Technol. A12, 2308 (1994)
39. Th. Berghaus, H. Neddermeyer, W. Radlik, V. Rogge: Phsica Scripta T4, 194
(1983)
40. J.C. Bertolini, J. Massarddier, P. Delichere, B. Tardy, B. Imojik, Y. Jugnet,
Tran Minh Duc, L. Temmerman, C. Creemers, H. van Hove, A. Neyens: Surf.
Sci. 119, 95 (1982)
41. H.J. Kang, Y. Matsuda, R Shimizu: Surf. Sci. 134, L500 (1983)
42. T.A. Flaim: Research Publication GMR-1942, 1975, Research Laboratories,
General Motors, Warren, Michigan
43. RP. Frankenthal, D.L. MaIm: J. Electrochem. Soc. 123, 186 (1976)
44. D.G. Swartzfager: Anal. Chern. 56, 55 (1984)
45. M.A. Wheeler: Anal. Chern. 47, 146 (1975)
46. P. Bertrand, J.-M. Beuken, M. Delvaux: Nucl. Instrum. Meth. 218,249 (1983)
47. J.-M. Beuken, P. Bertrand: Surf. Sci. 162, 329 (1985)
48. RC. McCune: Anal. Chern. 51, 1249 (1979)
49. D.J. O'Connor, RJ. MacDonald: Radiat. Eff. 45, 205 (1980)
50. M. Aono: Nucl. Instrum. Meth. B2, 374 (1984)
51. O. Oen: Surf. Sci. 131, L407 (1983)
52. M. Aono, Y. Hou, R Souda, C. Oshima, S. Otani, Y. Ishizawa: Phys. Rev.
Lett. 50, 1293 (1983)
53. J. Moller, H. Niehus, W. Heiland: Surf. Sci. Lett. 166, Ll11 (1986)
54. H. Niehus: Surf. Sci. Lett. 166, LI07 (1986)
55. J.A. Yarmoff, RS. Williams: Surf. Sci. Lett. 165, L73 (1986)
56. M. Aono, Y. Hou, C. Oshima, Y. Ishizawa: Phys. Rev. Lett. 49, 567 (1982)
57. R Souda, M. Aono, C. Oshima, S. Otani, Y. Ishizawa: Surf. Sci. Lett. 128,
L236 (1983)
58. M. Aono, C. Oshima, S. Zaima, S. Otani, Y. Ishizawa: Japan. J. Appl. Phys.
20, L829 (1981)
59. H. Niehus, W. Raunau, K. Besocke, R Spitzl, G. Comsa: Surf. Sci. Lett. 225,
L8 (1990)
60. S. Priggemeyer, A. Brockmeyer, H. Dotsch, H. Koschmeider, D.J. O'Connor,
W. Heiland: Appl. Surf. Sci. 44, 255 (1990)
61. RC. McCune: Anal. Chern. 51, 1249 (1979)
12 Reflection High Energy
Electron Diffraction
C.L. Price
Reflection high energy electron diffraction (RHEED) was first used in the
study of a cleaved calcite crystal by Nishikawa and Kikuchi in 1928 [1]. They
observed diffraction spots, and lines attributed to diffuse scattering. Other
early workers included Germer (1936) [2] who took diffraction patterns from
galena and Miyake (1936) [3] who examined oxide surfaces. Uyeda et al. [4]
used RHEED with metal films in 1940 and with adsorbed organic molecules
in 1950. Commercial RHEED equipment was developed through the 1950s
but this operated at 10- 4 to 10- 6 Torr and hence only on dirty surfaces.
However its application was as an alternative to X-ray diffraction: RHEED's
forward scattering nature and comparatively high scattering cross section
(10 8 :1) made it competitive. As ultra high vacuum equipment became com-
mon in the 1960s, systems were equipped with RHEED guns, but good LEED
was then possible, and LEED largely displaced RHEED as a diffraction tech-
nique for clean single crystal surfaces. The main reasons for this neglect was
that RHEED gives quantitative results only on extremely flat surfaces which
were not easily prepared and offered no theoretical advantages over LEED
in the central surface science question of the surface atomic structure. Also
the LEED apparatus was easily constructed and much cheaper than the tra-
SUBSTRATE
RHEED
PHOSPHOR
SCREEN
RHEED
ELECTRON
GUN
--OOCe
LIQUID NITROGEN
--
VACUUM
VACUUM
Fig. 12.1. Schematic diagram of an MBE system showing the RHEED geometry
308 G.L. Price
LEED because of the high electron energies. The glancing nature keeps the
apparatus well clear of other techniques in the chamber.
MBE and its variants such as metalorganic and gas source MBE, have
wide application throughout material science for the growth of single crys-
tals [6]. It is used to grow metals, semiconductors and ceramics and is a very
general method of crystal growth with particular application where extreme
purity and precision are required. MBE is simultaneously a mass production
and an advanced research technique. It is a unique combination of surface
science and production technology. As RHEED is the essential surface science
tool for routine MBE, it has been raised to new prominence and very detailed
studies have been done in an endeavour to exploit its potential. In this chap-
ter, the examples will be confined to III-V semiconductors as these were the
prototype materials for MBE. However the technique applies generally to all
other MBE growths.
In an MBE apparatus as shown in Fig. 12.1, the single crystal substrate
faces an array of eight or more ovens, each containing elements such as gal-
lium, indium, silicon, aluminium and arsenic. If as an example, a layer of
n-type GaAs is required, a single crystal GaAs substrate is held at rv 580° C;
the Ga, As and Si dopant shutters are opened and the growth commences.
The Ga and Si has unit sticking probability but the As simply supplies an
overpressure and is incorporated as needed. The crystal alloy layers are grown
epitaxially at about one atomic layer per second. These materials are subse-
quently transformed into electronic and optoelectronic devices. Generally the
ovens are ~ 1200° C, the substrate is ~ 700° C and the chamber may contain
10- 5 Torr of highly corrosive arsenic vapour. The quality of the semiconduc-
tor is critically dependent on the exact surface reconstruction. RHEED is
uniquely fitted to this role because of its robustness and geometry. It is one
of the few techniques of any kind that can monitor crystal growth in situ.
12.1 Theory
The relation between the surface reciprocal lattice and the diffraction pattern
is shown in Fig. 12.3a. The RHEED Ewald sphere takes a section through
the surface rods and streaks are observed on the phosphor screen rather than
the spots of the LEED case. To obtain the full reciprocal lattice map which
is given directly by LEED, the crystal substrate must be rotated about its
normal; the Ewald sphere then sections each plane in turn. The LEED pat-
tern shown in Fig. 12.3b is a complex gallium rich reconstruction referred to
as the centred 8 x 2. The three main azimuthes which should be observed in
RHEED are also drawn. The fine detail of the eighth order is often not ob-
served by RHEED and the pattern is referred to as a 4 x 2. Practically this is
of little consequence as the growth conditions are established to avoid this re-
construction by decreasing the metal/arsenic ratio and obtaining the arsenic
rich 2 x 4 (LEED c(2 x 8)): the fine detail is not required. When the transition
310 G.L. Price
a
- -- - -tl
(2.0)
Streak
(2.3) (2.2) (2,1)
• • • (2,0)- -
- - -- (1,0)
• •
(1.3) (2,2) (1,1)
• (1,0) - - - -
•
(0,3) (0,2) (0,1)
• • (0,0) - - - - - _ _ _ _ _ _ (0,0)
• • •
(1.3) (1,3) (1,3)
(1,0) ____ _
RECIPROCAL
LATTICE
--- (1,0)
RODS
ELEVATION
-- -
-- - ---
- - -- --
(0,0)
streak
SPECIMEN
",..
:~I'(
b ,%>" O"",<i>y fl?
~ "(
(~ O"",<i>v. +9
<i>v."'r. /' /
x 2) "''''~0
(4 c(8 x 2)
"",
I I I I I I / /
I I I I II I I II
I I I I I I I I I [110]
xoxoxoxoxoxoxoxoxoxoxoxo --
"
OJ
x
~
Fig. 12.3. (a) Sectioning of the two-dimensional reciprocal lattice rods give the
observed RHEED streaks (after [6]). (b) Comparison of the LEED and RHEED
patterns for the GaAs c(8 x 2) reconstruction (after [7])
12 Reflection High Energy Electron Diffraction 311
from the metal to arsenic rich reconstruction occurs, the four streaks of the
[110] change to two. If the surface roughens, then transmission rather than
reflection patterns are obtained as shown in Fig. 12.2c. The beam, skimming
over the surface, penetrates peaks and ridges. The streaks are replaced with
points since the surface reciprocal lattice rods are now replaced by the re-
ciprocal lattice itself. Roughness of order of an atomic layer can be detected
by RHEED. This is extremely useful when monitoring crystal growth and it
cannot be easily done by LEED.
To estimate the depth sensitivity of RHEED refer to the mean free path
(A) curve of Fig.lo8. At 10keV, the mean free path normal to the surface
is A sin I where I rv lOis the glancing angle. The result is about one to
two atomic layers. The electron wavelength is rv 0.1 A which is an order of
magnitude smaller than an atomic layer. Thus the diffraction is sensitive to
the surface and the electrons are easily scattered by surface steps and terraces.
The diffraction, like LEED, is dominated by multiple scattering and simple
kinematic arguments cannot be applied to the streak intensities.
Both LEED and RHEED are surface sensitive, but they differ in their
properties parallel to the surface. There are three aspects: the area which can
be examined, the coherence length of the diffraction and the ratio of ordered
to diffuse scattering. LEED is usually used at normal incidence with a beam
size of rv 1 mm 2 . The glancing nature of RHEED means that it is imprecise in
the direction of the beam, of order of a mm with a simple electrostatic gun,
while accurately positioned perpendicular to the beam. More sophisticated
high energy and magnetically focussed guns can improve the accuracy by
an order of magnitude or more. The coherence length is a measure of the
sensitivity of the diffraction to the long range order of the surface. It is the
maximum distance between reflected electrons which are able to interfere. By
the uncertainty principle, Ox = 21T 10k. The electron momentum uncertainty
has two contributions, oka due to the beam convergence half angle 0, and
dkE due to the finite energy spread oE. Using E = h2 k 2 /2m and assuming
a very small, it can be shown that [8]
100eV, ,=
where I is the glancing angle. In LEED, 0 rv 10- 2 rad, oE rv 0.5 eV, E rv
1T/2, and ox rv 100A. In RHEED, 0 rv 1O- 4 rad, oE rv 0.5eV,
E rv 10 4 eV, I rv 1O- 2 rad, and ox rv 1000 A. Thus RHEED is more sensitive
to long range order. Variations in RHEED streaks have been attributed to
monatomic steps 2000 A apart on a GaAs crystal in an MBE system [9]. A
large coherence length does not necessarily make the diffraction pattern more
susceptible to disorder. Disorder gives a diffuse background spread evenly over
a large solid angle. Large increases in this background have little affect on the
sharp, high intensity diffraction features. Thus the diffraction process tends
312 G.L. Price
to select out the ordered parts of the surface. For this reason and because of
the beam spread across the crystal, and the glancing angle geometry, RHEED
can often give a better diffraction pattern then LEED off a patchy or dirty
surface.
12.2 Applications
A B
>-
!::
en
z
w
f-
Z
o 50 100 o 50 100
TIME (sec)
Fig. 12.4. Observed oscillations on a GaAs crystal for increasing flux rates. In A the
flux is so slow that the substrate steps can scavenge the adatoms before nucleation.
The vicinal angle is l.0 mrad (after [10])
oscillations being equal to the time taken to grow a single monolayer of mate-
rial. Typical oscillations are shown in Fig. 12.4 and a schematic explanation is
given in Fig. 12.5. The first few layers are essentially complete before another
begins. Since the layer thickness is much larger than the de Broglie wavelength
of the electrons (for GaAs 2.83A»0.lA), the electrons are easily scattered
out of the specular beam by the step edges. The step edge concentration is
a minimum for a completed layer and a maximum for half a monolayer cov-
erage - hence the oscillations. This simple explanation must be modified to
explain the damping of the oscillations as the growth continues. If adatoms
fall in a 'ravine' between two steps of width less than the migration length,
they will be gettered by the steps edges and nucleation and creation of more
step edges will not be possible. As the coverage increases this becomes more
likely but so does nucleation of a second layer on top of the first. The result
is an increase in the concentration of step edges with growth, tending to a
limiting equilibrium value. In RHEED this is seen by the damping of the
oscillations and their disappearance, corresponding to the steps gettering all
arriving adatoms. A good illustration of this process is shown in Fig. 12.6.
Al has a smaller migration length than Ga. GaAs is grown until there are
no oscillations. When the aluminium is added oscillations are seen again as
the smaller migration length initiates nucleation. When the Al is removed,
nothing happens as the growth continues to be by step propagation. An es-
timate of the migration length can be found by growing at a fixed flux but
for differing temperatures on a vicinal surface. On such a surface, cut at a
slight angle to a (100) plane, there are terraces of known length (Fig. 12.9).
The result is shown in Fig. 12.7. The migration length increases with tem-
perature, and at a certain temperature, growth is only by step propagation.
314 G.L. Price
7
e=o.d £P
e=o'g i!I!?
e =o.~;;Y
6=1/ 6/
From this, values of lo and El can be derived. One estimate for Ga atoms on
a GaAs (100) surfaceis lo = 4A and El = 0.3eV.
It is expected that the best quality growths will be obtained if step prop-
agation is the main growth mode as the randomness of island nucleation and
dendritic growth is reduced. One method is to adjust the growth conditions
for minimum damping, endeavouring to keep the rate of step nucleation to
a minimum. Claims have also been made that vicinal surfaces produce bet-
ter quality materials. An extreme application of this principle is migration
enhanced epitaxy (MEE). If growing GaAs by this technique, first a layer of
gallium is grown, then separately a layer of arsenic and so on. The method
assumes that the migration length of Ga on a bound As layer or As on a
bound Ga layer to be much greater than if free Ga and As coexist in the
same plane. This is borne out by the RHEED data of Fig. 12.8. Here the
MEE oscillations are caused by specular reflection changes between Ga and
As terminated surfaces, not scattering by step edges. There are smooth tran-
sitions from As to Ga terminated surfaces and the oscillations do not damp
12 Reflection High Energy Electron Diffraction 315
I
t I
I I
lhY
(a)
. .
I I
time
Fig. 12.6. Intensity oscillations across GaAs/ AIGaAs/GaAs interfaces. The short
migration length of the Al restarts the oscillations (after [12])
TS (0C)
bt
t=t
~\~/584
596
598
-Tc=590°C
588
~~/580
570
560
E
ctS
Q) 550
.0
Oi 540
"S
u
Q)
Q.
en
Fig. 12.7. Growth of GaAs on a vicinal sur-
face as a function of temperature. At Te, the
migration length is of order the step separa-
Time - tion (after [13]
316 G.L. Price
580°C
H 10s
MBE
MEE
Fig. 12.8. RHEED oscillations by MBE followed by MEE. The same 2 x 4 recon-
struction remained through the growth (after [14])
2rc IL
a b
o 5
(mrad)
Fig. 12.9. (a) Schematic of a regular staircase and its reciprocal lattice. Curve A
is part of the Ewald sphere for a beam directed down the staircase with a RHEED
streak split into two. (b) Intensity profiles along a (00) streak as a function of
glancing angle I (after [9])
out with growth. With this technique, high quality materials can be grown
at much lower temperatures than normally used, 300 0 C instead of 600 0 C.
As previously mentioned, RHEED is highly dynamic; multiple scattering
dominates the diffraction and simple kinematic models can not in general
be used. One exception is the application of RHEED to measure the terrace
widths on vicinal surfaces. The direct and reciprocal lattice for a terraced
vicinal surface (not to scale) is shown in Fig. 12.9a. The diffracted intensity
is the product of the diffraction due to a terrace of atoms times the diffrac-
tion due to a grating of step edges. The rod shown by the dashed lines has
the reciprocal width of the terrace; the slashes have the reciprocal length of
12 Reflection High Energy Electron Diffraction 317
s A A s
the risers and the angle of the vicinal surface. Depending on the angle of
incidence, the Ewald sphere will intersect one or more of the slashes. Figure
12.9 shows intersection with two slashes, producing splitting in the observed
RHEED streak. The vicinal angle can be straightforwardly estimated from
this data using
(12.2)
,=
where (3 is the measured splitting angle, Be is the angle between the vicinal
surface and the low index bulk plane, 7r /2 - B is the glancing angle and
¢ is the azimuthal angle with ¢ = 0 down the staircase.
So far we have ignored multiple scattering, but one very important mul-
tiple scattering phenomenon which is observed in RHEED and LEED is
Kikuchi lines. They appear as sharp dark or light lines crossing the phos-
phor screen (Fig. 12.2a) and they move rigidly with the crystal. Because of
this they are used to check the crystal orientation and, with experience, their
sharpness provides a qualitative guide to surface cleanliness. Their origin is
shown in Fig. 12.10. Diffusely scattered electrons are diffracted inside the
bulk crystal. These diffracted electrons lie in cones which intersect the phos-
phor screen as arcs of large radii. Whether the line is light or dark depends on
whether the diffraction is in or out of the diffuse background. This diffraction
is from the bulk; the Kikuchi electrons carry underlying information of the
solid and not the surface. Their effect is often noticed in oscillation growth ex-
periments. If a Kikuchi line intersects that part of the specular streak which
is being monitored, a 7r out of phase component is observed in the inten-
sity oscillations. This is caused by electrons, which have been scattered from
the specular beam by surface steps, contributing to the population of diffuse
electrons and being diffracted by the Kikuchi process back into the specular
streak. At half a monolayer coverage when the specular beam intensity is a
minimum, the scattering in from the Kikuchi line is a maximum. An example
318 G.L. Price
References
1. S. Nishikawa, S. Kikuchi: Nature 122, 726 (1928)
2. L.H. Germer: Phys. Rev. 50, 659 (1936)
3. S. Miyake: Nature 139, 457 (1936)
4. R. Uyeda: Proc. Phys. Math. Soc. Japan 22, 1023 (1940); ibid 24, 809 (1942);
Y. Kainuma, R. Uyeda: J. Phys. Soc. Japan 5, 199 (1950)
5. E. Bauer: In Techniques of Metals Research, Vol. 2, ed. by R.F. Bunshah (In-
terscience, New York 1969) p.501
6. E.H.C. Parker (ed.): The Technology and Physics of Molecular Beam Epitaxy
(Plenum, New York 1985)
7. J.H. Ncave, B.A. Joyce: J. Cryst. Growth 44, 387 (1978)
8. D.P. Woodruff: In The Chemical Physics of Solid Surfaces and Hetemgencous
Catalysis, Vol. 1, ed. by D.A. King, D.P. Woodruff (Elsevier, Amsterdam 1981)
p.108
9. P.R. Pukite, J.M. Van Hove, P.I. Cohen: J. Vac. Sci. Technol. B2, 243 (1984)
10. J.M. Van Hove, P.I. Cohen: J. Cryst. Growth 81, 13 (1987)
11. B.A. Joyce, P.J. Dobson, J.H. Neave, K.Woodbridge, J. Zhang, P.K. Larsen,
B. Bolger: Surf. Sci. 168, 423 (1986)
12. B.A. Joyce, J. Zhang, J.H. Neave, P.J. Dobson: Appl. Phys. A45, 255 (1988)
13. P.J. Dobson, B.A. Joyce, J.H. Neave, J. Zhang: J. Cryst. Growth 81, 1 (1987)
14. Y. Horikoshi, M. Kawashima, H. Yamaguchi: Jap. J. Appl. Phys. 27, 169 (1988)
13 Low Energy Electron Diffraction
One of the most powerful techniques available for surface structural analysis is
low energy electron diffraction (LEED). It is widely used in materials science
research to study surface structure and bonding and the effects of structure
on surface processes. However because it usually requires single crystals and
ultrahigh vacuum conditions it has limited value for applied surface analysis,
which is often concerned with polycrystalline or amorphous materials. LEED
has many similarities to X-ray and neutron diffraction but it is preferred for
surface studies because of the short mean free path of low energy electrons
in solids.
A= !!. ~
p
J150
V
A (13.1)
where V is in electron volts and V ;5 1 keY. Thus for an electron with a kinetic
energy of 150 eV the de Broglie w-;;velength is ~ 1 A, which is similar to the
spacing between rows of atoms in a crystal. In 1927 Davisson and Germer [2]
carried out a systematic study of the scattering of electrons from Ni(1l1) and
found that the maximum in the reflected intensity of the elastically scattered
electrons at any angle a satisfied the plane grating formula
nA = asine (13.2)
,, ' ' , ,
\
Fig. 13.1. Typical LEED system showing the fluorescent screen S and the hemi-
spherical grids Gl, G2 and G3
•
• •
•
• • Fig. 13.2. LEED spot pattern for normal incidence on the
• (111) surface of eu with a primary beam energy of 80 eV
number of diffracted beams of electrons with the same energy as the incident
beam are produced in the backward direction. The spatial distribution of
these beams and their intensities as a function of angle and energy of the
incident beam provides information which can be used to analyze surface
structure. Thus, in contrast to X-ray diffraction in which the wavelength is
usually held fixed, in LEED there is this extra degree of freedom of changing
wavelength via changing energy.
It is customary in LEED to photograph spot patterns produced by the
surface layers as shown in Fig. 13.2 and to measure the intensity of one
or more beams as a function of the incident beam energy for a fixed angle
and azimuth of incidence as shown in Fig. 13.3. The LEED spot intensity
also varies as a result of other influences such as annealing and adsorption.
The analysis of LEED spot profiles at constant energy and incidence angles
can provide information about surface roughness, critical points for phase
transitions and surface structural changes during chemisorption. [15]
In practice the LEED experiment, like all surface experiments, is difficult
to carry out and difficult to analyze for the following reasons:
~
CJ)
z
UJ
f-
~
surface - (generally one of the lattice directions). The intensities are normally
collected by means of a spot photometer or by means of a microprocessor-
controlled video camera. The data must be processed before it can be used
for structure analysis. This involves:
a) normalisation to crystal incident current (since gun current varies with V)
b) contact potential difference correction (between sample surface and crystal
surface)
c) background subtraction
d) corrections for grid transparency as the spot moves across the surface.
reciprocal
~ lattice rods
-20 -10 40
10
Thus the LEED pattern which consists of the intersections of all of the
diffracted beams k' with the Ewald sphere is an image of the reciprocal
net when it is viewed along the normal direction to the surface at infinite
distance.
As the energy of the incident electrons increases the radius of the sphere
will increase and the spot pattern will shrink. Actually this form of the Ewald
sphere construction is only valid for scattering by a single layer. For 3-D scat-
tering, as in X-ray diffraction, the rods become points in the 3-D reciprocal
lattice. LEED also has some 3-D features as the electrons do penetrate some
distance into the crystal and the rods can therefore be thought of as having
3-D lattice points embedded in them. When the Ewald sphere intersects the
rod near a 3-D lattice point there will be a maximum in the intensity of that
spot. Such maxima are called Bragg peaks.
(13.8)
326 P. J. Jennings and C.Q. Sun
p
d UJ
Fig. 13.5. Formation of Bragg peaks
p in LEED
This is the condition for a Bragg peak in the diffracted beam v. For the
00 or specular beam it becomes
(13.9)
The Bragg peak locations are dependent on the surface structure of the crys-
tal and thus the aim of LEED intensity analysis is to deduce surface structure
from the measured intensity curves [10,11]. The usual approach is to postulate
a structural model of the surface and to use this as the basis for a simula-
tion of the LEED intensity spectra. The simulated and measured spectra
are compared visually or qualitatively and the model is refined to obtain the
13 Low Energy Electron Diffraction 327
best possible fit. The simulations are complex and the analysis is difficult and
indirect. The procedure involved is basically:
This produces a set of intensities for each of the diffracted beams as a function
of the energy and diffraction angles and for the assumed surface structure.
The analysis proceeds using simulation techniques in which the crystal model
is optimized to obtain the best fit of the computed and measured LEED
spectra as shown in Fig. 13.7. The process involved in the calculation is
(a)
(b)
~
~A
ii5
z
w B
~ ~~~~~~~~~~~L-~~~
(c)
~ B
(d)
similar to that used in band structure calculations for solids. In fact the
LEED intensity curves are dependent on the band structure for the solid
along a particular direction through the Brillouin zone corresponding to the
direction of the primary beam k o. The Bragg peaks occur at gaps in the
band structure (well above the Fermi level) while the secondary Bragg peaks
occur at partial band gaps. If an electron is incident on the crystal surface
at an energy where a band gap is present there are no propagating states in
the solid for it to enter and thus it is strongly reflected. Thus Bragg peaks
correspond to absolute gaps in the band structure. Secondary Bragg peaks
occur where the density of propagating states in the solid is low.
The LEED intensity curves also show a number of narrow, complex features
called LEED fine structure at very low primary energies. These features arise
from the scattering of diffracted beams by the surface barrier. They are found
only at low energies and they usually form a series of peaks converging on
the threshold energy for a new diffracted beam as shown in Fig. 13.8 [12].
The mechanism of production of these features is shown in Fig. 13.9. It
involves a new beam which has begun to propagate in the crystal but still
does not have sufficient energy to surmount the surface barrier. This beam is
totally internally reflected by the surface barrier and after subsequent diffrac-
tion by the crystal it may be diffracted back into the incident beam where it
~
(j)
z
W
f-
Z
W
>
~
--l
W
a:
10 14
ENERGY (eV)
Fig. 13.8. LEED fine structure for Cu(OOl) for an angle of incidence of 60° along
the (11) azimuth
13 Low Energy Electron Diffraction 329
can produce interference effects in that beam. The spacing of the interference
fringes is an indication of the structure of the surface barrier. The analysis of
the LEED fine structure enables us to deduce the structure of the barrier and
to map the valence band structure of metals and chemisorbed surfaces [16-
18]. This is an important component in the theory of many types of electron
emission phenomena from solids such as photoemission, thermionic emission
and field emission.
LEED is the basic structural technique for solid surface analysis for crystalline
materials. It has been applied to many different types of surface but usually
it is best suited to structural studies on clean, well-ordered surfaces of simple
metals and semiconductors.
The LEED pattern provides a map of the reciprocal lattice of the surface.
The (00) spot or specular beam can be identified as the spot towards which
all of the others gravitate as the energy of the electron beam is increased. At
normal incidence we can determine the rotational symmetry of the pattern by
inspection. We can also determine the size of the unit mesh from the pattern.
To do this we photograph the LEED pattern and measure b1 and b2 on the
photograph. From a knowledge of the geometry of the LEED chamber we
e
can calculate the diffraction angle given by
e
Hence from a knowledge of and)" we can calculate the direct mesh
lengths al and a2, and compare them with the predicted values. Thus we
can determine whether the surface and substrate nets are identical. This is
usually the case for most simple transition metal surfaces. However most
semiconductor surfaces show reconstructed unit meshes which are formed
330 P. J. Jennings and C.Q. Sun
because of the covalent bonds which are broken during the fracture of the
surface.
• • • • • •
• • • • • •
• • • • • •
Fig. 13.10. LEED patterns for normal incidence on clean Ni(OOl) and for c(2 x 2)
OjNi(OOl) with a primary beam energy of 80eV
The analysis of LEED intensities generally proceeds from the measured in-
tensity versus energy curves for the specular and several non-specular beams.
Because of the strong multiple scattering in LEED a dynamical theory is re-
quired. The simple kinematical analysis used in X-ray diffraction is unsuccess-
ful in LEED. Dynamical calculations involve heavy computational effort and
sophisticated simulation packages are widely available [19]. Studies have ex-
tended from the simplest metals to some complex structures using techniques
such as tensor-LEED [20] and diffuse-LEED [21]. A referential database for
ordered surface structures is also avalable on-line [22]. For example the posi-
tion of adsorbed selenium atoms on the v'2 x v'2 R45° Sel Ag(OOI) has been
13 Low Energy Electron Diffraction 331
studied by several authors. They all find that the adsorbed selenium atoms
are located at alternate fourfold silver sites [13]. Recently it has been shown
that LEED spectra can be inverted to produce the three-dimensional coor-
dinates of atoms neighbouring a reference atom without prior knowledge of
what types of atoms are present [23].
