CasaXPS User’s Manual

for Version 2.0, October 2001

Product design: Casa Software Ltd.

Documentation: Acolyte Science.

Manual version 1.2

CasaXPS and the User’s Manual are copyright © 2001 Casa Software Ltd.

All rights reserved.

Information in this document is subject to change without notice and does not
represent a commitment on the part of the copyright holder. The software
described in this document is furnished under a license agreement. No part
of this publication may be copied, reproduced, stored in a retrieval system,
or transmitted in any form or by any means, electronic, mechanical, photo
reproduction, recording or otherwise without the prior written consent of
Casa Software Ltd.
 Acknowledgements

CasaXPS wishes to thank all those connected with the

production of this manual and the software system it describes.
In particular, thanks are due to those who provided experimental
data, results, and interpretations which form the bulk of the
examples here. Notable among these contributors are: Morgan
Alexander, Graham Beamson, David Briggs, Peter Cumpson,
Kevin Harrison, Len Hazel, Fran Jones, Simon Page, Roy
Paynter and Martin Seah

All third party Trade Marks are acknowledged and used without
prejudice.
Any references to company names in sample output and tem-
plates are for demonstration purposes only and are not intended
to imply any endorsement, or refer to any particular existing or-
ganisation or equipment.

CasaXPS and the User’s Manual are copyright © 2001 Casa Software Ltd.
 Contents

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Experimental Variable . . . . . . . . . . . . . . . . . . . 22

Getting Started . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Transmission Correction and Quantification . . . . 22
Installing CasaXPS . . . . . . . . . . . . . . . . . . . . . . 6 Transmission Functions . . . . . . . . . . . . . . . . . . 23
Starting CasaXPS . . . . . . . . . . . . . . . . . . . . . . . 6 Quantification . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Data Display and Browser . . . . . . . . . . . . . . . . . . . . 25
A Quick Tour of CasaXPS . . . . . . . . . . . . . . . . . . 7 Understanding the Data . . . . . . . . . . . . . . . . . . . . 25
Step 1: Load the experiment file . . . . . . . . . . . . 8 Selecting the ISO 14976 Blocks . . . . . . . . . . . . 26
Step 2: Select a spectrum for processing . . . . . . 8 Zooming the Data . . . . . . . . . . . . . . . . . . . . . . . 26
Step 3: Identify peaks. . . . . . . . . . . . . . . . . . . . . 9 Tiles of Spectra . . . . . . . . . . . . . . . . . . . . . . . . . 26
Step 4: Annotate peaks using element markers 11 Tile Display . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Step 5: Create quantification regions . . . . . . . 11 Colours . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Step 6: Adjust region start and end points . . . . 12 Fonts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Step 7: Print the results . . . . . . . . . . . . . . . . . . 14 Display Parameters and Scrolled Tiles . . . . . . . . 29
Data Display and Browser Windows . . . . . . . . . 15
Graph Annotation . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Element Library . . . . . . . . . . . . . . . . . . . . . . . . . 16
Quantification Tables . . . . . . . . . . . . . . . . . . . . . 31
Graph Annotation . . . . . . . . . . . . . . . . . . . . . . . . 17
Moving Annotation and the History Mechanism 32
Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Peak Labels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Quantification . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Text Annotation . . . . . . . . . . . . . . . . . . . . . . . . . . 33
ISO 14976 (“VAMAS”) Files . . . . . . . . . . . . . . 17
Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
ISO 14976 File Format . . . . . . . . . . . . . . . . . . . . . . 19
Smoothing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Partially Encoded Format Versions . . . . . . . . . . 19
Differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
File Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Experiment header . . . . . . . . . . . . . . . . . . . . . . . 21
Energy Calibration . . . . . . . . . . . . . . . . . . . . . . . . 35
Data Blocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Intensity Calibration . . . . . . . . . . . . . . . . . . . . . . 36
Binding vs Kinetic Energy . . . . . . . . . . . . . . . 22
 Contents

Processing History . . . . . . . . . . . . . . . . . . . . . . . 36 Asymmetric Line-Shapes . . . . . . . . . . . . . . . . . 61

Element Library . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Modifications to the Doniach Sunjic function . 62
Library File Structure . . . . . . . . . . . . . . . . . . . . . 38 A New Line-Shape . . . . . . . . . . . . . . . . . . . . . . 64
Loading an Element Library . . . . . . . . . . . . . . . . 40 Background Subtraction . . . . . . . . . . . . . . . . . . . 65
Identifying Peaks . . . . . . . . . . . . . . . . . . . . . . . . 41 Adjusting the universal cross-section . . . . . . . . 68
Simulating Spectra using Tougaard methods . . 69
Quantification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Using the Calculator and Comparing Spectra . . . 71
Synthetic Components . . . . . . . . . . . . . . . . . . . 44 Creating a Comparison File . . . . . . . . . . . . . . . . . 71
Quantification Calculation . . . . . . . . . . . . . . . . 45 The Calculator . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Quantification Report . . . . . . . . . . . . . . . . . . . 45 Principal Component Analysis . . . . . . . . . . . . . . . . 80
Optimisation of the Peak Parameters . . . . . . . 47 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Propagating Quantification . . . . . . . . . . . . . . . 49 Theory of Principal Component Analysis . . . . . . 80
Data Editor . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Residual Standard Deviation (“Real Error”) . . 82
Derivatives and Peak Envelopes . . . . . . . . . . . 50 Chi square . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Quantification by Example . . . . . . . . . . . . . . . . . 51 Target Factor Analysis . . . . . . . . . . . . . . . . . . . . . 83
Report Files and Excel . . . . . . . . . . . . . . . . . . . . 54 Target Testing . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Line Shapes and Backgrounds . . . . . . . . . . . . . . . . 55 Principal Component Analysis by Example . . . 84
A List of Line-Shapes . . . . . . . . . . . . . . . . . . . . 57 Principal Component Analysis and Real Data . 90
Gaussian/Lorentzian Product Form . . . . . . . . . 57 PCA and CasaXPS . . . . . . . . . . . . . . . . . . . . . . 94
Gaussian/Lorentzian Sum Form . . . . . . . . . . . 57 Viewing the Data in Factor Space . . . . . . . . . . 96
Alternative Asymmetric Line-Shapes . . . . . . . 58 Monte Carlo Methods
Line-Shapes Based upon Backgrounds . . . . . . 59 - Uncertainties in Intensity Calculations . . . . . . . . 97
Line-Shapes Available in CasaXPS . . . . . . . . . . 59 Monte Carlo Data Sets . . . . . . . . . . . . . . . . . . . . . 98
Further adjustments to the basic shapes . . . . . 60 Error Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
 Contents

Monte Carlo: A Simple Example . . . . . . . . . . . . 99 SSI M-Probe Files . . . . . . . . . . . . . . . . . . . . . . . 129

Quantification . . . . . . . . . . . . . . . . . . . . . . . . . . 102 PHI MultiPak ASCII files . . . . . . . . . . . . . . . . . 130
Monte Carlo End Point Determination . . . . . 102 Command Summary . . . . . . . . . . . . . . . . . . . . . . . 131
Integration Region Limits . . . . . . . . . . . . . . . 102 Main menubar . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Intensities Determined by Peak-Fitting . . . . . . 104 File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 View . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Analytical Applications . . . . . . . . . . . . . . . . . . . . 107 Window . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Organic Polymers and Curve Fitting . . . . . . . . 107 Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
Quantification using “Tagged” Regions . . . . . . 111 Help . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 111 Toolbar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
“Tagging” regions in a survey scan . . . . . . . . 112 Options Bar . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
“Tagging” for different line shapes . . . . . . . . 113 Processing Dialog windows . . . . . . . . . . . . . . . . 136
Trend Analysis for Metal Oxidation . . . . . . . . 114 Page Layout . . . . . . . . . . . . . . . . . . . . . . . . . . 136
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . 114 Tile Display Parameters . . . . . . . . . . . . . . . . . 136
PCA for Aluminium Oxidation Sequence . . 116 Quantification . . . . . . . . . . . . . . . . . . . . . . . . . 138
Synthetic Model for an Oxidation Sequence . 117 Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Peak Fit Using Doniach Sunjic Line Shapes . 118 Annotation . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Adjusting a Custom Quantification Report . . . 120 Library . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Using Different File Formats . . . . . . . . . . . . . . . . 123 Appendix 1: ISO 14976 format files annotated . . . 143
Kratos DS800 Binary Files . . . . . . . . . . . . . . . 123 Appendix 2: ISO 14976 and the World Wide Web 148
VG Eclipse Files . . . . . . . . . . . . . . . . . . . . . . . . 124 Appendix 3: Definitions and Formulae. . . . . . . . . . 149
VGX900 (Ron Unwin) . . . . . . . . . . . . . . . . . . . 125 Glossary of terms . . . . . . . . . . . . . . . . . . . . . . 149
“Dayta” System Files (Bristol IAC system) . . . 127 Formulae: Peak shapes . . . . . . . . . . . . . . . . . . 149
Kratos Vision 1.x/2.x ASCII files . . . . . . . . . . 128 Gaussian/Lorentzian Product Form . . . . . . . . 149
Gaussian/Lorentzian Sum Form . . . . . . . . . . . 149
 Contents

Doniach Sunjic: . . . . . . . . . . . . . . . . . . . . . . . 150 References from the Text Footnotes: . . . . . . . 151

Appendix 4: References and other Resources . . . . . 150 Appendix 5: Quick Reference Card . . . . . . . . . . . . 151
World wide web links: . . . . . . . . . . . . . . . . . . 150
Introduction CHAPTER 1

Computer Aided Surface Analysis for X-ray Photoelec- and use as data by other popular Microsoft packages such as
tron Spectroscopy (CasaXPS) has been designed and writ- “Word” or “Excel”. Additional spectrum input filters for
ten for analysts and research scientists who use XPS for other commercial file formats are also available, supple-
surface characterisation. menting the ISO standard and broadening its usefulness.
In contrast to the special purpose control systems that form
part of commercial XPS instruments, CasaXPS offers a Casa XPS has been written entirely in native C++ using the
compact, portable, efficient and user-friendly processing Microsoft Developers Studio programming environment. It
system to anyone with an IBM compatible (Pentium) PC employs Microsoft Foundation Class (MFC) libraries to
running Microsoft Windows 95 (or later) or suitable emula- provide a standard graphical user interface as well as much
tor. It incorporates much of (in many cases more than) the of the data management required by the system.
processing functionality of the instrument linked packages The software suite as a whole may be regarded as being
without recourse to unfamiliar operating systems or hard- structured from five interacting components:
ware or proprietary file formats. It is designed from the out- • Data display and browser (an “experiment window”)
set on the basis of the ISO 14976 Surface Chemical Analysis
• Element library (a “dialog window” as are the others)
Standard Data Transfer Format and so by design has a uni-
versal, “cross platform” approach, independent of anything • Graph annotation
other than a reasonable adherence to the ISO standard. Spec- • Data processing
tra collected in the standard format may be selected, viewed • Quantification.
and processed in a simple yet powerful way and the results This manual provides a detailed description of the system,
of CasaXPS data reduction, presented in a variety of graph- its operation, command and function set, and its use in prac-
ical and tabular formats, are available for incorporation into tice in “real life” analytical applications.

5
 Introduction

ISO 14976 data files in an easily accessible directory (or

Getting Started have the demo data on the installation CD available) Ensure
also that you have an appropriate element library available
(one needs to be installed before the first use of “Library”
Installing CasaXPS system - see “Element Library” on page 38 - a simple “ge-
You must install the application from the CasaXPS CD onto neric” version is available on the installation disk, but check
your hard disk. “Drag and Drop” installation is all that is re- with your instrument operator which is the most appropriate
quired for Windows 98. Instructions are available, if re- file for the data acquired by that particular instrument). At
quired, in the “Read Me” file on the installation CD. some point you may wish to install and use the on-line
Installation should occupy no more than 5MB on your hard “help” files (see note on your CD inlay) and you should have
drive. You will need perhaps an equivalent space for storage a “.def” directory prepared if you wish to store synthetic
of data files. Make sure that you have your serial number to components and other items for future use
hand when you install the application, or else the package
will revert to “Demonstration Mode” and will not allow you
to save the results of your work. The serial number is avail-
able from your registration documents (which may be e-
mailed to you or included with the CD). You can “validate” Terminology
a registration at any time after installation, turning a “demo’
version into a fully licensed product, by using the “About” CasaXPS component windows appear on screen as shown in
button in the Toolbar. Figure 1, where the annotation defines their names as used
If you are using an emulator package on a processor other throughout this manual.
than an Intel Pentium, ensure that you obey the emulator’s It is assumed throughout that the user is familiar with the
instructions for installation of applications packages, that “normal” IBM PC and Microsoft Windows operation and
you understand the limitations (if any - e.g. mouse button terminology.
availability) of the system, and that you have sufficient
space and processing power for the convenient use of The following section should be used as a “hands on” intro-
CasaXPS. duction to CasaXPS to provide an overview of key features.
Subsequent sections enlarge briefly on the role of the system
Starting CasaXPS modules, and the remaining chapters deal in depth with spe-
Simply click on the desktop short-cut (if, as recommended, cific topics and applications.
you installed one) or double click on the programme icon in
“Windows Explorer”. Ensure that you have stored your

Getting Started 6
 Introduction

(alternative names in brackets)

CasaXPS (Program Frame)

Toolbar

Options bar
Experiment
Frames

Display
Window

a Dialog
Window

Block
Window

(Browser)
Splitter bar

Figure 1: A Windows 98™ screen showing CasaXPS in operation

quantify a survey spectrum in CasaXPS. The intention is to

A Quick Tour of CasaXPS explain how the interacting components of the system aid
the analyst rapidly to produce a quantification table suitable
This section describes one of the many possible ways to for inclusion in a customer’s report.

A Quick Tour of CasaXPS 7

 Introduction

Step 1: Load the experiment file Step 2: Select a spectrum for processing
CasaXPS is designed for ISO-14976 formatted XPS spectra. Once the data file has been selected via the Open VAMAS
A data file containing spectra stored in the ISO format may file dialog window, the spectra held within the file will be
be selected via the Open VAMAS file dialog window shown displayed in a logical array where each data region is la-
in Figure 2. To select a VAMAS file press the toolbar button belled by the VAMAS block identifier string and these block
indicated in Figure 3 to invoke the Open VAMAS file dialog labels are arrange according to the element/transition strings
window recorded in the VAMAS file. Figure 4 shows the state of the
CasaXPS windows after a spectral data file has been loaded.
.

Figure 2: Open VAMAS file Dialog Window

Figure 4: CasaXPS frame windows after a file has been

loaded

Figure 3: Toolbar. The Open file toolbar button is circled. Note the Multiple Document Interface MDI architecture.
When opened, the VAMAS file appears as sub-frame win-
dow, which is managed by the CasaXPS main frame and it

A Quick Tour of CasaXPS 8

 Introduction

in turn displays the file as a logical set of spectral blocks.

These logical blocks labelled by name allow spectral regions
to be selected and displayed in the left-hand-side of the sub-
frame (in the Display window). The current selection is that
set of labels highlighted blue (default colour scheme), and
when the file is first opened the top row of spectra regions
will be highlighted and displayed in the left-hand sub-frame.
If more than one spectrum appears on the first row of the
Browser, then the spectra will be displayed in the Display
using a scrolled list of display tiles.
The left-hand mouse button and the cursor are used to make
a new selection. Point the mouse at a block label within the
array of labels on the right-hand-side and left-click. Any
blocks currently selected will be deselected and the block
under the mouse will become the current selection. Press the
Figure 6: CasaXPS ready to identify spectral features.
toolbar button indicated in Figure 5 to display the selection
in the scrolled list on the left-hand-side of the sub-frame.
Step 3: Identify peaks.
CasaXPS will load an element library called CasaXPS.lib
located in the same directory as CasaXPS.exe. The element
Figure 5: Toolbar button to show spectra in the Display library is used to identify the peaks seen in the data and is
window (left hand sub-frame). central to rapidly creating quantification regions for a survey
spectrum.
In this example, the spectrum of interest is the last in the se- The Element Library dialog window is invoked or brought
quence of survey spectra recorded in the data file. Figure 6 to the front of any CasaXPS dialog windows by pressing the
shows the state of the CasaXPS windows after a specific sur- toolbar button indicated in Figure 6. The first property page
vey spectrum has been selected and displayed. Also note in of the element library dialog window is a scrolled list of line
Figure 6 that the user has maximized the Experiment Frame positions where the list is ordered by the energy of the en-
within the main frame of CasaXPS, hiding others below. tries. If the mouse is left-clicked on the spectrum displayed

A Quick Tour of CasaXPS 9

 Introduction

in the left-hand-side of the sub-frame shown in Figure 6, again using the mouse, a set of markers will appear on the
then the scrolled list of element library entries will scroll to spectrum positioned at the energies at which spectral lines
the energy indicated by the mouse click. Therefore, by should appear for the chosen element. Figure 7 shows the
pointing at the peaks in the spectrum via the mouse, the spectrum after the O 1s and C 1s lines have been activated
names for the various features can be brought into view and via the element library scrolled list.
if the name field displayed in the scrolled list is selected,

Figure 7: Peak identification via the element library.

A Quick Tour of CasaXPS 10

 Introduction

If the name field of an element library entry is selected a

second time, the element markers are removed from the dis-
play. The current set of active element markers can be view
on the Periodic Table property page also located on the ele-
ment library dialog window. The element markers may be
activated and deactivated via the Periodic Table as well as
the scrolled element list.

Step 4: Annotate peaks using element markers

Once all the spectral lines are identified and markers have
been placed on the displayed spectrum, the peaks can be an-
notated using the names stored in the element library. This
step not only creates a labelled spectrum, but these annota-
tion peak labels are used to create a set of quantification re-
gions at the press of a button.
The annotation dialog window includes a property page en-
titled Peak Labels (Figure 8). All the peak markers active on
a spectrum are listed on the Peak Labels property page and
may be selected using the mouse and the control key (to add
to the selection). When the Apply button is pressed those la-
bels selected in the list of names on the Peak Labels property
page become annotation labels on the spectrum. The peak Figure 8: Peak Labels Property Page on the Annota-
labels are positioned on the spectrum according to the loca- tion Dialog Window
tion of the nearest peak, so it is important that the spectrum
is calibrated before this procedure is performed. tries is offered. Each column in the scrolled list represents
the set of parameters that define a quantification region.
Step 5: Create quantification regions The annotation for the peaks is used by the “Create from La-
The Quantification Parameters dialog window is available bels” button indicated at the bottom of the Quantification
from the top toolbar button indicated in Figure 9. Integration Parameters dialog window in Figure 9. Regions created in
regions are created and adjusted via the Quantification Pa- this way have the name and RSF set from the entries in the
rameters dialog window, where a scrolled list of region en- element library. The background type for each of the newly

A Quick Tour of CasaXPS 11

 Introduction

created regions is determined from the last background type entered on the Regions property page.

Figure 9: Creating Quantification Regions

ing a field within the table using the left-hand mouse button
and the cursor position. If a field can be altered, then select-
Step 6: Adjust region start and end points ing that field with the mouse will both highlight the column
The Regions Property Page is used to manually adjust the of parameters for that region and cause the selected field to
values for the integration regions. The table of parameter become a text-edit field. Any changes are accepted only af-
fields offered in the scrolled list may be adjusted by select- ter the return-key is pressed.

A Quick Tour of CasaXPS 12

 Introduction

Although manual input is achieved through the Regions gration regions on the active spectrum will be marker by
property page, an easy way to check the start and end points vertical bars. The presence of these bars indicates that the
for each integration region is to use the Zoom List toolbar limits for the regions can be adjusted under mouse control.
buttons and adjust the integration limits under mouse con- If the cursor is dragged starting at the position of one of these
trol. Provided the Regions Property Page is the visible page vertical bars, the corresponding region end point will be ad-
on the Quantification Parameters dialog window, the inte- justed when the mouse button is released.

Zoom Reset

Zoom Out button

Drag Box
Marker

Figure 10: Cycle through the integration regions using the Zoom Reset button fol-
lowed by the Zoom Out button.
The integration regions on a survey spectrum are typically zoomed display. To facilitate the use of the mouse, CasaXPS
too narrow to be adjusted in this way without the use of a allows all the current integration regions to be entered onto

A Quick Tour of CasaXPS 13

 Introduction

the Zoom-List and then using the Zoom Out toolbar button the set of zoom states and therefore sequentially display the
(or Ctrl right-click of the mouse) the current set of integra- regions.
tion regions can be zoomed into and adjusted.
Press the Zoom Reset toolbar button indicated in Figure 10. Step 7: Print the results
The spectrum will be displayed showing the full set of data Once the elemental composition has been defined using the
and any integration regions currently defined for the spec- integration regions, the results may be added to the spectrum
trum will be entered onto the Zoom-List. Now pressing the display. This is achieved via the Annotation Dialog window
Zoom-Out toolbar button also shown on Figure 10 will cycle (Figure 11), where the Regions property page allows a table

A Drag Box

Figure 11: Annotating the spectra using the Quantification results.

of atomic concentrations to be positioned on the spectrum. changed are made to the set of integration regions so there is
The entries in the quantification table are updated when no problem about out-of-date quantification results being re-

A Quick Tour of CasaXPS 14

 Introduction

ported over a spectrum. If the Annotation History property

page is the active page on the Annotation dialog window
(Figure 10), then each annotation item will be displayed
with a small position box attached. The annotation may be
repositioned using these boxes. This is achieved by pointing
the cursor at a box and then, with the left mouse button held
down, the cursor is dragged to a new position. The annota-
tion will move once the mouse button is released.

Page Layout toolbar button.

Figure 11 shows the settings

used to achieve the display
layout for the spectra.

Figure 13: “Tiling” a printed page is achieved using

the Page Layout dialog window

Figure 12: Page Layout dialog window. Now that we’ve completed “the tour”, we’ll consider all the
CasaXPS modules in outline in the summaries below and in
much greater detail in the chapters that follow.
The spectrum shown in Figure 11 is now ready to be printed.
Any spectra visible in the left-hand-side of the experiment
frame (in the Display window) will be printed on a single Data Display and Browser Windows
page. If the Display window contains a scrolled list of spec-
tra then each tile within the scrolled list will be printed as a
separate page. If more than one spectrum is required on the Modern surface analysis instruments produce data at a high
same printed page then the Page Layout dialog window of- rate that makes efficient data reduction an important consid-
fers the means of arranging up to sixteen tiles on one page eration when assessing the sample throughput for a labora-
(Figure 12 and Figure 13). tory. The time spent processing the data can become a

Data Display and Browser Windows 15

 Introduction

serious problem when an automated acquisition sequence

has produced hundreds of spectra and spectrum regions all
of which fall into sets with related features. The data reduc-
tion often consists of the same processing/quantification op-
erations tediously repeated for each of the spectra within a
data set. The results of these calculations often have to be
graphed or tabulated to show the variation of concentrations
as a function of some experimental variable (e.g. etch time
or sample tilt).
CasaXPS has been designed based on the ISO 14976 stand-
ard format. The structure imposed by this data format per-
mits acquisitions to be saved with the ancillary information
necessary for understanding the context of the data, and so
processing software now has the opportunity to operate on
the data with the same degree of automation that is available
to the acquisition routines.
The Browser in CasaXPS presents the ISO blocks in a tabu-
lated (matrix) form, in which each row of the table repre- Figure 14: The CasaXPS Program Frame, containing the
sents data acquired with the same experimental variable Toolbars and a single Experiment Frame, which shows
(e.g. ion gun etch time). The columns are spectral regions both display and browser (splitter) windows.
that have the same chemical species label (element symbol
or formula) and transition (detection energy or level nota-
tion).
Element Library
Individual spectra or complete sets of them can be selected
and displayed in a scrolled list, and all processing and quan-
tification operations can be applied to a current selection. An element library is maintained by the CasaXPS system. It
The Browser thus forms the basis for automatic processing allows the user to manage the information that describes the
of the data. X-ray induced transitions, such as binding energy, peak
It is also through Browser operation that quantification re- widths or relative sensitivity factors. These quantities, once
ports are generated again reducing the time to complete the entered, can be accessed through either a periodic table in-
analysis of both simple and complex samples. terface or an energy-ordered scrolled list. Information se-

Element Library 16
 Introduction

lected via these mechanisms can be used to annotate the instrument to be included in the ISO 14976 file as one of the
spectra or create quantification objects such as regions or corresponding variables within a data block. Some manufac-
synthetic peak shapes. turer’s data acquisition systems routinely provide output in
this important form.
Principal Component Analysis and Target Factor Analysis
Graph Annotation are also available as a processing options, as is a Monte Car-
lo method for uncertainty analysis.
Although Casa XPS provides Microsoft enhanced metafiles
for displaying spectra via Word and Excel, XPS specific an-
notation and other important information can be added to the Quantification
display through the Casa XPS options. These include:
• Tabulated results and summaries extracted from the Spectra can be divided into quantification regions which
quantification regions and components. form the basis for estimating the background shape associat-
• Peak labels derived from the element library. ed with the recorded data. A sample may be analysed based
• Basic text overlays. on these user defined energy ranges or fine structure within
Annotation may be positioned on the graphs either with re- a data envelope investigated using synthetic components.
spect to the axes or the data. This allows annotation to be po- Intensities for the elemental composition can be combined
sitioned at a fixed location relative to the axes, independent from regions, components or via an arithmetic expression
of zoom-state, or to follow the data as a peak is expanded to that involves any combination of the region or component
fill the display area. intensities.
Tables generated from the quantification process are saved
to disk in a TAB separated ASCII file. Spreadsheet pro-
Processing grams such as Excel will read these files and provide many
options for printing the data in either tabulated form or as a
Understanding the data envelope - the peak shape for any graphical profile.
transition - can sometimes be assisted by smoothing, differ-
entiating or integrating the data. Techniques for providing
these alternative views of the spectra are made available to ISO 14976 (“VAMAS”) Files
the users of CasaXPS through the processing option. The
spectra can be calibrated with respect to energy and intensi- The ISO Surface Chemical Analysis Standard Data Transfer
ty. The latter requires the transmission characteristics of the Format (ISO 14976) is a prescription for presenting the ac-

Graph Annotation 17
 Introduction

quisition parameters and data for a range of instruments and ty, Surf. Interface Anal. 13, 63 (1988). (The present ISO
techniques. CasaXPS is only concerned with X-ray photo- standard is closely related to the VAMAS original, but not
electron spectroscopy (XPS) and as such whenever the term quite the same). It is not intended to mean here that every
ISO spectrum or “VAMAS file” is used in this manual it is variation of ISO XPS format is supported within CasaXPS -
intended to mean, that subset of ISO 14976 corresponding to there are many allowed variations even though the basic
the fields used by XPS normally stored in ISO format. The skeleton structure is well defined and invariant. While every
“VAMAS” appellation often used loosely and interchange- effort has been made to support a range of XPS formats
ably with ISO 14976, arose because the early work on stand- (VAMAS “flavours” as interpreted by particular systems) it
ard data transmission format was sponsored and encouraged has not been possible as yet to accommodate the entire set of
as part of VAMAS, the Versailles project on Advanced Ma- combinations and permutations that are allowed by the com-
terials And Standards1 from the early 1980’s - see for exam- plete file specification.
ple “VAMAS Surface Chemical Analysis Standard Data
Transfer Format with Skeleton Decoding Programs” W.A. Many of the processing techniques rely on equally spaced
Dench, L.B.Hazell, M.P.Seah, and the VAMAS Communi- abscissa. For this reason, irregular ISO files are not support-
ed. That is to say, although the full set of ISO file formats
1. http://www.vamas.org - TWA-02 is the Surface Chemical can be read by CasaXPS (excluding the annotated form),
Analysis work area with secretariat in the USA only a subset can be displayed and manipulated.

ISO 14976 (“VAMAS”) Files 18

ISO 14976 File Format CHAPTER 2

The ISO 14976 Surface Chemical Analysis Standard Data One natural consequence of defining a standard data format
Transfer Format2 was designed to enable uniformly encoded independent of data origin is that the file must contain all the
transmission of data originating from a variety of instru- information required to process (interpret) the data. An
ments, which in turn might employ a variety of techniques, ISO 14976 file is therefore much more than just a conven-
to any appropriate processing system, local or remote, for its ient way of uniformly encoding output from surface analysis
interpretation within specific applications. A desire to unify instrumentation; it is also a blueprint for anyone designing a
descriptions for SCA data and so reduce the number of pro- data acquisition system.
grams necessary to manage the increasing number of propri- The strength of the ISO file format lies in its ability to store
etary formats, led to a specification of how data files should all the information needed to process the data at a later time.
This is also a weakness since its completeness makes the de-
be ordered (formatted) and encoded. This work, identified
tailed description formidable to the uninitiated. Neverthe-
and steered by a Technical Working Party of the VAMAS
less, in the fullness of time, anyone designing a file format
project (see “ISO 14976 (“VAMAS”) Files” on page 17) in for XPS data will surely move towards a structure that es-
the early 1980’s, was taken up by the International Stand- sentially holds the same information as ISO 14976; albeit
ards Organisation (ISO - Technical Committee TC 201, perhaps in a different order or layout.
Sub-committee 3) and an internationally agreed standard
was produced.
Partially Encoded Format Versions
2. ISO 14976:1998 - Surface Chemical Analysis - Data
Transfer Format (ISO/TC 201) ISO Marketing Services,
Case Postale 56, CH-1211, Geneva 20, Switzerland. ISO 14976 is based on the earlier VAMAS format but with
http://www.ISO.org three major changes:-

19
 ISO 14976 File Format

• the spectrometer geometry description is now referenced point on the sample”. This partially encoded format is suita-
to a right-handed rather than left-handed co-ordinate sys- ble for experiments such as XPS depth profiles (either using
tem, angle resolved methods or ion bombardment etch sequenc-
• the number of entries in the “parameter inclusion or es) or surface chemical analysis over time with or without
exclusion list” has been set to zero in order to simplify temperature variations (see “Appendix 1: ISO 14976 format
the format, files annotated” on page 143).
• and the line separator sequence has been changed from
the single 7-bit ASCII character “carriage return” to the
more commonly used two character sequence of “car- File Structure
riage return” followed by “line feed”
The ISO format thus may be a sub-set of the VAMAS defi-
nition. What is defined by ISO 14976 is an experiment. This con-
Although both the VAMAS and ISO formats are very gen- sists of a series of parameters which apply to the measure-
eral, CasaXPS is only concerned with files generated from ment procedure as a whole, followed by a number of blocks
X-ray Photoelectron Spectroscopy (XPS) data. That is to of data (often called data blocks), followed by an experiment
say, while CasaXPS will read files in the format defined by terminator. Each block consists of a set of parameters that
ISO 14976, display and processing within the system are only apply to that block, followed by a series of ordinate val-
strictly only available for a subset of the Format. This should ues which may represent a curve (e.g. a depth profile), a
pose no significant difficulty for virtually all practical situa- spectrum, or a map. The experiment header contains the
tions. context for the acquisition sequences (blocks) that follow,
while the information held in the blocks is specific to a par-
Partially encoded versions of the Format - essentially “tem-
ticular determination (e.g. a spectrum region) recorded in
plates” for the most common (and most useful) instances -
the course of an experiment (Macintosh computer users may
were developed as annexes to the VAMAS definition and
see some resemblance to the “resource fork/data fork” file
other early reports: e.g. Appendix 2 of the National Physical
structures of MacOS here).
Laboratory Report DMA(A) 164 (July 1988). This is main-
tained in the full ISO standard (Annex C) The first of the An experiment described in this way is suitable for both
partially encoded formats (C2) is particularly appropriate to complex and simple data sets. The header and block struc-
CasaXPS. It describes the fields for defining “an experiment ture may seem perhaps over-elaborate for a single spectrum
involving a number of regular-scanned spectra or spectral experiment yet even the most complex profiling analyses are
regions for one technique as a function of one experimental well supported by the same organisation in the data; the only
variable, the analysis not being at a specifically-addressed difference lies in the number of blocks within each file.

File Structure 20
 ISO 14976 File Format

signal recording parameters

Experiment header sputtering
sample orientation
The parameters which apply to the experiment as a whole additional parameters and future-upgrade parameters
occupy the first section of a file written in the ISO format. (followed by the corresponding data values)
They fall into the following groups in the order in which Each block thus provides the acquisition parameters togeth-
they appear: er with a set of “corresponding variables”. The correspond-
identity of the experiment in its (analytical) environment ing variables may contain more than just recorded counts.
optional comments Additional corresponding variables can include values such
experiment mode as transmission function (correction factors) for adjusting
number of blocks and how they are arranged the recorded counts with respect to a reference spectrum.
pointers to manually-entered parameters Use of a second corresponding variable is the method pre-
future-upgrade parameters ferred by at least one commercial manufacturer to record
transmission function data in a “VAMAS” format. The
transmission characteristics for their instruments accompa-
The layout and interpretation of these parameters is straight-
ny the data point by point within each block.
forward.
Although blocks contain a comprehensive set of parameters,
it is permitted to replace many by a value that indicates no
Data Blocks information is present (“1E37”(1037) is the defined value).
Use of this feature can reduce the ability of a target system
to process the data, but it can help to limit the complexity of
The parameters which apply to a block (the “business end” writing a valid ISO file for an acquisition system.
of the format) fall into the following groups, in the order in
Comment lines may be included in each block. These are
which they appear:
useful for summarising technique specific information asso-
identity of the block in the experiment ciated with a particular instrument. For example, names cor-
identity of the sample responding to lens modes (magnification) could be entered
date and time as a comment line. Operational parameters could accompa-
optional comments ny such text, where the state of the charge neutraliser or any
technique other relevant descriptive information may be added.
analysis CasaXPS provides an option on the “Tile Display” dialog
analyser window for the display of block comments as a header for

Experiment header 21
 ISO 14976 File Format

the graphical display. Suitably formatted comment lines CasaXPS rely on these values to organise the blocks so that
may be used to annotate the display with, e.g. acquisition pa- profiles are meaningful
rameters for a spectrum.

Binding vs Kinetic Energy Transmission Correction and Quantification

An important piece of information that is not explicitly stat-
ed in the ISO format is the nature of the energy scan.
Modern XPS instruments are relatively complex devices,
CasaXPS needs to know whether the energy is in binding or
characterised by numerous operating modes and methods.
kinetic energy. The ability of the display to switch between
An ability to control and configure analyser input lens oper-
binding and kinetic energy is dependent on the “analysis
ation enables use of a range of energy resolutions, spatial
source characteristic energy” containing the correct value
resolutions, and magnifications, but for each operating
and the correct assignment for the energy type.
mode of an instrument, transmission characteristics will dif-
A REGULAR scan is defined by “abscissa start” and “ab- fer and also perhaps vary in a complex fashion within any
scissa increment” energy values. To differentiate between particular scanning mode, causing energy-dependent differ-
the two types of energy the associated “abscissa label” field ences to occur between spectra taken from the same sample
is used by CasaXPS as an indicator of which type has been but using different operating conditions. These differences
used to describe the data. If the label includes one of the sub- in data translate into quantification results that depend on the
strings “B.E.”, “BE”, “Binding” or “binding”, then the ener- instrumental settings as well as the sample itself and so an
gy type is assumed to be binding energy. Otherwise the en- intensity calibration procedure is required if consistent re-
ergy type defaults to kinetic energy. sults are expected from the same instrument operating with-
in its range of appropriate settings.
Experimental Variable
The most common method for comparing peak intensities is
The numerical value for the experimental variable is record- to use tabulated Relative Sensitivity Factors (RSF) to adjust
ed in each block. Associated information for the value en- the calculated values for physical effects. Without intensity
tered in the block is specified in the experiment header calibration procedures based upon transmission characteris-
section of the file, where a label and the units for the variable tics these RSF values are only appropriate for a specific op-
are found. erating mode. It is therefore essential to map out the
A proper use of the experimental variable is essential when transmission functions that accompany the lens modes and
multiple blocks are recorded in a file with the intention of pass energies before consistent results can be obtained from
quantifying trends in the data. Reports generated by a particular sample.

Transmission Correction and Quantification 22

 ISO 14976 File Format

Transmission Functions tion) procedure where the output takes the form of a rational
Transmission functions are dependent on instrumental set- function of two polynomials. The background of inelastical-
tings that can be adjusted by the user and other settings that ly scattered electrons measured from pure gold, silver and
change with age. A useful calibration procedure should copper provides a vehicle for comparing the response func-
therefore be easy to perform and apply. Moreover since tion of an instrument under test with that of a well character-
transmission functions vary with time, the transmission ised “metrology spectrometer”. Given spectra acquired at
function for a particular spectrum should be included as part the various operating modes of the instrument to be calibrat-
of the data file. ed and standard spectra from the reference machine, then
State-of-the-art instruments typically record the measured relative transmission functions are readily constructed and
intensities and extract some functional form for the trans- consistent quantification becomes possible.
mission characteristics; both pieces of information are saved It is clear that there is no absolutely correct way of represent-
to disk. The methods adopted vary: for example, a piece- ing transmission characteristics for a given set of intensities,
wise linear approximation is used by one manufacturer, although the importance of this information should not be
where the nodes may be determined either from peak inten- underestimated. Proper intensity calibration opens up the
sities or from the inelastically scattered background. Anoth-
opportunity for consistent quantification not just across op-
er manufacturer uses a polynomial form for the logarithmic
erating modes of a single instrument but also for instruments
shape of the transmission as a function of retard ratio, where
from the same manufacturer and possibly instruments from
the coefficients of the polynomial are determined using peak
different manufacturers. Analysts reliant on data taken from
area measurements rather than background information. A
a range of instrumentation will be aware of this problem.
third manufacturer again uses peak area measurements to
determine two analyser-dependent constants a and b as fol- Although there is no prescription for transmission functions
lows: within the ISO 14976 file format standard, the mechanisms
2 b are in place that enable recording of more than one corre-
a
T ( R ) = ---------------------- .......................... Eqn. 1 sponding variable per ISO block and therefore the energy
(a + R )
2 2
dependence of the transmission can easily be saved along
with the energy dependence of the signal. ASCII files have
R is the retard ratio(3). come-of-age with the increasing size of hard disks and “zip”
technology. There is no longer a serious concern about the
NPL offers an empirical intensity calibration (standardisa- size of (XPS/AES) spectroscopic files and the benefits of re-
cording open and explicit data outweighs any drawbacks
3. Tanaka A., J. Surf. Analysis, 1 189 (1995) that may accompany ASCII format spectra.