Similar studies were done for 0 on Ni but at first they disagreed about
the Ni-O bond length, [12]. Finally after considerable dispute it was found
that the 0 layer is located about 0.9 A above the topmost nickel layer [14].
This result has recently been disputed by researchers working with ion scat-
tering spectroscopy. In general the field is still in an immature state. LEED
intensity analysis is the most widely accepted surface structural technique
and it has been extended to the study of complex surfaces, although the
analysis requires considerably more computational effort and sophisticated
analytical skills [24]. Generally the adsorption site can be found readily and
clearly but the accuracy of the bond lengths and surface layer spacings are
unsatisfactory. Very little progress was made with the analysis of the struc-
ture of surfaces with large unit meshes or surfaces with adsorbed molecules
until recently when a computer code for multi-atom and multiple-diffraction
was developed [25]. In conjunction with scanning tunneling microscopy and
ultraviolet photoelectron spectroscopy, the band structure, bonding and the
non-uniform surface potential barrier for the O-Cu(OOl) surface have been
investigated [26,27]. On the basis of work done to date we can make some
useful observations about surface structure:
a) the most closely packed surfaces fcc(lll), bcc(llO) of clean transition
metals exhibit very little or no contraction of the top layer;
b) surfaces with intermediate packing density [e.g. fcc(OOl), bcc(OOl)] have
a contracted first interlayer spacing. The contractions are typically 2% to
10% depending on the material;
c) the more open surfaces such as fcc(llO) and bcc(lll) are strongly con-
tracted (of the order of 10-15 %). The contraction of surface layers provides
a justification for the bond-order - bond-length - bond-strength correla-
tion mechanism that may have important impact on the size dependency
of nanosturctured materials [28];
d) clean semiconductor surfaces are mostly reconstructed - although some
such as Si(l11) can be stabilized with minute amounts of impurity. Such
reconstruction is usually obvious in the patterns [e.g. Si(OOl) 2 xl, Si(l11)
7 x 7]. The structures are often very complicated and may involve four or
five layers;
e) the study of adsorbate structures has shown that surface bond lengths
are similar to gas phase covalent bond lengths and the surface is often
distorted in the process of adsorption.
f) for surfaces with chemical adsorbates, the first interlayer spacing often
expands by 10%-25%, and the second layer spacing contracts by 5%-
10%. However, the exact vertical positions of the adsorbates an~ still under
debate.
332 P. J. Jennings and C.Q. Sun
- Vo if z 2": Zo (a pseudo-
1 + A exp[-B(z - zo)] Fermi-z function)
{ (13.12)
Re V(z) = 1 - exp[.\(z - zo)]
if z < Zo (the classic
4(z - zo) image potential)
1m V(z) = 1m [v(z) x V(E)]
r
r
= 'Y x p(z) x exp [ E - E )]
'Y x exp [ E - E) ]
(13.13)
z-
l+exp [- - a -
Zl]
z
o Bulk
Fig. 13.11. One-dimensional model of the surface potential energy barrier showing
the real and imaginary parts as functions of the distance z from the surface
13 Low Energy Electron Diffraction 333
13.7 Conclusion
To a large extent LEED has fulfilled the expectations of its early advocates
such as Germer [4]. However it is now clear that it has certain strengths and
weaknesses. Its strengths include:
a) the ability to provide direct and accurate information about surface con-
tamination;
b) its suitability for studies of processes on ordered surfaces including chemi-
sorption, phase transitions and epitaxy;
c) its sensitivity to the surface barrier structure.
For these reasons LEED is one of the most popular research techniques in
surface science. However because of its limitations it is advantageous to use it
in conjunction with other compatible techniques such as scanning tunnelling
microscopy, ultraviolet photoelectron spectroscopy, thermal desorption spec-
troscopy, x-ray photo emission spectroscopy, Auger electron spectroscopy and
work function measurements. Together these techniques are capable of pro-
viding a comprehensive understanding of the atomic and electronic processes
on ordered surfaces.
References
1. C.J. Calbick: The Physics Teacher, May 1963, pp.I-8
2. C.J. Davisson, L.H. Germer: Phys. Rev. 30, 705 (1927)
3. G.P. Thomson: Engineering 126, 79 (1928)
4. L.H. Germer: J. Chern. Ed. 5, 1041 (1928)
5. W. Ehrenberg: Phil. Mag. 18, 878 (1934)
6. H.E. Farnsworth: Surface Chemistry of Metals and Semiconductors (Wiley,
New York 1959)
7. F. Jona, J.A. Strozier, Jr., W.S. Young: Rep. Prog. Phys. 45, 527 (1982)
8. G. Hitchen, S.M. Thurgate: Surf. Sci. 24, 202 (1985)
9. S. Thurgate, G. Hitchen: Appl. Surf. Sci. 17, 1 (1983)
10. J.B. Pendry: Low Energy Electron Diffraction; The Theory and Its Application
to the Determination of Surface Structure (Academic, London 1974)
11. M.A. Van Hove, S.Y. Tong: Surface Crystallography by LEED, Springer Ser.
Chern. Phys. Vol. 2 (Springer, Berlin, Heidelberg 1979); M.A. Van Hove,
W.H. Weinberg, C.-M. Chan: Low Energy Electron Diffraction, Springer Ser.
Surf. Sci. Vol. 6 (Springer, Berlin, Heidelberg 1986)
12. R.O. Jones, P.J. Jennings: Surf. Sci. Rep. 9, 165 (1988)
13. A. Ignetiev, F. Jona, D.W. Jepsen, P.M. Marcus: Surf. Sci. 40, 439 (1973)
14. J.E. Demuth, D.W. Jepsen, P.M. Marcus: Phys. Rev. Lett. 31, 540 (1973)
15. S. J. Murray, D. A. Brooks, F. M. Leibsle, R. D. Diehl, R. McGrath: Surf. Sci.
314, 307 (1994)
16. R. O. Jones, P. J. Jennings, O. Jepsen: Phys. Rev. B 29, 6474 (1984)
17. V. N. Strocov, R. Claessen, G. Nicolay, S. Hufner, A. Kimura, A. Harasawa,
S. Shin, A. Kakizaki, P. O. Nilsson, H. 1. Starnberg: P. Blaha. Phys. Rev. B
63, 205108 (2001)
13 Low Energy Electron Diffraction 335
R. Leckey
Elastically
Scattered
Electrons
(or unscattered)
--
hv
"
Vacuum Ek .... Scattered
Level _ - -- Electrons
---
No. of Electrons!
Unit Energy
Interval
Occupied
Density of States
Core
Level
Fig. 14.1. Illustrating the photo excitation of electrons from a valence band due
to monochromatic photons of energy hv. An idealised energy distribution is also
shown with primary emission distinguished from secondary electrons
Pd
EF 4 8 EF 2 4 6 8 10
Binding Energy (eV) Binding Energy (eV)
Fig. 14.2. UPS spectra of a series of Ag/Pd alloys taken with (a) 21.21 eV and
(b) 40.81 eV photons under angle integrated conditions. From [3]
340 R. Leckey
(14.1)
As the photon energy increases we may expect to find less structure in the
unoccupied density of states. At XPS energies, the photo-excited (final) den-
sity of states is assumed to be featureless and this assumption holds quite
well also at 40.81 eV. Thus both XPS and He 11 spectra may be expected to
mirror variations in the initial (occupied) density of states reasonably accu-
rately whereas a He 1 spectrum will represent a mixture of these initial and
final state functions which is in general difficult to interpret by inspection.
The strength of a transition depends on the orbital nature of both initial and
final states and experience tells us that emission from d band states is much
stronger relative to emission from s or p derived states at 40.81 eV than at
21.21 eV.
Armed with the above information, we can now assert that the He 11
Ag/Pd alloy spectra of Fig. 14.2 are likely to be a good representation of the
width and of the density of states of the conduction band. We can also pre-
dict that the most prominent structure comes from band structure states of
predominately 'd' nature. As a trivial example of the type of information that
follows from these results, we note the manner in which the onset of strong
'd'-band emission moves relative to the Fermi edge as the alloy composition is
varied. This in turn controls the optical absorption of these alloys and readily
explains the alteration in the colour of these materials with composition.
analyser &
z detector
····t:A
~--------- Sample
..,,-- -- - Input
Zoom Lens
hv
Toroidal
Sector
,
,
...-;ssssss "s,ss . . Intermediate
r
55
~22277
J 2 ( 22Z( ~
Focal Plane
~
I Output Lens
I
Chevron
I Channel Plates
"S~~~~!iii;===t==~~!i~tS~~J---
~ Position
Sensitive
Detector
Fig. 14.5. Angle resolved UPS data from eu (110) in the [112] azimuth. Structures
due to d- and s, p- derived bands are clearly visible together with a surface state at
r as described in the text
Unstrained Ino.2'1Gao.73As
I I I I I 1 I I I I 1 I
~~ b
~~
,........
....'j;!
til
0 f'6u
~~
::!
..ci
J... -1
.!.
....>.
'in
~~
.:: -2
.....\I) '>
.E .!. XSu
.:: ~ -3
~:
0
'in 1
Vl
'8\I) 11.0
~
-4
...,0
~::
0
.c:
A..
-5
-10 -5 0 r t. X
E - Ev reV]
Fig. 14.6. (a) Normal emission energy distributions curves from InGaAs (100) for
a range of photon energies. (b) Valence band dispersion curves derived from this
data
The small feature which shows some dispersion and is seen close to the Fermi
e
energy around = 0° is due to a surface state. Such states are not a feature
of the bulk band structure but exist as separate solutions of the Schrodinger
equation specifically associated with the surface itself. Surface states have
an important role to play in all surface sensitive situations e.g. catalysis,
adsorption and in the energy alignment of metal/semiconductor interfaces.
The major features of the E(k) band structure of a material can most
readily be experimentally determined by utilising a synchrotron radiation
facility. An example of a set of energy distribution curves obtained from
InGaAs over a wide range of photon energies at normal emission is shown
in Fig. 14.6, together with the experimental valence band structure derived
from this data. Details of the analysis needed to convert the raw data to E(k)
data may be found in [6].
344 R. Leckey
(a)
(b) w K
r x
Fig. 14.7. (a) Full hemisphere intensity pattern showing emission at the Fermi
energy from Cu (111) using 22 eV p-polarised photons. The presentation is shown
on a linear wave vector scale for direct comparison with the Fermi surface topology.
(b) The Fermi surface of Copper as determined by angle resolved photoemission
methods shown on the irreducible sector of the Brillouin Zone
14 Ultraviolet Photoelectron Spectroscopy of Solids 345
as good as de Haas van Alphen, but the ARUPS method can be performed
on standard crystals at elevated temperatures thereby opening up the study
of phase transitions. Similarly, the ARUPS method may be applied to sub-
stitutionally disordered alloys. An example from our laboratory is included
as Fig 14.7. Recently, the ARUPS method has been successfully applied to
the measurement of high Tc superconductors [9], to an investigation of the
order-disorder transition in CU3Au [10] and to the ferro- / para-magnetic
transition in Ni [11].
In this brief overview, there has not been space to consider UPS experi-
ments on adsorbates nor from ow dimensional samples, heavy fermion systems
etc. Similarly, it has not been possible to consider advances in understanding
the magnetic properties of materials using polarised photon sources and po-
larised electron detectors. More complete descriptions of the capabilities of
ARUPS may be found in the review articles [12,13].
References
15.1 Introduction
Fig. 15.1. Schematic XAFS experiment showing transmission detection using ion
chambers and a fluorescence detector. A typical ion chamber detector counter chain
is shown for the reference ion chamber
technique. A few laboratory based systems have been developed [9] depend-
ing on the Bremsstrahlung output of a rotating anode tube for a continuum
source, but are generally restricted to samples with a high concentration of
the absorbing element. From its inception, EXAFS for all intents and pur-
poses has been a purely synchrotron based technique, and is one of the tech-
niques which is made possible, as distinct from being greatly enhanced, by
the relatively recent availability of synchrotron radiation.
The critical energy of a nominal 2.5 GeV storage ring with a 1 Tesla bending
magnet field (a typical second generation facility) is thus around 4 keY. As a
rule of thumb, useful X-ray intensity can be obtained to approximately five
times the critical energy, or to 20-25 keY from the above 2.5 GeV machine.
Synchrotron radiation has a number of advantages for EXAFS over a
conventional source:
0, - Ity
11.--
Beamline Acceptance
U: -----\
X
::::J
• Wigglers have high fields or long magnetic periods so that the amplitude
of the oscillations exceeds the transverse size of the electron bunch. The
synchrotron emission from each oscillation adds and the resulting spec-
trum is that of a bending magnet of the same field multiplied by the
number of poles in the wiggler .
• Undulators have smaller magnetic periods, so that the amplitude of the
electron beam oscillation is less than the beam size. The emission from
each oscillation adds coherently resulting in a energy spectrum composed
of a series of harmonic peaks and an X-ray intensity proportional to the
square (in the ideal case) of the number of poles.
------a
Ring Monochromator
#f.("-------.e::::=->-~
Sample
"" Mirror
Fig. 15.4. Schematic synchrotron XAFS beamline
1. The high order reflection is intrinsically narrower, and very slightly offset
in angle from the fundamental. Thus a small angular misalignment of the
two monochromator crystals, called detuning, results in a small decrease
in the intensity of the fundamental and a much greater decrease in the
harmonic intensity. Reductions of over 100 times in the harmonic fraction
of the beam can be easily achieved with this technique.
2. A beamline mirror acts as a low pass energy filter. An ideal mirror has unit
reflectivity for X-rays below the critical angle for total external reflection,
and the reflectivity falls rapidly above this angle. The critical angle varies
with X-ray energy, and with the mirror coating, so that the mirror high
energy cutoff can be tuned to reject the unwanted harmonic energy.
uniform samples are required for example, and many samples are not suited
to transmission measurements. Additionally, small-angle scattering from the
sample superimposed on the energy "fan" of the transmitted beam can se-
riously degrade the energy resolution and render the data unusable. This
problem can be addressed by inserting a scanning slit after the sample at the
expense of some time resolution (the slit can also allow fluorescence detection
to be used). Despite these potential problems, structural changes can often
be tracked quite successfully from a known starting point.
QEXAFS is in essence fast scanning conventional EXAFS [13,14]. Instead
of the step-count-step approach of normal EXAFS, the monochromator is
continuously scanned and data recorded "on the fly". QEXAFS can equally
easily be performed in transmission or fluorescence modes. The achievable
time resolution depends on the available X-ray flux and on the stability of
the monochromator mechanism which will determine how fast it can be driven
and still produce a stable X-ray beam. Time resolutions of seconds to tens of
seconds are typical.
15.2.3 Detectors
A typical hard X-ray (i.e. photon energy above:::::; 4 ke V) EXAFS experimen-
tal setup is shown schematically in Fig. 15.1. The simplest EXAFS technique
is the transmission configuration, where the incident and transmitted X-ray
intensity is measured by gas ionisation chambers before and after the sample.
The X-ray absorption in the gas filled detector produces electron-ion pairs,
which are swept apart by a voltage applied between two parallel plates. If the
voltage is above the saturation level where recombination of the electron ion
pairs is prevented, and below the avalanche region, then the output current is
proportional to the X-ray intensity. The applied voltage is usually in the 500-
1000 V range for a 1 cm electrode separation. As can be seen in Fig. 15.1, the
output current is amplified and digitised by a voltage to frequency converter.
For optimum EXAFS data collection, the 10 detector should absorb about
25% of the X-ray beam, and the 1 detector 100% of the beam transmitted
by the sample. This can be achieved over a wide range of X-ray energies by
changing the gas mixture in the detector. Inert gases (He, Ne, Ar, Kr and
N 2 ) and their mixtures are normally used. In some cases a third ion chamber,
labelled the reference ion chamber in Fig. 15.1, is used in combination with
a reference foil to correct for any drift in the energy scale due to instability
of the monochromator, or other beamline elements.
Transmission EXAFS using ion chamber detectors is the method of choice
due to its superior signal-to-noise ratio but is limited in application to samples
containing a relatively high weight fraction of the absorbing element. Typ-
ically if a sample contains a weight fraction of absorber in excess of about
5%, it can be examined by the transmission method. For more dilute ele-
ments (and in some cases for surface species on highly absorbing samples)
fluorescence detection is normally used. This is almost always the case with
15 EXAFS 355
0 -- - 0
represent the outgoing photo-electron. The dot-dashed
circles are the backscattered electron wave from a
neighbouring atom
358 R.F. Garrett, G.J. Foran
O.B
Q)
C)
Q
til 0.6
.0
1-<
o
rn
~
0.4
0.2
12.0 12.5 13.0
Photon Energy (keV)
Fig. 15.6. Raw transmission XAFS spectrum of amorphous GaAs, taken at the
arsenic kedge [52J
10
Ul
~~
5
'"d
.3 01------..
..c1
....tlO
Q)
Ii: -5
I
'".!:s::
-10
-15~~~~~~~~~~~~~-L~~~~~~~
o 5 10 15 20
Photoelectron Momentum (X-1)
Fig. 15.7. k-weighted XAFS Extracted from the raw spectrum in Fig. 15.6. [52J
15 EXAFS 359
°Ob=~~~~1~~~-L2~~~~3~~~=4~--~~5
15.3.1 EXAFS
Theoretical attempts to explain EXAFS began soon after the first observa-
tions of X-ray absorption spectra [24]. EXAFS theory is now very mature,
and excellent and detailed reviews are available [1,3,25,26].
The usual qualitative explanation of EXAFS is illustrated in Fig. 15.5. An
incident X-ray photon is absorbed by an atom which emits a photo-electron.
The outgoing waves are backscattered by the neighbouring atoms and inter-
fere at the emitting atom, resulting in a small oscillation superimposed on
the X-ray absorption spectrum. The oscillations derive from the interference
being progressively constructive and destructive as the photo-electron energy
increases away from the absorption edge. EXAFS is seen for all but isolated
atoms (although recent work has identified fine structure due to scattering of
the photoelectron from the absorber itself i.e. atomic EXAFS [27,28]), and
was first observed in the 1920s [24]. The scattered electron wave undergoes
phase shifts at the scatterer and absorber atoms; if this were not the case
a simple Fourier transform would peak at or near the interatomic distances
of the neighbouring shells of atoms. Much of the task of EXAFS analysis
deals with determining the phase and amplitude functions of the scattering
processes.
360 R.F. Garrett, G.J. Foran
Here N is the number of atoms in the ith shell, A is the scattering amplitude
function, k is the photoelectron wave vector, ).. is the magnitude of the pho-
toelectron mean free path, a is the Debye Waller factor and ¢ is the phase
function. As can be seen from (15.2), the X-ray energy is converted to k in
all EXAFS data analysis, via:
Where me is the electron rest mass, E is the X-ray energy and Eo is the
absorption edge or threshold energy. Equation (15.2) consists of a sinusoidal
term including phase shifts, and exponential terms reducing the oscillation
amplitude due to the mean free path of the photoelectron, the thermal and
static disorder (represented by the De bye-Waller factor), and with k 2 . EX-
AFS data analysis uses (15.2) as a fitting equation, with experimentally or
theoretically determined phases and amplitudes, to fit a model structure to
the measured EXAFS.
Note that the use of a Debye-Waller factor to represent disorder assumes
that the disorder can be approximated by a Gaussian distribution, i.e. it is
symmetric. While this assumption is adequate in the majority of situations,
a more accurate expression which can incorporate asymmetric distributions
is sometimes required. This is clearly the case for surface atoms for exam-
ple, whose bonding environment is likely to be significantly asymmetric. In
these cases a commonly used technique is the cumulant expansion method
[26,30,31]' where higher order terms are summed to include asymmetry in
the disorder distribution.
Polarisation effects are not included in the above treatment. XAFS is
in fact polarisation sensitive [32], with a cosine squared dependence on the
angle between the electric field vector of the X-rays and the absorber-scatterer
axis. This can normally be ignored, as most EXAFS samples are randomly
oriented e.g. polycrystalline powders or solutions. With single crystal samples
or surface species, however, polarisation must be included in the analysis. In
this case EXAFS can give direct information on bond angles and molecular
orientation.
Equation (15.2) is essentially a single scattering plane wave approxima-
tion, which is usually sufficient to characterise the first shell of neighbouring
atoms. Beyond the first shell, multiple scattering is often significant. Fortu-
nately multiple scattering can now be routinely included in EXAFS analysis
15 EXAFS 361
15.3.2 XANES
ing the incidence angle of the X-ray beam, the orientation of the adsorbed
molecule on the surface can be easily determined.
The next step in the EXAFS analysis process is the extraction of the EXAFS
oscillations from the raw absorption edge spectrum. In principle, this involves
the removal from the observed data of the atomic (isolated atom) contribu-
tion to absorption. In practice, however even if the atomic contribution were
known, the situation is complicated by the contributions to background ab-
sorption or scatter from other elements in the beam path. For this reason
background subtraction is most often achieved via curve fitting means rather
than by theoretical calculation of the atomic contribution.
364 R.F. Garrett, G.J. Foran
The EXAFS signal, extracted and k-weighted, consists at this point of sums
of damped sine waves corresponding to the different shells of atoms repre-
sented in (15.2). The next step in the analysis is therefore to Fourier transform
the oscillations, producing a pseudo radial distribution function as shown in
Fig. 15.8, with peaks whose position represents the interatomic distance of
each shell of atoms, and whose amplitude is roughly proportional to the co-
ordination number. Were it not for the phase shifts in the electron scattering
process, the Fourier transform would yield a true radial distribution func-
tion. In addition to peaks corresponding to actual shells, extra peaks can be
present in the Fourier transform due to the truncation of the EXAFS oscilla-
tions and multiple scattering contributions, particularly at larger interatomic
distances.
the number of fitting parameters, the data can be decomposed into compo-
nents corresponding the contribution from individual shells by the process of
Fourier filtering. Fourier filtering also permits decomposition of the EXAFS
oscillations into amplitude and phase components.
In Fourier filtering, a window function is applied to the FT of the observed
EXAFS in such a way so as to isolate the peak in the FT corresponding to a
single shell. All Fourier components outside the window function are set to
zero. As a result, when the windowed FT components are back transformed,
only the EXAFS deriving from that shell of atoms remains and this much
simplified signal can be more easily handled.
Consecutive Fourier filtering, back transformation and refinement of indi-
vidual shells greatly simplifies the structural refinement and assists in avoid-
ing false minima during refinement which can often result when too large a
number of parameters are simultaneously refined.
~
·iii
~ 0.8 c
·iii 2
c E
Q)
C Q)
()
Q) c
() Q)
c ()
en
tl
Q)
~
0
0.4 :::J
'ai
a: u:
o~==~~~------~----------~====--------~o
o 100 400 500
Fig. 15.9. Titania sol gel film X-ray reflectivity and Ti K ex fluorescence yield
variation with incidence angle at an X-ray energy of 5.2 keY
critical angle, has been used for some time to study surfaces and interfaces
[44-48]. EXAFS is present in both the reflected beam and fluorescence emis-
sion, although fluorescence detection is generally superior [40,41]. The main
limitation of the technique is that flat optically smooth samples or substrates
are required for optimum surface sensitivity, and relatively large samples are
needed because of the large beam footprint at the grazing incidence angles,
which are generally below 0.5 degrees.
Glancing angle EXAFS has been used by Hanley etal. [49,50] to study
nanocrystalline titania thin films, which have applications in photovoltaic
cells, batteries, electro chromic devices and sensors. There are also possible
quantum size effects in these films where the particle size is small. The films
were prepared by dipping glass substrates into colloidal anatase sols with a
particle size distribution in the tens of nano-metres. The resulting films were
50 nm thick, and were fired at various temperatures. EXAFS spectra were
measured at the Ti K-edge at grazing incidence using a 100 micron high beam,
and Ti fluorescence was measured with a Ge SSD. The calculated penetration
depth of the X-ray beam at the critical angle of 0.36 degrees was about 10 nm.
Figure 15.9 shows the X-ray reflectivity and Ti fluorescence emission versus
the angle of incidence and is a good illustration of the grazing incidence
technique. The sharp fall in reflectivity at the critical angle can clearly be
seen, which coincides with a peak in the fluorescence yield as the X-ray beam
begins to fully penetrate the film. At higher angles the fluorescence yield
drops as the X-ray beam penetrates further into the substrate and less is
368 R.F. Garrett, G.J. Foran
10r---~----~~------~----------~--~
-10
0 4 8 12
k
10
I,
8 I
I
I
I
Q) I
<1l 6
E
I
l-
Ll..
R+.1.(k')
Fig. 15.10. Measured (solid) and theoretical (dashed) EXAFS from an air dried
titania film. The top panel is the k-weighted EXAFS and the bottom is the Fourier
Transform. This and Fig. 15.11 illustrate how the measured EXAFS can be fitted
in either k-space or real space
absorbed in the thin film. To obtain maximum signal to noise, the EXAFS
was measured at the peak of the fluorescence yield, at about the critical angle.
Figures 15.10 and 15.11 compare Fourier filtered observed and calculated
data for films dried in air and fired at 800 C respectively. Anatase, the tita-
nium mineral in the sol, has four coordination shells (0- Ti-Ti-O) at distances
of 2-4 A from the absorbing Ti atom. The EXAFS at room temperature is
consistent with a distorted anatase structure, with high degrees of disorder in
all but the first shell. Increased crystallinity is apparent in films fired at higher
temperatures, as shown in Fig. 15.11 for the film fired at 800°C where the
decrease in Debye-Waller factors for this higher shells produces much higher
15 EXAFS 369
10r-------~~----------~----------~--_,
-5
10.-------------------------------------_,
C>
ro
E
~
LL
R+t.(k')
Fig. 15.11. Measured (solid) and calculated (dashed) grazing incidence Ti K-edge
EXAFS of titania film fired at 800°C. Top: k-weighted EXAFS; Bottom: Fourier
transform
Fig. 15.12. Fourier transforms (non phase corrected) of XAFS data showing the
transition from crystalline to amorphous Ge induced by ion implantation. The ion
dose ranges from zero (crystalline Ge) to 2 x 10 14 ions/cm 2
15 EXAFS 371
<c2.~5 .-----------~--------~-r----,
..c
"&
~ 2.~0
=0
c
o
.0
~ 2.455
o
.0
..c
Cl
.~ 2.450
Vi
Q)
ro
Q) 2.445
c
Q)
<.'J
1013 10 14
Ion Dose (ions/em1
Fig. 15.13. Change in nearest-neighbour bond length in amorphous Ge with im-
planted ion dose
References
5. See for example the web site of the International XAFS Society at
http://ixs.csrri.iit.edu/IXS /index.html
6. D.E. Sayers, E.A. Stern, F.W. Lytle: Phys. Rev. Lett. 27, 1204 (1971)
7. see for example "Applications of Synchrotron Radiation", eds. H. Winick,
D. Xian, M. Ye, T. Huang, Gordon and Breach, New York, 1989
8. B.M. Kincaid, P. Eisenberger: Phys. Rev. Lett. 34, 1361 (1975)
9. Y. Udagawa, [3, p. 130]
10. J. Wong, Z.U. Rek, M. Rowen, T. Tanaka, F. Schafers, B. Muller, G.N. George,
I.J. Pickering, G. Via, B. DeVries, G.E. Brown, M. Froba: Physica B208/209,
220 (1995)
11. T. Matsushita, R.P. Phizackerley: J. App!. Phys. 20, 2223 (1981)
12. M. Hagelstein, C. Ferrero, M. Sanchez del Rio, U. Hatje, T. Ressler, W. Metz:
Physica B208/209, 223 (1995)
13. R. Frahm: Nuc!. lnstrum. Methods A270 (1988) 578; R. Frahm: Synchrotron
Radiation News 8 38 (1995)
14. L.M. Murphy, B.R. Dobson, M. Neu, C.A. Ramsdale, R.W. Strange, S.S. Has-
nain: J. Synchrotron Rad. 2, 64 (1995)
15. F.W. Lytle, R.B. Greegor, D.R. Sandstrom, E.G Marques, J. Wong,
C.L. Shapiro, G.P. Huffman, F.E. Huggins: Nucl. Instrum. Methods 226, 542
(1984)
16. S.P. Cramer, O. Tench, M. Yocum, G.N. George: Nuc!. Instrum. Methods
A266, 586 (1988)
17. J. Stohr, R. Jaeger, S. Brennan: Surf. Sci. 117, 502 (1982)
18. F. Farges, J.-P. Itie, G. Fiquet, D. Andrault: Nuc!. Instrum. Methods BIOI,
493 (1995)
19. D.J. Thiel, P. Livins, E.A. Stern, A. Lewis: Nature 362, 40 (1993)
20. F.A. Schultz, B.J. Feldman, S. Gheller, W.E. Newton, B. Hedman, P. Frank,
K.O. Hodgson: "Redox Mechanisms and Interfacial Properties of Molecules of
Biological Importance V", Proc. Vo!' 93-11, F.A. Schultz and I. Taniguchi, eds.