Transmission Correction and Quantification 23

 ISO 14976 File Format

Quantification approaches can be outside the precision estimates for the

Historically transmission correction has been performed via quantitative values. A consistent approach to transmission
adjustments to the RSF values used to quantify the peak in- correction therefore goes hand in hand with proper proce-
tensities. While at first sight this would seem a reasonable dures for monitoring the response of an instrument.
approach, the nature of transmission functions means that Provided a transmission function is included in the data file,
errors are introduced by this procedure. The RSF attempts to CasaXPS offers both methods for adjusting spectra with re-
allow for the set of factors that cause peak intensities to vary spect to the transmission function. The default method is to
for reasons other than sample composition. The influences use a single point in the energy scale to evaluate the trans-
of the transmission function are usually assumed to be con- mission function. This single value is then used to scale the
stant across a peak, but this assumption is often false in the RSF value before applying this value to the calculated peak
case of the relative transmission functions. The response of area. The main reason for offering this as the default mode
the instrument can vary rapidly across a peak width, partic- is that most manufacturers systems perform the quantifica-
ularly when portions of spectra are compared at opposite tion by this means. The results from CasaXPS in the default
ends of the resolution or energy scale. mode for transmission correction therefore will agree with
Empirically determined RSF values necessarily compensate these other systems, although it should be stated that correct-
for instrumental response in the reference mode. Alternative ing spectra point by point prior to calculating the intensity is
modes need to be adjusted relative to the reference mode by probably the better method.
dividing spectra point by point by the transmission function
rather than choosing a representative value for the transmis- An example of the way in which transmission can be encod-
sion (typically evaluated at the peak maximum). The differ- ed within the ISO 14976 file format (and thus used by
ence between intensity values determined by these two CasaXPS) is given in Appendix 1 (see page 143).

Transmission Correction and Quantification 24

Data Display and Browser CHAPTER 3

CasaXPS uses a multiple document interface (MDI) as de- frame, allocating a larger or smaller area to the graphical dis-
fined by Microsoft. That is, it conforms to the same style as play (the spectrum) or the block structure, as appropriate.
programs such as Microsoft Word or Excel and therefore Data held within a VAMAS (ISO) file appears in rows and
presents each file (Document) with a document view. The columns in the Block window, each row corresponds to data
main frame of the program manages a set of document/file acquired under the same experimental variable value while
frames. These file frames may be minimised or maximised, the columns correspond to spectra (regions) resulting from
or arranged using the Window menu allowing the frames to measurements on the same chemical species or transition
be tiled with respect to one another or to create additional (energy range). The mouse is used to select a set of one or
views of the same file. more blocks from the right-hand side (the Browser) and
these data blocks are displayed by selecting one of the two
leftmost toolbar buttons shown in Figure 15
Understanding the Data

The ISO 14976 files used by CasaXPS are viewed through a

“splitter window” called the Experiment Frame. The right-
hand section of the Frame (the Block window) represents the Figure 15: Spectrum Display Options.
logical structure of the experiment; showing the relationship The first icon will display a scrolled list of spectra (or re-
of the spectra to each other and to the experimental variable. gions), one per tile. The second of the two allows (many)
The “splitter” terminolgy is a reminder that the relative sizes spectra to be overlaid in the same tile. In both cases the x and
of the two components of the frame may be adjusted by y scale values are determined by the maximum range values
“dragging’ the central divider - the “splitter bar” in side the for the spectra or regions requested. The corresponding but-

25
 Data Display and Browser

tons on the right of the toolbar provide display and overlay from the current selection.
within the already set scale values. If there are no points in Selecting a block without the use of the “Shift” key or the
the range, then a message indicating “No Data Points in “Ctrl” key clears the current selection and replaces it by the
Window”appears (this facility enables the overlay of (de- indicated block.
tailed) regions onto a wide scan, for example).
Zooming the Data
Selecting the ISO 14976 Blocks Once a spectrum has been entered into the left-hand side
Each ISO file requires at least one Experiment Frame. Ex- scrolled display window the data can be examined under
periment Frames are created using the “New” option on the mouse control. In order to zoom into a range of the data, the
file menu. Once an empty Frame is selected, the “Open” op- mouse is used to mark a rectangular region and then the third
tion from the same menu offers a File Dialog window for button is pressed (Figure 15) The result of this action is to
choosing the ISO file. cause the graph axes to be computed from the defined area
The Browser view (right hand pane of splitter window) of- and the data is redisplayed. Zooming may be performed a
fers the data as described above. Selecting the data is number of times and the system maintains a list of zoom
achieved via the mouse in combination with the “Shift” and states. These zoom states can be revisited using the fourth
“Ctrl” keys. Positioning the cursor over a name in the brows- toolbar button. To reset the list of zoom states press the fifth
er view then clicking the left-hand mouse button selects that toolbar button. The axes are returned to the state when the
block. If a range of contiguous blocks is required (as seen in data was first displayed in the scrolled list.
the browser view), then select the first block in the range,
followed by holding down the “Shift” key and selecting the Tiles of Spectra
last item for the range. All blocks between the first block and A printed page corresponds to the visible area associated
the last will be selected. The range selection works across with the left-hand scrolling window. Each click within the
columns as well as within rows or columns. That is to say, scroll bar moves the display by one page whilst dragging the
rectangular sets of block can be selected by this mechanism. scroll button allows the pages to be moved at will. The arrow
Additions to the current selection require the use of the buttons on the scrollbar permit fine adjustments to the posi-
“Ctrl” key. If the “Ctrl” key is held down and a new name is tion of the spectra in view.
selected, the existing selection is retained in addition to the Pages can be divided into sub units referred to as tiles. The
block just chosen. If the block indicated with the mouse is format for the page is defined using a dialog window avail-
already selected then it is removed from the current selec- able from the main frame menu headed “Options”. If “Page
tion. The “Shift” key and the “Ctrl” key when used together Layout” is selected, the dialog box presents property pages
allow rectangular sets of blocks to be added or removed that allow the user to choose the number of tiles per page as

Understanding the Data 26

 Data Display and Browser

well as how many rows and columns of such tiles should ap- area to exceed the range of a signed 16-bit number (32767). Sor-
pear on the page. These pages represent an initial format for ry! If more tiles are required to view a set of spectra, then the
page layout mechanism can be used to reduce the overall size of
the number-of-tiles per page; however, the user is at liberty
the scrolled list.
to adjust these as seen fit.
There is a maximum of sixteen tiles per page. The layout is
specified first by selecting whether the tiles are to be dis- Tile Display
played in “rows” or “columns”, (this permits the layout to be
transposed at a click of a button) and then by specifying the
number of rows or columns, and finally by indicating, using Many of the attributes for displaying the spectra can be ad-
the “radio button” array, the number of tiles in each row or justed. The menu item labelled “Tile Display” on the “Op-
column (see “Page Layout” on page 136). Sixteen prede- tions” menu displays a dialog window for modifying the
fined layouts are provided, corresponding to arranging spec- appearance of the spectra.
tra in symmetrical rows up to the maximum (16) allowed. Ranges for the axes, labels and fonts used for the labels can
All these predefined formats may be changed to provide a be adjusted on the first two property pages found on the
“custom option” library of preferred layouts, as required. “Tile Display” dialog window. The user may also choose
between displaying the spectra using binding energy or ki-
netic energy, as well as counts per second or recorded
counts.
Figure 16: Main Toolbar An option on the “Y Axis” property page allows overlaid
Figure 16 shows the Main Toolbar. The dialog boxes avail- spectra to be offset with respect to one another. A check box
able from the main menu “Options” are also displayed using enables the offset mechanism and a numeric-field permits a
buttons on this Toolbar (see, e.g. “Options Bar” on percentage to be entered which changes the separation of the
page 134). Move the cursor over a toolbar button; a hint de- traces. Other check boxes offered on the “Y Axis” property
scribing the action associated with the button will pop up page alter the appearance of the display. The axis label and
and a slightly longer description will be displayed in the scale may be toggled on or off; also a spectrum may be plot-
main window status bar (see page 132). ted with the value of the experimental variable displayed at
either side of the graph area, and the spectrum background
Problem: The first release of CasaXPS limits the number of pag- may be subtracted from or included with the display (see
es that can appear in the scrolled view. Although the modern PC deatail of command options on page 136).
hardware is 32-bit, some 16-bit code still remains in MS Win-
dows software, in particular the class used by the routine that An alternative way of displaying overlaid spectra is via the
sets the scroll bar attributes, does not like the logical scrolled “Geometry” property page. Here the user can select between

Tile Display 27
 Data Display and Browser

2-D and 3-D views of the spectra. The latter option plots the
spectra as a function of the experimental variable. The first
spectrum in plotted in the front plane of a cube while the last
spectrum in the selection is plotted at the back plane. All
spectra in between appear at a plane that is determined from
the value of the experimental variable for the VAMAS
block.
The 3-D plot is adjusted using parameters that shift and scale
the two planes associated with the front and back of the
cube. Figure 3 shows a set of spectra plotted against etch Figure 17: 3-D plot.
time. The front plane is 75% of full size in both X and Y di-
rections, while this reduced axes area has been shifted left by
the maximum amount by specifying –100% for the shift. Colours
Note that the front plane can only be shifted around in X
while the back plane may be shifted in both X and Y. The “Colours” property page provides the means to set the
colours used throughout the system. Spectra, backgrounds
to spectra, synthetic components, regions and residual plots
A third option for data display allows a set of spectra to be may have the colours adjusted. Fill colours used to display
viewed in terms of abstract factors. This involves perform- 3-D plots are also changed through the options on this page.
ing a Principal Component Analysis (PCA) on the spectra. Selecting a button on the “Colours” property page brings up
Please see the section on Principal Component Analysis for a dialog window that allows the existing colours to be
a description of PCA and the use of this display option. viewed and new values set. In the case of spectra, sixteen
colours may be chosen. These colours are displayed in the
“Custom Color” section of the dialog window and changed
Other information drawn on a tile can be changed through using the “Define Custom Colors” expanded form of the di-
the “Display” property page. The font and text for the title of alog window. To change a colour within the “Custom
a tile may be adjusted. The optional header information font Color” set, first click on the colour you wish to change, then
may be changed similarly. A range of optional display items select a new colour, either from the default palette or using
can be toggled on/off allowing the spectrum to appear, for the colour values on the expanded section of the window. On
example, with or without quantification regions present. pressing the button labelled “Add to Custom Colors” the

Colours 28
 Data Display and Browser

colour square first selected is set to the colour just defined.

Each colour subsequently specified and the “Add to Custom Fonts
Colors” button pressed will cause the next custom colour
cell to change. The cells are updated in a top to bottom, left Typefaces (“fonts”) are managed in a similar way to colours.
to right order. However, the colours assigned to the graphs Buttons are provided for fonts associated with axes labels,
are assigned on a left to right, top to bottom order. the title and header text. The procedure for adjusting a font
Some colour selections are for a single colour. In these cas- is to select one of the buttons labelled “Fonts”. The title, for
es, clicking on any colour cell, so that the focus box sur- example, is located on the “Display” property page. To ad-
rounds the intended cell, followed by selecting the “OK” just the font, press the button next to the text entry field that
button on the dialog window will activate the “Apply” but- offers the current value for the title; a font dialog window
ton on the property sheet for the tile parameters. The colour appears. Then select the font parameters from the dialog
changes only take effect when the “Tile Display” window is window and press the “OK” button. Once again, like the col-
applied, either by the “OK” button or the “Apply” button. our dialog window, this action will activate the “Apply” but-
ton on the property sheet for the tile parameters. The font
will only be changed if the “OK” or the “Apply” button is
pressed.

Display Parameters and Scrolled Tiles

Tile display parameters can be applied to all the tiles in a

scrolled list. A property page headed “Global” allows a tog-
gle to be set that causes the settings currently active to be
transferred to all tiles when the “Apply” or “OK” button is
Figure 18: Colour definition window selected.
Two additional toggle buttons enable the transfer of ranges
for the X-axis and Y-axis. It is not always desirable to in-
clude these display parameters when the other characteris-
tics are globally applied, but on occasion it can be useful.
For example, setting the Y-axis range to that of a specific
size for a set of similar spectra can provide an interesting

Fonts 29
 Data Display and Browser

visual effect when scrolling is used to view the data. The ble is highlighted in this way.
variation in intensity as a function of an experimental varia-

Display Parameters and Scrolled Tiles 30

Graph Annotation CHAPTER 4

An important feature of CasaXPS is the ability to write En- The options for adding annotation are found on the main
hanced Metafiles. This permits data to be visualised through menu “Options” or via the main toolbar (Figure 19 - magen-
the data system, then exported in to other programs in order ta square).
to provide the final form used in a document or report. En-
hanced Metafiles supply the drawing information used to
display the data in CasaXPS to a word processor such as Mi-
Quantification Tables
crosoft Word, which can then be used to edit the drawing in
remarkable detail. The button to save these graphics files is
indicated below (red circle). After quantification regions have been defined for a spec-
trum, a tabulated form of the information gathered from data
may be added to the graphical display. The procedure to in-
Figure 19: EMF & Annotation buttons in Main Toolbar
Although the tile display options permit text, such as the ti- clude region information is similar to that used for synthetic
tle, to be adjusted and enhanced metafiles allow sophisticat- components. Both types of tables are added using the “Re-
ed visual editing, XPS spectra can sometimes best be gion” or “Component” property pages found on the “Anno-
annotated with information derived directly from the data. tation” dialog window. The font used for the table may be
The most obvious being tables related to the quantification selected before pressing the "Apply" button on the page of
of the chemical state. CasaXPS therefore provides some op- interest. A table of the current values derived from the
tions that allow the inclusion of such information as part of source in question will appear in the display area, and it is
the annotation. then repositioned via the "Annotation History" property

31
 Graph Annotation

page found on the same dialog window. which the change is required. Then adjust the settings using
the options found below the history list. Press "Apply" to see
the result of the changes.
Text lines may have the orientation adjusted in addition to
altering the contents via the text edit box. Fonts and colour
are reselectable through the history list and the edit options
on the "Annotation History" property page.
Entries on the "Annotation History" property page are in the
format used to save the annotation in the ISO 14976 file. A
block of data within the file includes a number of comment
lines. It is within these comment lines that processing, anno-
tation and quantification information is stored. The idea of
offering the full description used to save the data to file is
that the parameters can be understood (if desired). It is then
Figure 20: Annotation using region information
possible to save the data and include CasaXPS information
from the start.
Moving Annotation and the History Mechanism
Peak Labels
Annotation, either tables or text, can be moved using the list
of annotation entries currently defined for the active spec- Spectra can be labelled using the names from the element li-
trum. The spectrum display will be updated with the annota- brary. The element markers currently displayed are made
tion now at the new location. Position for annotation items available for annotating the spectral features on the "Peak
may be referenced to the display tile frame or the data itself Label" property page of the "Annotation" dialog window.
(for convenient placement e.g. when multiple blocks are dis- The element marker text is entered in the scrolled list, found
played in overlay) and labels may be orientated vertically or on the page of interest. To annotate the spectrum, the ele-
horizontally. Positioning targets (small squares with central ment names must be selected within the scrolled list and the
dots) may be set with the mouse anywhere within the display desired attributes chosen. The annotation text entries are
tile and disappear when the annotation window is dismissed. added to the annotation history list when the "Apply" button
Other attributes for the annotation can be adjusted using the is pressed. Again, the exact location for the annotation will
history list. Select the line describing the annotation for need to be assigned through the normal positioning proce-

Moving Annotation and the History Mechanism 32

 Graph Annotation

dure.
Note that “Peak Labels”, derived from the entries in the Text Annotation
Element Library, are different from “Peak Markers”,
generated by the “Find Peak” button in the Element Peak labels are merely multiple instances of text annotation.
Table tab of the Element Library module, ( see Individual lines of text can be added to the display using the
page 141). These markers serve only as a guide to the "Text" property page. A text edit box allows the desired text
presence (or absence) of a feature in the spectrum (a to be entered and any attribute, such as, vertical/horizontal,
peak maximum) which you may want to identify, for font or colour may be chosen before pressing the "Apply"
example, by using the Element Library routines, as button.
described on page 9. Markers may be set and cleared
independently of Peak Labels (use the “Clear Elements”
button for the Labels)

Text Annotation 33
Processing CHAPTER 5

An XPS spectrum represents an envelope that is derived

from many unresolved peaks. The relative intensity of these Smoothing
peaks may vary as a function of the kinetic energy imparted
to the electrons being recorded as well as the characteristics Data smoothing may be performed using a range of Savitz-
of the electron optics (lens mode/pass energy). In addition, ky-Golay algorithms or by weighted averaging the data us-
the position of the peaks may differ from the expected value ing a set of weights that are a normalised Gaussian
due to a combination of instrument energy calibration and distribution.
charging effects on the sample. CasaXPS therefore provides Smoothing algorithms should always come with a statistical
health warning. Smoothing can seriously affect your curve
a number of processing options that assist in understanding
fitting! The act of fitting synthetic peaks is not only the best
the data envelope and allow adjustments to the spectrum to
form of smoothing, but also makes assumptions about the
enable compensation for these analysis variables. distribution of errors within the data as well as the statistical
independence of the measurements. Smoothing the data be-
The processing options are available from the "Options" fore curve fitting invalidates these assumptions.
main menu item or from the main toolbar button shown in A Savitzky-Golay filter is derived by approximating the data
Figure 21. with a polynomial of degree less than or equal to the chosen
number of 2n + 1 channels. The coefficients for the selected
polynomial are determined in the linear least square sense
and the datum at the centre of the 2n + 1 channels is replaced
by the value for the polynomial at that point.
Figure 21: Main toolbar processing options. To perform a smoothing operation using Savitzky-Golay

34
 Processing

methods it is necessary to select the degree of the polynomi- Cotes or other quadrature methods, as these generally in-
al and the width of the data-channels over which the approx- volve forcing a function to take on the data values at the cor-
imation is to be made. responding nodes. The virtue of putting a quadratic through
Gaussian smoothing merely needs the width of the data- three points containing noise is somewhat doubtful. Inte-
channels to be used when averaging the spectrum. grating the data using a least square fit of a quadratic to more
than three points seems better, however from a practical per-
spective, if there is a significant difference between these
Differentiation two operations, then there’s trouble somewhere!
To summarise, integration is performed using the Savitzky-
Differentiation of spectra is also achieved using the Savitz- Golay polynomial in which a quadratic is used to approxi-
ky-Golay polynomial. This is appropriate since noise inter- mate 5 data channels. From a user’s point of view, integra-
ference would cause problems to an algorithm used for tion is performed simply by pressing the “Apply” button on
analytical functions, and so the implied smoothing operation the "Integration" property page, with the single option of be-
involved in Savitzky-Golay differentiation, would explicitly ing able to reset the zero point of the integraal, if required.
be required with other techniques.
The parameters used in differentiation are, therefore, identi-
cal to those for smoothing the data. To get a feel for what Energy Calibration
was actually done to the spectrum during the differentiation
procedure, a smooth operation with the same parameters,
applied to the original spectrum can provide an insight into A spectral line may appear at an energy position that is not
the shape of the data that produced the derivative. the expected value for a transition. The reason for this may
be due to spectrometer calibration or sample charging, how-
ever, for presentation purposes, the energy scale can be ad-
Integration justed using parameters on the "Calibration" property sheet.
Two values can be entered, the recorded position of the line
A more novel use of the Savitzky-Golay polynomial is in and the required energy for the peak. On pressing the "Ap-
calculating the integral for a spectral range. Integration by ply" button a shift is computed and the display is updated.
any means has implied smoothing involved since it is essen- Note that if preferred, the spectrum can be shifted by an
tially an averaging process. However, when noise is in- amount by setting the measure value to zero and entering the
volved there is little benefit in using sophisticated Newton- required shift in energy into the field for the true energy.

Differentiation 35
 Processing

Intensity Calibration Processing History

A spectrum recorded on a particular instrument represents Each VAMAS file data-block can be processed using any of
only the electrons as a function of kinetic energy that the in- the available options. The "Processing History" property
strument was capable of sampling. The transmission charac- page provides the means of viewing the operations that have
teristics for an instrument are important if the spectra are been applied to the data. Further, the processing may be re-
used to quantify the chemical composition of a sample. This versed entirely or selectively. The "Reset" button removes
extends even as far as calculating the background for the da- all the processing operations and restores the data to that
ta, where techniques presented by Tougaard can be only ap- originally supplied in the ISO 14976 file. Alternatively, by
plied after accounting for such variations in the intensity. selecting a set of lines within the scrolled list, then pressing
The "Intensity Calibration" property page allows the inten- the "Apply Selection" button, the indicated set of processing
sity to be adjusted using a function of a constant power of the operations are performed on the original data. (Selection of
energy. The same page on the "Processing" dialog window items follows the normal Microsoft conventions - left mouse
allows the inclusion of a relative transmission function to be button for a single item, left click then shift left click for an
specified. inclusive list, or control click for separated items). Thus the
The ISO 14976 file format does not specify how the trans- previous processing history is replaced by only those actions
mission function for an instrument can be incorporated with selected prior to pressing the button.
the data, however CasaXPS does allow the use of a second
corresponding variable per block to define the transmission Processing for a block of data can be globally applied to a
behaviour for the instrument (See “Appendix 1: ISO 14976 set of ISO blocks. Select these blocks within the browser
format files annotated” on page 143.) This is particularly view, then right click on the corresponding display view to
important when a single relative sensitivity factor (R.S.F) is reveal a dialog window for propagating operations such as
used to quantify results from a range of magnifications and curve fitting and processing. Choose the required actions
pass energies. Without first correcting for the differences then press the "Apply" button. A second window appears
due to operating modes of the instrument, a single R.S.F. showing the progress of the propagation and offers a "Stop"
would not yield consistent quantification results for the button. If the "Stop" button is pressed, the propagation will
same sample. terminate following the completion of the current action.
If a valid corresponding variable index is specified, the act Note that some actions, for example curve fitting, may take
of calibrating the intensity modifies the data by dividing the a significant period of time to complete.
raw spectrum by the indicated transmission function. Not all actions for processing the data may be applicable to

Intensity Calibration 36
 Processing

all the selected blocks. To prevent, for example, the energy Please note that although the “Calculator”, “PCA” (Prin-
calibration for the current block from being propagated to cipal Component Analysis), and “Test Data” tabs appear
other blocks, the history item can be flagged so that it is not (rightly) in the Processing module, they are of sufficient im-
propagated along with other appropriate actions. portance and power to merit their own topic sections (see
To exclude a processing operation from a propagate action, “Using the Calculator and Comparing Spectra” on page 71,
select the item in the history list, then press the button la- “Principal Component Analysis” on page 80, and “Optimi-
belled "Propagate Flag". The entry in the list for the selected sation of the Peak Parameters” on page 47).
processing option will change to indicate that the flag for The same is true of the Monte Carlo procedures, the entry
propagating the action is set to “false” (“don’t perform to which appears under the “Components” tab of the “Quan-
this”). Repeat the procedure (toggle) to remove the exclu- tification” module. (see “Monte Carlo Methods - Uncertain-
sion. ties in Intensity Calculations” on page 97).

Processing History 37
Element Library CHAPTER 6

The "Element Library" dialog window provides the means ysis) a library file must be present. The file may be changed
of managing the information used to identify spectrum fea- at any time using the controls described below (so that sev-
tures and then create regions and components with prede- eral library files may be held on disk), but only one library
fined values. Strictly it is not an “element” library at all, but (of unlimited size) may be active at any time.
a list of spectrocopic lines which, while derived from char- CasaXPS is normally shipped with a relatively simple ge-
acteristic transitions within the electonic structure of atoms neric library. Instrument or analysis specific libraries must
of a particular element, can also reflect the exact chemical be generated or imported as required by the user. the module
environment of those atoms - that is provide chemical com- recognises both CasaXPS and Kratos (proprietary) formats.
pound specificity. Three property pages within the module
offer two views of the data held in the element library file,
and a means of loading and merging data into the current Library File Structure
session.
The module is launched from the main toolbar, as shown be- The element library file is an ASCII (plain text) file. As such
low it is open to simple modification or change by the user, so
long as changes are made within the (casaXPS) format out-
lined below.
Figure 22: Library button in Main Toolbar. The file is a series of single item lines, each terminated by a
In order to function correctly, CasaXPS should have an ele- (single character) “return”.
ment library present in the system (although some basic fea- The first line is the version number.
tures will function without one) - but for many commands Each subsequent entry in the file has items that are included
and operations (involving qualitative and quantuitative anal- one per line as follows.

38
 Element Library

1. Element GL(30)
2. Transition 2.000000e-004
3. Label/Name Mg
4. Mass (Daltons) He
5. Energy Type (B.E. or K.E.) 1s
6. Energy (eV) He 1s
7. F.W.H.M. 0.000000e+000
8. Line shape (e.g. GL(30)) BE
9. Relative Sensitivity Factor 2.500000e+001
10. Ionisation source (X ray anode) 1.000000e+000
Photoelectric transitions should be entered with item 5 (en- GL(30)
ergy type) set to "BE"; this allows the same energy to be 8.200000e-003
used for different X-ray anodes. Auger lines always appear Mg
at the same kinetic energy independent of the excitation ....
source and these values should be specified with energy type ....
"KE".Note that the ionisation source is also specified so that
source specific information can be offered for spectra that An entry allows the species and transition to be specified but
match the acquisition characteristics found in the library. also permits a name other than these two labels to be also en-
line shapes are specified using the abbreviations listed in tered. For example, a carbon line corresponds to a chemical
“Line Shapes and Backgrounds” on page 55 state arising from PVC is identified by entering, "C" for the
The beginning of the “default” (CasaXPS) library file is species, "1s" for the transition and "C 1s PVCa" for the
shown below: name. It is the name that appears as the first item in the
scrolled list of transitions.
0 Although the expected energies for X-ray induced transi-
H tions are universal constants, quantification data will be (in
1s practice) specific to an instrument. An element library can
H 1s be customised through the energy-ordered scrolled list on
0.000000e+000 the property page labelled "Element Table". Right clicking
BE on an item name will produce a dialog window that allows
1.400000e+001 the items associated with the indicated line to be updated.
1.000000e+000 New entries can be created using this mechanism but for

Library File Structure 39

 Element Library

large numbers of new entries a text editor with cut and paste different R.S.F. values for different acquisition conditions.
can be the best method of modifying the element library.
Items which are specific to an instrument or instrument con-
figuration include the relative sensitivity factors (R.S.F.),
full width half-maximum (F.W.H.M.) and the line shape. In-
strument manufacturers will supply some of this informa-
tion, but individual experience will determine appropriate
values for items such as the F.W.H.M. and line shapes.
These parameters clearly vary from analysis to analysis and
specific figures are present here only to provide “sensible”
initial values for the creation of synthetic components - val-
ues that are sometimes useful!

Loading an Element Library

The property page labelled "Input" (see “Library” on

page 141) allows the session element library to be modified Figure 23: Element Markers.
by reading different element library files. A browser button
The Element Table window provides a scrolled list of the
offers a file selection dialog and the chosen file may be load-
ed or merged with the current session element library. If the contents of the Library File: clicking on a name (toggle)
"Load" button is selected, the element library in the current transfers a labelled position marker to the display for the se-
session is replaced by the contents of the specified file. Al- lected item Position markers may be removed from the dis-
ternatively, the "Merge" button causes the contents of the play with the Clear All Elements button. Find Peaks adds
file to be added to the existing information for the session. coloured markers to the display for all major peaks in the
A user can maintain a set of element library files and load/ displayed block, according to a defined algorithm: the Clear
merge only those appropriate for a type of sample or operat- Markers button removes these. Take care to note the dis-
ing mode of the instrument. In the absence of transmission tinction between element markers (from the library) and
function data this mechanism allows the user to maintain peak markers (from the Find Peaks routine)

Loading an Element Library 40

 Element Library

down. The energy order list will scroll with the energy posi-
Identifying Peaks tion of the cursor. Release the mouse button when over a
peak of interest and then use the table to toggle the peak
Peak identification is performed using the scrolled list of markers until a match is established.
transitions and the "Periodic Table" property page. A spec- A more detailed example of how to approach peak identifi-
trum may be viewed with markers added at energies where cation is given in “A Quick Tour of CasaXPS” on page 7
peaks might be expected (Figure 23 - note that the position
line sizes (heights) are scaled to reflect the RSF values
stored in the library) To add and remove these markers, se- Peak and element markers are maintained within a session
lect an element through either the periodic table interface or on a spectrum by spectrum basis. Altering the markers on
from the element table. The markers are toggled on and off one spectrum does not change those previously chosen for a
by repeatedly selecting the same element. different spectrum (block) even within the same experiment.
An alternative method for examining a spectrum is to have Simply select the tile used to display the data and the period-
the scrolled element table visible and then run the cursor ic table will update to reflect the state of the markers chosen
over the tile display window with the left mouse button held for that spectrum.

Identifying Peaks 41
Quantification CHAPTER 7

Regions
Data can be quantified at a number of levels. The most basic
form of quantification is using regions applied to a wide
scan spectrum; the results are added to the display using an The first step in quantifying a spectrum is defining one or
more energy regions. These specify the data channels that
annotation option. Alternatively, both regions and synthetic
are to be included in the intensity calculation.
components can be prepared over a set of narrow scan spec-
Each region requires the following information:-
tra and the results specified through a set of expressions in-
volving the names used to label the objects. A report • Name The name given to a region is signifi-
generated in this way can be viewed within CasaXPS or ex- cant as a quantification object. A report is specified in
ported as a TAB separated ASCII file for use with a spread- terms of these names and intensities are summed for
sheet program such as Excel. Figure 24 shows the main regions with identical names.
toolbar button for the dialog window used to perform these • R.S.F. Relative Sensitivity Factor for the data
two extremes plus a number of reporting mechanisms in be- within a region. The integrated data are divided by
tween. this value.
• Start Start point for the integration interval.
• End End point for the integration interval.
• Background Tougaard, Shirley, Linear, None
Figure 24: Main Toolbar.
• Average WidthNumber of channels used to fix end
The three property pages used to quantify spectra are la- points for the backgrounds.
belled "Regions", Components" and "Report Spec".

42
 Quantification

Armed with the values described in the above table, a region end points of the region. Noise in the data often makes the
is created by either pressing the "Create" button on the "Re- value at the end point unrepresentative of the desired back-
gion" property page or by clicking the column buttons on the ground limits. To reduce this problem, a number of channels
scrolled list of regions. may be used to define the background limits. The "Average
The start and end points for a region are taken from the zoom Width" parameter defines the number of data channels to be
parameters currently in effect when the region is created. averaged when calculating the tie-in points for the back-
Before pressing a create button, zoom into the spectral fea- ground.
ture until the display shows the range which characterises The backgrounds are calculated using algorithms based on
the intensity in question. The new region created at this stage those presented in a number of published articles. The "Lin-
will have start and end energies defined by the display. ear" case is simply a straight line between the end points of
After creation, the limits can be adjusted under mouse con- the region. "None" is just a constant set to the lowest data
trol. To perform this operation it is necessary that the "Quan- channel in the region. The more complicated backgrounds
tification" dialog window is visible and the active-page are those due to Shirley and Tougaard.
corresponds to the "Region" property page. Position the cur- The procedure due to Shirley is essentially a weighted-aver-
sor near an end point for a region, then hold down the left age of the background limits chosen to tie in with the spec-
mouse button. A box will appear showing the extent of the trum at the end points of the region. The weighting is
region. The end under the cursor is altered when the mouse determined from the area between the background and the
is moved whilst continuing to hold down the mouse button. data. Since the weighting is determined using the quantity
Should the same procedure be followed, but instead of ini- being computed, a sequence of iterations are required to ar-
tially positioning the cursor towards an end point, the mouse rive at the desired result.
button is pressed when the cursor is towards the middle; the
Tougaard has extensively studied the subject of back-
result is both end points adjusted simultaneously. A box out-
grounds to XPS spectra, however the background generated
lining the region is displayed and moving the mouse causes
by CasaXPS is simply calculated using the Universal loss
the box to shift both end points for the region.
function. For this algorithm to work it is necessary to re-
Backgrounds to spectra can be selected from Tougaard, move all instrumental contributions from the spectral shape
Shirley, Linear or None. To specify a new background type, before calculating the background. This is seldom possible
it is sufficient to type the first character for the background for practical situations. To allow the procedure to provide a
name. For example, to change from a Tougaard to a Shirley, background under less than ideal conditions, an adaptive
replace the name "Tougaard" by "S" then press return. The procedure has been adopted that attempts to fit the back-
regions will be recomputed with the new background type. ground to the given spectrum. The result is a background
The initial and final values for the background are tied to the that "looks" plausible in situations when the data has not re-

Regions 43
 Quantification

ceived the necessary pre-processing and equal to the back-

ground proposed by Tougaard for practical applications
when the appropriate adjustments have been made to the da-
ta.
To delete a region, first select the region in the scrolled list
of regions. Then press the "Delete" button.
Each region defined for a spectrum has a number of statistics
gathered from the data and displayed in the scrolled list on
the property page in question. The raw intensity (CPS eV)
between the background and the data, an estimate for the
F.W.H.M. and the position of the maximum count rate re-
corded are all list below the parameters that define the re-
gion. In addition, a percentage concentration is shown at the
Figure 25: Region plus Synthetic Components
bottom of each column. The raw intensity is calculated di-
rectly from the data, however, the percentage concentration Line shapes and R.S.F. values are stored in the element li-
includes R.S.F. adjustments. The results of quantification brary and these values are used when a new component is
based upon a wide scan and a set of regions are therefore created. A new component is created using the buttons
available through the "Regions" property page. above the list on the "Components" property page. If the el-
ement library dialog is active and a transition is selected
from the "Element Table" property page, then when a "cre-
Synthetic Components ate" button is pressed the values for the synthetic component
The property page labelled "Components" manages the cre- are read from the transition so specified. If the element li-
ation and optimisation of synthetic components. brary page is not active then the ISO species/transition label
is used to pick out the parameters for the component from
Synthetic components are specified by name, line shape, the element library.
R.S.F., position, F.W.H.M. and area (CPS eV). Line shapes that are available at present are either Gaussian-
The name is the means of identifying a component. A quan- Lorentzian product or sum functions augmented by an
tification report specified using expressions makes use of asymmetric tail shape (for more detail, see “Line Shapes and
the name to define how the intensities are to be combined. Backgrounds” on page 55). These are specified as follows,
Further, components with the same name are summed to- GL(n) where 0<=n<=100 for a product function or SGL(n)
gether before an expression is evaluated. for a sum function. A value of "n = 0" provides a pure Gaus-

Regions 44
 Quantification

sian while "n = 100" results in a pure Lorentzian. A tail is in- Quantification Calculation
troduced by appending T(x) where x > 0 and is typically Percentage atomic concentration (XA) calculations are per-
between 0.5 and 10.0. For example, an asymmetric line- formed using the formula
shape for Al 2p might be "GL(30)T(2.0)", see Figure 25 for
an example of such a component. ( IA E α ) ⁄ ( RA T ( E ) )
Clicking on the corresponding item in the scrolled list and X A = ------------------------------------------------
-
entering new values can specify peak positions, widths and ∑ i ( I E α) ⁄ (R T(E))
i
areas. Alternatively, when the "Components" property page
is on top of the dialog window, the mouse can be used to Where Ri is the relative sensitivity factor for the measured
pick up a component and move it to a new position, area or intensity Ii and T(E) is the transmission function for the in-
width. If the cursor is located near the top of a peak then the strumental operating mode used to measure the intensity Ii at
position and area are adjusted. If the cursor is located near kinetic energy E. The term involving raising the kinetic en-
the side of the peak then the width will change. Note that the ergy E to a power α allows traditional adjustments for ana-
height also changes when the width is adjusted using the lyzer behavior for those who wish to quantify spectra
mouse. These two parameters are modified simultaneously according to a manufacturer’s specification. The value for
so as to preserve relationships between peak intensities. the exponent a is entered on the Regions Property Page of
Constraints may be specified for position, width and area. the Quantification Parameters dialog window and is used
These may be either an interval (for example "0.5,1.3") or a whenever the accompanying check box is ticked. If the
constraint that is measured with respect to another compo- check box is ticked then both the exponent and the transmis-
nent. In the latter case, the parameter for one peak can be sion function located in the VAMAS block will be automat-
fixed with respect to another by entering the column label ically used during quantification, however a transmission
into the constraint box. For example, to fix the position of a function must be present for either of these corrections to be
peak in column "B" with respect to a peak in column "A", included in the calculation. Note that intensity calibration
type "A" for the position constraint in column "B". The sys- for a spectrum can be performed using the “Intensity Calib”
tem responds by setting the constraint to be "A + x", where processing option where an exponential adjustment to the
x is equal to the initial offset between the two peaks. The data can be made regardless of the presence or absence of at-
value for x can be modified by hand. To provide an offset of ransmission function
2 eV, column "B" constraint item should read "A + 2". Area
may be constrained similarly, but instead of plus use times, Quantification Report
e.g. "A * 0.666". Read, column B component is 0.666 times There are several types of quantification report. The proper-
the area of column A. ty page labelled "Report Spec" allows peak intensities to be