Pennington N.J.: The Electrochemical Society, Inc. pg. 108, 1993
21. I. Ascone, A. Cognigni, M. Giorgetti, M. Berrettoni, S. Zamponi, R. Marassi:
J. Synchrotron Rad., 6, 384 (1999)
22. A. Yoshiasa, T. Nagai, O. Ohtaka, O. Kamishima, O. Shimomura: J. Syn-
chrotron Rad. 6, 43 (1999)
23. D. Bazin, H. Dexpert, J. Lynch: [3, p. 113]
24. An interesting and informative history of XAFS is given by Lytle, J. Syn-
chrotron Rad. 6, 123 (1999)
25. S.J. Gurman, J. Synchrotron Rad. 2, 56 (1995)
26. E.D. Crozier: Nuc!. Instrum. Methods B133, 134 (1997)
27. D. Koningsberger, B. Moget, J. Miller, D. Ramaker: J. Synchrotron Rad. 6,
135 (1999)
28. D.E. Ramaker, W.E. O'Grady: J. Synchrotron Rad. 6, 800 (1999)
29. E.A. Stern: Phys. Rev. B 10, 3027 (1974)
30. G. Bunker: Nuc!. Instrum. Methods 207, 437 (1983)
31. E.D. Crozier: [1, Chap. 9]
32. C. Bourder: J. Phys. Condens. Matter 2, 701 (1990)
33. J.J. Rehr, R.C. Albers, S.1. Zabinsky: Phys. Rev. Lett. 69, 3397 (1992)
34. S.1. Zabinsky, J.J. Rehr, A. Ankudinov, R.C. Albers, M.J. Eller: Phys. Rev. B
52, 2995 (1995)
15 EXAFS 373
R.St.C. Smart
Structural Information
Imaging Latices:Sites
Optical Microscopy XRD
1 1
SEM:SAM XRD (glancing angle)
1 1
FESEM XAFS:LEED
1 1
AFM; STM; TEM LEIS
Chemical Information
Near Surface Surface Molecular
EDS:XRF XPS FTIR:Raman
1 1
XAFS:RBS:NRA SAM
1
SIMS (ToF)
1
STS(STM):LEIS
16.1 Minerals
Fig. 16.1. High resolution TEM micrograph of part of a large gibbsite (AI(OHh)
crystal, from bauxite pisolites, showing lattice fringes (striations). Some fringes are
fragmented due to either disorder in the lattice or surface roughness (or both).
The surface shows pits (Fig. 16.4) and protusions over distances of order 2-3nm
consistent with the high surface area of this material. The dark patches (ca. 2nm
diameters) show high Fe/AI ratio consistent with iron oxide particles (as hematite
and goethite) attached to the gibbsite surface. From [1]
16 Minerals, Ceramics and Glasses 381
200nm with very small (i.e. 1-10 nm) iron-containing regions dispersed across
their surfaces. Colloidal particles of hematite and goethite adhere to the sur-
face of the larger platey minerals. They are not removed by ultrasonic agita-
tion but reductive dissolution in dithionite dissolves them (at different rates
for hematite and goethite). XPS confirms removal from the surface [1].
The efficiency of methods for separation of iron oxides can thus be moni-
tored using XPS with electron microscopy.
In very many cases (perhaps even a majority), the surface of a mineral has nei-
ther the same composition nor structure as its bulk. This can be due to reac-
tion with air (e.g. oxidation of sulfides), groundwater (e.g. silicate, carbonate
depositions), contamination (e.g. humics, hydrocarbons), intergranular lay-
ers (e.g. graphitic) exposed on grinding, or adsorbed species (e.g. sulfates).
These alterations to the surface can profoundly influence the behaviour of
the mineral in a process.
For instance, a glacial sand deposit to be used for glass-making had unac-
ceptable dewatering and melting properties very different from the dune sand
it replaced. XRD and SEM/EDS found only quartz with very little clay and
no significant impurities in the quartz after washing. XPS consistently re-
vealed 2.5-5.0 at.% AI, with an aluminosilicate binding energy (not adsorbed
AI3+) in the surface. Ion etching suggested that this layer was no more than
200 A thick. It could not be removed by any practical base-hydrolysis reac-
tion but choice of a surfactant suitable for an aluminosilicate surface reduced
dewatering to an acceptable time and adjustment of melting conditions for
the presence of the layer allowed acceptable processing [2].
Valuable sulfide minerals are usually separated from their waste (or gangue)
minerals by froth flotation. The process involves chemical conditioning of
the valuable mineral surface by selective adsorption of collector molecules
designed to render the particle sufficiently hydrophobic to attach to a bubble
and collect in the froth concentrate. Minerals separated in this way include
pyrite FeS2, chalcopyrite CuFeS2, galena PbS, sphalerite ZnS and pentlandite
(Fe, Ni)g S8.
The process is complicated by such considerations as:
Hence, the complex mixture of minerals and their interactions via dis-
solved species in most real ores makes the measurement of surface changes
difficult to interpret using any technique. Nevertheless, a respectable body of
literature has now been developed (e.g. [3-12]) in which some useful obser-
vations from surface analysis can be found. Studies, particularly using XPS
and FTIR, of the effects of conditioning agents (e.g. metabisulphite, sulfide,
cyanide), activators (e.g. Cu 2+) and collectors (e.g. alkyl xanthates, dithio-
phosphinates) on samples derived from processing plants and from synthetic
minerals and ores have elucidated some of the mechanisms of flotation sepa-
ration. A recent review [13] has summarised the application of surface ana-
lytical techniques to studies of sulfide mineral surface oxidation and collector
adsorption.
An example can be seen in the effect of xanthate collector on an ore com-
prised predominantly of chalcopyrite, Cu(I)Fe(III)S2' pyrite Fe(II)S2' quartz,
dolomite and iron oxides [14]. Table 16.2 shows the composition of the sur-
face at Eh + 200 mV before (Test 2) and after (Test 4) addition of sodium
butyl xanthate (NaC 4 H g OCS 2) at 0.04 kg/tonne. Analysis of the flotation
concentrates and tails (non-floating) is compared before and after ion etch-
ing to a depth of 25 nm. In the concentrates, several changes are evident after
xanthate addition:
The tails show little evidence of S (and only as sulfate species) or Cu but
Fe(III) (as hydroxides only) is found in both cases. Much more carbonate is
found in the tails samples. The xanthate appears to have a cleaning action
partly removing surface hydroxides and exposing the underlying mineral sur-
faces. Other studies, at different Eh values in solution (i.e. -400, -200, 0,
+400 mY) show that Eh + 200 mV is the most favourable condition for this
effect of the xanthate. More detailed examination of the elemental regions in
the XPS spectra can identify, for instance, the xanthate contributions to the
C, 0 and S regions, sulfide/sulfate ratios, and different forms of Fe.
16 Minerals, Ceramics and Glasses 383
attenuation of signal
'.
5
,.. , \." ... ~, ........... .+-
......
~ 4 .....~ ''',
: ~
..... ., ... , , ........ ,. ... : ......... .
w ~
..~ .. i.: ....
o ~~~~~~~~~~~~~~~~~~~~~~~~~.u~~~~~..~.
-740 -736 - 732 - 728 -724 -720 - 716 - 712 -708 -704 - 700
BINDING ENERGY, EV
Fig. 16.2. Fe2p (4 at.% abundance) XPS spectra from two copper ore flotation
concentrates from Mt. Isa (Australia) conditioned for 10min. at Eh - 400 mV with
no collector addition (Julliine) and Eh + 200 mV, 0.04 kg/t butyl xanthate collector
addition (dotted line). The low Eh sample shows only broad iron hydroxide species
on the surface. (Fig. 16.5) The collector-modified ore reveals new signals near 707 eV
and 720 eV from clean chalcopyrite (CaFeS2) and pyrite (FeS2) surfaces.
384 R.St.C. Smart
SAM, SIMS and STM have been used to image and analyse oxidation
products in a number of different forms and processes, namely:
a,
c. d.
Fig. 16.3. STM images from cleaved galena (PbS) surfaces after (a) 270 min in air
(70 x 70 nm area; upper image, top view; lower image 3D rotated; z scale 3.8nm). (b)
60 min air-purged pH 6 solution (500 X 500 nm area; top view image; z scale 2.6 nm).
(c) in acetate buffer pH 4.9 , applied potential +281 mV SHE (2000 x 2000 nm area;
side view image; z scale 450 nm). (d) 30 min in 10- 3 M Pb 2 + at pH 7 (500 x 500 nm
area; top view image; z scale 13.7 nm).
386 R.St.C. Smart
I I I I
3000 2500 2000 1500
Wavenumbers (em")
A. Styrene / Kaolinite C. Acrylic Acid/Kaolinite
B. Methyl Vinyl Ketone/Kaolinite D. Acrylic Acid/Talc
Fig. 16.4. FTIR spectra from polymer-coated clay particle surfaces (i.e. kaolinite,
talc) activated in a low-temperature, low-density Ar plasma (10- 1 torr; 50W) for 5-
30 mins followed by admission of the monomer vapour (i.e. styrene, methylvinylke-
tone, acrylic acid).
16 Minerals, Ceramics and Glasses 389
AFM imaging of the lattice structure of the kaolinite surface at the molecular
level has revealed details of these sites of attachment of the silane molecules.
Improvement of the bonding of mineral particles into polymer matrices
is reliant on improvements in the mechanical and chemical properties at the
particle/polymer interface. With unmodified clay particles, this interface is
often subject to disbonding (or "pull-out"), hydrolysis resulting in weaker
bonding, preferential fracture and, in some cases, reduced brightness or even
discolouration. The low-temperature plasma process can also be used to poly-
merise a variety of monomeric species and induce improved bonding to the
activated mineral surface. The presence of the polymer films on the mineral
surfaces can again be confirmed using both XPS and FTIR. Figure 16.4 gives
some examples of FTIR spectra from polymer coatings applied to kaolinite
and talc surfaces. Polystyrene vCR absorptions in the range 2800-3150 cm- 1
and corresponding chain vibrations below 1800 cm- 1 confirm the presence
of the surface layer and can be correlated with C Is XPS spectra from this
layer. The polymer can be chosen to provide chemical compatibility with the
corresponding polymer matrix, e.g. polyacrylate, polymethylvinylketone [35].
16.2 Ceramics
a - - - - unleached
----leached Ca
/ 80
_--6-------a.-----......o. - Ca 11... 70
7
o
•
6 Ca
--'-L -0 o. 60
Zr
t
Nominal
§
5
./ Stoichiometry 50
"0
.0 §
'" 4 40 {ij
#- #-
o--_-..ll------c
3 z~r
_ _ _ _ -C--;-~AI,. - Zr
30
- AI v,.
AI Ti
O. 20
.
2 11~=--- - Ti
--- -
¢-
~
-~--~~-~----~
/
Sa
. • • •
-6a<>.
10
10 100 1000
Dose ("A-min)
,. ,.
Fe
/
x 0.5
,.
,.
0
Sr -----
---
0_ _ -
0
i
"
••
------~==~==~==~=~~==~
0.1
F~_--
.~
_-- v v 9'l
.... ~ $Ii
/ "IF-~
eNS,.
-- _-
OL-~~--L--~~---1~0L-------~10~0~------~1~0~00~---J
N/
,
Dose ("A-min)
Fig. 16.5. (a) XPS depth profiles from the ceramic Synroc C matrix (i.e. major
phase) elements before (open symbols) and after (closed symbols) hydrothermal
attack for 25 days at 350 0 C. Nominal average bulk stoichiometries are marked as
bars on the right with % abundance scales for Ba, AI, Zr, Ca at left and Ti, 0 at
right. 1~A.min ion etch is ca. 0.2nm removal. (b) XPS depth profiles for simulated
waste species in the same specimen as Fig.16.5a recorded under the same conditions.
Uncertainties in % abundance are of order 0.2at.%. From [42].
392 R.St.C. Smart
Fig. 16.6. TEM micrograph of Ti02 (anatase A and brookite B) crystals formed
in-situ from perovskite (CaTi0 3 ) grains in the surface of a Synroc B ceramic. An
ion beam-thinned disc of Synroc B was subjected to hydrothermal attack in water
at 190°C for 1 day. The original perovskite grain size was ca 1 [lm in diameter.
The other less-reacted phases are hollandite (H, BaAbTi 6 0 16 ) and zirconolite (Z,
CaZrTb07).
Surface layers of ceramic materials are often applied to metal, oxide and poly-
mer surfaces to impart corrosion resistance, thermal or electrical insulation
and, in bioceramic applications, layers that protect the substrate, harden the
interface and induce biocompatibility for tissue or bone growth. A variety of
methods for applying these ceramic surface layers are now used including: pre-
cipitation or sol-gel deposition followed by calcination; physical vapour depo-
sition; chemical vapour deposition; and plasma reactions either in low density,
low temperature processes or as high temperature plasma sprays. The com-
position, particularly stoichiometry and impurity levels, structure, integrity,
continuity, conformity, interfacial bonding, interfacial flaws and porosity de-
termine the performance of the ceramic layer and, in turn, the coated device
for material in application. For this reason, detailed information is required
16 Minerals, Ceramics and Glasses 393
1611r-----~------~------7r~_,~~----~
o silicates
I'W alumino·silicates
L...3 part tetrahedral AI
• alumino·silicates tetrahedral AI
1~~~L-~7-----~------~
10~
2------------~101
Fig. 16.7. Modified Auger parameter plot for a variety of silicates (from [47]) and
TEOS plasma-reacted oxidised nickel surfaces.
50nm
Fig. 16.8. TEM photograph of edge of silica thin film produced on nickel by plasma
assisted chemical vapour deposition of TEOS:Water (Etch Type Specimen).
16.3 Glasses
Surface analysis has been used extensively, particularly SEM, XPS, SIMS and
FTIR, to characterise glass surfaces before and after weathering or aqueous
attack (e.g. [49-51]). Like ceramics, attack by water at glass surfaces proceeds
by four processes: ion exchange in the first few atomic layers; the formation
of an amorphous, so-called siliceous layer from base-catalyzed hydrolysis dis-
rupting the network bonding; surface segregation of particular elements into
the hydrogen-bonded, hydroxylated siliceous layer; and recrystallisation in-
situ or precipitation from solution [41]. These reactions are important in con-
trol of glass technology such as surface powdering, mould transfer, opacity,
surface hardening and corrosion.
100~
(b)
;g
~
~ O~----~-- __~____~__~____~__~
c
11O~rv:s:J
100~
(d)
OL-________~____L __ __ L_ _~_ _~
4000 3000 2000 1600 1200 800 400
Wavenumber (cm-1)
Fig. 16.9. Infrared spectra from the surface of a nuclear waste glass (a) initially;
(b) after evacuation (::::0 10- 5 Torr) at 170° C for 1 h; (c) after hydrothermal reaction
at 200°C for 18 h, drying at 200°C for 1 h in air and cooling in a vacuum dessicator;
(d) sample (c) evacuated at 170°C for 1 h. Note the changes in the hydrogen-
bonded Si(OH)x vibrations in the broad absorptions near 3500-3300 cm- I . The
sharp shoulder near 3700cm- I arises from free (not H-bonded) OH groups. The
broad absorption in (a) and (d) near 1640cm- I is the bending vibration of trapped
H20 molecules in the reacted layer. From [50].
(or 1l.7 at.%) with 14 other cations, including Na, Mg, Ca, Fe and Ni,
was reacted at 170-200°C in water. Figure 16.9 illustrates IR spectra, from
both in-situ (ATR) and detached (transmission) samples of the altered hy-
drolyzed layer. The initial v(OH) absorption on the unattacked surface near
3500 cm- I is entirely removed by evacuation at 170°C. After aqueous reaction
at l70°C for 60 h, strong, broad, hydrogen-bonded v(OH) (3200-3500 cm- I )
and 'Y(H 2 0) (near 1600 cm- I ) absorption is found. Evacuation removed some
of the 'Y(H 2 0) absorption but not the v(SiO) region around 1000cm- 1 [52]
indicating possible crystalline products within the siliceous layer.
Figure 16.10 shows XPS profiles for the same glass before and after a
relatively short duration of aqueous attack at 200°C for 1 hour. Note that
16 Minerals, Ceramics and Glasses 397
50
+
y--e
40 40 _ _- - - e O
e - e - - -__ eO
"E
e I
Q)
~30
<>
·E
+
f~·-- \..-+-- +-
0
as
~
r"-"
~ 20 -+Zn
-+Zn Q)
[j
-----.Si
• / " _____ " ________ " 5;
10 10
•
•
0
o 2 4 6 8 10 0 20 40 60 80 100
a Milling time b Milling time
~~ ~~
Fig. 16.10. XPS depth profiles (~lnm min-I) of 0, Zn and Si in the surface of
a polished disc of a zinc-containing nuclear waste glass (a) before and (b) after
hydrothermal attack in water at 200°C, 1 h. Note change of depth scale from (a)
to (b). From [50].
the Si and 0 signals are relatively unaltered in the siliceous layer but there is
a strong accumulation of Zn in the first 10 nm of the surface. SIMS profiles of
longer duration attack have shown Zn accumulation in the surface to depths
greater than 100 nm. Other elements, notably Mg, Fe and Ni, are also found
segregated into the siliceous layer in this and other glasses after aqueous
attack whilst the alkali metal cations (e.g. Na+, K+, Rb+) are almost always
lost from the reacted layer.
SEM studies of the surface after extensive reaction show a relatively uni-
form (conducting) surface layer with small crystallites embedded in it as
illustrated in Fig. 16.11. TEM with electron diffraction, however, established
that more than 90% of the layer is amorphous. The crystalline forms were
indexed to the mineral zinc hemimorphite Zn4(OHhSi207.H20. In other
zinc-containing glasses, different crystal forms: e.g. Zn2Si04 have been iden-
tified. No crystalline forms containing the segregated elements of Fe, Mg or
Ni were found so that it appears that they are at least partly preferentially
bound in the siliceous layer instead of dissolving in the solution.
398 R .St.e. Smart
Table 16.4. The composition of E-glass as determined from the components used
in the manufacturing process and by XPS. The XPS percentages were adjusted so
that the carbon contamination was taken as zero [53].
(16.2)
where lea is the measured calcium concentration on the treated sample, lea is
the calcium concentration for the E-glass surface before treatment and ACa is
the attenuation length of the Ca 2p photoelectron. The factor 2/7f accounts
for the curvature of the fibre which can increase the effective path length
for photoelectrons in the surface layer. A variety of functionalised siloxanes
with different numbers of polymer repeat units have been studied in this way
producing layers varying in thickness from 1.2 to 4.1 nm [35]. Confirmation
of the layer structure can be obtained from the inverse correlation ofthe C Is
and Ca 2p elemental concentrations in static depth profiling using an argon
ion beam.
Long-chain (i.e. C I2 , Cl4, C I8 ) alcohols have been shown to strongly ad-
sorb on glass surfaces and may offer a cheaper alternative to silanes for some
applications. Reacted with E-glass fibres at 130°C, it is found that these al-
cohols adhere to the glass surface after more than 10 extraction processes
using hexane or cyclohexane. Advancing contact angles in the range 85-97°
are measured on these surfaces corresponding to hydrophobic methylene or
polyethylene-like surfaces. Quantitative DRIFT measurements correspond to
400 R.St.C. Smart
References
28. M.F. Hochella, Jr: in Mineral Surfaces, eds. D.J. Vaughan and R.A.D. Pattrick,
Min. Soc. Series No.5, (Chapman and Hall, London 1995) p.17
29. R.J. Atkinson, R.L. Parfitt, R.St.C. Smart: Infrared study of phosphate ad-
sorption on goethite. J. Chern. Soc., Faraday I 70, 1472 (1974)
30. R.R. Martin, R.St.C. Smart: X-ray photoelectron of anion adsorption on
goethite. Soil Sci. Soc. Am. J. 51, 1 (1987)
31. D.J. Vaughan and R.A.D. Pattrick (eds). Mineral Surfaces, Min. Soc. Series
No.5, (Chapman and Hall, London 1995)
32. W.H. Casey: ref. [31], p185ff
33. J.F. Banfield, B.F. Jones, D.R. Veblen: An AEM-TEM study of weathering and
diagenesis, Abert Lake, Oregon II: diagenetic modification of the sedimentary
assemblage. Geochim. Cosmochim. Acta, 55, 2795-810 (1991)
34. R.A. Eggleton: Formation of iddingsite rims on olivine: a transmission electron
microscope study. Clays Clay Miner., 32, 1-11 (1984)
35. R.St.C. Smart with P. Arora, B. Braggs, H.M. Fagerholm, T.J. Horr, D.C. Ke-
hoe, J.G. Matisons, J.B. Rosenholm, J. Ralston: Modification of Oxide, Glass,
Mineral and Metal Surfaces Using Adsorption, Plasma and Sol-Gel Reactions,
in Interfaces of Ceramic Materials: Impact on Properties and Applications, eds.
K. Uematsu, Y. Moriyoshi, Y. Saito and J. Nowotny, Trans Tech Publications,
361-404 (1996)
36. R.L. Segall, R.St.C. Smart, P.S. Turner, T.J. White: Microstructural charac-
terisation of Synroc C and E by electron microscopy. J. Am. Ceram. Soc. 68,
64 (1985)
37. J.A. Cooper, D.R. Cousens, J. Hanna, R.A. Lewis, S. Myhra, R.L. Segall,
R.St.C. Smart, P.S. Turner, T.J. White: Intergranular films and pore surfaces
in Synroc C: structure, composition and dissolution characteristics. J. Amer.
Ceram. Soc. 69, 347 (1986)
38. T.J. White, R.L. Segall, P.S. Turner: Radwaste immobilisation by structural
modification the crystallochemical properties of Synroc, a titanate ceramic.
Angew. Chemie (Int. Ed. Eng!.) 24, 357 (1985)
39. P.E. Fielding, T.J. White: Crystal chemical incorporation of high-level waste
species in aluminotitanate-based ceramics: Valence, location, radiation damage
and hydrothermal durability. J. Mater. Res. 2, 387 (1987)
40. W.J. Buykx, K. Hawkins, D.M. Levins, H. Mitamura, R.St.C. Smart,
G.J. Stevens, K.G. Watson, D. Weedon, T.J. White: Titanate ceramics for
the immobilization of sodium-bearing high-level nuclear waste. J. Am. Ceram.
Soc. 71, 678 (1988)
41. S. Myhra, R.St.C. Smart, P.S. Turner: The surfaces of titanate minerals, ce-
ramics and silicate glasses: surface analytical and electron microscope studies.
Scanning Microsc. 2, 715 (1988)
42. S. Myhra, A. Atkinson, J.C. Riviere, D. Savage: A surface analytical study of
Synroc subjected to hydrothermal attack. J. Am. Ceram. Soc. 67, 223 (1984)
43. T. Kastrissios, M. Stephenson, P.S. Turner, T.J. White: Hydrothermal dissolu-
tion of perovskite: implications for Synroc formulation. J. Amer. Ceram. Soc.
70, C144 (1987)
44. S. Myhra, D.K. Pham, R.St.c. Smart, P.S. Turner: Surface reaction and dis-
solution of Ceramic and high temperature superconductors, in Science of Ce-
ramic Interfaces, ed. J. Nowotny Mat. Sci. Monographs, 75, (Elsevier, Ams-
terdam 1991) p.569
16 Minerals, Ceramics and Glasses 403
45. P.S. Arora, R.St.C. Smart: Formation of silicate structures in oxidised nickel
surfaces using low temperature plasma reaction. Surf. Interface Anal. 24, 539-
548 (1996)
46. M. Steveson, P.S. Arora, R.St.C. Smart: XPS studies of low temperature
plasma-produced graded oxide-silicate-silica layers on titanium. Surf. Interface
Anal 26, 1027-1034 (1998)
47. C.D. Wagner: Faraday Discuss. Chern. Soc., 60, 291 (1975)
48. RSt.C. Smart, P.S. Arora, M. Steveson, N. Kawashima, G.P. Cavallaro,
H. Ming, W.M. Skinner: New approaches to metal-ceramic and biocerarnic
interfacial bonding, in Ceramic Interfaces: Properties and Applications II, ed.
S-J Kang, H-I Yoo and J. Nowotny, Inst. Materials Press (UK), 293-326 (2000)
49. D.E. Clark, L.L. Hench: An overview of the physical characterisation ofleached
glass surfaces. Nucl. Chern. Waste Manage. 2,93 (1981)
50. RA. Lewis, S. Myhra, RL. Segall, RSt.C. Smart, P.S. Turner: The surface
layer formed on zinc-containing glass during aqueous attack. J. Non-Cryst.
Solids 53, 299 (1982)
51. N.S. McIntyre, G.C. Strathdee, D.F. Phillips: SIMS studies of the aqueous
leaching of a borosilicate glass. Surf. Sci. 100, 71 (1980)
52. V.C. Farmer (ed.): The infrared spectra of minerals. Mineral. Soc. Monograph
4, London 285-303 (1974)
53. H.M. Fagerholm, J.B. Rosenholm, T.J. Horr, RSt.C. Smart: Modification of
E-glass fibres by long chain alcohol adsorption. Colloids and Surf. 110, 11-22
(1996)
54. H.M. Fagerholm, C. Lindsjo, J.B. Rosenholrn, K. Rokman: Colloids and Surf.
69, 79 (1992)
1 7 Characterization of Catalysts
by Surface Analysis
Solid catalysts are the basis of many important industrial processes. Reac-
tion of gases or liquids to form particular products occurs at specific sites
on the catalyst surface. The structure and composition of the catalyst sur-
face is critical in determining the reactivity and selectivity of a catalyst. The
techniques of surface analysis provide the means of characterizing a catalyst
in terms of the actual composition and structure of the surface rather than
by its bulk properties. The objective of such studies is to provide a scientific
basis for improving catalyst formulations and understanding the processes of
activation and deactivation which the catalyst undergoes. Supported cata-
lysts, the type most widely used in industry, consist of an active component
dispersed on the internal surface of a porous inorganic oxide. High area solids
and light loadings very highly dispersed are often employed to maximize the
catalytic activity of expensive components. Metals and oxides may be formed
on a support by decomposing or reducing a salt which has been introduced
by solution impregnation. For studies of model catalysts under UHV condi-
tions metals can be deposited in situ on to oxide surfaces usually formed on
metal substrates. The preparation, pre-treatment and actual catalytic reac-
tion conditions may result in reaction between the components including the
support. It is the purpose of the surface analysis to reveal these processes.
Specific objectives of the surface analysis of catalysts are:
1) to determine the surface composition of the catalyst in reactive form;
2) to identify the valence state of elements present at the surface of the active
catalyst, and to identify the catalytically active sites;
3) to monitor the extent of interaction between the catalyst components and
the support and possibly explain the effects of varying the support;
4) to determine the effects on surface composition of various catalyst prepa-
ration and pretreatment procedures.