Regions 45
 Quantification

compared to one another either via information derived from in the Browser view of the active data file. The list entitled
regions, components or both regions and components. In ad- "Quantification item names" shows the set of names current-
dition, a set of named formulae allows intensities to be ly available for generating a report.
mixed and matched in any way that an arithmetic expression
Clicking on the column header buttons labelled "Name" or
can define.
"Formula" creates a new name/formula pair. The entry is ad-
A report generated from regions and/or components is pre- justed via a dialog window. Right click over the name item
sented in the form of a scrolled list. For each experimental for which changes are required. The dialog window enables
variable from the ISO file, the set of quantification items is the name and/or the formula to be modified or deleted.
listed. Information including peak position, F.W.H.M.,
R.S.F. and raw-area appear in separate columns. The final Each entry in the "Custom Report" corresponds to the exper-
column is the percentage concentration for each set of quan- imental variable in the active data file. The columns are en-
tified items with the same experimental variable. tered two per named formula, one for the raw intensity
When the report is based upon regions and/or components, calculated from the expression and one for the percentage
each item is included in the report on a separate line even if concentration. Thus a table as follows (Table 1 ) would re-
a region has the same name as another region or component. sult in a report with five columns. One for the experimental
This allows the statistics associated with each quantification variable and two pairs headed "Oxygen CPS eV", "Oxygen
item to be included in the list. Items with the same name are %" and "Aluminium CPS eV", "Aluminium %".
treated in a different way when formulae are used to gener- Table 1:
ate the report.
If a report is based on regions and/or components, and not Oxygen O 1s region
all the regions are to be included in the calculation for the
Aluminium Al 2p + Al 2p Oxide
percentage concentration, then an R.S.F. of zero should be
entered for those items that are to be excluded from the re-
sults. The variable names derive from an Oxygen region "O 1s re-
gion", and three synthetic components, two "Al 2p" (from a
The second type of report is generated using the table of doublet of aluminium metal peaks both named "Al 2p") and
named formulae (see, for example, “Adjusting a Custom one "Al 2p Oxide". The "Al 2p" name was repeated so that
Quantification Report” on page 120). Each formula takes the intensities calculated for both peaks for the metal doublet
the form of an arithmetic expression. Variables in these ex- were implicitly summed before the explicit sum defined via
pressions may include names given to the regions and com- a formula computes the total contribution from the alumini-
ponents, such as the data blocks which are currently selected um.

Regions 46
 Quantification

Optimisation of the Peak Parameters

Fitting synthetic components to a data envelope is probably
Table 2: Artificial Peak Structures.
the most useful tool in XPS data reduction. It is also one of Carbon C 1s peaks are used to form data envelopes with the offsets and
the more difficult procedures to perform owing to the com- relative sizes list below
plex nature of the underlying peak structure and the question
of how to account for the background structures in the data. PMMA PVA PVC PIB
The line-shapes used to characterise a peak are only approx-
imations and there can be no doubt about the uncertainty as- Number of 4 4 2 3
sociated with the different background algorithms. It is peaks
therefore entirely possible to fit a set of synthetic compo- peak 2 2 2 2
nents with good statistics, yet without any chemical or phys- offset (eV) 2.2 2.6 1.1 0.4
ical meaning. factor 1.238 1.04167 1.0 2.0
To provide a guide to peak fitting, an option on the process- peak 3 3 3
ing dialog window labelled "Test" can be used to replace the offset (eV) 3.3 3.7 0.6
true data using one of a set of known peak structures. These factor 1.238 1.04167 1.0
structures derive from work presented by Seah and Brown.
peak 4 4
The relative intensities and separations are those presented
offset (eV) 4 4.2
in Table 4 in the Seah and Brown publication4; a GL(50) line
factor 2.4762 1.08333
shape has been used to generate the components. It is useful
to exercise the fitting procedure using these structures with
different options for the synthetic components, especially It is very useful to try to fit the artificial peaks without any
when a background has been added. constraints with respect to the other peaks and then gradual-
ly include the relationships shown in the above table. The
The following table provides the characteristics of the artifi- chi square value for the curve fit should go to zero. This is
cial peak structure. Note that the data replaced determines the case since the artificial data does not contain noise. Or-
step size, energy position and count rate. dinarily, in the presence of noise, the chi square value should
be about equal to the number of degrees of freedom (also
4. M.P. Seah and M.T. Brown, J. Elec. Spec., 95 (1998) 71-93 shown on the "Components" property page).

Regions 47
 Quantification

The option for switching between the two-optimisation

methods is on the "Components" property page.

Figure 26: PIB artificial peak structure.

A good exercise is to attempt to reproduce the peak envelope
for PIB. What should be observed is how difficult it is to Figure 27: PVC artificial structure.
produce a good fit for PIB. The three peaks are very hard to The optimisation procedure labelled "Marquardt" is actually
identify unless additional information is given to the fitting not a pure Levenberg-Marquardt method. Linear and non-
procedures. If the offsets are first provided then the fit im- linear parameters are separated. Each step of the optimisa-
proves, but the exact match does not materialise until both tion procedure includes the solution of sub-problems intro-
the correct intensity ratios are supplied as well as the relative duced by the presence of constraints. The Marquardt method
positions. is used to establish the next set of non-linear parameters for
Another useful feature of fitting this artificial data is that the the current set of optimal linear ones. Although no optimisa-
strengths of the two fitting algorithms on offer can be seen. tion procedure should be viewed as universally applicable,
PIB is a stern test for the Marquardt method. The uncertainty these procedures provide a robust method for determining
as to which of the many combinations of similar peak pa- the peak parameters.
rameters presents a plateau in the parameter space that fails The Marquardt method uses information about both the
to give any good direction towards the optimum values. On function and its derivative. There are situations where the in-
the other hand, the Simplex algorithm, once close and suffi- formed steps produced by this method are slower that the
ciently constrained, will march straight to the exact fit. The "lets guess here" approach of the Simplex method. In fact,
Marquardt method works best when it has a clear view of the once the Marquardt method stops making significant im-
target such as with PVC (Figure 27). provements in the chi square value there is no harm in

Regions 48
 Quantification

switching to the Simplex method for one last try. This can plex method unless the parameters are already close to the
sometimes push the parameters away from a local minimum optimal values.
that has trapped the Marquardt method or on occasion serve
to confirm the optimum has been found. Propagating Quantification
Marquardt method with constraints can help with the Once a representative spectrum has been modelled using re-
progress towards physically significant peak parameters. gions and components, the next question is how to transfer
However, if the constraints prevent the synthetic peaks this model to similar spectra within the same ISO 14976 file
reaching an obvious optimum without an alternative availa- and other files. CasaXPS provides a propagation mechanism
ble then the algorithm tends to be slow. That is, if no natural to accomplish this task.
optimum lies within the range of the constrained parameters The spectrum for which the component model has been con-
and yet an optimum is visible outside the parameter range, structed must be displayed in the active tile view. Any ISO
then the algorithm will labour. The problem lies in the nu- blocks that require the equivalent model must be selected via
merous probes outside the range that will cause backtrack- the Browser View for each of the files involved. Once these
ing to the boundary values. Understanding this fact is useful. conditions are met, the mouse cursor is placed over the ac-
A well-posed problem will tend to converge quickly, while tive tile and the right mouse button pressed. A dialog win-
an inadequate model may result is sluggish behaviour. dow appears for propagating the quantification objects.
The Simplex method on the other hand works based upon Check boxes are offered on the "Propagate" dialog window.
maintaining a set of function values at the vertices of an N- These allow a choice of what actions are propagated through
dimensional simplex. One of a set of prescribed transforma- the selected ISO 14976 blocks. To transfer the quantifica-
tions for the simplex is employed depending on what func- tion model the "Regions" and the "Components" check box
tion value is founded at a probe point. should the ticked. Then press the "OK" button. A progress
As it is prescribed, there is very little that can go wrong with dialog window appears that allows the propagation process
the Simplex algorithm. However, convergence to the opti- to be terminated (see “Processing” on page 34).
mum set of N-parameters is not guaranteed especially when
a non-smooth function is used (i.e. when constraints are in- Data Editor
troduced). It does seem to work well for the situations typi- Curve fitting assumes a statistical model for the noise re-
cally found by CasaXPS, although it has been pointed out corded with the data. This model can sometimes be invali-
that for some optimisation problems, the Simplex method dated by the presence of unrepresentative spikes that are due
fails to converge when the number of parameters exceeds a to the detector system and have nothing to do with the true
problem-dependent value (even for smooth functions). The electron yield. It is best to remove such artefacts before at-
conclusion is, don’t try to fit too many peaks using the Sim- tempting to optimise the parameters for the synthetic com-

Regions 49
 Quantification

ponents.
A data editor is provided on the quantification dialog win-
dow. The abscissa and ordinates for the data displayed in the
active tile are listed on a property page labelled "Data Edi-
tor". The value for an ordinate can be changed. By right-
clicking the mouse whilst the cursor is over the correspond-
ing abscissa a dialog window is brought up that allows the
ordinate to be edited.
A check box allows the user to specify that an ordinate value
should be permanently altered. That is to say, if the data is
written back to disk then the adjusted value will be used in
the file. It should be noted that some actions in the process-
ing window can cause edits made to the data to be “undone”.
This occurs if the check box is not ticked and the processing
history is used to change the state of the processing (either
by resetting or applying a selection of processing to the da- Figure 28: PMMA peak structure with first and second de-
ta). The history mechanism always refers to the unprocessed rivatives.
data before taking the requested action. PMMA has four peaks, two well resolved and two that
merge together to from a broad structure towards the lower
Derivatives and Peak Envelopes
binding energies. Spotting the shoulder associated with the
A number of authors have proposed methods for identifying third and fourth peaks and determining the locations for
the underlying peaks responsible for a measured spectrum. these poorly resolved peaks is critical for constructing a
Some of these ideas can be tried out on the known peak physically meaningful model. The presence of a background
structures for PMMA, PVA, PVC and PIB. also interferes with identifying the true structure, although
the data in Figure 27 has been prepared without this compli-
cation.
Differentiating the artificial data once and then twice can
help to see the underlying peak structure (see Figure 28).
The shoulder of the third peak in the PMMA envelope can
be seen in the second derivative where a kink in both deriv-
atives highlights the shoulder in the data.

Regions 50
 Quantification

the same etch time.

Quantification by Example
Table 3: Structure of the Browser view for a depth profile.
The procedure for quantifying a set of spectra will be ex- Etch
plained using an ISO 14976 file. This file includes measure- C 1s O 1s Al 2p Mg 2p
time (s)
ments for four transitions taken after etch cycles have been
used to change the state of the sample. The problem is to cre- 0 C 1s/1 O 1s/2 Al 2p/3 Mg 2p/4
ate suitable models for each of the transitions using a com-
100 C1s/6 O 1s/7 Al 2p/8 Mg 2p/9
bination of regions and components, then to define the way
these quantities should be combined in the form of a report. 200 C 1s/11 O 1s/12 Al 2p/13 Mg 2p/14
The multiple-document-interface (MDI) used by CasaXPS 300 ...... ...... ...... ......
requires a new document before a file can be selected. When
CasaXPS is first run, the initial state is an empty document ...... ...... ...... ...... ......
and the "Open" menu option on the "File" menu is in an ac- 2000 C 1s/101 O 1s/102 Al 2p/103 Mg 2p/104
tive state. If all the documents have been used or none are on
offer then the "Open" option as well as the corresponding
The first step towards creating a model for each of these
toolbar button will be inactive. To activate the "Open" op-
transitions is to identify a representative spectrum for each
tion, press the "New" menu button; an empty ISO document
column of the browser. Click the column header for "C 1s".
frame will be created.
All the carbon spectra then become the browser selection.
Pressing the toolbar button for displaying the blocks one per
tile; the spectra are displayed in the left-hand scrolled view.
Figure 29: Main Toolbar
A file dialog window enables browsing of the file system
and choice of an ISO file. Take care within the file dialog. It
is the standard MFC dialog with drag and drop functionality. Figure 30: Spectrum Display Toolbar.
It is therefore possible to move files around simply by drag- Choose a carbon spectrum that shows a well formed C 1s en-
ging the icon for the file over the icon for a folder. velope. The limits for the energy range and hence the back-
Once read, the ISO file structure is displayed in the right- ground to the spectrum needs to be well defined. A good
hand side of the Document view. Each transition appears as idea of where the carbon peaks are located is essential for es-
a column and each row is populated by spectra acquired with tablishing a realistic background shape.

Quantification by Example 51
 Quantification

Before proceeding further ensure that your element library is the parameters until the background and the integration
loaded. (See “Element Library” on page 38). limits look right (whatever that may mean).
Another important consideration is intensity calibration. Be- 6. Ensure that only the set of carbon transitions is still
fore quantifying the raw data it may be necessary to apply a selected in the Browser view of the ISO file. Then right-
transmission adjustment. See “Processing” on page 34 for a click with the mouse cursor over the active display tile.
discussion of the issues involved with intensity calibration. The "Propagate Actions" dialog window will appear.
The next decision determines what values are used in the re- 7. Select the check box for "Regions" and press "OK". The
gion. In the event that only the total counts for carbon is re- result is that each spectrum in the set of carbon spectra
quired, no synthetic peaks are necessary and the name plus now has a region defined to be the same as the spectrum
the R.S.F. values will be the key information used by the displayed in the active tile.
quantification report. On the other hand, if the chemical 8. Scroll through the carbon spectra and check that sensible
states within the carbon envelope are to be profiled, then the regions have been defined and that the backgrounds are
R.S.F. within the region can be set to zero. The name for the indeed representative of the data. Repeat steps 6 through
region should be chosen to be different from any that will be 8 until satisfactory results are obtained.
used to identify the components. Once the above sequence of steps have been performed for
Let us say the only chemical state information required is "C s", "O 1s" and "Mg 2p" it is then time to generate syn-
that related to the "Al 2p" transitions. For all other transi- thetic components for the "Al 2p" data. Figure 25 shows a
tions, only the total intensity is needed to construct the pro- representative spectrum taken from a sample for which alu-
file. So for "O 1s" and "Mg 2p" repeat the following steps minium metal and aluminium oxides were present.
described for "C 1s": To create a model for an aluminium envelope, the first step
1. Select all the spectra for "C 1s" by clicking the column is to create a region. In this case the region is best named
header. "Re: Al 2p" and a value of zero should be entered for the
2. Enter the spectra into the scrolled list by pressing the R.S.F. value. By entering zero for the R.S.F. value, any
button that displays one spectrum for each of the tiles quantification report will not include the intensity calculated
(Figure 30). from the region in the percentage concentration values.
3. Scroll to the spectrum that best illustrates the character Two pairs of components named "Al 2p Metal" and "Al 2p
of the data and click in the tile with the left-hand button Oxide" describes the data seen in Figure 25 These corre-
of the mouse. This makes that tile the active tile. spond to two sets of doublets for aluminium. This knowl-
4. Bring up the dialog window for quantifying the data edge can be introduced into the model via the constraints for
(Figure 24). the component parameters. The relative intensity and spac-
5. Add a region via the "Region" property page and adjust ing of the components of a spin-orbit split (doublet) are usu-

Quantification by Example 52
 Quantification

ally well known, and inclusion of this information results in "Combined" button on the "Report Spec." page. Regions and
a chemically more meaningful model and also one in which components can be reported in a similar format by pressing
the optimisation routines will work more effectively. the buttons on the same property page but baring the appro-
Armed with the component model for one spectrum from the priate names.
"Al 2p" set, the other members of the set can be given simi- A custom report for the same data is usually in a format more
lar models automatically. Ensure that the "Al 2p" set is se- suited to showing trends. To generate such a report it is nec-
lected in the Browser view then, in the tile view, right-click essary to complete a table of names and formulae.
on the active "Al 2p" spectrum. Choose both the "Region"
and "Component" check boxes, then press the "OK" button. The first column for a custom report is a list of the experi-
Each spectrum within the set is fitted with the same region mental variable values. In this example the column will con-
and components as the model which was prepared earlier. tain etch time in seconds; the units are defined by the
When components are propagated the parameters are auto- VAMAS file format. Subsequent columns are defined by the
matically fitted to the target data. If there is a trend in the da- formulae, and are labelled by the corresponding names.
ta, for example, the oxide components dominate at the Pressing the "Reset" button initialises the list of names/for-
surface but smoothly diminish with depth, then the compo- mulae (Table 4 ). The entries are taken from the regions de-
nent parameters for the previous spectrum in the set of fined for the spectra currently selected within the Browser
"Al 2p" narrow scans are a better starting point for the opti- view. Typically this provides the right number of quantifica-
misation process. The "Propagate Action" dialog window tion names, but additional ones may be created or existing
includes a check box labelled "From previous block". If the ones removed via the edit dialog window. Right-click the
check box is ticked, the parameters for successive blocks are mouse button over a name field to display the edit dialog
taken from the preceding block. Otherwise the parameters window.
derive from the block displayed in the active tile.
Again, it is advisable to view the results of these automatic Table 4: Name/Formula List.
fits by scrolling through the spectra in the tile view. Once
Name Formula
satisfied, the ISO 14976 file is now ready to generate a
quantification report. C 1s C 1s
The first step is to ensure that all the spectra to be included
in the report should be selected within the Browser view. O 1s O 1s
A full report detailing parameters such as position, Re: Al 2p Re: Al 2p
F.W.H.M., R.S.F., raw-area and percentage concentrations
for all the quantification units is obtained by pressing the Mg 2p Mg 2p

Quantification by Example 53
 Quantification

In the current example, three of the entries are appropriate umns represents the percentage concentrations for the same
for the custom report; only the "Re: Al 2p" region needs to items.
be edited. Move the cursor over the name field and then The values within this report may be written to file in an AS-
press the right-hand mouse button. The dialog window for CII format. Each column of data is TAB separated. This al-
editing the name/formula field appears and the current val- lows the text file to be read into a spreadsheet program such
ues for these two items are entered into the text-edit fields. as Excel, where the data can be formatted for printing either
Change the name from "Re: Al 2p" to "Al 2p", then edit the as a table or in graphical form.
formula to read "Al 2p Metal + Al 2p Oxide". Note that the
list above the name/formula table contains all the names de-
fined for the report. On the edit dialog window, press the Report Files and Excel
button labelled "OK"; observe that the name/formula for
aluminium has changed to the desired values (Table 5 ). When a report is generated the data are displayed in a
Table 5: Name/Formula List after editing. scrolled list. This represents an alternative view for a VA-
MAS file document. If a report view is the active MDI frame
Name Formula (indicated by the frame title bar colour), then the main menu
only offers "File" and "Window" menus. The options on the
C 1s C 1s "File" menu under these circumstances are restricted to
"Save As …"; this allows a File Dialog to be used to specify
O 1s O 1s
a text file to receive the report.
Al 2p Al 2p Metal + Al 2p Oxide Once a report has been saved to disk, Excel can open the text
Mg 2p Mg 2p file simply by selected the file name via the "File/Open" op-
tion on the Excel main menu. A Wizard for loading text-files
into Excel guides the way through the available options. On
The custom report is generated by pressing the "Apply" but- completion the columns of data appear in a spreadsheet for-
ton on the "Report Spec" page. Nine columns of data are dis- mat.
played in a scrolled list view. The first column is the Excel provides many tools for presenting spreadsheets in
experimental variable and is followed by eight columns cor- both tabulated forms as well as graphically. These far ex-
responding to the names/formulae previously prepared. The ceed the sophistication that could be implemented in
first four of these eight columns list the raw areas that were CasaXPS and so the use of a spreadsheet for presenting the
specified via the formulae, and the second set of four col- results seems most appropriate.