The most useful techniques are X-ray photoelectron spectroscopy, XPS
(Chap. 6), Auger electron spectroscopy, AES (Chap. 6), and ion scattering
spectroscopy (ISS). Others include SIMS (Chap. 5), FT-IR (Chap. 8), RBS
and NRA (Chap. 9), LEIS (Chap. 11), LEED (Chap. 13) and UPS (Chap. 14).
The application of these surface techniques to the study of catalysts poses a
number of experimental problems:
i) The active components of the catalyst are generally present in low con-
centration and are at the surface of a high area porous solid. Most of
406 N.K. Singh and B.G. Baker
To overcome these problems the spectrometer for catalyst studies should have
a sample preparation chamber which provides for heating, reactant gas expo-
sure and vacuum transfer to the analysis chamber. The method of mounting
the sample needs to be compatible with these processes and particular atten-
tion must be given to the effects of electrostatic charging in order to quantify
chemical shift effects.
Whilst all techniques noted above are excellent for characterising cata-
lysts materials, a combination of XPS and FT-IR is particularly powerful,
as they provide complementary information. XPS provides information on
the chemical environment of the catalyst components. Infrared spectroscopy,
by virtue of its inherent sensitivity to chemical functional groups and their
conformations within a molecule or particular sites in framework structure,
is ideally suited for identifying adsorption sites on catalyst surfaces, and fol-
lowing catalytically controlled reactions. A URV spectrometer, equipped to
perform XPS, AES UPS and ISS and which has been modified for catalyst
studies is shown in Fig. 17.1. Based on the Leybold LRS10 electron spec-
LHS 10
SAMPLE
TREATMENT
CHAMBER
LOADLOCK
r:
(
SAMPLE
HOT
STAGE
TURBO PUMP
PUMP
2
TURBO
PUMP
1
trometer, this system has a rod with a heated stage [1]. The sample may be
heated in the treatment chamber in either a gas flow or static atmosphere to
create conditions comparable to practical service of the catalyst. High vac-
uum is then achieved before opening the gate valve and transferring to the
spectrometer chamber. The sample stage can be heated during analysis if
required. Other systems provide for transfer of a sample holder and allow
the treatment chamber to be isolated during analysis. Systems combining a
micro-reactor with facilities for kinetic studies and electron spectroscopies
have also been employed [2]. Vibrational spectroscopy requiring UHV con-
ditions, especially monitoring of reactions on single crystal transition metal
and semiconductor surfaces, are also performed in similar spectrometers but
incorporating oppositely mounted differentially pumped infrared transparent
windows. One such spectrometer is described in reference [3], which is set
up for Fourier transform reflection absorption infrared spectroscopy (FT-
RAIRS) (also known as infrared absorption spectroscopy, IRAS), LEED and
AES. In RAIRS the infrared beam is focussed from a source, using focussing
mirrors, at a glancing angle onto a surface held inside an ultrahigh vacuum
chamber through a KBr window. The reflected beam from the surface is fo-
cussed, again using mirrors, through another KBr window onto an external
liquid nitrogen cooled MCT detector. The KBr windows, the input and the
output optics, including the detector, are all contained within cells. The two
cells are continually purged with dry nitrogen gas to minimise absorption of
infrared beam by atmosphere. The reader is referred to articles by Greenler
[4,5], Pritchard [6] and Horn [7] for a more detailed description of this par-
ticular mode of vibrational spectroscopy. High surface area catalysts do not
require UHV environment, and hence these studies are usually performed
in high vacuum utilizing glass cells j compartments or under an inert gas
environment, as discussed in Chap. 8.
The objective in each experimental set-up is to examine catalysts in their
reactive form. The mounting of powder or granular samples of catalysts for
analysis must allow for good thermal and electrical contact with the holder
and ensure mechanical stability so that sample is not spilled during evacua-
tion. An effective method is to press a pellet of sample in a small stainless
steel cup. The top of the pellet can be cleaved off to expose fresh surface which
has not contacted the plunger of the press. The cup is then mounted on the
stage of the spectrometer providing good electrical and thermal contact. The
latter is not a problem when oxide thin films, e.g. Ab03, are formed on a
metal or semiconductor substrates in vacuum. A fresh film is formed prior
to XPSjRAIRS measurements which then eliminates the cleaning, cleaving
requirements. In XPS, an insulating sample tends to acquire a positive charge
which decreases the observed kinetic energy of the photoelectrons. This shift
is in the same direction as a chemical shift where the element has a more
positive valence state. Catalysts frequently show both effects and in addi-
tion often exhibit differential charging, i.e., some parts of the sample charge
408 N.K. Singh and B.G. Baker
a
1072.3
10756
. Na
1s b
~-------
Fig. 17.2. X-ray photoelectron spectra for sodium (Is) and aluminium (2p) in (a)
'I-Alumina (Merck) and (b) heat treated (1200°C, 30 min) alumina. The apparent
binding energies are displaced ~ 3.4 e V except for the sodium peak in the initial
'I-alumina. Sodium in this material is present in a ,6-alumina surface phase which
conducts. Heat treatment destroys the ,6-alumina leaving sodium compounds on
the surface
while others are uncharged. This can be of use in identifying particular el-
ements as being in combination in a separate phase. Charging is detected
by comparison of observed energies with values tabulated for reference peaks
in the spectrum. A low energy electron "flood gun" may also be used to di-
rectly neutralize charge at a surface. This is particularly useful in cases of
differential charging.
17.1.1 Alumina
A variety of forms of alumina are used as catalyst supports. The valence state
of aluminium is invariant at +3 and the oxide is insulating. Analysis by XPS
provides a good measure of the extent of charging. This analysis also reveals
surface impurities which are likely to be incorporated into a catalyst prepared
on this support. Gamma alumina prepared by the Bayer process has an area
17 Characterization of Catalysts by Surface Analysis 409
d
Fig. 17.3. XP spectra of un-
c
supported W03 reduced in wet
b hydrogen (pH 2 /pH 2 0 = 40) at
405°C. (a) W03 before reduction:
a
(b) after 5 h reduction: (c) after
45 40 35 30 10 h reduction: (d) after 15 h re-
BINDING ENERGY (eV) duction
410 N.K. Singh and B.G. Baker
40 35 30
(a) (c)
(b) (d)
40 35 30 40 35 30
Electron binding energy (eV)
Fig. 17.5. XP spectra (a) of sample from Fig. 17.4d after exposure to wet hydrogen
at 400°C. (b) Subsequent exposure to dry hydrogen at 400°C (15% reduction). (c)
of sample from Fig. 17.4a heated in air to only 200°C. (d) Subsequent exposure to
dry hydrogen at 400°C (75% reduction)
./..........
...... \ .....<iil
...... Fig. 17.6. A sample of unreduced
......
W03/HT-Ab03 (air, 450°C) serves
as an example of how charging by
X-rays can complicate a spectrum.
(i) Spectrum recorded without using
the flood gun. (ii) Flood gun used to
1210 1215 1220 discharge sample: 2.7eV charge (see
ELECTRON KINETIC ENERGY (eV) text)
d _ __
Fig. 17.7. He+(Eo = lkV) ion scattering spectra of (a) bulk W03; (b) HT-Ab03:
(c) 6% W03/HT-Ab03, untreated; (d) 6% W03/HT-Ab03, heated in air at 400°C
for 20h: (e) bulk Ab(W04h
as EjEo, where E is the measured kinetic energy of the scattered ion and
Eo the incident energy. Electrostatic charging of insulators is a problem in
applying the technique to catalysts and there are difficulties in quantifying
the data [10]. The technique is however more surface sensitive than XPS and
is of particular use in defining the location of a component on a catalyst
support.
Analysis of the tungsten oxide catalyst system using ion spectroscopy is
shown in Fig. 17.7. Oxygen, aluminium and tungsten are identified. The rel-
ative intensities of tungsten and aluminium in the prepared catalyst indicate
that tungsten oxide is well dispersed on the support effectively covering the
surface.
a
o in MgO
534.6
by XPS (MgKa) are shown in Fig. 17.8. Oxygen Is peaks from Pd~ and
MgO differ in binding energy by 1.8 eV for chemical reasons. The observed
spectra in Fig.I7.Sa differ by 4.5 eV due to electrostatic charging of the MgO
only. The complicated palladium 3d spectrum in Fig. I7.8b is interpreted as
a superposition of three sets of the 3d3 / 2 3d s/ 2 doublet; the metal, palladium
oxide which is conducting and palladium in the insulating magnesium oxide.
414 N.K. Singh and B.G. Baker
x1
x1
Ui
t:
z
::::>
ai 773(25)
II:
~
~
en
x1
z
w
I-
Z
148(12.5)
Co2p Co-Th02-MgO-Kieselguhr
Co-MgO-Kieselguhr
REDUCED 400°C 1HR.
(j)
!:::
z
::::>
cO Co-Th0 2-Kieselguhr
a::
~ Co-Kieselguhr
REDUCED 400°C 1HR.
~
Cf)
Z Total Cobalt Spectrum
W (either catalyst)
I-
Z
A-B
region shows a complex structure due to the overlapping of the various oxide
components. The binding energy and shape of the Co 2P3/2 peak identify the
presence of C0 3 0 4 . The largest component of the 0 Is peak is also identified
with C0 3 0 4 .
After reduction of the catalyst, the Mg Is and Th4f lines indicated no
reduction of Mg2+ and Th4+ whereas the cobalt and oxygen spectra were
markedly changed. Four catalysts, reduced at 400°C for 1 hour gave cobalt
spectra as in Fig. 17.10. Difference spectra are plotted to reveal that the
amount of unreduced cobalt is large in catalysts containing MgO, but very
small when MgO is absent. The oxygen spectra showed overlapping peaks
and they required a curve fitting procedure to resolve the contributions. An
416 N.K. Singh and B.G. Baker
015 Co-Kieselguhr
UNREDUCED
~
Z
::::>
ai
a:
~
~
u;
z
W
I-
Z
example is shown in Fig. 17.11 for the unpromoted catalyst. The reduced
form contains only small contribution of CoO to the 0 Is peak, consistent
with the observation that cobalt is reduced to metal in this catalyst. The
corresponding result for MgO-promoted catalysts shows CoO present after
reduction. It is concluded that there is a strong interaction of MgO with
cobalt and that CoO possibly forms a solid solution which is resistant to
reduction.
Catalysts containing iron are used in the synthesis of ammonia and for the
Fischer-Tropsch synthesis of hydrocarbons from carbon monoxide and hy-
17 Characterization of Catalysts by Surface Analysis 417
Pr 3d AI2p o 15
Treatment
CO+H2
280°C
t
H2 320°C
~~,L - I \: 3hr
r--r--r-.-~--oL-.--~--.-~
730 720 710 700 960 950 940 930 80 75 70 535 530 525
Electron binding energy (eV)
Fig. 17.12. Precipitated iron catalyst. XPS analyses at various stages of reduction.
The reference marks on the energy scales indicate the peak position for an uncharged
sample. In the final state of reduction the sample is uncharged but iron is not in
the metallic state (see text)
for catalyst service. Very light loading of metal is necessary to ensure that
excess metal does not obscure the analysis of the interface. The XPS analy-
ses for iron on alumina are shown in Fig. 17.13. Trace (b) after deposition of
the iron film shows two charge states; Fe2p3/2 at 706.7eV and 709.6eV. The
former, probably represents a conducting, continuous iron film and the latter
represents islands or clusters of discontinuous Fe on the alumina. After reac-
tion with H2 at 320°C for 3 hours, the spectra are markedly changed [trace
(c)]. The binding energies of Fe(2P3/2) XPS spectra can then be attributed
to Fe(II) and Fe(III) at 709.5eV and 711.0eV, respectively, with charging
rv 4 eV. The iron has reacted with alumina under conditions known to reduce
730 720 710 730 720 710 730 720 710 700
Electron binding energy (eV)
Fig. 17.13. XPS Fe 2p peaks from an iron film on an oxide support. Traces ( b) as
deposited; (c) after heating in hydrogen at 320°C
Fe I alumina Fe I Ti0 2
40 50 60 70 80 40 50 60 70 80
Kinetic energy (eV)
Fig. 17.14. Iron on alumina and on Ti02 Auger spectra from the thin iron film. (a)
before, and (b) after taking the XP spectra, and (c) after heating in H2 at 320°C
occurred. Auger analysis of the reduced sample was difficult due to electro-
static charging but a weak broad peak at 56 eV and a peak near 48 eV are
consistent with the conclusion that part of the iron is reacted.
Reference was made previously to the effect of praseodymium oxide as
an additive to iron Fischer-Tropsch catalysts. To test the reactivity of this
oxide towards iron, a sample of the oxide (Pr6011) was reacted with an iron
film under reducing conditions. The XPS results in Fig. 17.13 are free of
the effects of electrostatic charging: Pr6011 is a semiconductor. The iron 2p
peaks are well defined for the metal film (trace b). After heating in hydrogen
at 320°C most of the iron is in a higher valence state. Extensive reaction
of iron has occurred. Possibly the compound formed is PrFe03 however the
amount present as a surface phase is too small for structural identification.
This series of results, after the reduction of the deposited iron layers in
H2 at 320°C, can be summarized as follows:
• Fe on Ab03 reacts totally to form a new surface phase in which the Fe
is in oxidized form.
• Fe on Ti0 2 does not react and remains as separate Fe metal particles in
islands on the Ti0 2 surface;
• Fe on Si0 2 is partly oxidized suggesting that the role of this oxide as a
catalyst promoter is in the formation of a compound with iron.
The extreme differences in reactivity of the metal with different oxide sub-
strates correlates with the behaviour of the supported metal catalysts where
alumina is the preferred substrate for selective Fischer-Tropsch activity. Silica
is less effective and titania is not useful as a substrate.
active sites can be modified during the synthesis of the material, and indeed
it is this attribute of these solids that make them such versatile catalysts for
a multitude of catalytic reactions.
One such zeolite is a high silica aluminate, ZSM-5, composed of linked
Si0 4 and AI0 4 tetrahedra. The presence of aluminium in the lattice makes
the framework negatively charged, which is compensated by mobile extra-
framework cations, such as hydrogen and metal ions. In the former case the re-
sulting zeolite is then referred to as HZSM-5, and in the latter case it would be
M-HZSM-5. The acid centres in these zeolites can be easily characterised by
transmission FT-IR using zeolite samples pressed into self-supporting wafers
and weighing approximately 10-20mgcm- 2 .
In the following example we will show how IR spectroscopy can be used to
identify acid sites in four proton-exchanged HZSM-5 zeolites [16]. These zeo-
lites, labelled Z12, Z16, Z27 and Z121 vary in their Si/ Al ratios, having values
of 12, 16, 27 and 121, respectively. The values of bulk aluminium (framework
aluminium) per unit cell, again respectively, are 7.7 (7.6), 6.0 (2.3), 3.5 (1.4)
and 0.8 (0.8), and the difference between these two sets of values yields the
values of extra-framework aluminium per unit cell. In Fig. 17.15 the IR spec-
tra in the v(OH) region of the four zeolites, measured at 423 K after heating
in flowing nitrogen at 673 K, are shown. The spectra for the samples have
Q)
(.)
c
C13
.0
L- b
aC/) x
..c
«
d
3800 3600 3400
Wavenumber / em- 1
Fig. 17.15. Infrared spectra in n(OH) region of HZSM-5 samples: (a) Z12, (b) Z27,
(c) Z16, (d) Z121. X, A and B refer to SiOH, AIOH and Br0nsted acid hydrox-
yls, respectively. (Reprinted from [16]; copyright 1999 Elsevier Science Publishers,
B. Y.)
422 N.K. Singh and B.G. Baker
Absorbance
.5
.4
.3
Fe-ZSM-5 B (oxalate)
.1
o Fe-ZSM-5 (MeOH)
Wavenumbers / cm- 1
.25
/,.zs",,,,,,, "m'" NO
Fe-ZSM-5(Vec) 1 mbar NO
.2
.05
2300 2200 2100 2000 1900 1800 1700 1600 1500 1400
Wavenumbers I cm- 1
Fig. 17.17. Infrared spectra measured at 303 K for increasing NO pressure from
10- 3 mbar to 10 mbar over Fe-ZSM-5. (Reprinted from [17]; copyright 1999 Amer-
ican Chemical Society)
.1r-------------------------------------,
.08
.06
:.
:.:
Fe-ZSM-5A(vac., 02, vac.), 1e-1 mbar NO!:,
""
:-
:,
"
:1
Q)
u j:, Fe-ZSM-5A(vac., 02), 1e-1 mbar NO
c
co l: :
-e .04 :1
l'
I
5l
.0
<I:
.02
a
"
2300 2200 2100 2000 1900 1800 1700 1600 1500 1400
Wavenumbers / cm- 1
species acting as a charge balancing cation_ The interaction of iron with the
acidic protons are quite complex and not fully understood, as yet,
In summary, the infrared data presented above, in conjunction with XPSj
EXAFS data show that the exchange of iron with zeolite ZSM-5 leads to the
formation of both isolated iron atoms, and iron-oxo nanoclusters of typical
size Fe404 with very short Fe-Fe inter-atomic distance of ca. 2.50 A. These
iron species exist predominantly within the pore structure of the zeolite and
not on the external surfaces_
426 N.K. Singh and B.G. Baker
I 0.00025
a)
Q)
u Hydroxylated
C OD
ell AI 2 0 3 /Mo(100)
b)
....
.0
oC/)
.0
~~~~
«
Frequency / cm- 1
Fig. 17.19. RAIRS of (a) alumina grown on Mo(lOO) by reaction between alu-
minium and water (H 2 0), (b) alumina grown on Mo(100) by reaction between
aluminium and deuterium oxide (D 2 0), (c) alumina grown on Mo(100) by reaction
between aluminium and water (H 2 0), annealed at 900 K to remove surface hydrox-
yls, exposed to ammonia at 80 K and then annealed to 260 K before data acquisition
at 80K, (d) alumina grown on Mo(lOO) by reaction between aluminium and water
(H 2 0), then exposed to ammonia at 80 K, annealed to 225 K before data acquisition
at 80K. (Reprinted from [18]; copyright 1999 Elsevier Science Publishers, B. V.)
\ 90K
•
I·CO
13C+O
'~J
300K+90K
2200
2 117
12112
2 101
12113
2100
2058
13113
Energy (em-I )
2000 1900 ~2oo 2100 2000 1900
Energy (em-I)
mately equimolar mixture of 12CO and 13CO (bottom) at 90 K. The isotopic compo-
sitions giving rise to the three dicarbonyl bands are indicated below the correspond-
ing wavenumbers, (b) Infrared spectra recorded after CO saturation of rhodium
deposits at 90 K , along with room temperature STM images (500 Ax 500 A). (Top):
0.057 ML of rhodium deposited at 300 K , (Middle): 0.057 ML of rhodium deposited
at 90 K , (Bottom): 0.057 ML of rhodium at 300 K, followed by the same exposure
at 90K. (Reprinted from [19]; copyright 2000 Elsevier Science Publishers, B. V.)
such a deposit shows a band at 2117 cm-I, implying that Rh(COh species
will form as long as rhodium nucleation sites exist at oxide point defects.
In this section we provide another study which utilizes RAIRS for the char-
acterization of a metal supported catalyst, copper on silica [20] . CO was used
as a probe molecule to elucidate the structures of copper deposits on silica
support. Thermal desorption spectroscopy was used to support the infrared
data. Copper was evaporated onto a planar silica thin film (rv 100 A) sup-
ported on Mo(110) surface at 100 K , while the CO adsorption on the resulting
model catalyst was carried out at 90 K.
Fig. 17.21 shows the infrared spectra of CO saturated silica deposit an-
nealed to temperatures indicated. At 100 K an intense peak at 2099 cm- 1
with a shoulder on the low frequency side is observed. A new band appears
at 2070 cm- 1 when heated to 300- 500K, which gradually increases in inten-
sity as the catalyst is heated to 900 K. The 2099 cm -1, on the other hand,
splits into two peaks at 2108cm- 1 and 2094cm- 1 . Using literature values of
CO on Cu(111), the observed bands are attributed to CO adsorbed on several
430 N.K. Singh and B.G. Baker
I 0.25%
CO/Cu/Si02( 1OOA)
e(Cu) = 2.2ML
Ts = 90K
2099
Annealing T
100 K
Q)
()
c:
al 300 K
....
.0
o
en
.0
« 500 K
700 K
900K
1 0 . 16%
cO/Cu/Sio 2 (100A)
Annealed to 900 K
=
T8 90K
~2076
8(Cu)
65ML
2102
Q) 2094 2087
(.)
c
ctS
....
.0
15ML
0
en
.0
« 3.4 ML
2108 2.2ML
1.1ML
0.5ML
x2 0.3ML
reaches 3.4 ML. This band indicates the presence of (110)-facets of copper
deposits. The 2110 cm- 1 band remains constant till this coverage, but with a
further increase of copper coverage (15 ML) it becomes the main feature of the
spectrum. The 2094cm- 1 band is not evident for copper coverages> 15ML.
The dominance of the 2110cm- 1 mode in the whole CO spectral region sug-
gests that at high copper coverages continuous films with majority (111) ori-
entations form during deposition. The high index (211) and (311) planes may
also be present for these coverages as the spectral range for these planes fall in
the 2108-2110cm- 1 range. At even higher copper coverages (> 30ML) CO
exhibits three well-resolved absorption features at 2102, 2087 and 2076 cm- 1 .
The 2087 and 2076cm- 1 bands are due to CO adsorption onto the (111) and
(100) facets, respectively, and the 2101 cm- 1 band is attributed to adsorp-
tion at step/edge sites of the polycrystalline copper film that forms at this
coverage.
432 N.K. Singh and B.G. Baker
It should be noted that the results presented here do not provide any
evidence of interactions of reaction of copper with the support material silica.
However, a later paper of the same researchers [21] shows, using XPS and
temperature programmed desorption, that a small fraction of copper in the
copper-silica interface is partially oxidized and this phase desorbs at 1300 K.
In summary, this infrared study of a model copper catalyst shows that at
low effective copper coverages, the un-annealed copper deposit has at least
two types of small copper islands on the silica surface. As the copper coverage
increases the small clusters are merged into an extended smooth film. On the
basis of the CO stretching frequencies it has been proposed that the copper
atoms are initially aggregated into (l11)-like islands and high index islands.
Annealing the copper to 900 K following copper adsorption at 100 K causes
the formation of three-dimensional clusters.
17.3 Conclusion
The above examples show that surface techniques, such as XPS and vibra-
tional spectroscopy (FT-IR, FT-RAIRS), can reveal much of the complex
chemistry of a catalyst surface. The reactions which occur during the con-
ditioning or activation of the catalyst involve only materials contacting at a
surface. The surface phases which form are not readily characterized by bulk
techniques nor do they necessarily follow the thermodynamics of bulk prepa-
rations. Direct analysis of the surface is the only way to reliably characterize
the catalyst surface.
References
1. P.J. Chappell, M.H. Kibei, B.G. Baker: J. CatalysiH 10, 139 (1988)
2. D.J. Dwyer: In Catalyst Characterization Science, ed. by M.L. Deviney and
J.L. Gland (American Chemical Society 1985) pp. 124--132
3. R. Raval, M.A. Harrison, D.A. King: J. Vac. Sci. Techno!. A9 (2), 345-349
(1991)
4. R.G. Greenler: J. Chern. Phys., 44, 310 (1966)
5. R.G. Greenler: J. Vac. Sci. Techno!. 12, 1410 (1975)
6. J. Pritchard, M.L. Sims: Trans. Faraday Soc., 66, 427 (1970)
7. A. Horn: In Spectroscopy for Surface Science, ed. by R.J .H. Clark and R.E. Hes-
ter (John Wiley & Sons Ltd., 1998), pp. 273-339
8. B.G. Baker, N.J. Clark: New Fischer-Tropsch Catalysts. Report NERDDP
EG85/470 (Australian Govt. Printing Service 1985)
9. B.G. Baker, N.J. Clark, H. McArthur, E. Summerville: Catalysts. U.S. patent
4610975, New Zealand patent 216337, Australian patent application
No. 73419/87
10. J.C. Carver, S.M. Davis, D.A. Goetsch: In Catalyst Characterization Sci-
ence, ed. by M.L. Deviney and J.L. Gland (American Chemical Society 1985)
pp.133-143
17 Characterization of Catalysts by Surface Analysis 433
11. B.A. Sexton, A.E. Hughes, T.W. Turney: J. Catalysis 97, 390 (1986)
12. B.G. Baker, N.J. Clark, H. McArthur, E. Summerville: U.S. patents 4610975,
4666880, 4767792; New Zealand patent 207355; Australian patent 565954
13. R.St.C. Smart, P.S. Arora, B.G. Baker: Proc. Aust. X-ray Analytical Associ-
ation (AXAA-88) 219-228 (1988)
14. C. Klauber: Ph.D. Thesis, Flinders University (1984) p.7.42
15. J.M. Thomas, W.J. Thomas: Principles and Practice of Heterogeneous Catal-
ysis, (VCH Verlasgessellschaft, 1997), 347-364
16. S.M. Campbell, X.-Z. Jiang, R.F. Howe: Microporous and Mesoporous Mate-
rials, 29, 91-108 (1999)
17. R. Joyner, M. Stockenhuber: J. Phys. Chern. B, 103, 5963-5976 (1999)
18. M. Kaltchev, W.T. Tysoe: Surface Science, 430, 29-36 (1999)
19. M. Frank, R. Kuhnemuth, M. Baumer, H.-J. Freund: Surface Science, 454-456,
968-973 (2000)
20. X. Xu, D.W. Goodman: J. Phys. Chern., 97, 683-689 (1993)
21. X. Xu, J.-W. He, D.W. Goodman: Surface Science, 284, 103-108 (1993)
18 Application to Semiconductor Devices
a)
Fig. 18.1. Example of a display in SEM-VC dynamic mode showing images with
500 ns b etween states. The images have a (a) 1.2 IlS and (b) 1.7 IlS delay at a clock
speed of 4 MHz. (Courtesy of J. Rogers, Telstra Research Laboratories).
tical microscopy from 0.5 [lm to 0.25 [lm [12]. However, many of the defects
which can cause failure in ICs cannot be resolved even at this level by optical
microscopy. Hence the scanning electron microscope (SEM) has become a
standard tool in the analysis of failure in advanced ICs [8,13,14]. The sec-
ondary electron (SEM-SE) mode is used to reveal the topography of devices
with a lateral resolution down to :;:::j 4 nm while the backscattering electron
(SEM-BSE) imaging mode depicts the contrast between different materials
due to variation in electron density. Another useful mode is electron beam
induced current (SEM-EBIC) microscopy which can determine minority car-
rier lifetime and diffusion length at p-n doped junctions. Alternatively, the
optically beam induced current (SEM-OBIC) mode creates pairs of charge
carriers by bombardment with light to give visibility to the p-n junctions
(sources of current) or defects (sinks of current).
However, the most important mode of the SEM for the simple dc anal-
ysis of circuits and the measurement of time response is voltage contrast
(SEM-VC) microscopy [8,13,14]. In the static mode of voltage contrast at its
simplest level, a potential difference of +5 V is applied between the grid of
the detector and a conductor track on the chip. Any discontinuity in the con-
ductor results in a local electric field with fewer secondary electrons reaching
the detector and a darker appearance in that region. This technique is partic-
ularly effective in the location of faulty connections and defects which create
open circuit on the chip surface. In the dynamic mode of VC, the SEM beam
is chopped stroboscopically at exactly the same frequency as the modulation
of voltage in an operational device. Measurements of time-resolved waveform
can be performed from any part of the IC using a probe to explore the delays
440 P.W. Leech and P. Ressel
AFM (high resolution 3-dimensional
mapping of surface morphology,
feature dimensions)
SEM (measurement of critical
dimensions on interconnect)
--
,,
L
,
AESITEM (analysis of
contacUsilicon interface) TEM (measurement of critical
thickness/high resolution imaging of
SIMS (profle of dopants in S,02/Si interface)
junction)
TEM/Optical microscopy
(crystalline defects in silicon)
in signal propagation and allow comparison with the logic states in the design
of the circuit. This dynamic mode of SEM-VC is very useful in locating faults
in very large scale integrated circuits and examples of stroboscopic images
are shown in Fig. 18.2. The different logic states and state transitions can be
displayed in a colour voltage contrast (CVC) mode [15]. The techniques of
stroboscopic microscopy and waveform point analysis are collectively known
as "e-beam testing" . These testers are increasingly integrated into CAD sys-
tems for design validation as well as inspection for the validation of process
steps and the failure analysis of circuits.
Another technique increasingly used in conjunction with the SEM-VC is
the focused-ion beam (FIB). A finely focused beam (250A probe size) of
energetic ions is scanned over a specific area of an IC. The beam can be
used in the dual roles of imaging/ analysis and modification of features in
the circuit with very precise cross-sectioning or micro-milling on-chip and
rapid analysis at the point of a defect [7,16]. The location of FIB within the
fabrication line enables the repair or redesign of interconnects within a circuit
while remaining in production.
The atomic force microscope (AFM) with capabilities of atomic level reso-
lution and 3-dimensional mapping can perform measurements on the nm-scale
18 Application to Semiconductor Devices 441
Lateral Resolution
«1 [lm)
YES NO YES NO
Fig. 18.3. Decision tree for the application of compositional analysis techniques in
a device or circuit [30].
to defects in the crystal. While RBS has a poor sensitivity to light elements
on heavy substrates, the measurement of x-rays by particle induced x-ray
emission (PIXE) in conjunction with RBS provides complimentary informa-
tion about heavy elements. In plan-view and cross-sectional imaging, energy
dispersive x-ray (EDX) analysis provides a rapid method of multi-elemental
identification with good resolution (1 !lm) and a visual correlation with the
SEM in the region of the sample. The depth resolution of 1-5!lm means that
this is essentially a bulk technique with sensitivity of 100-2000ppm and a
limited ability to identify chemical states depending on the element.
An alternative form of protocol for the analysis of devices or circuits using
these preceding techniques is shown in Fig. 18.3 [30]. Here, the choice of
technique is based on the required lateral resolution, sensitivity and depth in
profiling. Where the techniques are listed in a horizontal series, the first shown
in the sequence is preferred on the basis of a number of factors including non-
destructivity, quantitative character and cost.
Schottky barrier theory predicts that to produce a good ohmic contact the
interfacial metal requires a work function which is smaller than the electron
affinity of the semiconductor i.e. 'Pm < Xs. However, in the case of wide band
gap semiconductors including III-V compounds, the properties are dominated
by the pinning of the Fermi level by interface states. This is due to the high
densities of surface states which appear in the band gap of the semiconductor
and effectively pin the Fermi level. The two main methods used to reduce the
barrier height involve additional processing steps either ex-situ or in-situ to
the epitaxial growth chamber. Ex-situ contacts are formed by the deposition
of a layer or the ion implantation of a dopant metal. Practical types of ex-
situ contacts to III-V compound semiconductors include a melting/alloying
reaction such as in the Ni/Ge/ Au system [33], a solid phase regrowth such
as in Pd/Ge/ Au or Pd/Si/ Au systems [34] or sintering as in Ti/Pt/ Au [35].
In-situ schemes use very heavy doping to > 10 19 cm- 3 in the growth of the
either the substrate or of a narrow bandgap overlayer.
An example of an epitaxial layer with a low barrier height on either GaAs
or InP is Inl-xGaxAs. The Inl_xGaxAs/InP heterojunction has been widely
used in the fabrication of photodiodes, multi-quantum well lasers, and high
electron mobility transistors because of the high values of electron mobility
and electron saturation velocity of Inl_xGaxAs. Furthermore, the composi-
tion of Inl-xGaxAs with a mole fraction of InAs = 0.53 can be grown as
lattice matched to InP and exhibits a low Schottky barrier height to both
n- and p-type InP. Over the range of applications of InO.53Ga0.47As, one of
the most challenging examples of an ohmic contact is formed to the thin, p-
type base layer « 100nm) on InP-based Heterojunction bi-polar transistors
(HBTs). This dimension of the base layer severely limits the available depth
of the metallurgical reaction allowing only a moderate level of doping (5-10
x 10 18 cm- 3) in the p-InO.53Ga0.47As while requiring a low specific contact re-
sistance of ~ 1 x 10- 6 Qcm 2 . For the purposes of this section, a review is made
of an ex-situ contact scheme, the Pd(20nm) / Zn(10nm) / Au (10nm) /
LaB 6 (100nm) / Au (100nm) metallisation to p-Ino.53Ga0.47As/InP. In this
scheme, the interfacial layer of Pd reacts with the p-InO.53Ga0.47As by a pro-
cess of regrowth in the solid phase accompanied by the release of the dopant
Zn. This reaction also allows the penetration of native oxides at the inter-
face. The intermediate film of LaB 6 is inserted in the contact as a barrier to
the indiffusion of the outer layer of Au. An understanding of the reactions
which occur in this complex, multi-layered structure requires a range of ana-
lytical techniques, each contributing information on the microstructure and
composition of the interfacial region. In the remainder of this section, the
Pd/Zn/ AuLaB 6 / Au to p-InO.53Ga0.47As contacts are used as an example of
the analysis of thin layered structures as encountered in modern semiconduc-
tor technology. Initially, the Pd/Zn/Pd/ Au system without a diffusion barrier
is examined by AES followed by characterisation of the Pd/Zn/ Au/LaB 6 / Au
system containing a LaB 6 barrier using several other techniques.
446 P.W. Leech and P. Ressel
10
\
'"
.~
0 40 1 \ In Ga
~ 0- ~ /~ +
6
e ~ Zn
20 ~As
'\l.l. 0 o
0
0 200 400 600 800
Etch Time (Sec.)
80
tiP
0 60
i 40
~
20
0
0 200 400 600 800
Etch Time (Sec.)
100
(c) Sample 3: Annealed 450°C
80
tiP
0 60
.~
0 40
~
20
0
0 200 400 600 800
Etch Time (Sec.)
Fig. 18.4. AES depth profile of the Pd/Zn/Pd/ Au to Ino.s3Gao.47As/lnP contacts
(a) as-deposited and after annealing at (b) 350 0 C and (c) 450 0 C [38].
Fig. 18.5. RBS spectra of the Pd/Zn/ Au/LaB6/ Au metallisation on Ino.53Gao.47As
as-deposited and after annealing at 425 and 500 0 C [39].
of the outer Au peak, indicating only a very limited indiffusion. The spectra
also show that during annealing there was little or no change in the In, Ga
and As plateau or Pd peak, with the interfacial reaction confined to a narrow
width of below 100 nm without extensive intermixing. In general, these RBS
spectra show the effectiveness of the LaBfi layer in preventing an indiffusion
of the outer Au layer and in enhancing the stability of the Pd/Ino.53Gao.47As
interface without a significant increase in Pc.
SIMS is particularly well-suited to the analysis of metal/semiconductor
contacts because of its high (ppm) sensitivity for trace concentrations such
as the dopant elements released into the semiconductor during annealing.
This technique also provides a good resolution of the compositional cross-
section of the contact, giving information on the depths of penetration of
the metals. However, the depth resolution of SIMS in multi-layer contacts
may be degraded by surface roughness which can be induced either by the
sputter process or by annealing of the metallisation. In addition, the ion
bombardment may cause a mixing or knock-on of metallic species into the
semiconductor. As a consequence, it is difficult to accurately detect by SIMS
the profile of a thin, low concentration region of dopant beneath a metal layer
of high concentration.
These effects can be circumvented by sputtering from the substrate or
backside of the contact following a chemical thinning of the sample [43,44].
18 Application to Semiconductor Devices 449
Au (100 nm)
Au (30 nm)
Zn(10nm)
Pd (20 nm)
_lno.6Ga0.4Aso.S6P014
marker layers
(8 nm thick)
IIII Cs+
Fig. 18.6. Schematic illustration of the structure of Pd/Zn/ Au/LaB 6 / Au met alli-
sation on Ino.5aGao.47As showing method of analysis by Backside SIMS [45J.
:::J As
cci
-
>.
'en
Ga
-
~ 1021
c
§ 1020
"C
Q)
~ 1019
E
.....
\1
\ reaction depth
~ 10 18
-100 0 100
Sputter Time, a.u. Sputter Time, a.u.
Fig. 18.7. Normalised backside SIMS profiles of Pd/Zn/ Au/LaB 6 / Au metallisation
on Ino.53Gao.47As with the Zn either a) implanted or b) evaporated. The sputter
time equals depth in nm within the semiconductor [45].
I ,200 run
Fig. 18.8. Cross-sectional transmission electron micrograph of a Pd/Zn/ Au/
LaB 6 / Au on Ino. 53 Gaa.47As contact annealed at 425°C for 30 s [45] .
polishing, dimpling and Ar ion milling. Figure 18.8 shows a cross sectional
TEM micrograph of a Pd/Zn/ Au/LaB 6 / Au contact to Ino.53Gao.47As an-
nealed at 425°C [45]. The technique allows a clear identification of the Au,
LaB 6 and Ino.53Gao.47As regions and of the reacted interfacial region formed
during the thermal treatments. Electron diffraction has shown the LaB 6 as
amorphous. Within the reacted interfacial region, TEM reveals 3 zones: (a) a
thin (~7 nm) almost continuous band adjacent to the LaB 6 (b) areas of pro-
nounced contrast which were amorphous with a 3-dimensional morphology
resembling a sponge and (c) large crystallites with a mean size of ~ 95 nm.
EDX analysis with a local resolution of < 20 nm is able to determine the com-
position of the patterns except for region (a) which was below the resolution
of the apparatus. The contrast-rich regions contain 31 at% Pd, 8 at% Au,
3 at% Zn, 6 at% In and 27 at% Ga and 23 at% As while the areas of large
crystallites comprise 25 at% Pd, 14 at% Au, 3 at% Zn, 14 at% In, 28 at% Ga
and 13 at% As. Both types (b) and (c) regions thereby contain all elements
from the metallisation and the semiconductor.
XRD analysis provides further information on the phases formed during
thermal treatment. Prior to analysis, the top layer of Au is etched off to avoid
a possible overlap of peaks. The spectra in Fig. 18.9 allow identification of
two compounds, cubic PdAs 2 and hexagonal Pd12Ga5As2 ' The identification
of the phases involves a comparison of the XRD peak family with calculated
values assuming a = 0.9468 nm and c = 0.372 nm and analysis of single
films of Pd on Ino.53Gao.47As [45]. These spectra, combined with EDX data,
suggest that the large crystallites in areas c) in Fig. 18.9 are Pd12Ga5As
where part of the Pd and Ga are replaced by Au and In. Combined TEM
and XRD data show the formation of (Pd, Auh2(Ga, In)5As2 phase during
annealing at > 360°C. The formation of this solid solution phase during
452 P.W. Leech and P. Ressel
::::l
10° • • Pd 12 Ga 5As 2
-
ctl
IJ PdAs 2
~
'wc
-Q)
C
"0
Q)
10-1
IJ •
•
.t:::!
ctl
E
.... 10-2
0
Z
20 30 40 50 60 70 80 90
28, degrees
Fig. 18.9. X-ray diffraction spectrum of a Pd/Zn/ Au/LaB 6 / Au metallisation on
Ino.53Gao.47As annealed at 425°C for 30s [45J.
18.3 Summary
References
36. C.J. Palmstrom, and D.V. Morgan, in Gallium Arsenide, Materials, Devices
and Circuits, Eds. M.J. Howes and D.V. Morgan, (J. Wiley and Sons), 195,
(1985)
37. T.C. Shen, G.B. Gao, and H. Morkoc, J. Vac. Sci. Techno!., BIO(5), 2113,
(1992)
38. P.W. Leech, G.K. Reeves and M.H. KibeI, J. App!. Phys., 76(8), 4713, (1994)
39. P. Ressel, P.W. Leech, G.K. Reeves, W. Zhou and E. Kuphal, App!. Phys.
Letts., 68(13), 1841, (1996)
40. P.W. Leech, P. Ressel, G.K. Reeves, W. Zhou and E. Kuphal, Mater. Res. Soc.
Symp. Proc., 406, 419, (1996)
41. P.W. Leech and G.K. Reeves, Thin Solid Films, 298, 1, (1997)
42. P.W. Leech, G.K. Reeves, W. Zhou and P. Ressel, J. Vac. Sci. Techno!.,
B16(1), 227, (1998)
43. J. Herniman, J.S. Yu and A.E. Staton-Bevan, App!. Surf. Science, 52, 289,
(1991 )
44. S.A. Schartz, M.A. Pudensi, T. Sands, T.J. Gmitter, R. Bhat, M. Koxza,
L.C. Wang and S.S. Lau, App!. Phys. Lett., 60(9), 1123, (1992)
45. P. Ressel, P.W. Leech, P. Veit, E. Nebauer, A. Klein, E. Kuphal, G.K. Reeves
and H.L. Hartnagel, J. App!. Phys., 84(2), 861, (1998)
19 Characterisation of Oxidised Surfaces
Metals, in general, critically depend on their surface oxide scales for envi-
ronmental stability, particularly in aggressive oxidising atmospheres at high
temperatures. The protective capabilities of oxides are dependent on many
physical and chemical properties, as well as on their mechanical adherence
to the metal surface. In summary, an "ideal" protective oxide would be:
• physically and chemically stable. An ideal oxide would not dissociate nor
melt at the temperatures and pressures of interest;
• mechanically stable. The scale would be capable of maintaining intimate
contact with the surface of the metal, particularly when sudden temper-
ature changes occur;
• a barrier to diffusion. The function of a protective oxide is to separate
the metal from the oxygen in the gas phase. The ideal oxide would, there-
fore, have a low diffusion rate for both oxygen and metal ions, otherwise
the oxidation reaction would proceed at the oxide/metal or oxide/gas
interfaces respectively;
• continuous and dense. When pores or cracks are present in the oxide scale
the protective capabilities of the oxide are lost.
All of these properties that are essential for oxidation protection are depen-
dent on the chemistry and morphology of the oxide scale, the oxide/metal
interface and the near-surface region of the metal.
So that alloys can be developed with improved oxidation resistance, it is
essential to know the mechanisms whereby oxidation occurs. Such knowledge
can only be obtained through characterisation of oxidised surfaces, particu-
larly of surfaces in the initial stages of the oxidation process. The character-
isation of thin oxide scales, particularly for oxides less than a micron thick,
has only been possible in recent years with the development of the array of so-
phisticated surface analytical equipment described elsewhere in this book. A
number of these techniques have been used by the authors to characterise ox-
idised surfaces of metals [1-3]. In this review, application of three techniques,
namely scanning Auger microscopy (SAM), microbeam Rutherford backscat-
tering spectrometry ([t-RBS) and extended X-ray absorption fine structure
(EXAFS), to the study of the oxidation of selected cobalt- and nickel-based
alloys is described. SAM and [t-RBS are powerful tools, particularly when
used together, for analysing multi-phase surfaces where spatial resolution of
better than 20 [tm is required, whereas EXAFS has great potential for the
non-destructive determination of atomic structure in thin surface layers.
456 J.L. Cocking and G.R. Johnston
The first example presented in this brief review of the application of micro-
analytical techniques to oxidation studies is the oxidation of Co-22Cr-llAI
(nominal composition, wt%) cast alloys. This alloy has a two-phase structure
consisting of a matrix of ;3-CoAI with a-Co solid solution precipitates ranging
in size from 5 to 20 !lm. The oxidation experiments were performed, in air,
at 700° C for times between 10 min and 96 h. One surface of each sample was
metallographically polished to a 0.25!tm diamond finish. Half of this polished
surface was implanted with 2 x 10 16 Hf (or Y) ions/cm 2 with an energy of
150keV, while the other half was shielded with a tantalum mask. This tech-
nique allowed the in situ comparison of the oxidation of non-implanted and
implanted alloys on the same surface.
Fig. 19.1. Secondary electron (a) and backscattered electron (b) micrographs of the
boundary region between the non-implanted (left of boundary) and Hf-implanted
(right of boundary) areas of a Co-22Cr-llAI surface oxidised for 9.5 h at 700°C.
These micrographs show the presence of voids in the non-implanted alloy and their
suppression following Hf implantation
458 J.L. Cocking and G.R. Johnston
The two most powerful microanalytical tools for analysing the oxides on both
the a and fJ phases are scanning Auger microscopy (SAM) and microbeam
Rutherford backscattering spectrometry (ft-RBS). These two techniques are
extremely powerful tools when used in combination. High resolution SAM
gives chemical composition data in the top 4 to 10 atomic layers of the sur-
face. For depth information from SAM, ion beam milling must be used. This
technique, however, is at best semi-quantitative with respect to depth because
of many experimental problems including:
With the present experimental technique where both the implanted and
non-implanted surfaces are on the same alloy face (and in some cases at
the boundary, on the same grain), SAM depth profiles give accurate relative
depth profiles for the oxides.
RBS and [t-RBS on the other hand are non-destructive techniques that
give quantitative depth information from the backscattered spectra. The two
techniques (SAM and ft-RBS) therefore provide complementary information
for very detailed surface characterisation.
above, RBS is used to confirm relative thicknesses. Ion beam milling was
performed on areas that included the interface between the implanted and
non-implanted regions. Information obtained using the SAM technique is now
illustrated with results obtained for oxidised yttrium-implanted alloys.
Figure 19.2a shows a scanning electron micrograph of the area of interest
for a sample which was oxidised for 60 min in air at 700°C. The Y-implanted
region is on the right hand side of the micrograph. A number of analysis points
were chosen in both the a and the (3 phases on both the implanted and non-
implanted regions. The five elements of interest - Co, Cr, AI, Y and 0 - were
monitored before, during and after ion milling through the oxide into the
underlying alloy. Milling was interrupted at selected times to allow elemental
area maps to be made. These times were chosen at, or near, the interface
between the oxide and the alloy for each phase, and examples are shown in
the remainder of Fig. 19.2. The left hand series are the area maps obtained
after 5 minutes of milling (near the oxide/alloy interface for Y-implanted (3-
phase) and the right hand series after 14 minutes milling (near the oxide/alloy
interface for the Y-implanted a-phase). Depth profiles of the five elements, at
each analysis point, were also constructed, using sensitivity factors to convert
the Auger signal to elemental concentrations. Examples of depth profiles for
non-implanted (3-phase and Y-implanted (3-phase are given in Fig. 19.3a and
b respectively. The sensitivity factor used to construct Fig. 19.3 pertain to the
elements in the oxide and therefore the concentrations in the metal are not
accurate. It is possible to construct composite concentration profiles using
sensitivity factors for elements in both the oxide and the metal, but this was
not done in Fig. 19.3.
The depth profiles are strikingly similar to the concentration profiles gen-
erated from RBS spectra (see next section). Preliminary correlations between
the two techniques have been published [2], with a more detailed publication
in preparation. The area maps in Fig. 19.2 show quite distinctly the very
uniform thicknesses of the oxides on the same phases. The area maps also
show the positions of the implanted layer in each of the a and (3 phases after
oxidation. The Y Auger signal was too weak to monitor during depth profil-
ing - it was only possible to detect it accurately during the compiling of the
area maps. In agreement again with RBS results, the Y is in the oxide scale
near the oxide/alloy interface. Because of the considerable difference in the
thickness of the oxide which formed on the two phases, the Y-rich layer on
the a-phase is located at a vastly different depth beneath the outer surface
than the corresponding layer on the (3-phase.
non-implanted Y implanted
a 5 minutes 14 minutes
Fig. 19.2. For caption see the opposite page
19 Characterisation of Oxidised Surfaces 461
b 5 minutes 14 minutes
Fig. 19.2. Scanning electron micrograph and Auger area maps at the interface
region between non-implanted (left of boundary) and V-implanted (right of bound-
ary) Co-22Cr-l1Al oxidised in air for 1 hat 700 0 • The Auger maps on the left side
were recorded after 5 min of ion beam milling, while the maps on the right side
were obtained after 14 min of milling. Depth profiles at points 1 and 2 are shown
in Fig. 19.3
462 J.L. Cocking and G.R. Johnston
point 2 Y - implanted
70
60
.. Co
50 a 0
AI
o Cr
z
o
'=
(3
... ...... .
....
".-
30 .....
Q.
::;: .. "
o :,: "
u 20
10
2 4 6 8 10 12 14
SPUTTER TIME (m inutes)
70
60
... eo
50 D 0
AI
40 o Cr
z
Q A
!::
(/)
...•... " \ '"
30
oQ.
::;: : ~_ .D>oO,.
o D "'~~ 0.<00.
u 20 :/. O.~.o.o..o."""Q>OO>"".o
~\
10 ~ ~
\. .
....
... . .,.
.-........ -.- ....•.~ .....-.-....... .
.~., .• •~:..•• "0 - -0 - - 0- _0 _ ~_.,o. .. o _ _ D _oO_ O .. _ _ _ .. ~
2 4 6 8 10 12 14
SPUTTER TIME (minu tes)
Fig. 19.3. Auger element depth profiles for Co, Cr, Al and 0 (Y concentration too
low to monitor in this mode) at the two points designated in Fig. 19.2, i.e. point 1 -
,6-phase, non-implanted; point 2 - ,6-phase, Y-implanted
19 Characterisation of Oxidised Surfaces 463
ENERGY (MeV)
0.5 1.0 1.5
70
60
0
...J
w 50
>=
0
w
rJ) 40
:::i
..:
::;;
II:
0 30
z
20
10
and non-implanted, are given in Figs. 19.4 and 19.5 respectively. Figures 19.6
and 19.7 give the respective elemental concentration profiles obtained by de-
convolution of the spectra using the RUMP programme [3]. The important
results obtained from Figs. 19.4 to 19.7 are:
Perhaps the most important contribution of RBS and I-t-RBS to this surface
characterisation exercise is that the thickness of the various oxide films are
quantitatively defined without recourse to milling techniques and without the
464 J.L. Cocking and G.R. Johnston
ENERGY (MeV)
0.5 1.0 1.5
.'....,:... ..:...
50
0 40
...J
W
>=
0
w
<f) 30
:J
«
::;;
a:
a
:z 20
10
" Co
60
o 0
* Hf
• AI
50 o Cr
30 ,<'
.•.. '-:: ...
"
0···.0·
"
j . . ."0••:••
20 • 0. •
0
••0 ......... ......... 0. ......... 0. ......... ........ .
0
10
Fig. 19.6. Elemental concentration profiles of the oxide and near-surface region of
oxidised, Hf-implanted i3-phase Co-22Cr-llAl iteratively generated to produce the
spectrum (solid line) in Fig, 19.4
" Co
o 0
60 AI
o Cr
50 Q
E
~
:!i
.§ 40
~ ... ... ..
Z
o . . .. ...... . ...........
",.
~ 30
oQ. , p•. D... "
r'
:; : p.... '0 ••'0. . ....
j~~..,.' ~ .".. . . . . . ·. ·.
o
() 20
0... • ..••• ....•• ....... '0 .........................0
~
10
,D"), ",
",,-
500 1500 2500 3500
THICKNESS (x 10 15 atoms I cm 2 )
Fig. 19.7. Elemental concentration profiles of the oxide and near-surface region
of oxidised, non-implanted ,6-phase Co-22Cr-llAl iteratively generated to produce
the spectrum (solid line) in Fig. 19.5
466 J.L. Cocking and C.R. Johnston
were collected from all samples. The EXAFS data were used in conjunction
with RBS, SAM and STEM data to monitor the changes induced by the
implantation and oxidation processes. Only the EXAFS data will be discussed
in this section.
1.8
1.6 Ni .......
Cr --
:2 1.4
a:
0
u.
(/) 1.2
z
«
a:
unoxidised NiCrAIY
f- 1.0 fcc matrix
u.
0
w 0.8
Cl
=>
t: 0.6
z
Cl
«
:2 0.4
0.2
2 3 4 5 6 7 8
a RADIAL CO-ORDINATE (iI)
1.8
1.6
::; 1.4
a:
oLL
~ 1.2
« unoxidised NiCrAIY
a:
t- 1.0 low dose V - implant
LL
o
w 0.8
o
::J
~ 0.6
(')
«
::; 0.4
0.2
2 3 4 5 6 7 8
b RADIAL CO-ORDINATE (A)
1.6 r
::;
1.4 I
is 1.2
V foil
LL bee matrix
UJ
~ 1.0
a:
t-
(; 0.8
w
o
~ 0.6
z
(')
~ 0.4
0.2
2 3 4 5 6 7 8
c RADIAL CO-ORDINATE (A)
Fig. 19.8. Fourier transforms of (a) the Ni and Cr edges of unoxidised Ni-18Cr-
6Al-O.5Y; (b) the V edge of V-implanted, unoxidised NiCrAlY; and (c) the V edge
of pure vanadium
468 J.L. Cocking and G.R. Johnston
NiO Ni I NiCrAIY
=E Nil NiCrAIY
II:
0 [
I.L - high dose V
2,0
'"..:z oxidised
...cc
I.L 1.5
0
W
a
:;)
I-
.'.
Z 1,0
(!)
..:
=E
0,5
2 3 4 5 6 8
a RADIAL CO-ORDINATE (A)
' .2
1.0
=E
a::
0
I.L
'"..:z 0,8
...cc
I.L
0 0,6
UJ
a
:;)
t::::
z 0.4
(!)
..:
~
0,2
2 3 4 5 6
b RADIAL CO- ORDINATE (A)
1.4
0.2
...
....
.
2 3 4 5 6 7 8
C RADIAL CO-ORDINATE (A)
V20 S V I NiC,AIY
0.6
::; 1.0 - - VPs
a: V I NiC,AIY
0
u.. [
rfl -- high dose V
z 0.8
-0:
a: oxidised
I- 0.4
IL
0 0.6
w
0
=>
'z=
Cl
-0: 0.2
::;
2 3 4 5 6 7 8
d RADIAL CO-ORDINATE (A)
Fig. 19.9. Fourier transforms of (a) the Ni, Cr and V edges of V-implanted Ni-18Cr-
6AI-0.5Y oxidi~ed for 10 min at 900 D C; (b) the Ni edge of oxidised, V-implanted
NiCrAIY (solid line) compared with the Ni edge of pure NiO; (c) the Cr edge of
oxidised, V-implanted NiCrAIY (so lid line) compared with the Cr edge of pure
Cr203; (d) the V edge of oxidised, V-implanted NiCrAIY (solid line) compared
with the V edge of pure V 20 5
470 J.L. Cocking and C.R. Johnston
able differences that are readily apparent. Detailed analyses of these results
are yet to be performed, but it is possible to characterise the oxides scale on
the implanted alloy by following one or both of two paths. Firstly transforms
of mixtures of oxide standards may be obtained experimentally and com-
pared with the oxide formed on the oxidised alloys. Alternatively, computer
generated transforms may be obtained by modelling the oxides. In order to
do this, information provided by other analytical techniques is necessary,
particularly at this stage in the development and application of the EXAFS
technique. The information provided by STEM will enable comparisons of the
transforms obtained experimentally with transforms generated from models
of oxides to assist in the characterisation of the oxides. The present results
highlight the potential of the EXAFS technique for characterising surfaces.