Report Files and Excel 54

Line Shapes and Backgrounds CHAPTER 8

A range of physically possible line profiles in core-level trons from the data. Figure 31 and Figure 32 show examples
XPS is possible, and simple Gaussian or Lorentzian func-
tions are only rarely entirely adequate. In the case of metal
samples it has been shown that asymmetric profiles should
be expected on theoretical grounds [6], however recorded
spectra exhibit deviations from idealized profiles due to a
range of instrumental and physical effects:
• The response function of the electron analyzer (which
may be asymmetric),
• The profile of the x-ray line shape (predicted to be asym-
metric for non monochromatic lines from metal anodes
[6]),

• Intrinsic life-time broadening of the core-level hole state

(usually assumed to be Lorentzian in nature),
• Phonon broadening,
• Differential surface charging of the sample. Figure 31: Au 4f doublet fitted using an offset linear back-
In addition to these instrumental considerations the shape of ground. The line-shape used in the fit is a Doniach Sunjic form con-
a synthetic peak is also influenced by the choice of back- voluted with a Gaussian that results in an asymmetry index [6] of
ground algorithm used to remove so called “extrinsic” elec- 0.49 for each peak

55
 Line Shapes and Backgrounds

of peak fits where the difference between the two outcomes background is used. This is best measured by the asymmetry
lies in the choice of background. Doniach Sunjic profiles index [6], which changes from 0.49 in the case of a linear
have been used to model the recorded doublet, where it has background to 0.008 when a Shirley background is em-
become necessary to adjust the asymmetry parameter as ployed.
well as the width of the Gaussian used to broaden the under- Although theoretically based, the Doniach Sunjic profile
lying Doniach Sunjic shape. The most obvious difference suffers from ill-defined areas. The integral of the Doniach
Sunjic function is infinite for non-zero values of the asym-
metry parameter [7] and therefore any direct use of this line-
shape requires a somewhat arbitrary use of cut-offs to allow
finite peak areas to be reported. Changing the shape of the
peaks through adjustments to the asymmetry parameter and
the width of the broadening function has altered the relative
proportion of the spin-orbit split peaks in the Au 4f doublet
(Figure 31 and Figure 32). The background choice has al-
tered the reported intensity as well as FWHM for the indi-
vidual peaks. These factors emphasize the importance of
comparing like-with-like when quantification results are
used in practice.
The choice of line-shape is determined by the nature of the
problem. The best fit is not always as important as producing
quantification results that can be compared to historical re-
sults. On the other hand if the results represent a sequence of
experiments for which a trend is more important than the ab-
Figure 32: Au 4f doublet fitted using an iterated Shirley back-
ground. The line-shape used in the fit is a Doniach Sunjic form con- solute values then a good model may take precedence over
voluted with a Gaussian that results in an asymmetry index of 0.008 the need to supply numerical values consistent with the past.
for each peak. Provided the models are self consistent the use of Doniach
Sunjic line-shapes is usually acceptable.
between the two peak fits is the reduction in the amount of Analysis driven by precise quantification lies at the root of
asymmetry required in the line-shape when a Shirley [5]
6. Doniach S. and Sunjic M., J. Phys. 4C31, 285 (1970)
5. Shirley D. A., Phys. Rev., 55, 4709 (1972 7. Evans S., Surf. Interface Anal., 17, 85 (1991)

56
 Line Shapes and Backgrounds

many ad hoc procedures used in XPS. Shirley backgrounds

were introduced precisely [8] to remove as much asymmetry A List of Line-Shapes
as possible from recorded data (Figure 32) in a well-pre-
scribed fashion, so that near symmetric synthetic models can The line-shapes offered in CasaXPS are based around the
be used to characterize the intensity under a peak. A Gaus- following fundamental functional forms.
sian/Lorentzian line-shape is finite and, with the appropriate
relative sensitivity factors (RSF), can be used to compare in- The Voigt functional form has been the basis for most
tensities from fitted peaks to those calculated from integra- quantitative analysis of XPS spectra and is simply a convo-
tion regions. The key factor that has popularized Shirley lution of gaussian and lorentzian primitives.
backgrounds is the availability of well-characterized RSF Unfortunately this convolution proves to be mathematically
values coupled with a specific background and line-shapes. intractible and an exactly accurate analytical form for a
gaussian - lorentzian convolution is not available[7], so prac-
tical systems have adopted two approximations to the true
Some of the line-shapes introduced in CasaXPS have been Voigt function, as described below.
constructed to allow Doniach Sunjic asymmetric behaviour
to be associated with an underlying Gaussian/Lorentzian Gaussian/Lorentzian Product Form
shape. The Voigt approximation is used to characterize the
area, position and FWHM, while the asymmetric form ap-
 (x – E) 
2
proximates the rise in the signal much in the same way that exp  ( – 4 ln 2 ) ( 1 – m ) ------------------ -
 F
2

the Shirley background is used to reduce the data to symmet- GL ( x, F, E, m ) = --------------------------------------------------------------------------
2
ric shape. The advantage of retaining a separate asymmetry (x – E)
1 + 4m ------------------ 2
-
parameter in the synthetic model is apparent when a Tou- F

gaard background is used to remove the extrinsic contribu-

tion to a metal spectrum. Gaussian/Lorentzian Sum Form
To facilitate both trend analysis and basic quantification a
wide range of line-shapes is required.
 (x – E) 
2
m
SGL ( x, F, E, m ) = exp  – 4 ln 2 ------------------
- ( 1 – m ) + --------------------------------
 F
2
 (x – E)
2

8. Castle J. E. et al., J. Electr. Spectr. Related Phenom,106, 65 1 + 4 ------------------2

-
F
(2000)

A List of Line-Shapes 57
 Line Shapes and Backgrounds

Exponential Asymmetric Blend based upon Voigt-type

line shapes A ( x, a, b, F, E ) =

 GL ( x, F, E, m ) + w ( a, b ) [ AW ( x, F, E, m ) – G ( F, E ) ] x≤E

Given either of the above Gaussian/Lorentzian symmetric  GL ( x, F, E, m ) x>E

line-shapes, an asymmetric profile is obtained from a blend

Where
function as follows.

(x – E) 2
AW ( x, a, F, E ) = exp –  ------------------------------
 F + a ( x – E )

Y ( x ) = GL ( x ) + ( 1 – GL ( x ) ) × T ( x )

and

Where 0.3
w ( a, b ) = b 0.7 + ------------------------
( a + 0.01 )

 (x – E)
 exp  – k ----------------- x≤E The parameters a and b determine the shape of the asymmet-
T ( x, k, F, E ) =  F
 otherwise ric portion of the curve.
 1

Doniach Sunjic: a theoretically based asymmetric line

shape
Alternative Asymmetric Line-Shapes Doniach and Sunjic [6] performed an analysis for both pho-
toemission and X-ray line-shapes, both of which result in an
underlying profile given by the expression below. The for-
An asymmetric line-shape due to Ulrik Gelius (Uppsala, mula includes an asymmetry parameter a that characterizes
Sweden) offers a class of profiles by modifying the Voigt the asymmetry for a particular metal-like material. F is relat-
function via an ad hoc adjustment. The profile is given by ed to the FWHM and the position E is again related to but

A List of Line-Shapes 58
 Line Shapes and Backgrounds

not equal to the position of the maximum intensity for the form for the Shirley background can be determined for each
line-shape. It is therefore difficult to relate optimization pa- of the line-shapes and the sum of these backgrounds plus
rameters determined from the Doniach Sunjic profile to sim- line-shapes is used to approximate the variation in the spec-
ilar quantities determined from Voigt-type line-shapes. trum. A simple constant background is all that is required for
this procedure although other forms for the background are
still an option.
πα
cos  ------- + ( 1 – α ) tan-1  ------------ 
x–E
 2  F  Castle et al. [8] have developed a Shirley-type adjustment to
DS ( x, α, F, E ) = ----------------------------------------------------------------------------
2 2 (1 – α) ⁄ 2 a Voigt line-shape. The Shirley approximation is calculated
(F + (x – E) )
from the current gaussian/lorentzian shape and a polynomial
b0+b1(x-E) is used to scale the background in order to pro-
Nevertheless, the Doniach Sunjic profile offers an asymmet- vide a fit to the observed spectra. The procedure yields a
ric shape that is particularly appropriate for non-monochro- “Kappa” parameter (given by b0) that characterizes the “in-
matic X-ray induced transitions; the profile is potentially trinsic” step in the spectrum observed for a particular sam-
present in both the photoemission process as well as the ex- ple.
citation source.

Line-Shapes Based upon Backgrounds

Line-Shapes Available in CasaXPS
The above profiles are assumed to be entirely due to intrinsic
electron energy variations, where a background subtraction
algorithm is required before these line-shapes can be used to Curve fitting in CasaXPS is performed via the Quantifica-
model the spectra. An alternative approach is to include the tion Parameters dialog window where the synthetic line-
background shape as part of the model [8]. An analytical shapes are defined from the scrolled-list on the Components

Line-Shapes Available in CasaXPS 59

 Line Shapes and Backgrounds

property page (Figure 33). SGL(p): Gaussian/Lorentzian sum formula where the mix-
ing is determined by m = p/100, GL(100) is a pure Lorentz-
ian while GL(0) is pure Gaussian.
GL(p)T(k): Gaussian/Lorentzian product formula modified
by the exponential blend.
SGL(p)T(k): Gaussian/Lorentzian sum formula modified
by the exponential blend
GL(p)K(b0,b1): Gaussian/Lorentzian product formula
modified by a Shirley-type background prescribed by Castle
et al. A linear polynomial determined from b0 and b1 adjusts
the step in the Shirley background.
SGL(p)K(b0,b1): Gaussian/Lorentzian sum formula modi-
fied by a Shirley-type background prescribed by Castle et al.
A linear polynomial determined from b0 and b1 adjusts the
step in the Shirley background.

Further adjustments to the basic shapes

Figure 33: Property Page used to define synthetic line-shapes The basic shapes that result from the various functional
forms can be further modified by numerically convoluting
the profile with a Gaussian. For example, GL(100,50) will
The line-shape used to describe a photoelectric transition is convolute a pure Lorentzian with a Gaussian characterized
entered in the row labelled “Line Shape” and takes the form by a width of 50. The value for the width is the number of
of a text string. digital nodes used to describe the Gaussian and is therefore
an arbitrary unit characteristic of the numerical form for the
GL(p): Gaussian/Lorentzian product formula where the Gaussian.
mixing is determined by m = p/100, GL(100) is a pure N.B. Any line-shape that requires a digital convolution will
Lorentzian while GL(0) is pure Gaussian. cause slower performance than a shape that does not involve
this procedure.

Line-Shapes Available in CasaXPS 60

 Line Shapes and Backgrounds

Asymmetric Line-Shapes lute the basic shape of the profile.

The exponential tail described by Sherwood in the book
Practical Surface Analysis edited by Briggs and Seah at-
tempts to approximate asymmetric line-shapes commonly
found in photoelectric peaks. Several researchers have of-
fered alternative deformations of the symmetric Voigt pro-
file however none but Doniach Sunjic backs their
approaches with a theoretical basis. Nevertheless, practical
surface analysis requires practical solutions and the need for
a rigid line-shape model that matches the observed profiles
makes ad hoc forms acceptable tools. Philosophically, theo-
retically based solutions are preferred and hence, at least for
metals, Doniach Sunjic profiles ought to offer “better” fits.
A practical solution offered by Ulrik Gelius (QSA 7)[9] uses
a Voigt-like function as the underlying shape and modifies Figure 34: Asymmetric form due to Ulrik Gelius
the lower kinetic energy electrons using the equations given
above. CasaXPS offers these line-shapes in the form: The line-shapes defined by A(a,b,n)GL(p) and
GL(p)K(b0,b1) are different from the Doniach Sunjic pro-
A(a,b,n)GL(p): Gaussian/Lorentzian product formula file in that the Doniach-Sunjic profiles asymptotic limit is
modified by an asymmetric form define above. The param- zero, while both of the ad hoc forms allow steps to be mod-
eters a and b allow the asymptotic form of the asymmetric elled within the line-shape itself. Another characteristic of
tail to change whilst also altering the shape of the asymme- these ad hoc line-shapes is that the peaks are defined in
terms of intensity, position and FWHM taken from the un-
try prior to attaining the asymptotic shape. Figure 34 shows
derlying Voigt function. This is not unnatural in view of the
one such class of profiles where the parameter a has been fact that the area beneath all of these curves would be infi-
held fixed at 0.35 while b varies between 0.2 and 1. The third nite without the introduction of some arbitrary cutoff limits.
parameter n defines the width of a Gaussian used to convo-
DS(a,n): The basic Doniach Sunjic profile is defined in
9. G. Wertheim, J. Electron Spectrosc. 6, 239 (1975) terms of the asymmetry parameter a and a convolution width

Line-Shapes Available in CasaXPS 61

 Line Shapes and Backgrounds

n. The profile defined above is numerically convoluted with moves intensity modelled by the line-shape away from the
a Gaussian whose width is determined from n to produce the primary peak position and towards the cutoff regions (Fig-
final line-shape. ure 35). For this reason it is essential to use the Doniach Sun-
Although the Doniach Sunjic line-shape allows very good jic line-shape in situations where the asymmetry parameters
fits to experimental data, the infinite intensity defined by the are close to the same value so that calculated areas can, in
curve makes this profile difficult to use under practical situ- some sense, be compared.
ations. Any scheme that limits the area through cutoff func-
tions or ranges introduces inconsistencies into the peak Modifications to the Doniach Sunjic function
parameters that make relationships such as stoichiometry The advantages of the Doniach Sunjic profile are lost when
invalid. The degree of asymmetry used to describe the data synthetic models are required for quantification purposes. It
is necessary to introduce ad hoc cutoff behaviour before fi-
nite intensities can be reported and, once obtained, the peak
areas can only be used in a quantification reports when rela-
tive sensitivity factors are know for the specific cutoff crite-
rion adopted.
Empirical observations suggest that pure Doniach Sunjic
profiles do not always yield good fits for electrons with
higher kinetic energy (KE) when monochromatic X-ray
sources are employed. For example, Al 2p doublets record-
ed using a monochromatic X-ray source appears less
Lorentzian in nature (to the higher KE side of the peak) than
the Doniach Sunjic shape will allow (Figure 36), however
the same line measured using an Aluminium anode does ex-
hibit the predicted (but somewhat broader) shape. The latter
Figure 35: Doniach Sunjic profiles for asymmetry parameter
equal to 0.1 and 0.2. Note how the asymptotic curves cross over illus- is thought to be due to the double influence of the Doniach
trating how intensity is transferred away from the peak maximum Sunjic shape from the photoemission process as well as the
X-ray line profile, while monochromatic X-rays have lost
the Doniach Sunjic distribution as a consequence of the fil-

Line-Shapes Available in CasaXPS 62

 Line Shapes and Backgrounds

tering process. try index [6] measures the departure of the line-shape from a
symmetric form. Such a regime fits well with the philosophy
adopted by Shirley, where the background shape was used to
reduce an asymmetric step to a form that could be character-
ized by a Voigt function. The asymmetry index is given by

fwhm left
a = 1 – ------------------------
fwhm right

The asymmetry index is reported on the Components Prop-

erty Page for each line-shape.

H(a,n)GL(p): Hybrid Doniach Sunjic/Gaussian-Lorentzian

(product) line-shape.
Figure 36: Al 2p Doublet from a monochromatic X-ray source.
H(a,n)SGL(p): Hybrid Doniach Sunjic/Gaussian-Lorentz-
ian (sum) line-shape.
In order to fit spectra such as that shown in Figure 36 the
pure Doniach Sunjic shape has been modified. The lower
KE asymmetry derives from the functional form for the Do- The “H” form of the DS/GL hybrid line-shape is identical to
niach Sunjic profile while the higher KE side of each com- the DS(a,n)GL(p) form except the position, FWHM and area
ponent is a Voigt-type function. A numerical convolution is parameters are all determined from the Voigt-type shape.
applied to the combined profiles to produce the line-shapes Compare the area values reported in Figure 37 to those from
in Figure 36. Figure 36. The area taken from the Voigt-type portion is sig-
Given an approximation to the Doniach Sunjic profile of the nificantly less than the intensity taken from the Doniach
form shown in Figure 36 then the possibility of characteriz- Sunjic curve where a cutoff has been used. Another differ-
ing the line-shape by the associated Voigt-type profile be- ence between the two forms is that the area reported by the
comes feasible. The Voigt-type portion of the line-shape “H” is much less sensitive to adjustments in the asymmetry
provides the position, width and intensity while an asymme- parameter (numerical convolution does move shape infor-

Line-Shapes Available in CasaXPS 63

 Line Shapes and Backgrounds

mation between the two halves of the line-shape). proximates a step function as the asymmetry parameter in-
creases, while for small values of this parameter the cosine
term is responsible for the movements in the DS maximum
and for a = 0 results in a pure Lorentzian form. The formula
for the “F” profile combines these characteristics by blend-
ing the asymptotic behaviour of the DS profile with one of
the Voigt-type functions using a linear mapping.

Figure 37: Hybrid DS/GL line-shape characterized by the

Voigt-type portion.