References
1. G.R. Johnston, J.L. Cocking, W.C. Johnson: Oxid. Metals 23, 237 (1985)
2. G.R. Johnston, P.L. Mart, J.L. Cocking, J.W. Butler: Materials Forum 9, 138
(1986)
3. J. Saulitis, G.R. Johnston, J.L. Cocking: Thin Solid Films 166, 201 (1988)
4. C.S. Giggins, B.H. Kear, F.S. Pettit, J.K. Tien: Met. Trans. 5, 1685 (1974)
5. J.G. Smeggil, A.W. Funkenbusch, N.S. Bornstein: Met. Trans 17A, 923 (1986)
6. J ..A.. Spragl!e, G.R. Johnston! F.A. Smirlt~ S.Y. Hwang, G.H. Meier, F.S. Pettit:
In High Temperature Protective Coatings, ed. by S.C. Singhal (Warrendale, PA,
TMS of AIME 1983) pp.93-103
7. F.A. Smidt, G.R. Johnston et al.: In Surface Engineering, ed. by R. Kossowsky,
S.C. Singhal (NATO ASI Ser., Ser. E 1984), pp.507-523
8. J.L. Cocking, J.A. Sprague, J.R. Reed: Surf. Coatings Technol. 36, 133 (1988)
20 Coated Steel
R. Payling
suitability for forming, welding and painting, in their high temperature sta-
bility, and their suitability for containing food stuffs, etc.
The production of coated steel sheet is a multistage process and the
changes in surface composition at each stage of processing must be deter-
mined if a knowledge of the complete operation is to be obtained [2]. Such
information is useful when trying to trace a particular product feature, such
as surface segregation, to a particular processing stage(s). Since each pro-
cessing stage will leave telltale chemical traces which are then overlaid by
subsequent processing, knowledge of the composition of near surface layers
of the product is often more useful than the immediate surface composition.
Such information can often separate the manifestation of a problem, such as
a corrosion product, from the underlying processing variables which caused
the problem.
A typical processing sequence (popularly called a route, or routing) for
painted, metallic coated, steel strip, starting with the steel coil from a hot
strip mill, is [3]:
c) cold reduction, to reduce the steel thickness to the desired level- here, an
emulsion of oil in water is applied to the strip to aid lubrication and heat
dissipation;
d) metallic coating by immersion in a liquid metal bath of controlled com-
position;
e) gas-jet stripping, to control metallic coating thickness;
f) temper rolling and tension levelling, to improve base steel properties and
surface flatness;
g) a paint pretreatment process, involving alkaline cleaning, hot and cold
water rinsing, a proprietary conversion coating, and a final chromate rinse;
h) a multistage painting process, including the application of primer and top
coats, with stoving and water quenching.
20.1 Applications
eo 0
20
~~ ~~
CO 0
0
:> 0
.s
00
~ 0
~ 0
Z
10
0 Boron nitride
+ Nitrogen 0
0
340 400
Fig. 20.1. Change in NKLL peak shape between boron nitride and adsorbed nitro-
gen on steel
as painting, will give problems, providing of course that the sample(s) chosen
is representative of the whole surface. Common to any off-line analysis, one
is always concerned whether the small area analyzed is representative of the
whole 12 tonne steel coil, especially since only the leading or trailing ends of
the coil are readily accessible.
Various surface studies have shown the complexity of residual carbon
forms on steel. Authors differ in their interpretation of the dominant form
[12-15] but carbon may be present as organic molecules, graphite, amorphous
soot or smut, or iron carbide. One common problem with coil annealing of
oiled steel strip is redeposition of carbon near the edge of the strip, so-called
"snaky" edge. While surface analysis is useful here, ultimately the solution
to such problems is found in the proper control of the production process, in
controlling oil levels and furnace conditions, for example.
In XPS studies of oiled steel surfaces, the carbon Is peak is typically found
at a binding energy of 286 eV. Degreasing the sample, to remove excess oils,
shifts this peak to 284.7 ± O.4eV. This peak is present on most real surfaces
and corresponds to a form of tenacious hydrocarbon, often called adventitious
hydrocarbon, which has a reference value of 284.geV [16]. Unfortunately, the
carbon Is peaks for graphite, amorphous carbon, and iron carbide, all occur
around 284.3 e V; so that positive identification of carbon on steel by XPS
is uncertain. Since ion-bombardment of all these forms of carbon on steel
results in the carbon Is peak shifting to 284.2 ± 0.2 e V [3], information on
any variation in the type of carbon with depth is therefore also unobtainable.
An alternative XPS method for characterizing oxygen-containing hydro-
carbons, such as those found in lubricants used for rolling steel, involves a
comparison of the binding energy of the oxygen Is peak with the kinetic en-
ergy of the X-ray excited oxygen Auger K LL peak [3]. Both peaks appear in
the same XPS spectrum and the results are called XPS chemical state plots,
or Wagner plots [17]. One possible source of residual carbon is the formation
of iron soaps between the acid groups in rolling oils and the steel. The Wag-
ner method has been used to demonstrate that organic residues can survive
closed coil annealing at 700°C, though energy shifts indicate some structural
modification does occur.
Auger analysis is capable of distinguishing carbides from other forms of
carbon due to a characteristic peak shape [18]. The CKL L (272eV) peak
shape changes markedly from hydrocarbon (or graphite) to carbide, and the
sensitivity increases by a factor of about 3 [18-20]. When an uncoated steel
sample is ion-bombarded to remove most of the oxide film (typically to a
depth of 10-15 nm), the surface carbon K LL peak shape, typical of either
graphite or hydrocarbon, is replaced by a carbide peak shape [13]. But this
does not necessarily indicate an underlying iron carbide. The measured car-
bon level (rv 6%) is 20 times the bulk value and diffusion of carbon from
the surface or bulk and incorporation of oil during iron oxide formation have
been proposed to explain this apparent carbide enrichment [21]. Steel sam-
20 Coated Steel 477
pIes abraded to mid-thickness, however, show exactly the same behavior and
the explanation may lie in a combination of other factors. First, in differ-
ential spectra a narrow peak gives a greater apparent peak-to-peak height,
so that the peak shape change to the narrow carbide shape exaggerates the
amount of carbide present, by a factor of about 3; secondly, carbon has a
very low sputtering rate and the AES technique analyzes what is left on the
surface during or after sputtering, which exaggerates the total carbon con-
centration measured below the surface; and thirdly, carbon mixed with iron
- possibly as a result of the sputtering process - may give a carbidic AES
peak shape without its being a true carbide, because of short-range Coulomb
interactions.
In the enamelling of steel, for the manufacture of enamelled stoves, bath
tubs, hot water systems, etc., a frit is applied to the steel and then fired
to form the enamel surface. Good adhesion of the enamel requires a suffi-
cient reaction between the frit and the steel surface. In an early study of the
factors affecting the adhesion of enamel to steel, it was found that better
adhesion was related to a thinner surface oxide [3] (Fig. 20.2). Other factors
which have been found to be relevant to enamel adhesion are surface carbon
(also associated with blistering of the enamel coating), silica, and steel grade.
Titanium killed steels, for example, have exceptionally thin oxides (typically
:s; 4 nm) but may give enamel adhesion problems, presumably because the
smf"re m(ine formed on titanium killed steels inhibits the bonding reaction
between the steel and the frit.
4 0
0
I
f-
~
z
w 3 0
a:
f-
Cf)
0
z
0
Ol
...J
W
::;; 2
..:
zw
0
3 9
OXIDE THICKNESS (nm)
Fig. 20.2. Variation in enamel bond strength on steel in one study as a function
of surface oxide thickness
478 R. Payling
Method Composition
Top Phase Bottom Phase
Al Fe Zn Si Al Fe Zn Si
Electron microprobe 54.2 3l.9 7.7 6.2 43.1 48.6 6.8 l.4
Chemical 55.1 32.5 7.2 5.2 52.3 39.3 4.8 3.6
AES 54.2 3l.9 7.7 6.2 56.3 36.0 4.4 3.2
Table 20.2. Standard free energy of oxide formation, llF, at 400°C (673K)
the coating compete at the surface for oxygen. Various mechanisms, such as
minimum surface energy, charge diffusion, or rejection by the solid coating,
may all influence surface composition; but, because of the relatively large
energies involved in oxidation, the oxygen reaction would appear to domi-
nate [25]. Table 20.2 lists the standard free energy of formation for a number
of oxides, referenced to 400°C [26]. This temperature was chosen as repre-
sentative of the high oxidation temperatures of the coatings after exiting the
pot. Within the simple theoretical framework considered, the more negative
the free energy the more likely the species will be found at the surface. For
example, the large negative value for Ab03 compared with ZnO explains the
o 20 20
(lo91Oe).AF/RT
-1
Fig. 20.3. Surface enrichment, (38, for a range of hot-dipped metallic coatings on
steel as a function of the difference in free energy of oxide formation, llF, between
solute (inside brackets) and solvent metals (outside bmckets)
480 R. Payling
aluminum oxide on galvanized coatings. The large negative value for MgO
justifies current interest in magnesium additions. Defining the surface enrich-
ment factor, f3B, by:
S XSjY s
f3 = xcjYc (20.1)
(b) 55%AI-Zn
Fig. 20.4. AES depth profiles of the surfaces of (a) galvanized steel, and (b) 55%
Al-Zn coated steel. The AES signals (pph) have been normalized with modified
elemental sensitivity factors (S')
the variation largely depending on surface carbon levels which are typically
10-30 atomic %. Though significant levels of hydrocarbons are present on all
industrial surfaces, the hydrocarbons on fresh metallic coatings are generally
very thin, typically less than 1 nm.
The aluminum oxide surface on galvanized steel could be expected to form
an excellent barrier to oxygen diffusion and hence a barrier to oxidation of the
zinc coating beneath. Unfortunately, if galvanized steel is left unpassivated
and stored in a wet environment, the aluminum oxide soon converts to alu-
minum hydroxide and an unseemly "white" corrosion of the zinc forms [25].
The lead content of galvanized coatings varies greatly (typically from
o to 1.5 mass %), depending on the manufacturing process. Lead has an
extremely low solubility in solid zinc and so is rejected during solidification
both towards grain boundaries and towards the surface of the coating. This
rejection mechanism is separate from the oxidation mechanism presented
above, as during solidification the lead coalesces within the remaining liquid
to form discrete metallic lead particles at the final solid interfaces rather
than a continuous thin oxide layer on the surface. The absence of lead at the
immediate surface, however, indicates the lead is under the surface aluminum
oxide layer [3].
Since aluminum oxide is a tightly bound, stable oxide in dry conditions,
its stability is likely to be affected by defects in the oxide and these in turn
be affected by trace impurities. Trace levels are beyond the sensitivity lim-
its of Auger analysis (typically limited to concentrations above about 0.5%
atomic). SIMS does not share this limitation and elements detected by SIMS
on the surface of a series of galvanized steel samples but not detected by AES
482 R. Pay ling
were arsenic, indium, lithium, gallium, and fluorine, though copper, tin and
bismuth could not be detected at low levels with the experimental conditions
used because of peak overlap. Clearly, knowledge of the surface composition
is vital in such work.
:§'
·c
:::J
6
.ri
~
>-
I-
en
z 4
lJ.J
I-
~
nearly uniform centre of the coating and the complex zinc, aluminum, silicon,
iron alloy region, described earlier, joining the steel to the coating. Such coat-
ings are now often covered with a very thin (0.5 ~m) non-conductive polymer
coating to aid final fabrication and are best analysed with rf GD-OES.
(c) QUV: 200 light hours (d) QUV: 1000 light hours
1 !--1m
3000
EXPOSURE (light hours)
0 0
+ 50
2500 200
" 1000
~
CJ)
*
C
0
::::>
2000
~
1:
OJ
.Qi
I 1500
.>0::
ttl
Q)
0..
E 1000
::::>
·c
ttl
i!:
500
I I
00 200
Fig. 20.7. Changes in XPS depth profiles for titanium from a commercial polymer
following exposure in a QUV weatherometer
steadily increases with depth, thus providing a possible depth marker for
paint film loss during weathering [32]. The ion-bombardment of polymers is
known to create compositional artefacts under certain conditions, so special
care is required with the technique [35].
As part of a larger program, described in [36], metal panels were coated
with a commercial white silicone modified polyester (SMP) and kept unex-
posed or subjected to 50,200 or 1000 light hours (2000 total hours) exposure
in a cyclic accelerated weathering instrument (QUV). White was chosen be-
cause it is a relatively simple pigmented system based on rutile (titanium
dioxide). Unexposed and exposed samples were then examined by scanning
electron microscopy (SEM) and XPS. The SEM micrographs, in Fig. 20.6,
show that the surface morphology changed markedly with exposure, with
the smooth unexposed surface becoming quite rough following erosion of the
polymer resin and exposure of the pigment particles. XPS analysis combined
with argon ion etching indicated the presence of a pigment-free layer (about
20 nm thick) on the unexposed polymer surface which quickly disappeared
during accelerated testing. The elements detected by XPS were carbon, ni-
trogen, oxygen, silicon and titanium. The surface carbon and nitrogen con-
centrations both dropped during testing, their loss representing the loss of
the polymer and of the melamine cross-linking agent in the polymer, respec-
tively. The nitrogen/carbon ratio decreased during testing, with rapid loss of
486 R. Pay ling
nitrogen in the first 200 hours; whereas, the oxygen/carbon ratio increased
rapidly in the first 200 hours. An increase in oxygen/carbon ratio could in-
dicate increased oxidation or hydrolysis of the polymer, however, the carbon
peak width decreased slightly (from 3.6eV unexposed to 3.3eV at 1000 ho-
rus) which does not support an increase in the number of C-O bonds. The
increasing oxygen signal was therefore more likely to be associated with the
exposed pigment particles. This exposure of pigment is best illustrated by
changes in the titanium depth-profile in XPS (Fig. 20.7) or GD-OES [37].
20.2 Conclusion
References
1. M.P. Seah: Surf. Interface Anal. 2, 222 (1980)
2. P.D. Mercer, R. Payling: Nuclear Instrum. Meth. 191, 283 (1981)
3. P.D. Mercer, R. Payling: Proc. 4th Australian Conf. on Nuclear Techniques of
Analysis, AINSE, Lucas Heights, NSW, 6-8 Nov. 1985, p. 132
4. M.T. Thomas, R.H. Jones, D.R. Baer, S.M. Bruemmer: The PHI Interface
3(2), 3 (1980)
5. A.J. Garratt-Reed, G. Cliff, G.W. Lorimer and R. Pilkington, Interfacial En-
gineering for Optimized Properties Symp., Boston, pp 103-108 (1997)
6. R.H. Edwards, F.J. Barbaro, K.W. Gunn: Metals Forum 5(2), 119 (1982)
7. A.P. Coldren, A. Joshi, D.F. Stein: Metall. Trans A 6a, 2304 (1975)
8. M.P. Seah: Surf. Interface Anal. 1, 86 (1979)
9. G. Hanke, K. Muller: Surf. Sci. 152/153, 902 (1985)
10. R.H. Stulen, R. Bastasz: J. Vac. Sci. Technol. 16(3), 940 (1979)
11. J. Angeli, K. Haselgrbler, E.M. Achammer and H. Burger, Fresenius J. Anal.
Chern. 346, 138 (1993)
12. J.A. Slane, S.P. Clough, J. Riker-Nappier: Metall. Trans. A 9, 1839 (1978)
13. P.L. Coduti: Metals Finishing 78, 51 (1980)
14. V. Leroy: Mater. Sci. Eng. 42, 289 (1980)
15. R.A. Iezzi: PhD Thesis, Lehigh University (1979)
16. L.E. Davis, N.C. MacDonald, P.W. Palmberg, G.E. Riach, R.E. Weber: Hand-
book of Auger Electron Spectroscopy, 2nd ed. (Physical Electronics Div.,
Perkin-Elmer Corp, Eden Prairie, MN 1978)
17. C.D. Wagner, D.A. Zatko, R.H. Raymond: Anal. Chern. 52, 1445 (1980)
18. S. Craig, G.L. Harding, R. Payling: Surf. Sci. 124, 591 (1983)
20 Coated Steel 487
G.C. Morris
The increasing importance of thin films for new technologies has encouraged
fundamental and applied research on their physical and chemical structures
and on the interfaces made with them [1]. Their physical structures (e.g. mor-
phology, topography, crystallite properties, extent and type of defects) are
explored by diffraction and microscopic techniques including X-ray and elec-
tron diffraction, scanning tunnelling microscopy and ultrasonic microscopy.
Their chemical structures (e.g. element type, concentration and spatial distri-
bution) are explored by microanalytical techniques such as Fourier transform
infra-red spectroscopy, secondary ion mass spectrometry (SIMS), X-ray pho-
toelectron spectroscopy (XPS), Auger electron spectroscopy (AES), ion scat-
tering spectroscopy (LEIS, HEIS), as well as dispersive X-ray analysis and
electron energy loss spectrometry with scanning and transmission electron
microscopy. As an ultimate objective, researchers desire a three-dimensional
elemental map on an atomic scale for the thin film and its interfaces. Some
progress towards that aim has been made, but achievement is some time
away.
Thin film analysis may be considered as characterizing materials about
one micron thick in which the outer few atomic layers probably have a differ-
ent structure and composition for a number of reasons especially atmospheric
contamination. A wide diversity of application areas, some of which are noted
in Table 21.1, need such analysis to solve problems and understand phenom-
ena which occur. The analytical techniques used for the particular application
must be carefully selected using criteria such as information depth probed,
lateral resolution required, detection limit needed, elemental type and chemi-
cal state investigated. Complementary methods are recommended to confirm
or deny conclusions made using one technique alone. Thus, a typical analy-
sis might include e.g. X-ray diffraction data, a physical image from electron
microscopy, AES and XPS analyses of specific areas exposed by ion sputter
removal of successive layers (depth profile analysis). The compositional data
from the two techniques, AES and XPS, could be compared. Analytical elec-
tron microscopy could also give useful comparative data. Elsewhere in this
book, each of the main analytical techniques have been reviewed so that their
advantages/disadvantages for a particular application can be assessed.
The aim of this chapter is to illustrate how surface sensitive techniques
provide a data base to understand processes and phenomena which occur with
particular thin films. Example from several of the different application areas
490 G. C. Morris
Table 21.1. Some application areas benefiting from thin film analysis
criteria of efficiency and stability require that the effects which arise from
contaminations at surfaces, in the bulk and at interfaces, must be minimized,
chemical reactions and compositional changes induced by humidity, light and
temperature must be inconsequential, and time dependent processes such as
impurity segregation and interdiffusion must be insignificant. Surface sensi-
tive analytical methods (SIMS, XPS, AES), combined with structural, mor-
phological (XRD, SEM) and microelectrical characterization (electron beam
induced current - EBIC) play a key role in probing the processes which
occur during preparation of the thin films and when fabricating them into
photovoltaic devices. This will be illustrated using one of the most promis-
ing devices for solar energy conversion, viz. that based on the heterojunction
cadmium sulphide/cadmium telluride (nCdS/pCdTe) [2].
Photons Photons
Electrons Electrons
Ions Ions
Cadmium Telluride
Cadmium Sulphide
Glass
Cd Te
102 .............. :· S i ... .
-o
(/)
C
::l 10 1
<.>
>--
I-
(1)
Z .... CRYSTAL ......... :.................. :.............. .
W 103
t-
Z
50 100 150
MASS, amu
Fig. 21.2. Positive SIMS spectra for an electrodeposited CdTe film and a CdTe
single crystal supplied as 5N pure. From [9J
a beam energy of 4 or 5keV, the secondary ion signals which arose from
neutrals at distances greater than 3 mm from the beam centre was 1.6% of
those generated by the primary ion. At distances greater than 6 mm and 9 mm
from the beam centre, the values were 0.3% and 0.1% respectively. Hence, to
minimize the contribution to the 'true' SIMS signal of fringe neutral signals,
two strategies should be used. Firstly, sample mounting hardware such as
cover plates, screws, etc., on the front of the sample should be avoided. A
small stub projecting about 2 cm from the sample mount covered by the
sample has been found useful [9]. Secondly, a large crater such as 9 mm x 9 mm
should be etched during initial profiling and then the analysis taken over a
smaller area such as 3 mm x 3 mm so that data representative of the film's
bulk composition could be collected. Thus the SIMS signal whether from the
primary ion or the fringe neutrals would originate initially from an area at a
similar depth and with similar composition.
The use of an ion gun with a curved ion path can significantly reduce
fringe neutral effects. Analysis of the CdTe films with such a gun (using Cs+
and 0- primary ions) detected each element observed with the simpler in-line
ion gun thus validating the procedures used.
The thin film used to provide the SIMS data of Fig. 21.2 was made using
an A.R. grade (2N) source of cadmium sulphate in the deposition solution
although this was subsequently electrolytically purified. Purer films can be
made with better starting material as illustrated in Fig. 21.3 where SIMS data
for films prepared by A.R. grade (2N) and a 4N grade of cadmium sulphate
are compared. Although the data of Fig. 21.3 show that the film made from
the 2N grade was 'dirtier' over all regions of the mass spectrum, cells as
efficient as those made with the purer starting material were obtained. That
result suggested that the cleaning procedures [5] removed the electrically
active impurities.
Although SIMS is primarily a qualitative technique, it is useful for com-
parative work provided similar matrix samples are used and the precautions
stated above are taken. Useful information as illustrated by the data of
Figs. 21.2 and 21.3 can be obtained. However, quantitation is often needed
and other techniques must be used to complement the SIMS data. Quan-
titative XPS and AES data can be useful, although restricted to elements
with concentrations above 0.1%. Inductively coupled plasma-atomic emis-
sion and atomic-absorption spectrophotometry have been used to determine
the magnitude of impurities in the components of the plating bath solution
and six different supplies of cadmium sulphate [12]. The origin of an impurity
and in some cases an upper limit to its concentration in the films has been
determined.
21 Thin Film Analysis 495
1.5
1.0
~
C
::::!
8
"0
0.5
CD
.!::!
Cii
E
c
o
~ 1-20 21-40
o--I
MASS, amu
-0.5
-1.0
Fig. 21.3. Intensities of positive SIMS signals in various mass regions for electrode-
posited CdTe films prepared using a 2N grade (0) and a 4N grade (0) of cadmium
sulphate. The SIMS counts have been normalized to the combined Cd + Te count
With the type of studies outlined above, the purity of the films produced
under varying conditions could be monitored to ensure that the films were
purer than 5N single crystals purchased from commercial suppliers.
104
.£l
'c
~
OJ
>
~
~
>"
l-
10 3
ii)
Z
UJ
I-
~
102
100 200 300 400 500
DEPTH, nm
function of the depth into the cadmium telluride. Note that the PO+ species
had a similarly shaped profile to the p+ species and was considerably more
intense, making it the preferred ion species to monitor the phosphorus level.
0.2
z
o
i=
~
a:
~ 0.1 l\-~--_________--+-~b'-l
a:
::::J
(f)
0>
I
+
a
15 30
DEPOSITION TIME, min
Table 21.2. Angular resolved XPS data from CdTe films exposed to the atmo-
sphere
f\
.:!~""'" : ~~\
::
f . . . . :#.\. . . . •. : \
.,.! \./
C ..
~
......
......... " '"
,"" ,
,,
,',
I'
,,, ,,,
w ,, '' : '
-.. ,
I
W b ,,
----
Z ,
' ,, , ,,
,,
,
1-.
."\
ii
i i
.I'I
I I ,.
.\
I
\ .r.\
I .
I
.I 'I
I I
ai i I
I Fig. 21.6. XPS spectrum in the Te 3d
\,.._.' . I
..,
I. .
I. _.I
. I region of the surface of a Cd Te crystal ( a)
'-.-
'.i \
" after air anneal 10 min, 350° C, ( b) after
590 580 570 20s etch with 80°C KOH, 30% (w/w) ,
(c) after ion sputtering the oxidized film
BINDING ENERGY, eV for 2 min
When the n-CdTe was type converted by heating the film in air at 350~
400°C for about 10~ 15 minutes, an oxygen-rich film, readily removed by a
basic etch, was formed on the surface as shown by the XPS spectrum for
the Te 3d region (shown in Fig. 21.6) [17]. The oxygen-rich layer, said to be
CdTe03 [18] was shown to be a few nm thick by depth profiling until the
atomic concentrations of Cd and Te were 50%. The change in the Te 3d peak
position as a function of ion sputter time was another useful way of displaying
the surface nature of the oxygen-rich region as shown in Fig. 21.6c.
590 580
Br Cd
1
38 4d
11
Br
3p Te
4d
lJTe I I
Cd3p
l
Te
I) I l
Cd Te 0
I
Cd
MNN MNN 3 3d 18 3d
llJ
W d
z
580
1000 500 o
BINDING ENERGY, eV
Fig. 21. 7. XPS survey spectra of CdTe film after statically etching for 5 s in 0.1 %
bromine-methanol then (a) blow-dried with nitrogen; (b) washed each in methanol
(1 min) and Milli-Q water (1 min) then blow-dried with nitrogen; (c) Te 3d region
from spectrum (a); (d) Te 3d region from spectrum (b)
ing and before and after washing with water. The bromine 3d peaks were
observed near 6geV when the film was not properly washed. Washing also
altered both the intensity and position of the cadmium 3d and tellurium 3d
peaks. The unwashed film showed a strong cadmium 3d signal consigned to
cadmium bromide on the surface and also showed a split tellurium 3d signal
assigned to Teo /Te 2 - and tellurium bromide. Washing removed the bromide
species as evidenced by the disappearance of the Te4+ peaks (Fig. 21.7d) and
500 G. C. Morris
2 ~-<>--od c
a
~
(3 1
o
E
c:
r:-
0.
Ql
o
°O~----------~6~O~----------~1~2~O------------~180
Etch time, S
Fig. 21.8. Depth of Cd depleted surface with etch time for CdTe single crystals
statically etched by BM of different concentrations: (a) 0.1%; (b) 5%; (c) 10%;
(d) 50%. From [19J
a change in the relative intensity of the Cd:Te 3d peaks from 0.7 (unwashed)
to 0.5 (washed).
The depth of the cadmium depleted region was determined using XPS
with ion profiling as a function of chemical etch time and concentration.
A sputter rate of about 0.5nm min- 1 was used with XPS analyses after
successive 1 min sputter times. The atomic concentrations of Cd and Te were
plotted as a function of sputter time with the endpoint reached when the
Cd:Te ratio was 1.00, the value it was in the crystal bulk. Figure 21.8 shows
the result of such a study. The amount of material removed depended on the
time and the concentration of the chemical etch but even though the crystal
was etched at a rate of 14±4nms- 1 by a static 0.1% BM solution, the depth
of the Cd depleted region was limited to about 2 nm. Note that even though
the surface material was not CdTe the depth of the Cd depleted region has
been expressed as if the surface species sputtered at the same rate as CdTe.
This is a convenience, but not accurate e.g. elemental Te sputters about twice
as fast as CdTe.
100~----------------------------~
~_--Au(week)
~ 80
~
Q)
Cl.
Z
o 60 Au (year)
~
a::
I-
z
UJ
o 40
z
o
o
o
~ 20
o
!;i:
Fig. 21.9. Depth profiles of
the atomic concentrations of
C and Au in Au deposited on
SPUTTER TIME, min CdTe films
mally be used to avoid such problems. Note that the atomic concentrations
of carbon and gold in the data of Fig. 21.9 are less than 100% after several
minutes of ion beam etching. Tellurium species are observed in this region,
but a detailed study of the diffusion of the near surface species has not been
published.
21.5 Conclusion
Surface analytic methods are a fruitful way of learning some details of thin
film structures as shown by examples in this chapter mainly dealing with
chemical structures of thin film photovoltaics. The usual methods used to in-
vestigate these devices are by current-voltage-time-temperature relationships
to probe device parameters and current generation and trapping mechanisms.
Various spectroscopies such as time- and energy-resolved luminescence, quan-
tum yield, admittance provide further details, including the type and extent
of traps. Ideally, data from these experiments should be correlated with de-
tails of the physical and chemical structures available from surface analytic
methods. That task is currently too complex for completion for even one
type of thin film device made by one preparative method. However, some
progress has been made and further development of analytic methods of near
atomic resolution will hopefully lead to inexpensive, durable, efficient thin
film photovoltaics.