A New Line-Shape
The hybrid form of a DS/GL line-shape is blended using a
numerical convolution. The maximum for the two shapes di- Figure 38: A new line-shape that blends the asymptotic behav-
verge for larger asymmetry values, therefore characterizing ior of a Doniach Sunjic profile with a Voigt-type function.
the peak parameters by the Voigt-type portion is more diffi-
cult as the asymmetry becomes more pronounced. A new
class of functions can be introduced to overcome this prob- F(a,m,n)GL(p): Asymptotic Doniach Sunjic/Gaussian-
lem, namely, a set of line-shapes that are a blend of the as- Lorentzian (product) linear mapping line-shape.
ymptotic behaviour from the DS line-shape with a
Lorentzian. F(a,m,n)SGL(p) : Asymptotic Doniach Sunjic/Gaussian-
The cosine term in the Doniach Sunjic profile rapidly ap- Lorentzian (sum) linear mapping line-shape.

Line-Shapes Available in CasaXPS 64

 Line Shapes and Backgrounds

The linear mapping between the two functional forms is de- profile (asymmetry parameter equal to 0.09) for a range of
fined as a percentage through the m parameter. Figure 38 the linear mapping parameter. The characteristic shape of
shows the same Al 2p data envelope as is seen in Figure 37, the Doniach Sunjic profile can be retained to some degree
but the line-shapes are defined using the “F” functional but is supplemented by a class of shapes that go beyond what
form. In Figure 38, the same asymmetry value has been used is possible with a DS profile. The principal difference be-
as the one in Figure 37 however the linear mapping param- tween the DS profile and the “F” function lies in the static
eter is set to 32% DS asymptotic form 68% Voigt-type func- position of the maximum for the class of profiles. The mov-
tion. The resulting profile is convoluted with a relatively ing maximum is a feature of the DS profile and can be
achieved to some extent by use of the numerical convolution
option. Increasing the width of the Gaussian caused the line-
shape to lean in towards the asymmetric side of the peak
(Figure 38).
A further variation on a theme is also possible where the “F”
function can be replaced by:

(1 – α) ⁄ 2
E ( x, α, m, F, E ) = m ( ( Voigt ( x ) ) ) + ( 1 – m )Voigt ( x )

The “E” function is merely allows the shape to the right to

be a part of the asymmetric shape to the left.

Figure 39: “F” function plotted over a range of values for the linear Background Subtraction
mapping parameter, a constant asymmetry parameter (0.09) and no nu-
merical convolution applied.
The line-shapes described above are very dependent on the
availability of background subtraction algorithms that com-
wide Gaussian (characterized by the number 380) and ac-
plement their use. If a background is incorrectly removed
counts for the Lorentzian bias in the Voigt approximation.
from a spectrum then the merit of attempting to model the
Figure 39 shows how a pure Voigt-type function, SGL(70), result with theoretically correct synthetic line-shape is less
is merged with the asymptotic form for a Doniach Sunjic clear.

Background Subtraction 65
 Line Shapes and Backgrounds

Tougaard and co-workers [10] have devoted much time to has a characteristic loss function of varying complexity and
understanding the backgrounds that are present in XPS spec- Tougaard has written many papers describing methods for
tra. The transport of electrons through a material after exci- establishing both generally applicable approximations (so
tation by X-rays can be described by equations [11] that called, universal cross-sections) and specific forms for indi-
involve a one-sided convolution of the recorded data with an vidual materials.
energy-loss probability distribution. This loss function of- The principal methods used by Tougaard for calculating the
loss functions are theoretical dielectric response models and
empirical estimates computed from spectra acquired using
reflected energy-loss spectroscopy (REELS). Tougaard fol-
lows these calculations by fitting the results to a rational
function, which models the resonance structure in the calcu-
lated loss-functions and allows the essential distribution to
be described by a simple formula. These loss-functions are
defined in terms of four parameters, three of which describe
the shape of the rational function and one to allow for vari-
ous band gaps in different materials. Tougaard refers to the
distribution below as a “three-parameter universal cross-
section” and has established values for a number of materi-
als including aluminium, silicon, silicon dioxide and others
[10]. The form of this universal cross-section is

 ---------------------------------------
BT
- T > T0
 2 2
P ( T, B, C, D, T 0 ) =  ( C – T ) + DT
2
Figure 40: Three-parameter universal cross-section for Silicon 
 0 T ≤ T0

fers a prescription by which electrons leaving the sample CasaXPS offers several ways of using the three-parameter
with an initial energy may have their characteristic energy universal cross-sections. The background type defined on
altered by the interactions with the surface. Each material the Regions Property Page of the Quantification Parameters
dialog window (Figure 40) may be chosen from the follow-
10. Tougaard S., Surf. Interface Anal., 25 137 (1997) ing list:
11. Tougaard S., Surf. Interface Anal., 11 453 (1988)

Background Subtraction 66
 Line Shapes and Backgrounds

U Si Tougaard (short form U Si) : Three-parameter cross- a cutoff limit. It is therefore possible to retain the shape of
section determined for silicon. the distribution but move the position of the maximum with-
out changing the C and D parameters.
U SiO2 Tougaard (short form U SiO2) : Three-parameter
cross-section determined for silicon dioxide. The custom Tougaard cross-sections offer a chance to
change the background in response to the spectrum under in-
U Al Tougaard (short form U Al ): Three-parameter cross- vestigation. Such an approach is not unprecedented [12].
section determined for aluminium. “Direct, simultaneous determination of XPS background
and inelastic differential cross sections using Tougaard’s al-
gorithm” by Dr. M. Jo describes a method that extracts loss-
U Ge Tougaard (short form U Ge) : Three-parameter
functions from the XPS data using optimization procedures
cross-section determined for germanium.
applied to a cross-section defined in terms of spline polyno-
mials.
U Poly Tougaard (short form U Poly) : Three-parameter
The rigid model offered by the three-parameter cross-sec-
cross-section determined for polymers.
tion provides more guidance in the regions most influential
in peak fitting, but will struggle to reproduce the asymptotic
U 4 Tougaard (short form U 4) : Custom three-parameter behavior for the cross-section as well as the adjustments
cross-section where the parameters are entered on the Re- needed to describe the near peak region of the background.
gions Property Page (Figure 10) in the Cross Section text- Nevertheless, a practical approach to the use of Tougaard
field. The four numerical values entered in a comma separat- backgrounds in peak fitting is to choose a background that
ed list correspond to B,C, D and T0 in the energy-loss func- works in conjunction with the synthetic peaks. The custom
tion above. form for the cross-section allows the background to develop
at the same time as the peak parameters and choice of line-
C 4 Tougaard (short form C 4) : Alternative custom three- shapes. Figures 36 through 38 all use a custom three-param-
parameter cross-section where the parameters are entered on eter universal cross-section to describe the background un-
the Regions Property Page (Figure 10) in the Cross Section der an Al 2p metal doublet; typically, researchers would
text-field. The four numerical values have the same meaning choose Shirley background for such spectra.
as those found in the U 4 Tougaard form, however T0 is used
to shift the resonance structure rather than simple acting as 12. Jo M., Surface Science, 320 191 (1994)

Background Subtraction 67
 Line Shapes and Backgrounds

Adjusting the universal cross-section respect to a real variable. The D parameter principally
moves the poles of the Tougaard cross-section away from
The rational function for the three-parameter cross-section is
the real-line, while C changes the projection of the pole onto
defined in terms of three constants B, C and D. This approx-
the real-line. Small D corresponds to near singular behavior
imation is the quotient of a linear polynomial and a quadratic
in the cross-section and therefore results in shape structures
in terms of T2, but it is the denominator that determines the
located, to some degree, by C while the magnitude of the
position and strength of the resonance in the cross-section.
function can be moderated by adjusting the B parameter. It
The parameters C and D both determine the position, width
should be noted, however, that B adjusts the behavior of the
and height of the resonance, but to a first order (small D) the
cross-section in the region close to zero energy-loss. This is
position is influenced by the square root of C and the sharp-
precisely the region of interest for modeling a peak and so
ness of the cross-section is determined from the size of D.
the values given to B and D can be influential in obtaining a
reasonable synthetic model.
The three-parameter universal cross-sections for specific
materials may be used as a basis for determining a back-
ground. A wide scan spectrum allows the essential form for
a background to be determined in the case of materials for
which cross-sections are not available. The resulting back-
ground is easily transferred to any high-resolution spectra
from the same sample (Figure 41 and 42).
The meaning attributed to custom backgrounds should be
viewed in the same way a synthetic model for a peak enve-
lope is viewed. That is to say, only chemistry and knowledge
of the sample allows a synthetic model to be constructed
which has any meaning and this should also be the case for
custom Tougaard backgrounds. The use of these back-
Figure 41: Wide scan spectrum from a PAA sample with a custom grounds is entirely at the discretion of a researcher and bad
three-parameter cross-section applied. backgrounds can be constructed in much the same way that
synthetic models are open to abuse. However, a carefully
The distance of a complex pole from the real-line in the do- constructed background can reveal features that may be
main of a complex function is instrumental in determining overlooked or not accessible when a linear or Shirley form
the width of a resonance when the function is plotted with is applied. Figure 42 shows an O 1s structure from PAA (po-

Background Subtraction 68
 Line Shapes and Backgrounds

ly(acrylic acid)) after a custom Tougaard background has an associated background. The synthetic PAA spectrum in
been subtracted. The structures to the higher binding energy
side of the peak may be real or an artifact of the background
algorithm, but a linear background applied to the same data
would fail to ask any questions at all regarding these fea-
tures.

Figure 43: Synthetic PAA spectrum constructed from a set of

Gaussian/Lorentzian line-shapes and a custom Tougaard cross-sec-
tion.

Figure 42: High-resolution spectrum from the same PAA sam-

ple shown in Figure 41. The custom Tougaard background has been Figure 43 has been constructed from a set of Gaussian/
subtracted from the data. Lorentzian line-shapes approximately positioned at the ex-
perimental values for the valance band, two C 1s structures,
two O AES peaks and one O 1s photoelectric peak. The
Simulating Spectra using Tougaard methods widths and intensities are consistent with results for such a
Given a set of synthetic peaks and a three-parameter univer- sample measured using a Scienta ESCA-300 at RUSTI,
sal cross-section it is possible to construct a spectrum with Daresbury Laboratory, UK, and the cross-section has been
chosen similar to the one used in Figure 42. A sequence of
iterations using these peaks and the three-parameter univer-

Background Subtraction 69
 Line Shapes and Backgrounds

sal cross-section shown in Figure 44 converges to the syn- a perfectly understood data set.
thetic spectrum in Figure 43.
Everyday procedures such as curve fitting can be evaluated
This technique for generating theoretical spectra offers a
using simulated spectra and therefore the ability to produce
way to understand the consequences of choosing a particular
spectra with known shapes can enhance understanding for
cross-section and if nothing else provides a basis from which
students and researchers alike.
real spectra can be viewed. Surface morphology and nano-
structure causes most real spectra to exhibit background be-
havior that is not characteristic of a material but the environ-
ment in which the element is found. Attempting to construct
a spectrum based purely on a resonant cross-section model
clearly shows the limitations of such approximations when
applied to real samples.
Although synthetic modeling has little direct use for real
samples, it does offer an analyst the opportunity to investi-
gate algorithms used to probe real spectra. For example, nu-
merous techniques have been proposed from Multivariate
Statistics for reducing XPS spectra to meaningful chemical
information [13] [14]. Procedures for generating simulated
spectra allow data sets to be constructed with similar charac-
teristics to the experimental data but with well-defined in-
formation. These algorithms can be applied to the simulated
spectra in order to establish what information is extracted for
Figure 44: Initial synthetic components used to construct a PAA
13. Fiedor J.N., Proctor A., Houalla M. and Hercules D.M., Surf. spectrum (green). The background curve (brown) represents the first
Interface Anal.,20 1 (1993) iteration of the cross-section with the model data.
14. Do T., McIntyre N.S., Harshman R.A., Lundy M.E. and
Splinter S.J., Surf. Interface Anal., 27, 618 (1999)
[

Background Subtraction 70
Using the Calculator and Comparing Spectra CHAPTER 9

CasaXPS provides a “spectrum Calculator” which enables 2. Load the separate files for which a comparison or calcu-
the normal arithmetic operations (adddition, subtraction, lator operation is required (each in its own experiment
multiplication an,d division) to take place between spectra frame).
or spectral regions.The Calculator is a means of managing 3. Select the ISO blocks of interest via the loaded experi-
and extracting information used to identify spectral features ment frame browser windows. To add to the selection the
and then to take that information on to quantify regions and Control key must be pressed when new blocks within a
components. Calculator operations may be performed in a file or from different files are selected.
“one-to-one” or “one-to-many” fashion. 4. Move the cursor over the right-hand-side of the empty
experiment-frame window and click the right-hand
mouse button. A dialog window will appear and offers
Creating a Comparison File the list of ISO blocks currently selected. Provided these
data are compatible (in the ISO sense), the selected spec-
CasaXPS enables calculation operations on spectra provid- tra will be copied into the previously empty window. If
ed that the data blocks are located in a single ISO 14976 the tick box labeled “Data only” is ticked then the spec-
file. It is clearly not always convenient for data acquired at tra will be copied using the results of any processing,
different times to be stored in the same file, but spectra from otherwise an exact image of the original data will be
one or more files may be easily copied into a single file in transferred to the new file.
order to make such comparisonsor calculations. An empty frame represents an empty VAMAS file and by
The procedure for moving the spectra is as follows: copying spectra into an empty frame one creates a new (uni-
1. Create an empty experiment frame window (use the form) experiment within which it is possible to compare the
“New” button or shortcut). spectra, but in addition the transferred data may subsequent-

71
 Using the Calculator and Comparing Spectra

ly be saved to disk so that the results of any data manipula-

tion performed on the spectra are preserved. For example,
the difference of two spectra can be copied to a new file by
performing the subtraction and then copying the VAMAS
block altered by this operation using the “Data Only” check
box option. Any charge correction or other processing can
be preserved using this method.

The Calculator

The spectrum calculator is accessible from the Processing

Dialog window.
An important use of the calculator is that of constructing dif-
ference spectra where the intention is to use the processed Figure 45: Calculator Operations and Spectrum Acquisition
data as part of a quantification procedure. There is a funda- Channels after charge correction
mental requirement for these operations to maintain the in-
tegrity of the data and therefore a limited set of calculation Figure 46 shows a set of spectra in various states of oxida-
options is available to support this end. The principal limita- tion. The spectrum in the left-side display tile ( Figure 46(a))
tion is that the number of acquisition channels must be iden- represents a metallic surface and may be subtracted from the
tical in each spectrum used in the calculation. The data may set of spectra selected in the right-hand side of the same ex-
be charge corrected, but only those channels for which the periment-frame to produce a set of difference spectra show-
charge correction lies within the original range of channels ing the trend in the Al 2p oxide peak. Figure 46(b) illustrates
will the results be meaningful. Figure 45 illustrates how the the result of the subtraction operation shown on the Calcula-
data is adjusted to account for charge shift corrections prior tor Property Page, where each of the spectra selected in the
to performing one of the calculator operations. right-hand-side browser frame is displayed, showing the re-
The Calculator is designed to allow a single spectrum to act sidual Al 2p oxide peak following the Calculator operation.
upon a set of spectra. Any spectra selected in the right hand Note the processing history indicates that the active tile on
(browser) frame will be altered by the operation defined by the left-hand-side shows data that has been modified by
both the Calculator property page and the current data in the “vb1” or VAMAS BLOCK index 1 (where the indices start
active display tile. at vb0).

The Calculator 72
 Using the Calculator and Comparing Spectra

Figure 46: (a) The spectra selected in the browser will be Figure 46: (b) Result of the operation shown alongside.
modified by the spectrum displayed in the active display tile.

Normalizing the data (which are the result of a calculator op- background types are appropriate. Figure 47 shows two
eration) with respect to one another is sometimes necessary. spectra taken from a Cr 2p region. The O 1s plasmon struc-
One such situation arises when a peak lies on a plasmon loss ture is very evident in both spectra and clearly interferes
structure. Chromium oxide supported on SiO2 catalyst sam- with a consistent choice for the background to the Cr 2p
ples produces a spectrum where the O 1s line includes a pro- lines. The first step in quantifying the Cr 2p lines is to charge
nounced loss structure associated with SiO2 which impedes correct one spectrum with respect to the other. Once per-
measurements for Cr 2p owing to the complex nature of the formed the scale factor representing the difference in count
background, particularly when the amount of chromium is rates may be estimated from the relative size of the oxygen
small. If a “standard” spectrum is available representing the peaks and is used as a factor that is applied to the standard
SiO2 catalyst material prior to modification by the Cr, sub- spectrum before subtracting the two Cr 2p data intervals.
tracting the standard may result in data for which common First the data regions for the unknown and the standard must

The Calculator 73
 Using the Calculator and Comparing Spectra

page window. Use the mouse to drag out a box from the
top of one peak to the top of the second. The limits of the
box define the “Measured” and the “True” energies
entered on the Calibrate property page, pressing Apply
will calibrate the two spectra with respect to one another.
3. Once satisfied with the O 1s calibration, the values for
the Measured and True energies may be applied to the
corresponding Cr 2p region. Note that only one of the
two overlaid O 1s spectra will have an entry in their
processing history following the calibration step. The
Cr 2p spectrum from the same file as the O 1s with the
calibration entry is the Cr 2p spectrum for which these
calibration values are appropriate.
Figure 47: Cr 2p regions acquired from a
Figure 51 and Figure 52 show how to define the calculation
catalyst before and after an uptake of Cr.
for removing the underlying trend from the Cr 2p peaks.
be copied into a new experiment frame. In Figure 48, six re-
gions from each experiment have been copied into an empty The quantification report printed over the wide spectrum
file initially called CasaXP1. This file may be subsequently (Figure 48) for the unknown sample shows the results com-
saved with an appropriately changed name. piled using a Shirley background applied to the difference
spectrum for the Cr 2p region. Note that the quantification
The steps used to calibrate the energy scale are as follows report is compiled using the results from the high-resolution
(refer to Figure 49 and Figure 50): spectra even though it is displayed over the survey spectrum.
Without subtracting the reference spectrum the validity of
1. Overlay the two O 1s peaks. the Shirley procedure would be in question for the Cr 2p da-
2. Select the Calibrate option on the processing property ta.

The Calculator 74
 Using the Calculator and Comparing Spectra

Figure 48: Quantification based upon the Cr 2p data after subtracting the corresponding spectrum from a standard sample

The Calculator 75
 Using the Calculator and Comparing Spectra

Figure 49: Calibrate the spectra with respect to one another. The calibration point is determined from the overlaid O 1s
peaks

The Calculator 76
 Using the Calculator and Comparing Spectra

Figure 50: Regions after the calibration has been applied to both the O 1s and the Cr 2p regions using the “Apply
to Selection button on the Calibration Property Page, Figure 46

The Calculator 77
 Using the Calculator and Comparing Spectra

Figure 51: Calculator page setup up to subtract the standard data away from the Cr modified spectrum. The standard is
the spectrum labeled B and must be the spectrum in the active display tile. The modified data (A) must be selected via
the right-hand-side

The Calculator 78
 Using the Calculator and Comparing Spectra

Figure 52: Cr 2p region after the equivalent data from the standard has been subtracted.