References
1. H. Oechsner (ed.): Thin Films and Depth Profile Analysis, Topics Curr. Phys.
Vol. 37 (Springer, Berlin, Heidelberg 1984)
2. U.S. Department of Energy, National Photovoltaics Program Five Year Re-
search Plan, 1987~ 1991, DOE/CHI000093-7, 1987
3. W.J. Danaher, L.E. Lyons, G.C. Morris: Solar Energy Materials 12, 137 (1985)
4. G.C. Morris, P.G. Tanner, A. Tottszer: 21st Photovoltaic Specialist Confer-
ence, IEEE (1990)
5. L.E. Lyons, G.C. Morris, D.H. Horton, J.G. Keyes: J. Electroanal. Chern. 168,
101 (1984)
6. B.M. Basol: Solar Cells 23, 69 (1988)
7. G.C. Morris, L.E. Lyons, P. Tanner, C. Owen: Technical Reports of the 4th In-
ternational Photovoltaic Science and Engineering Conference, Sydney (1989),
p.487
8. A 10% efficient solar array covering about 5% of the area of Australia would
generate more electrical power than the world's power stations
9. G.C. Morris, L.E. Lyons, R.K. Tandon, B.J. Wood: Nucl. Instrum. B 35, 257
(1988)
10. C.W. Magee, R.K. Honig: Surf. Interface Anal. 4, 35 (1982)
11. G.C. Morris, B.J. Wood: Materials Forum 15, 44 (1991)
12. L.E. Lyons, G.C. Morris, R.K. Tandon: Solar Energy Materials 18, 315 (1989)
13. D.S. Simons, P. Chi, R.G. Downing, J.R. Ehrstein, J.F. Knudsen: Proc. Sixth
International Conference on Secondary Ion Mass Spectrometry, ed. by A. Bon-
ninghoven, A.M. Huber, H.W. Werner (Wiley, Chichester 1988) p.433
14. National Physics Laboratory, Certified Reference Material NPL No. S7B83,
BCR No. 261
15. G.C. Morris, M. Marychurch: Materials Forum 15, 143 (1991)
16. C.T. Au, M.W. Roberts: Chern. Phys. Lett. 74,472 (1980)
17. W.J. Danaher, L.E. Lyons, G.C. Morris: Appl. Surf. Sci. 22/23, 1083 (1985)
21 Thin Film Analysis 503
B.G. Baker
The practical objective of most of the scientific interest in nitric oxide adsorp-
tion is the control of emissions of this gas into the atmosphere. The control of
506 B.G. Baker
0.2
_ _ _ _ r:i.. c..§.lc-,-
E
.~0.2
..c
~
CIl
Q)
c..
o calc.
Q; 0.2
Cl
::::s
«
Q)
>
~ 0.2
a:
I~',
,
I
,,,
\
\
,, ,,
\
'0'
,,
~,
I , \
, ,,
I ,
,I ,
" ,,
,,
,,
,
,, ,,
I
I
,, ,,
,, ,,
,, ,,
,,
1 ---'_L-...1........L
....1... 1 ---'_.L..
1 , 1 1
-2 -1 -3 -2 -1 EF -2
Energy (eV)
Fig. 22.2. Angle-resolved spectra from clean nickel (100) taken in three crystallo-
graphic directions (full lines); surfaces with adsorbed nitric oxide (broken lines)
The photon source for UPS is typically a helium discharge lamp providing
HeI radiation at 21.2eV (Chap. 14). This is introduced to the surface via
a windowless port with differential pumping between the lamp and work
chamber. The energy range of this form of UPS is rv 16 e V i.e. 21.2 e V less
the work function. The Fermi energy is usually detected and is taken as a
reference so that spectra have an energy scale relative to E F . The transition
metals important in nitric oxide adsorption have a high density of electrons
near the Fermi level. Chemisorption of a gas results in a decrease in intensity
in the spectrum near E F . This is shown in Fig. 22.2 for the adsorption of nitric
oxide on (100) nickel [5]. These spectra are taken in angle-resolved mode in
the apparatus. The surface orientation is (100). Three directions of emission
from this surface reveal differing d-band features. In fact, the shape for the
directions (110) and (111) correspond closely to those observed from surfaces
prepared with these orientations. The effect of adsorption of nitric oxide is
shown in each case to decrease the intensity at close to E F . This represents
electrons from the metal becoming involved in the process of chemisorption.
Nitric oxide may be considered to be intermediate electronically between
CO and O 2 , two molecules capable of reaction with NO. The series CO, NO
and O 2 represent occupancies of the outermost [J* anti bonding orbital of 0,
1 and 2 electrons. Back-donated charge from a metal surface is accommo-
dated in the vacant [J* orbital of CO without causing dissociation of the
chemisorbed molecule. The corresponding process in O 2 results in dissocia-
tion which is always observed as the initial result of chemisorbing oxygen on
22 Identification of Adsorbed Species 509
(a) (b)
300K
277K 368K
259K 356K
223K 341K
192K 329K
170K
114K 299K
92K 265K
-12 -8 -4 0 -12 -8 -4 0
Initial state energy (eV)
Fig. 22.3. (a) Initial adsorption of NO on Fe(llO) at 92K (0:1 state) followed by
a sequence of temperature flashes to form the (3 state. (b) Initial adsorption of NO
on Fe(llO) at 265 K ((3 state) followed by a sequence of temperature flashes to form
the dissociated (d) state
metals. But NO with one electron in the Jl* anti bonding may form molecular
or dissociated adsorbate depending on the conditions.
A study of the adsorption of NO on (110) iron by UPS has revealed
considerable complexity in the modes of molecular adsorption [6]. The initial
adsorption of NO at 92 K results in the UPS spectra shown in Fig. 22.3. The
broken lines mark features not present in the spectrum of the clean (110)
iron surface (see also Fig. 22.4). They are attributed to a molecular adsorbed
state of NO referred to as 001. Repeated temperature flashes show a series of
spectral changes until at temperatures above 330 K a single major feature is
centred at -5.5 eV (Fig. 22.3b). This feature indicates photo emission from
N 2p and 0 2p and hence atomic adsorbates. The important characteristic
spectra are summarized in Fig. 22.4 where they are represented as difference
spectra by subtracting the spectrum of the clean Fe(110) crystal.
Thus the adsorption of nitric oxide on an Fe(110) single crystal surface at
temperatures 90-350 K results in the formation of at least four distinguishable
adsorption states, depending on the substrate temperature. The molecular
adsorption states, 001 over 90-110 K and 0:2 over 110-170 K, are both thought
to involve NO chemisorb ed, N end down, possibly at an off-normal angle.
Only slight differences exist between the valence electronic structure of the a
states. The (3 adsorption state exists over a very wide temperature range of
170-290 K, although its concentration can be made to peak at 270 K. A model
for the (3 state is that the initial dissociative adsorption occurs randomly and
non-incorporatively, thereby allowing the formation of single vacant sites at
Fig. 22.4. HeI UPS difference spec-
tra, weight averaged and smoothed,
comparing the 001, 002 and (3 molec-
Clean ular adsorptions. An estimate of the
-12 -8 -4 0 inelastic scattering contribution is
Initial state energy (eV) shown by dotted curves
o o o
II II II
NON N N 0 0 N N N
-
0
399.0
l
Au
1.26
w (d)
z
(c)
Fig. 22.5. N Is photoelectron peak for (a) Au(CN);- adsorbed from alkali then
subjected to increasingly severe acid (1M HCI) reaction, (b) 15min at 298K, (c)
180 min at 298 K and (d) 15 min at 373 K. Curves normalized to constant gold
coverage
2.456 A
f-----i
C-Au(CN)2"
Au-C 2.12A
C-N 1.17 A
C-Au 4 (CN);;
Au-Au 5.09 A
Au-N 1.97 A
Au-C 1.97 A
C-Nb 1.15 A
C-N t 1.17 A
would have both terminal and bridged N and aN / Au ratio greater than unity.
The species AU4(CN)5 is proposed. This has N/ Au = 1.25 and Nb/Nt = 1.5
in excellent agreement with the experiment.
The monomer and tetramer are represented in Fig. 22.6. The Au(CN);-
species adsorbed on activated carbon from alkaline solution is adsorbed on
and parallel to the graphitic planes. The bonding mechanism proposed is a 'if-
donor bond from the graphite surface to the central cationic gold atom. The
Au 4fT/2 peak has a binding energy 0.3 to 0.5eV lower than that observed
for the compound KAu( CNh suggesting that the surface bond involves a net
charge transfer of this magnitude. A shift of the same magnitude exists for
the N Is peak from the adsorbate suggesting that there is also a transfer of
charge to the terminal nitrogens. This is not an indication of direct bonding
of N to the surface but a consequence of the 'if-donor bond to the gold.
A similar bonding model is applied to the tetramer AU4(CN);- formed
by the acid induced oligomerization of C-Au(CN)2' The geometry of the
adsorbate on the graphitic carbon lattice is shown in Fig. 22.6. All four gold
atoms are bound by 'if-donor bonds each resulting in the same BE shift for
Au 4fT /2 and hence the same charge transfer. In this case, however, there are
four gold atoms and only two terminal nitrogens: the observed BE shift in N
Is is larger.
The above study is remarkable in that a single technique, XPS, has been
applied to samples prepared in solution to yield results comparable in detail
to many in situ, single crystal, multitechnique experiments. The absence of
514 B.G. Baker
Fig. 22.7. Schematic of a 127° high resolution electron energy loss spectrometer for
vibrational studies of surfaces. The incidence angle is 60° from the surface normal
r - 1010
x100 Cu(100)
+CH 3 0H
370K
290
!
! ! lr
1450 1960
2830 2910
--......-- .......-.....".~
o 1000 2000 3000
Energy Loss (cm- 1)
x3
.,.mo
100 K(a)
>-
!=
C/)
z 100 K(b)
W
I-
Z
x10 Fig. 22.9. Energy loss spectra
of the reaction of methanol
(CH30H) with atomic oxygen on
Pt(111) to produce the methoxy
intermediate (CH 3 0). (a) Atomic
oxygen p(2 x 2) structure. (b) Ex-
cess methanol condensed on (a).
Annealing to 170 K produced the
methoxy spectrum (lower). The
ENERGY LOSS (cm- 1) beam energy was near 1 e V
A. Pt(111) + CH30H
A
~
oT
en
:!:
C I °
r'"
~ 104 Hz
100K
.0
'- x10
~
>-
!:::
en
z x30
w
I-
Z 00
00
'10 Fig. 22.10. Comparison of the
,T
I
104 Hz
energy loss spectra of multilayer
methanol (100 K) and monolayer
methanol (155 K) on Pt(111). In
the monolayer spectra consider-
able "softening" of the methyl
o 1000 2000 3000 4000 group CH stretching modes can
ENERGY LOSS (cm-1) be seen. The beam was near 1 eV
C-R bonds in the methyl group and the CO inverts to bind carbon to the
surface, the usual mode of adsorption of CO.
The breakdown of the methyl group at low temperature clearly precludes
the chance to produce formaldehyde. The mechanism may well occur On
transition metals generally which strongly chemisorb hydrogen and carbon
monoxide. Evidence for the effect of the platinum surface on methanol is
shown by the EEL spectra in Fig. 22.10 [11]. The upper trace is for bulk
methanol, i.e. a multilayer of methanol ice. No oxygen is added. The lower
trace is for a monolayer of physically adsorbed methanol at 155 K. In this case
vibrational modes are sharper and the OR stretch has moved to 3280cm- 1 .
This is an indication of decreased hydrogen bonding between molecules in the
two-dimensional layer. Also in the two-dimensional layer, new frequencies are
observed in the CR region near 2900cm- 1 . These occur at lower energies and
indicate that there is strong hydrogen bonding of the methyl group hydrogens
to the platinum surface. Thus the methyl group in the physically adsorbed
state is strongly influenced by the Pt(I11) surface. When this monolayer
was heated to > 200 K, complete decomposition to CO and hydrogen was
observed by EELS and no methoxy groups formed.
The distinction between copper and platinum as catalysts for the selective
oxidation of methanol to formaldehyde then depends on the stability of the
methoxy species. The EEL spectra show that this species is more stable on
518 B.G. Baker
22.2 Conclusion
The examples discussed above have been chosen to illustrate the detailed
nature of the information about adsorbates obtainable by surface analyt-
ical techniques. The application of LEED, AES, UPS, XPS and EELS to
particular adsorption systems has been discussed. These techniques are best
combined with other methods of detecting reaction or desorption by monitor-
ing the gas phase by mass spectrometer. It is also generally found that more
than one surface technique is needed to obtain a satisfactory description of an
adsorbate. The emphasis on single crystal techniques is inherent in this area
of surface science. A surface exposing multiple crystal planes would produce
mixed spectra from multiple adsorbate states. The behaviour on practical
surfaces is best deduced by measuring overall properties of adsorption and
reactivity of the practical surface and comparing with the detailed studies on
various crystal planes of pure substances.
References
1. B.G. Baker, R.F. Peterson: Proc. Sixth Int. Congress on Catalysis 2, 988-996
(1977)
2. G.L. Price, B.A. Sexton, B.G. Baker: Surf. Sci. 60, 506 (1976)
3. G.L. Price, B.G. Baker: Surf Sci. 91, 571 (1980)
4. H. Conrad, G. Ertl, J. Kiippers, E.E. Latta: Surf. Sci. 50, 296 (1975)
5. G.L. Price, B.G. Baker: Surf. Sci. 68, 507 (1977)
6. C. Klauber, B.G. Baker: Appl. Surf. Sci. 22/23, 486 (1985)
7. C. Klauber, B.G. Baker: Surf. Sci. 121, L513 (1982)
8. C. Klauber: Surf. Sci. 203, 118 (1988)
9. B.A. Sexton: Appl. Phys. A 26, 1 (1981)
10. B.A. Sexton: Surf. Sci. 88, 299 (1979)
11. B.A. Sexton: Surf. Sci. 102, 271 (1981)
23 Surface Analysis of Polymers
• the fact that most polymers contain various additives and/or are pro-
cessed with the aid of processing agents such as extrusion slip agents
• the marked susceptibility of most polymers to compositional alterations
("radiation damage") while the analysis is proceeding
• the inability to perform depth profiling by ion beam sputtering and
• the multifunctional nature of many polymeric surfaces
In addition, as for inorganic analyte surfaces, issues such as ready ad-
sorption of contaminants and charging of insulating surface layers must be
considered.
The two main issues of relevance are the covalent connectivities of poly-
mers and the intrinsic mobility of polymer chains. The multitude of covalent
bonds that connect the various atomic constituents of polymer materials lead
to unique effects such as the preferential formation of specific fragments in
SSIMS. The intrinsic mobility of polymer chains can lead to polymer surface
layers displaying compositions and properties that are time-dependent and
can vary with the environment that the polymer is exposed to.
While the range of elements that are typically found in polymers is quite
limited, with e, H, N, 0, and Si the most frequent and F, el, Br, and SIess
frequent, and a few other elements very occasionally found, the ways in which
these elements can be assembled into polymeric materials is literally unlim-
ited. Hydrocarbon polymers such as polyethylene, polypropylene, polybuta-
diene, polystyrene, and poly-para-xylylene possess very different properties
as a result of the ways in which the e and H atoms are connected. Moreover,
the properties also depend on the average molecular weight (there always
arises a distribution of chain lengths and thus molecular weights of polymer
chains as a result of the random nature of monomer addition processes),
crosslinking degree, and processing conditions (for instance polyethylene is
fabricated with different densities which affects mechanical properties). How-
ever, established surface analysis methods, particularly XPS, are best suited
to elucidating elemental compositions and consequently are relatively insen-
sitive to the long-range discrete structural connections that are so important
for polymers. It is necessary to study primary and secondary chemical shifts,
shake-up satellites, and valence band spectra in order to gain an apprecia-
tion of the composition and structure of polymer surfaces beyond the level
of nearest-neighbour elemental bonds.
As for inorganic materials, the spatial arrangements of the atomic con-
stituents of polymers also play a key role. Most polymers are composed of
regular repeat structures that are added during synthesis in a linear fash-
ion to assemble to long linear polymer chains. Some polymers also possess a
branched structure which confers crosslinks between linear segments. Exam-
ples of polymer structures are shown in Fig. 23.1. The covalent connections
23 Surface Analysis of Polymers 521
Poly(ethylene) Poly(styrene)
Poly(tetra Poly(dimethyl
fluoroethylene) siloxane)
CH3
iSliI -01:;n
CH3
Poly( urethane)
and the long-range order are responsible for many properties of polymers and
their surfaces and can be evident in spectral data. While nearest-neighbour
interactions are most prominent, often it is essential to also take into account
longer-distance effects. In XPS, secondary shifts can be quite marked and
may lead to erroneous interpretations of functional group signals particularly
in the C Is spectral region.
Polymeric surfaces tend to be more difficult to analyze than metals and
ceramics because polymer surfaces generally are mobile and unpredictable.
The atoms that constitute polymers can change their spatial positions with
time over considerable distances [5,6,1]. Motions that alter the spatial posi-
tions of constituent atoms in polymers are of two types, viz. rotational and
translational (reptation). Polymer chains have freedom of rotation around
the central polymeric backbone (the long chain of covalently connected atoms
that form the principal cohesive linkage) unless such rotation is hindered by
conjugation (n-bonds) or steric crowding. There also are concerted motions
of polymeric chain segments which, like rotational movements, are driven by
random Brownian motion. The relatively low packing density of polymers
(compared with inorganic materials) provides void spaces of atomic dimen-
sions which enable such motions to occur. Rotational and segmental motions
are, however, short distance; longer distances are not easily traversed by the
"sideways" motions typical of segmental flexibility. Long-distance diffusion
is achieved by reptation, a worm-like motion in which a polymer chain end
diffuses into a temporary gap in-between other chains and drags its entire
chain along. The two motions, rotational and translational, are illustrated in
Fig. 23.2.
522 H.A.W. StJohn, T.R. Gengenbach, P.G. Hartley, and H.J. Griesser
OH
Fig. 23.2. Two possible motions of polymer chains (see text for details).
70 (a)
60
50
~
c
:€:g 40
0.
S
0
u 30
20
10
0 10 15
Depth (A)
-- Fluorine
70 (b) - Oxygen
- Nitrogen
-- Carbon (CF2)
60 - Carbon (other)
50
~
c
:~ 40
0
0.
S
0
u 30
20
10
0 10 15 20 25 30 35
Depth (A)
Fig. 23.3. Compositional depth profile of an NH3 plasma treated fluorinated ethy-
lene propylene copolymer (FEP) determined by angle-resolved XPS (a) immedi-
ately after the surface treatment and (b) after storage in air for several months.
In the case of the aged sample, nitrogen-carrying polymer segments, formed at the
surface by the plasma treatment, have partially moved below the surface and have
been replaced by fluorocarbon segments. Oxygen which is being incorporated due
to oxidative processes during aging also moves toward the bulk.
23 Surface Analysis of Polymers 525
o Is
50000
CIs
40000
I 30000
Si 2s
Si2p
02s
20000
o (KVV)
10000
o
1000 800 600 400 200 o
Binding Energy (e V)
Fig. 23.4. XPS survey spectrum of a commercial contact lens (poly hydroxyethyl-
methacrylate). Photoelectron signals due to silicon are clear evidence for contam-
ination. The Si 2p binding energy of about 102 eV indicates the presence of an
organosilicon compound. Also detected is residual sodium from the saline solution
which is used to store the contact lens.
signals due to the polymer itself and can lead to erroneous interpretations of
surface segregation and mobility effects, for instance.
Silicones are the most common contaminants observed on polymer sam-
ples, particularly those from an industrial/commercial source. Fortunately,
the presence of Si signals in XPS provides a clear indicator, unless the poly-
mer itself contains this element; and the high positive ion yield of silicones in
SSIMS allows unambiguous identification. The pervasiveness of this class of
contaminants is linked to its common usage in a variety of applications such
as lubricants and mold release agents, its low surface tension, and its ability
to migrate across a surface [10]. XPS data obtained from a contaminated
contact lens are shown in Fig. 23.4. Other commonly observed contaminants
are additives, including plasticizers, catalysts, antioxidants, light stabilizers,
slip extrusion agents, and others which may be included at a very low con-
centration as a bulk additive, but preferentially migrate to the outermost
surface of the polymer. Some of these additives, particularly slip extrusion
23 Surface Analysis of Polymers 527
agents which act as lubricants, can be very surface active and are either ap-
plied to the surface or accumulate at the surface by diffusion. Thus, they
interfere with surface analyses and can cause erroneous interpretations of
the results of surface modification processes [11], since additives, being low
molecular weight organic molecules, often do not possess a unique elemental
signal. The propensity of these substances to surface segregate results in in-
creased distortion of results for techniques most sensitive to the outermost
atomic layer, such as static SIMS and contact angle measurements. Further-
more, the high ion yield of PDMS fragments in static SIMS relative to other
organic polymers means that even for trace contamination (e.g. 0.2 % Si as
measured by XPS) the PDMS spectrum swamps the signals of interest.
As outlined in Chapter 2, techniques such as solvent cleaning or ion beam
etching are often used to remove contamination from samples prior to anal-
ysis. These approaches are not suitable for polymers. Solvent cleaning can
lead to swelling of the polymer and enhance the mobility and rearrange-
ment of the surface region, or solubilise low molecular weight components.
Thus the cleaned surface analysed does not reflect the 'true' surface [12]. Ion
beam etching is also not suitable for cleaning polymer surfaces. Chemical
functionalities of organic materials are easily degraded leading to preferential
sputtering and subsequent chemical reactions. The type of chemical reac-
tions induced depend on the initial composition of the material. Argon ion
irradiation may result in dehydrogenation, loss of heteroatoms and loss of aro-
maticity. Therefore neither atomic concentration data nor molecular species
information reliably reflect the composition of the original material after ion
beam etching [13,14].
The surface of as-supplied polymers often is far from what one may ex-
pect it to be. In fact, experience shows that very few polymers possess clean
surfaces; most must be rigorously cleaned from contaminants and additives
before reliable analyses are possible. A good clean reference polymer is Teflon
FEP, which can be used for calibration of XPS elemental F IC ratios. Pre-
sumably its very low surface energy prevents adsorption of the typical adven-
titious contaminants (see Fig. 23.5). Teflon PTFE, on the other hand, is not
as suitable as it requires addition of a hydrocarbon compound for processing.
Therefore, surface analysis of polymer samples requires rigorous adher-
ence to the protocols of clean sample handling as outlined in Chapter 2. In
addition, prior to commencement of any surface modification experimenta-
tion it is worthwhile to remove any potential source of surface contaminants
from the bulk material itself: This can be efficiently achieved by sequential
Soxhlet extraction with a polar solvent followed by a non-polar solvent. In
all subsequent stages of sample treatment it is wise to adopt the philoso-
phy that tools, beakers, etc, are a potential source of contamination unless
proven otherwise. The time between sample treatment and analysis should
be kept as brief as possible to minimise the extent of surface reorientation or
out-diffusion.
528 H.A.W. StJohn, T.R. Gengenbach, P.G. Hartley, and H.J. Griesser
F (1s)
(a)
F(KVV)
L-A~
C 1s
I I
800 600 400 200
Binding Energy (eV)
Fig. 23.5. XPS spectra of Teflon FEP: (a) survey spectrum and (b) high resolution
carbon Is spectra. In the latter case data from two different specimens are plotted,
the first obtained from a section of clean FEP, the second from a slightly contam-
inated sample. The spectrum of the latter clearly shows a hydrocarbon signal at
around 285.0 eV (top spectrum).
23 Surface Analysis of Polymers 529
This is the simplest, least expensive and most rapid method for assessing
the surface of polymers. It is, however, chemically least informative, basically
providing only information about the surface energy of the material and thus
about the presence of hydrophobic (nonpolar) and hydrophilic (polar) groups
on the surface. From the difference between advancing and receding contact
angles, one may estimate the fraction of polar and nonpolar surface segments.
Contact angles are also affected by the surface topography of the material,
and hence, for chemically well characterized surfaces, one can also infer a
roughness factor as a measure of the topography [1,6].
Contact angle determinations are most valuable for rapidly detecting the
presence - but not identity - of surface contaminants, and for studying the
extent and stability of surface modification. Often contaminants which adsorb
from the vapour phase are less polar, and thus an air/water contact angle
higher than expected is an indication of contamination. Contact angles also
enable the rapid verification of the efficacy of surface modification procedures
which, by the insertion of new groups, usually alter the air/water contact
angles considerably.
1200 C-H
C-C
1000
800
W
"
0
-"-
.~ 600
<J)
c:
~
400
200
A= M. EK (23.2)
pn 14.llnEK - 25.5 -1500/EK
23 Surface Analysis of Polymers 533
where M is the molecular weight of the repeat unit, p the density, n the no.
of valence electrons in the repeat unit, A the IMFP in Angstroms, and EK
denotes the kinetic energy of the photoelectron. Tabulations of the Ashley
parameter (M/ pn) are available for common organic polymers, making the
calculation of A for a given energy straightforward.
To be scientifically rigorous, the best predicitive equation currently avail-
able is that developed by Tanuma, Powell and Penn, known as TPP-2M. This
is the refined version of their earlier equations, and is appropriate for organic
materials. It is based on a modified form of the Bethe equation for elastic
scattering [23].
A= E (23.3)
E;, [;9lnbE) - (G/E) + (D/E2))]
where A is the IMFP in A, and ,,(, ;9, G, D and Ep are material related
parameters that take into account atomic weight, density, number of valence
electrons and the bandgap energy (see reference for full definition of variables)
As this algorithm is somewhat unwieldy and the bandgap energy may
not be known, Gries has recently proposed a simplified predictive algorithm
(G1) based on an atomistic model of inelastic electron scattering [24]. It is
claimed to be more widely applicable than the TPP-2M equation. Though
Tanuma and coworkers have criticised this model for being inappropriate for
metals, semiconductors and inorganic compounds, they acknowledge it may
be relevant for molecular solids where the interactions between constituent
molecules are weak. As for the Ashley equation, it also predicts that the
IMFP will be inversely proportional to density.
(23.4)
section and is stable under the conditions of analysis. This approach is anal-
ogous to the derivatization methods employed in other disciplines such as gas
chromatography, or nitroxide derivatives used as spin labels in ESR.
A number of potential reagents have been proposed for derivatization,
and these have been summarised in reviews by Batich [34], Andrade [2] and
Briggs and Seah [3]. However, many of these have been proposed on the
basis of solution organic chemistry, and the analagous reactions may not
proceed as expected at a polymer surface. Evaluation of the reactions on
model polymer surfaces is required. The key criteria for surface analytical
derivatization reagents are summarised below, with examples drawn from
common derivatization reactions. These selected derivatization schemes are
listed in Table 23.l.
There are several key criteria in the design of a derivatization scheme:
the degree of substitution which will affect the ultimate yield. For example,
Alexander's study of the derivatization of carboxylic acid groups by vapour
phase trifiuoroethanol showed that acid groups of polyacrylic acid were com-
pletely labelled after 10 hours, while the conversion for polymethacrylic acid
was only around 60% even after 75 hours [35].
3. Selectivity. Ideally the reagent should react only with the functional group
of interest. At the very least, cross-reactions with other functional groups
must be known so that their contribution can be considered. This is particu-
larly important for complex multifunctional surfaces, but unfortunately very
few reagents have been compared systematically on a range of model poly-
mers. One instructive example is the investigation by Chilkoti and Ratner
on the reactivity of trifiuoracetic anhydride [36]. Previously it was thought
that TFAA could be used to selectively label hydroxyl groups on highly oxi-
dised surfaces (on nitrogen-free samples, as TFAA is known to also react with
amines, amides and ureas). However, in their study encompassing 15 different
polymer surfaces it was shown that TFAA also reacted rapidly and quanti-
tatively with epoxide groups, as well as to a lesser extent with ketone and
carboxylic acid groups. Non-specific adsorption can also result in significant
background signal in some derivatization schemes.
During a SIMS experiment the sample is irradiated with primary ions (Ar+
being the most widely used), and the secondary ions emitted from the sam-
ple surface are collected and mass analyzed. For polymer surface character-
ization it is important to operate the spectrometer in the "static" mode,
i.e. the accumulated ion dose during spectral acquisition must be kept suf-
ficiently low for the surface to be essentially unperturbed for the duration
of the experiment. Under typical conditions (2 keY < ion energy < 4 keY)
the primary ion penetrates to a depth of ,,-,3 nanometers below the surface.
Transfer of kinetic energy within the material, as well as electronic interac-
tions, lead to primary events such as bond cleavage reactions and the emission
of small fragments, followed by indirect processes such as the desorption of
large molecular fragments some distance from the primary ion impact site.
About 5 % of the material sputtered from the surface consists of positively
and negatively charged ions (atoms or molecular fragments), and these are
mass analyzed to produce positive and negative, respectively, SIMS spec-
tra. Quadrupole mass spectrometers are most commonly used at present but
their mass range is limited; they are not ideally suited to the surface anal-
ysis of polymers. State-of-the-art instruments employ time-of-flight (TOF)
538 H.A.W. Stjohn, T.R. Gengenbach, P.G. Hartley, and H.J. Griesser
Table 23.2. Assignment of major peaks observed in the positive ion spectrum of
a pHEMA contact lens (see Fig. 23.7).
15 CHt
27 C2 H t
29 C2 H t
39 C 3 Ht
41 C3 H t
43 C 3 Hi C2 H 3 0+
45 C2Ht (pendant hydroxyethyl group)
53 C 3 HO+
55
69 C 4 H 5 0+ (methacryloyl ion)
71 C 4 H 7 0+
89 C4 H g Ot
91 c7Hi
113 [M-OH]+ (M=backbone subunit of pHEMA)
43 r~:--:::::::--::--~-----: 113
:: CH :69 :
41 ,' 1 3, ,
-;'+-CH - C - : :
:: 2 1 : :
600 :: C=O::
27 : ~-------+----: :
: a :
: ,--1----------+----; 45
: : CH2-CH 2,OH:
39 45 :_______ ~ :::::: ::::: ::l----_.
29
55
69
53 113
200 15
91
groups and for which no similar, well characterized reference material is avail-
able.
The information depth of static SIMS is assumed to be of the order of
1 nm, i.e. the very surface layers of a polymer are probed. This high surface
sensitivity is of particular interest where one needs to characterize the surface
chemistry of a polymer on a depth scale of less than 1 nm, which is typically
the range of interfacial forces. The low information depth, high absolute sen-
sitivity, high molecular specificity, and a spatial resolution of the order of
several [lm which allows imaging to be done, make SIMS a relatively recent
but very valuable addition to existing polymer surface analysis techniques.
The complementary combination of XPS and static SIMS has already proven
to be powerful in the analysis of modified polymer surfaces.
Static SIMS is also a powerful method for detecting and identifying low
molecular weight organic contaminants on polymeric surfaces [15]; the mass of
the molecular parent ion, if observed, and characteristic SIMS fragmentation
patterns of classes of organic compounds can provide information on the
nature of the contaminant. Figure 23.8 shows an example [40] of a lipid
adsorbed onto a commercial hydrogel contact lens material. The peak at
284 m. u. is assigned to the molecular ion. The characteristic fragmentation
pattern of a series of peaks separated by 14 m.u. indicates an extended linear
540 H.A.W. Stjohn, T.R. Gengenbach, P.G. Hartley, and H.J. Griesser
120 284
60
142
170
156
268
184
.l!l
c
:::J
o 198
()
212
226 240
20 254
Fig.23.8. SIMS spectrum of a lipid adsorbed onto a contact lens (see text for
details) .
Fig. 23.9. AFM Tapping Mode Image of a Plasma Polymer film deposited on a
freshly cleaved mica surface. The step was created by masking one side of the sample
during the deposition, then removing the mask prior to imaging.
542 H.A.W. Stjohn, T .R. Gengenbach, P.G. Hartley, and H.J . Griesser
500.0 om
250.0 om
0.0 nm
I'm
20.0 deg
2.00
10.0 deg
0.0 deg
1.00
o
I'm
Fig. 23.10. 2.5 X 2.5 [.lm Tapping mode (top) and Phase mode (bottom) image of a
stretched polyethylene film, highlighting the increased two dimensional resolution
attainable using phase imaging.
mode of operation, in which the tip oscillates normal to the sample surface,
thereby reducing contact time (see Fig. 23.9). In this case, variations in both
the amplitude and phase of the tip oscillation contain useful information.
Indeed, so-called 'Phase Imaging ' has shown great promise in the analysis of
polymer surfaces due to its exquisite sensitivity to variations in composition,
adhesion, friction and viscoelastic properties (see Fig. 23.10) [42].
The control of tip-sample interactions has also been used to gain insights
into interfacial chemical and mechanical properties which are frequently not
accessible from the topology information. In force modulation imaging, the tip
is oscillated whilst in contact with the sample. By monitoring the compliance
23 Surface Analysis of Polymers 543
of the surface through the bending of the AFM tip, two-dimensional images
of the local elasticity of the sample surface can then be constructed. Such
imaging is particularly useful in detecting differences in crystallinity across
the surface of a polymer sample [43].
The torsional forces experienced by the AFM tip as it is scanned across the
sample also provide a measure of the frictional force between it and the sur-
face. This Lateral Force Microscopy (LFM) allows chemically distinct regions
on surfaces to be differentiated through their affinity for the tip. By modifying
the surface chemistry of AFM tips, it has also been possible to compare the
frictional forces resulting from the interactions operating between surfaces
bearing specific functionalities, and to construct two dimensional images of
chemically patterned surfaces despite minimal differences in topology [44].
Measurements of the long range forces operating between AFM tips of
well defined geometry and surfaces of interest also yields valuable informa-
tion regarding the interactions of polymer surfaces, particularly in solvent
environments. In this mode of operation, native AFM tips or AFM tips mod-
ified with micrometre sized colloidal spheres are brought into close proximity
with the surface, and force versus separation curves are constructed from the
deflection of the tip at different positions normal to the surface [45]. Such
force versus separation curves reveal information regarding the extension of
polymer chains from a surface, as well as providing insights into other prop-
erties of the polymer surface, such as charge density and hydrophobicity.
• 0.16
0.32
1.8
o
.~
([
c:
.8 1.6
<ii
~c:
"§
&:1.4
1.2 (a)
.gl
§ 1000
8
E
(;
.s ....
microscopes used in the contact mode; in this mode the tip being scanned
across the sample is in actual contact with the surface atoms. It may there-
fore penetrate the surface of soft organic specimens and even partially sweep
away material during a scan. Since a minimum load force is necessary for
the tip to stay in contact with the surface (up to 100 nN) a certain degree of
damage is inevitable. Several methods have been developed to reduce sample
damage (see for example [53,54]):
(1) In the non-contact mode of operation the tip 'floats' above the surface,
sampling the long-range surface forces operating between tip and sample.
(2) Tapping Mode AFM: in this mode the tip oscillates with a drive am-
plitude which is sufficiently high to ensure periodical contact with the
surface, thus providing the necessary feedback signal for the AFM. Elim-
ination of the friction force results in a reduction of the overall force by
at least an order of magnitude.
(3) Scanning under liquid removes capillary forces which are present in air
due to a thin hydrated layer on the polymer surface. The load force may
be reduced by up to two orders of magnitude.
either combined with a metal mesh above the specimen or with a magnetic
immersion lens which focusses the electrons on the surface.
As for all insulating materials, the use of a reliable energy reference is
crucial in determining binding energies. The most reliable method of spectral
referencing is the use of an internal standard; this is particularly well suited
to the analysis of most polymeric materials because of the prominence of a
signal with known binding energy originating from the polymer backbone. For
example, the binding energy of a Is photoelectron emitted by a carbon atom
in aliphatic structures bonded to carbon and/or hydrogen (285.00eV) is now
well established [16]. Various other methods of binding energy calibration
have been investigated, ego the use of adventitious carbon contamination
which is detected on almost every surface or the deliberate deposition of
a reference element onto the surface prior to analysis (evaporation of gold,
implantation of Ar ions). However, they all suffer from specific problems
such as ill-defined binding energy (adventitious carbon) or relaxation effects
[55,56]. Deposition of a reference element on the surface may even partially
degrade organic materials.
Charge neutralisation or, more appropriately, the control of surface poten-
tial and its effects on the information content of spectra, is considerably more
difficult in SIMS than in XPS. Here the main problem is not a shift of peak
positions but a shift of the energy distribution of secondary ions; as a con-
sequence, relative peak intensities change or a complete loss of signal might
occur [52,57,58]. The objective is to adjust the surface potential in order
to match the maximum of the secondary ion distribution to the acceptance
"window" of the mass analyzer. This is usually achieved by a combination of
an electron flood gun (electron energies of up to several hundred eV) and the
application of a bias potential to the specimen. The use of electron sources
employing rather high electron energies and current densities for charge neu-
tralisation, however, can also give rise to potentially serious problems such as
electron stimulated ion emission and sample degradation. In some cases using
a primary atom beam instead of ions can facilitate surface potential control.
Generally, controlling the surface potential is more difficult in the negative
ion mode compared to the positive ion mode and more difficult in imaging
mode compared to spectroscopic mode. Due to the substantially lower pri-
mary ion beam currents and its pulsed nature in TOF instruments, charging
is much less of a problem but the surface potential still needs to be controlled.
The most successful method is based on combining the pulsed primary ion
beam with a pulsed secondary ion extraction field and a pulsed electron flood
source in a precisely controlled sequence.
References
1. F. Garbassi, M. Morra, E. Occhiello: Polymer Surfaces - From Physics to
Technology, (John Wiley & Sons, Chichester, 1994)
2. J.D. Andrade in: Surface and Interfacial Aspects of Biomedical Polymers, Vol-
ume 1, Surface Chemistry and Physics, J.D. Andrade (Ed) (Plenum Press,
New York, 1985)
3. D. Briggs in: Practical Surface Analysis, Volume 1: Auger and X-ray Photo-
electron Spectroscopy, D. Briggs and M.P. Seah (Eds), 2nd Edn. (John Wiley
& Sons, Chichester, 1990),437-484
4. J.e. Vickerman (Ed), Surface Analysis - The Principal Techniques, (John Wi-
ley & Sons, Chichester, 1997)
5. J.D. Andrade (Ed.), Polymer Surface Dynamics, (Plenum Press, New York,
1988)
6. M. Morra, E. Occhiello, F. Garbassi: Adv. Coll. Interface Sci., 32, 79 (1990)
7. F.J. Holly, M. Refojo: J. Biomed. Mater. Res., 9, 315 (1975)
8. J.H. Chen, E. Ruckenstein: J. Colloid Interface Sci., 135,496 (1990)
9. T.R. Gengenbach, X. Xie, R.C. Chatelier, H.J. Griesser: J. Adhesion Sci. Tech.,
8, 305 (1994)
10. D.T. Clark, in: Photon, Electron and Ion Probes of Polymer Structure and
Properties, D.W. Dwight, T.J. Fabish and H.R. Thomas (Eds.), ACS Sympo-
sium Series, 162, 225 (1981)
11. M.A. Golub, T. Wydeven, R.D. Cormia: Langmuir, 7, 1026 (1991)
12. D.S. Everhart, C.N. Reilley: Surf. Interface Anal., 3, 126 (1981)
13. G. Marietta, S.M. Catalano, S. Pignataro: Surf. Interface Anal., 16,407 (1990)
14. G. Marletta: Nucl. Instrum. Methods Phys. Res., B46, 295 (1990)
15. D. Briggs, "Characterization of Surfaces", in: Comprehensive Polymer Science,
Vol. 1: Polymer Characterization, C. Booth and C. Price (Eds.), (Pergamon
Press, 1989)
16. G. Beamson, D. Briggs: High Resolution XPS of Organic Polymers. The Sci-
enta ESCA300 Database, 1st edition (John Wiley & Sons Ltd, (1992).
17. B.J. Tielsch, J.E. Fulghum: Surf. Interface Anal., 21, 621 (1994)
18. B.J. Tyler, D.G. Castner, B.D. Ratner: Surf. Interface Anal., 14, 443 (1989)
19. R.e. Chatelier, H.A.W. Stjohn, T.R. Gengenbach, P. Kingshott, H.J. Griesser:
Surf. Interface Anal., 25, 741 (1997)
20. M.P. Seah, W.A. Dench: Surf. Interface Anal., 1, 2 (1979)
21. A. Jablonski, C.J. Powell: Surf. Interface Anal., 20, 771 ( 1993)
22. J.C. Ashley: J. Electron Spectrosc. Relat. Phenom., 28, 177 (1982)
23. S. Tanuma, C.J. Powell, D.R. Penn: Surf. Interface Anal., 21, 165 1993)
550 H.A.W. StJohn, T.R. Gengenbach, P.G. Hartley, and H.J. Griesser
53. F. Sommer, T. Minh Duc, R. Pirri, G. Meunier, C. Quet: Langmuir, 11, 440
(1995)
54. J.P. Spatz, S. Sheiko, M. Moller: Langmuir, 13, 4699 (1997)
55. M.P. Seah in: D. Briggs and M.P. Seah: Proctical Surface Analysis. Vol. 1:
Auger and X-Ray Photoelectron Spectroscopy. 2nd edition (John Wiley & Sons,
Chichester, England, 1990) Appendix 2
56. A.E. Hughes, B.A. Sexton: J. Electron Spectrosc. Relat. Phenom., 50, C15
(1990)
57. F. Garbassi, M. Morra, E. Occhiello: Polymer Surfaces. From Physics to Tech-
nology. 1st edition (John Wiley & Sons, Chichester, England, 1994)
58. N.M. Reed, J.C. Vickerman in: L. Sabbatini and P.G. Zambonin: Surface
CharocteTization of Advanced Polymers. 1st edition (VCH Verlagsgesellschaft
GmbH, Weinheim, Federal Republic of Germany, 1993) chapter 3
59. T.R. Gengenbach, H.J. Griesser: J. Polym. Sci.: Part A: Polym. Chem., 36,
985 (1998)
24 G low Discharge
Optical Emission Spectrometry
24.1 Instrument
Plasma
chamber Incident
Anode
Sample
GO Source
slits
Rellected
Spectrometer zero order
e.g. a zinc coated steel sheet or a silicon wafer. The sample mounting places
some restrictions on the types of samples that can be analysed. In particular
they must be capable of withstanding evacuation, i.e. they cannot be loose
powders, and ideally they should have a nearly flat face at least as large
as the outside diameter of the anode, this eliminates some complex shapes.
In addition, they must be capable of withstanding some heating of their
surface by the plasma, perhaps 200°C, this eliminates some polymers and
low temperature metals.
An argon flow is maintained in the discharge cavity to stabilise the pres-
sure between 200-1400 Pa (2-12 Torr). An rf power of 10-80 W, typically
30-50 W, is supplied to the sample and a plasma is created within the tubular
anode. A dc voltage can also be used for conductive samples. The anode is in
a fixed position close to the sample surface, forming a gap of 0.1-0.2 mm, so
that no plasma is possible directly between the annular surface of the anode
and the sample surface. The inner diameter of the anode - usually 4 mm, but
2-8 mm are also used - thereby defines the analysis spot size, as the plasma
is restricted to this area. In rf operation, a negative potential, the auto bias
potential or self-bias, is established at the sample surface (on both conductive
and non-conductive samples) due to the difference in size between the two
electrodes.
Argon ions, created in the negative glow of the plasma, are accelerated
towards the sample where they bombard the surface. This bombardment
with argon ions induces an erosion process called sputtering. Atoms from the
sample surface diffuse through the cathode dark space of the plasma into
the negative glow. Here they are excited through collisions with high energy
electrons, metastable argon atoms and ions. Each element emits a charac-
teristic electromagnetic spectrum. This spectrum is recorded by an optical
spectrometer and the concentration distribution of the different elements is
deduced from the intensities of the characteristic spectral lines.
24 Glow Discharge Optical Emission Spectrometry 555
24.2 Theory
24.3 Applications
24.3.2 Coatings
GD-OES is the only technique capable of bulk analysis and of rapid, quan-
titative depth profiling of thin films and commercial coatings [13]. All of the
common industrial zinc-based coatings have now been successfully quantified:
two examples are shown in Fig. 24.2. Note the enrichment of Al and Pb at
the surface and interface of the galvanized steel sample and the uniformity
of the ZnNi coating.
24 Glow Discharge Optical Emission Spectrometry 557
100 100 1
~[
'--- Zn
Fe '<f!. 80
'<f!. 80
<II
<II <II
<II <0
<0 60
S 60 - S
....c:::
ca> \ Pb"OO
C
40 I, C
a>
I 0
=\:
0 u
u
20 Ni
0 o,
0 5 10 15 20 25 0 2 4 6 8
Depth( ~lm) Depth ( ~l m)
(a) (b)
Fig. 24.2. Quantitative GD-OES depth profiles of (a) galvanized steel and (b)
ZnNi coated steel.
1.6
TiN Ti AIO x AIOx+TiCN TiCN
CN
~
1.2
;:j
-...
~
>-
'iii 0.8
c:
CII
.!:
0.4 N
0
0 40 80 120 160 200 240
Time (5)
An area of growing interest for GD- OES is PVD jCVD coatings, especially
hard coatings, including tungsten and titanium carbide and carbonitride lay-
ers on tool steels. An example of a complex PVD coating is given in Fig. 24.3 ,
where for simplicity only the first five layers and some elements are shown.
GD- OES can be used for the measurement of the elemental concentration
profiles of thin films used in the manufacture of integrated circuits (lCs) [14].
The detailed information of element concentration profiles throughout the
layer and at the interface often gives valuable information on the process
558 T. Nelis and R. Payling
D
'IUU
90
80
o
..
~ 70
°
(/)
(/) - - Si
60
.s ~ P
cQ)
50
B
C 40 Si
0
u 30
20
p B
10
0
0 100 200 300 400 500 600 700
Depth (nm)
quality and stability. The homogeneity of the deposition process across the
wafer can be checked by analysing different spots on the wafer. The instru-
ment can be equipped with an automatic wafer handling system for large
wafers.
One typical application of rf GD-OES in the 'memory' production process
is the characterisation of BPSG (boron phosphorus silica glass) films, as these
films cannot easily be analysed by any other analytical technique already used
in IC manufacturing.
The boron and phosphorous concentrations can be determined precisely
as a function of film thickness (see Fig. 24.4). Figure 24.4 shows a simulated
depth profile because the original data is confidential. As only atomic species
are measured, the analysis is independent of the molecular structure of the
BPSG film and in particular of residual water. In addition to the control of
the concentration profiles of the major dopants, diffusion processes between
the silicon core and the BPSG film can be investigated. Effects at the layer
surface and particularly at the interface can be studied. One example of the
capabilities of GD-OES, though of only minor interest to IC manufacturing,
is the control of alkali elements, such as sodium, at the interface.
To optimise deposition processes and other treatments, elements such as
hydrogen, carbon, chlorine, fluorine, etc. can be measured. A few minutes
are usually enough to obtain an analytical result, and the optimal process
parameters can be quickly established.
A different application of rf GD-OES in semiconductors is the control
of ion implantation concentration profiles. The detection limits for most ele-
ments are in the ppm range or 10 17 atoms cm -3. In this case the advantage
of GD-OES compared to SIMS is the speed and cost of analysis rather than
the detection limit of some elements.
24 Glow Discharge Optical Emission Spectrometry 559
References
Appendix
Acronyms Used
in Surface and Thin Film Analysis
SC Surface Capacitance
SCM/EFM Scanning Capacitive Microscope/y
SDMM Scanning Desorption Molecular Microscope
SE SpectroEllipsometry
SEE Secondary Electron Emission
SEM Scanning Electron Microscopy (Chap. 3)
SEMPA Scanning Electron Microscopy with Polarization Analysis
SERS Surface Enhanced Raman Spectroscopy
SEXAFS Surface Extended X-ray Absorption Fine Structure (Chap. 1)
SFM Scanning Force Microscope/y
SHG Second Harmonic Generation (optical)
SI Surface Ionization
SlIMS Secondary Ion Imaging Mass Spectrometry (Chap. 5)
SIMS Secondary Ion Mass Spectrometry (Chap. 5)
SKP Scanning Kelvin Probe
SLEEP Scanning Low Energy Electron Probe
SNIFTIRS Subtractively Normalized Interfacial FTIR Spectroscopy
SNMS Secondary Neutrals Mass Spectrometry (equivalent to INA)
SNOM Scanning Near-field Optical Microscope/y
SP Spin Polarized (technique prefix)
SP Surface Potential
SPIES Surface Penning Ionization Electron Spectroscopy
SPIPE Spin Polarized Inverse PhotoEmission
SPM Scanned Probe Microscope/y
SPPD Spin Polarized Photoelectron Diffraction
Spy Surface Photovoltage
SRPS Synchrotron Radiation Photoelectron Spectroscopy
SRS Surface Reflectance Spectroscopy
SSA Spherical Sector Analyser
SSIMS Static Secondary Ion Mass Spectrometry (Chap. 5)
STEM Scanning Transmission Electron Microscopy (Chap. 3)
SThM Scanning Thermal Microscope/y
STM Scanning Tunnelling Microscopy (Chap. 10)
STS Scanning Tunnelling Spectroscopy (Chap. 10)
STS Surface Tunnelling Spectroscopy
SXAPS Soft X-ray Appearance Potential Spectroscopy
SXES Soft X-ray Emission Spectroscopy
TD Thermal Desorption
TDMS Thermal Desorption Mass Spectrometry
TDS Thermal Desorption Spectrometry (equivalent to TPD)
TE Thermionic Emission
TEAM Thermal Energy Atomic and Molecular beam scattering
TEAS Thermal Energy Atom Scattering
TED Transmission Electron Diffraction
568 Acronynms
Blakely, J.M. (ed.): Surface Physics of Crystal Solids (Academic, New York
1975)
Boehm, H.P., H. Knozinger: Nature and Estimation of Functional Groups on
Solid Surfaces, in Catalysis - Science f3 Technology, Vol. 4 (Springer, Berlin,
Heidelberg 1983)
de Boer, J.H.: The Dynamical Character of Adsorption (2nd edn.) (Oxford
University Press, London 1968)
Briggs, D., M.P. Seah (eds.): Practical Surface Analysis by Auger and X-ray
Photoelectron Spectroscopy (Wiley, Chichester 1983)
Briggs, D. (ed.): Handbook of X-ray and Ultraviolet Photoelectron Spectro-
scopy (Heyden, London 1977)
Brown, A., J.C. Vickerman: Static SIMS for Applied Surface Analysis, Surf.
Interface Anal. 6, 1 (1984)
Brummer, D., O. Heydenreich, K.H. Krebs, H.G. Schneider (eds.): Hand-
buch Festkorperanalyse mit Elektronen, Ionen und Rontgenstrahlen (Vieweg,
Braunschweig 1980)
Brundle, C.R, A.D. Baker (eds.): Electron Spectroscopy: Theory, Techniques
and Applications, Vols.1-4 (Academic, London 1977, 1978, 1979, 1981)
Brundle, C.R: Elucidation of Surface Structure and Bonding by Photoelec-
tron Spectroscopy, Surf. Sci. 48, 99 (1975)
Carbonara, RS., J.R Cuthill (eds.): Surface Analysis Techniques for Metal-
lurgical Applications (American Society for Testing and Materials, 1975)
Cardona, M., L. Ley (eds.): Photoemission in Solids I (General Principles);
Photoemission in Solids II (Case Studies), Topics Appl. Phys. 26 and 27
(Springer, Berlin, Heidelberg 1978, 1979)
Carlson, T.A.: Photoelectron and Auger Spectroscopy (Plenum, New York
1975)
Chu, K.W., J.W. Mayer, M.A. Nicolet: Backscattering Spectrometry (Aca-
demic, New York 1978)
Clarke, L.J.: Surface Crystallography - An Introduction to LEED (Wiley,
New York 1975)
Czanderna, A.W. (ed.): Methods of Surface Analysis, Methods and Phenom-
ena Vol. 1 (Elsevier, Amsterdam 1975)
Davis, L.A., N.C. MacDonald, P.W. Palmberg, G.E. Riach, RE. Weber:
Handbook of Auger Electron Spectroscopy, 2nd edn. (Perkin-Elmer Corp,
Eden Prairie 1979)
Delannay, F. (ed.): Characterization of Heterogoeneous Catalysts, Chern. Ind.
Ser. 15 (Dekker, New York 1984)
Surface Science Bibliography 571
Oechsner, H. (ed.): Thin Film and Depth Profile Analysis, Topics Curf. Phys.
Vol. 37 (Springer, Berlin, Heidelberg 1984)
Pendry, J.B.: LEED - The Theory and its Application to Determination of
Surface Structures (Academic, London 1974)
Prutton, M.: Surface Physics (Clarendon, Oxford 1983)
Rhodin, T.N., G. Ertl (eds.): The Nature of the Surface Chemical Bond
(North-Holland, Amsterdam 1979)
Roberts, M.W., C.S. McKee: Chemistry of Metal-Gas Interface (Clarendon,
Oxford 1978)
Roberts, M.W.: Photoelectron Spectroscopy and Surface Chemistry, Adv.
Catal. 29, 55 (1980)
Sickafus, E.N., H.P. Bonzel: Surface Analysis by Low Energy Electron Diffrac-
tion and Auger Electron Spectroscopy (Academic, New York 1971)
Somorjai, G.A. (ed.): The Structure and Chemistry of Solid Surfaces (Wiley,
New York 1969)
Somorjai, G.A.: Principles of Surface Chemistry (Prentice-Hall, New Jersey
1972)
Somorjai, G.A.: Chemistry in Two Dimensions: Surfaces (Cornell University
Press, Ithaca 1981)
Spicer, W.E., KY. Yu, I. Lindau, P. Pianetta, D.M. Collins: Ultraviolet Pho-
toemission Spectroscopy of Surfaces and Surface Sorption, Surf. Def. Prop.
Solids 5, 103 (1976)
Steele, W.A.: The Interaction of Gases with Solid Surfaces (Pergamon, Ox-
ford 1974)
Stohr, J.: Surface Crystallography by SEXAFS and NEXAFS, in R Vanselow,
R Howe (eds.): Chemistry and Physics of Solid Surfaces V, Springer Ser.
Chern. Phys. Vol. 35 (Springer, Berlin, Heidelberg 1984)
Teo, B.K, D.C. Joy (eds.): EXAFS Spectroscopy - Techniques and Applica-
tions (Plenum, New York 1981)
Teo, B.K: EXAFS - Basic Principles and Data Analysis (Springer, Berlin,
Heidelberg 1986)
Thomas, J .M., RM. Lambert (eds.): Characterization of Catalysts (Wiley,
New York 1980)
Thompson, M., M.D. Baker, A. Christie, T. Tyson: Auger Electron Spec-
troscopy (Wiley, New York 1985)
Tompkins, F.C.: Chemisorption of Gases on Metals (Academic, London 1978)
574 Surface Science Bibliography
- photoelectron spectroscopy (XPS), 49, 53, 56, 60, 63, 64, 168, 169, 405,
175 412-414, 442, 505, 511
- source, 179 of polymers, 529
X-ray absorption fine structure
spectroscopy (XAFS) 347-371 Young's modulus 272, 276
- Near edge (NEXAFS), 361
X-ray absorption near edge structure
(XANES) 22, 347, 348, 361 Zeolites 420-425
X-ray diffraction (XRD) 450, 451 Zinc
X-ray fluorescence (XRF) 155 - containing glass, 395
X-ray photoelectron spectroscopy - ZnO, 44, 50, 53
(XPS) 24, 26, 28-39, 41, 44, 46, ZSM-5 421-425
SPRINGER SERIES IN SURFACE SCIENCES
Editors: G. Ertl, H.Liith and D.L. Mills Founding Editor: H.K.Y. Latsch