The Calculator 79
Principal Component Analysis CHAPTER 10

ance in understanding the model is therefore of great value

Introduction and it is with this end that Principal Component Analysis is
offered as a supplementary tool.
XPS is a technique that provides chemical information about Factor analysis is a field that is as broad as it is deep. It is a
a sample that sets it apart from other analytical tools. How- mathematically challenging tool that requires knowledge of
ever, the key information sought by the analyst is locked into matrix algebra coupled with a feel for a statistical approach
a data envelope and as a consequence the need for powerful to data interpretation. A true understanding for the subject
algorithms is paramount when reducing the data to chemi- can only be obtained by studying the literature and through
cally meaningful quantities. Two approaches have been em- practical experience. Therefore the material presented here
ployed on XPS data: is only an introduction rather than a complete set of works.
• Curve synthesis and fitting (see “Quantification” on
page 42).
• Techniques from multivariate statistical analysis of Theory of Principal Component Analysis
which Principal Component Analysis (PCA) is the most
common form. Factor analysis is a multivariate technique for reducing ma-
trices of data to their lowest dimensionality by use of orthog-
Curve synthesis is probably the most widely used method onal factor space. The challenge is to identify the number of
for data analysis practised by XPS researchers. Unfortunate- significant factors (principal components) and use this infor-
ly, statistically good curve fits are not always physically mation to model the data using techniques such as Target
meaningful and, in many cases, great care must be exercised Transformations or curve fitting.
when choosing the model to describe the data. Any assist- In XPS the data matrix is composed of spectra where each

80
 Principal Component Analysis

acquisition channel is viewed as a co-ordinate in an r-dimen- has been performed by observation. Unfortunately real spec-
sional space; r is equal to the number of acquisition channels tra are not so simple and spotting the linear relationships be-
per spectrum. The problem addressed by PCA is that of de- tween the columns of the data matrix requires a more
termining the number of distinct spectroscopic features sophisticated approach.
present in a particular set of c spectra. PCA, also known as Eigenanalysis, provides a method for
identifying the underlying spectra that form the building
The following example tries to illustrate the nature of the blocks for the entire set of spectra. The data matrix is trans-
problem. formed into a new set of r-dimensional vectors. These new
Consider a set of three spectra; each spectrum has three ac- vectors span the same subspace as the original columns of
quisition channels: the data matrix, however they are now characterised by a set
s1 = ( 4, 3, 6) , s2 = (2, 3, 2) , s3 = (2, 0, 4) of eigenvalues and eigenvectors. The eigenvalues provide a
The data matrix is given by measure for the significance of the abstract factors with re-
spect to the original data. Various statistics can be computed
from these values that aid in identifying the dimensionality
422 of the subspace spanned by the spectra.
D = 330
The procedure for calculating the abstract factors has its
624
roots in linear least square theory. In fact the preferred meth-
These three vectors belong to a 3-dimensional space, how- od is to form a Singular Value Decomposition (SVD) for the
ever they do not span 3-dimensional space for the following data matrix.
reason. If a linear combination of the vectors s1, s2 and s3 is D = USV'
used to construct a new vector v, then v always lies in a plane Where D is the data matrix formed from c spectra, each con-
(a 2-dimensional sub-space of 3-dimensional space). The taining r channels. U is the same dimension as D, while S
fact that v lies in a plane is a consequence of the following and V are c by c matrices. S is a diagonal matrix; the diago-
relationships between the three spectra. nal elements are the square root of the eigenvalues of the
s3 = s1 – s2, correlation matrix
so v = a s1 + b s2 + c s3 Z = D'D
= a s1 + b s2 + c (s1 – s2) The abstract factors are computed from US. The rows of V
= (a + c) s1 + (b – c) s2. are the corresponding eigenvectors of Z; the co-ordinates of
Thus, two principal components exist for the set of three the eigenvectors represent the loading for the abstract fac-
spectra. tors and specify how linear combinations of these factors
The analysis of the data matrix in the above simple example can be used to reproduce the original data. Including all of

Theory of Principal Component Analysis 81

 Principal Component Analysis

the abstract factors with the appropriate loading enables the secondary factors are those factors that can be associated
data to be reproduced to an accuracy only limited by the pre- with the noise and, in principle, can be omitted from subse-
cision characteristic of the Eigenanalysis procedure. quent calculations. It is not possible to completely disasso-
The essential feature of the SVD procedure is to compute the ciate the true data from the error within the measured data,
abstract factors so that the factor corresponding to the largest however the statistics guide the analyst in choosing the most
eigenvalue accounts for a maximum of the variation in the appropriate number of abstract factors that describe the data
data. Subsequent abstract factors are generated such that 1) and therefore the "best guess" dimensionality for the data
as much variance as possible is accounted for by each new matrix.
factor and 2) the newest axis is mutually orthogonal to the In the case of XPS spectra the experimental error is known
set of axes already located. The procedure therefore com- to be the square root of the number of counts in an acquisi-
putes an orthogonal basis set for the subspace spanned by tion channel. Under these circumstances where the experi-
the original data matrix that is oriented with respect to the mental error is known, a number of statistics have been
data in a linear least square sense. proposed for determining the size of the true factor space.
In principle, the number of non-zero eigenvalues is equal to
the number of linearly independent vectors in the original
data matrix. This is true for well posed problems, but even Residual Standard Deviation (“Real Error”)
the presence of errors due to numerical operations will result An alternative name for the residual standard deviation
in small eigenvalues that theoretically should be zero. Nu- (RSD) used by Malinowski is “real error” (RE).
merical errors are an insignificant problem compared to the The RSD is defined to be:
one presented by the inclusion of experimental error in the
calculation. Noise in the data changes the underlying vectors c
so that almost every data matrix of c spectra with r acquisi- ∑ Ej
tion channels, where c <= r, will span a c-dimensional sub- j = n+1
RSD n = ----------------------
r(c – n)
space. This is true even though the underlying vectors
should only span fewer than c dimensions. where Ej is the jth largest eigenvalue, n is the number of ab-
Various statistics are available for identifying the most like- stract factors used to reproduce the data; c spectra each with
ly dimensionality of a data matrix. These statistics are de- r channels are used to construct the data matrix.
signed to aid partitioning the abstract factors into primary RSDn must be compared against the estimated experimental
and secondary factors. The primary factors are those corre- error. If the value computed for RSDn is approximately
sponding to the largest n eigenvalues and represent the set of equal to the estimated error then the first n abstract factors
abstract factors that span the true subspace for the data. The span the factor space. The dimensionality of the original

Theory of Principal Component Analysis 82

 Principal Component Analysis

data matrix is therefore n. size of the eigenvalues and weighted by the co-ordinates of
Two further statistics may be derived from RSDn, namely, the corresponding eigenvectors. The chi square value for a
IEn (Imbedded Error) and INDn (Indicator Function) given set of n abstract factors is computed using:
by:

χ
r c 2
2 ( d ij – d ij )
1 n = ∑ ∑ -------------------------
2
---
2 i = 1j = 1 σ ij
IE n = RSD n  ---
n
 c
where dij is an element of the data matrix, dij is the corre-
And sponding approximation to the data point constructed from
the first n abstract factors with the largest eigenvalues. The
standard deviation for XPS data σij is the square root of dij .
The expected value for each n is given by:-
χn2 (expected) = (r-n)(c-n).
RSD n
IND n = ------------------
-
2
(n – c)
A comparison between the expected value and the computed
value is the basis for determining the number of principal
IEn and INDn are statistics that should decrease as the components. Both χn2 and its expected value decrease as n
number of primary abstract factors is increased. Once all the increases. χn2 initially is larger than χn2 (expected) but as n
primary factors have been included, these statistics should increases a crossover occurs. The true dimensionality of the
begin to increase since at this point factors from the noise data matrix is chosen to be the value of n for which χn2 is
subspace start to interfere with the accuracy of the data de- closest to its expected value.
scription. This minimum is therefore an indicator of the di- Note that smoothing the data will alter the characteristics of
mensionality of the data subspace. the noise. Performing such pre-processing therefore invali-
dates the χn2 statistic.
Chi square
Bartlett proposed using the chi square criterion for situations
similar to XPS data, where the standard deviation varies Target Factor Analysis
from one data channel to the next.
The procedure involves reproducing the data matrix using Principal Component Analysis provides a set of basis vec-
the abstract factors. Each abstract factor is progressively in- tors that describe the original set of spectra. Although useful
cluded in a linear combination in the order defined by the as a means of characterising the data, these abstract factors

Target Factor Analysis 83

 Principal Component Analysis

are in general not physically meaningful. Target Factor trix is reproduced from the real factors (spectra) rather than
Analysis is concerned with identifying vectors that can also from abstract factors and by comparing the results for differ-
describe the data, but with the additional property that they ent sets of real factors, the best TFA solution to a problem
are recognisable as spectra rather than simply abstract vec- can be determined.
tors in an r-dimensional space. Testing a target vector x with respect to the chosen set of pri-
There are numerous methods for transforming the PCA ab- mary abstract factors involves forming the projection t onto
stract factors to provide vectors that are more open to chem- the subspace spanned by the PCA primary abstract factors.
ical interpretation. These involve constructing abstract The predicted vector x, calculated using the co-ordinate val-
rotation transformations that map the abstract factors into ues of t to load the corresponding abstract factors, is com-
one of the infinite number of alternative basis sets for the pared to the original target vector. A target vector that
factor space. Fortunately there is a technique which when belongs to the subspace spanned by the primary abstract fac-
coupled with curve synthesis, lends itself to the analysis of tors should result in a predicted vector that is identical to the
XPS data, namely, Target Testing. initial target vector. Errors in the original data matrix and
similar errors in the measured target vector mean that the
Target Testing predicted and target vector differ from each other as well as
Once a Principal Component Analysis has been performed, from the pure target vector x* (x but without error). Esti-
the mathematical bridge between abstract and real solutions mates for these differences allow a comparison to be made
is Target Testing. Individual spectra can be evaluated to as- between the predicted and target vector and a decision as to
sess whether the corresponding vector lies in the subspace which targets to include in the target combination step.
spanned by the chosen primary abstract factors. The essen- The apparent error in the test vector (AET) measures the dif-
tial feature of Target Testing is to form the projection of the ference between the test and predicted vectors in a root mean
target vector onto the subspace spanned by the primary fac- square (RMS) sense. Similarly two other RMS quantities es-
tors, then compute the predicted target vector using this pro- timate the real error in the target vector (RET) and the real
jection. Statistical tests applied to the predicted and test error in the predicted vector (REP). These error estimates
vectors determine whether these two vectors are one and the form the basis for the SPOIL function defined to be approx-
same. These tests serve as a mean of accepting or rejecting imately equal to the ratio RET/REP.
possible fundamental components of the sample.
Ultimately, the goal of target testing is to identify a set of Principal Component Analysis by Example
spectra that span the same subspace as the primary abstract The first example illustrating the characteristics of PCA uses
factors. Complete models of real factors are tested in the tar- a set of artificial data.
get-combination step. In the combination step the data ma-

Target Factor Analysis 84

 Principal Component Analysis

Three sets of spectra prepared from synthetic components

are used in the PCA. The structure of the artificial data de-
rives from Carbon 1s states within three compounds, name-
ly, PMMA, PVA and PVC (Figure 53). The proportion of
each compound varies throughout each set of ten VAMAS
blocks. The data is located in the files c1stest1.vms,
c1stest2.vms and c1stest3.vms. The underlying trends intro-
duced into each file are as follows: peaks corresponding to
PMMA and PVC obey quadratic adjustments in intensity
over the set of ten spectra (PMMA decreases while PVC in-
creases). The difference between the three files is the pro-
portion of PVA in each data envelope. The first file
(c1stest1.vms) has a constant level of PVA (Figure 54); the
second file (c1stest2.vms) varies linearly, first increasing
then decreasing; the third file (c1stest3.vms) includes a lin-
ear increase in the level of PVA.

The objective is to show how the statistics used in PCA be-

have for a known problem. Data matrices constructed from
the three sets of spectra should have a dimensionality of
three.

Figure 53: (a - c) Artificial C 1s Data

Target Factor Analysis 85

 Principal Component Analysis

Note that, although each compound is constructed from a

number of C 1s peaks (PMMA 4, PVA 4 and PVC 2), the No noise is present in the data; therefore eigenvalues be-
stoichiometry of these compounds masks the true number of longing to the primary set of three abstract factors should be
synthetic components actually present in the data. Hence the non-zero, while the remaining seven eigenvalues should be
dimensionality of the data should be three not ten (4+4+2). zero.
An additional twist to this example is that two of the under-
lying envelopes are similar in shape to each other, though The results of applying PCA to these data sets (Table 6) il-
not identical (see Figure 53 (a) and (b)). lustrate the uncertainty associated in estimating the dimen-
sionality of the data matrix from the statistics. The fourth
largest eigenvalue in each case is small but non-zero. Also
The trend throughout the first data set may be seen in Figure the statistics for IE and IND indicate a minimum at eigenval-
54. ues other than the expected result. Chi square is not a valid
statistic since no noise is present in the data, however it does
show that three abstract factors are sufficient to reproduce
the data to within reason.

Figure 54: c1s_test1.vms C 1s Spectra

Target Factor Analysis 86

 Principal Component Analysis

Table 6: PCA report for file c1stest.vms

Chi-sq
Factor Eigenvalue RMS RE (RSD) IE IND * 1000 Chi-sq Calc.
Expected

C 1s/1 10861690000 266.7944 615.8053 194.7347 7602.534 29055.89 1800

C 1s/2 684568100 13.0963 29.86145 13.35445 466.5852 172.5176 1592

C 1s/3 1433862 0.01064209 0.02964045 0.01623474 0.6049071 0.000190564 1386

C 1s/4 1.230771 0.000391784 0.002107629 0.001332982 0.05854525 9.86356E-08 1182

C 1s/5 0.003712633 0.000385433 0.001279202 0.000904532 0.05116807 1.20367E-07 980

C 1s/6 0.000545838 0.000230365 0.001168993 0.000905498 0.07306205 4.86336E-08 780

C 1s/7 0.000473059 0.000240069 0.001018602 0.000852223 0.113178 5.18512E-08 582

C 1s/8 0.000306354 0.000155331 0.000891207 0.000797119 0.2228016 2.20338E-08 386

C 1s/9 0.000200465 0.00012725 0.00076887 0.000729414 0.7688698 1.91008E-08 192

C 1s/10 0.000118823 4.68061E-13 0 0 0 6.13259E-23 0

It is interesting to see how the eigenvalues change with re- value. This statement has been made with the benefit of a
spect to the three data sets (Figure 55 and Figure 56). The good understanding of what is present in the data. In real sit-
same spectra varied in different ways results in slightly dif- uations such statements are themselves suspect and so re-
ferent orientations for the principal component axes and quire support from other data reduction techniques. For
hence different eigenvalues. example curve fitting using three sets of synthetic peaks all
The PCA statistics IE and IND have implied a dimensional- linked with the appropriate stoichiometric relationships
ity other than three (Table 6). The clue to the correct dimen- would lend support to the hypothesis. Curve fitting such
sionality of the data lies in the relative size of the structures is not an exact science and such fits themselves
eigenvalues. The fourth eigenvalue is in two cases better should be supported by statistics gathered from the fitting
than five orders of magnitude smaller than the third eigen- parameters.

Target Factor Analysis 87

 Principal Component Analysis

Figure 55: Abstract Factors and Eigenvalues. Figure 56: Eigenvalues for c1s_test2.vms Abstract Factors

The second example illustrates the effects of experimental

error on a PCA calculation.

Target Factor Analysis 88

 Principal Component Analysis

Table 7: PCA Applied to Artificial Data with Simulated Noise

Chi-sq
Factor Eigenvalue RMS RE (RSD) IE IND * 1000 Chi-sq Calc.
Expected

C 1s/1 10883620000 267.626 617.9728 195.4202 7629.294 50489.63 1800

C 1s/2 687172800 18.93302 47.75373 21.35612 746.152 3572.687 1592

C 1s/3 2714766 6.90718 26.01386 14.24838 530.8951 483.7729 1386

C 1s/4 946794.7 0.1649431 2.106561 1.332306 58.51557 0.5688374 1182

C 1s/5 5095.574 0.0622169 0.5048702 0.3569971 20.19481 1.247441 980

C 1s/6 226.9936 0.01656361 0.1904914 0.147554 11.90571 0.002773735 780

C 1s/7 29.13143 0.002688086 0.00847462 0.007090376 0.9416245 5.0833E-05 582

C 1s/8 0.03943089 0.000339289 0.003105158 0.002777337 0.7762894 3.05649E-07 386

C 1s/9 0.003409306 0.000206314 0.001523904 0.001445703 1.523904 9.22657E-08 192

C 1s/10 0.000466779 1.26006E-12 0 0 0 7.4912E-22 0

Real data includes noise. The effect of noise on a PCA cal- eigenvalues increase in size and lead to further uncertainty
culation can be see from Figure 57 together with the report with regard to which eigenvalues belong to the set of prima-
in Table 7. The data in the file c1stest1.vms has been used ry abstract factors. Note that the abstract factors in Figure 57
together with a pseudorandom number generator to simulate are plotted in the reverse order to the ones in Figure 55 and
noise that would typically be found in XPS data. The conse- Figure 56
quence of including a random element in the data is that the

Target Factor Analysis 89

 Principal Component Analysis

etch time and the common acquisition conditions provide

data in a form that is well suited to PCA.
An aluminium foil, when profiled using a Kratos Analytical
Axis ULTRA, provides a good example of a data set that can
be analysed using some of the features on offer in CasaXPS.
These data are not chemically or structurally interesting, but
do show how trends can be identified and anomalies isolat-
ed.

Figure 57: c1stest1.vms data plus artificial noise.

Fortunately, the chi square statistic becomes more meaning- Figure 58: Al 2p Depth profile
ful when noise is introduced into the problem. A comparison Figure 58 shows a plot of the Al 2p energy-region profiled
between the computed chi square and its expected values do against etch time. The data envelopes change in shape as the
seem to point to a 3-dimensional subspace. The crossover surface oxide layer is removed by the etch cycles to reveal
between the two quantities suggests the need for three ab- the homogeneous bulk aluminium metal.
stract factors when approximating the data matrix using the It should also be noted from Figure 59 that the data contains
results of PCA. an imperfection. One of the energy scans includes data ac-
quired during an instrumental event. Noise spikes are super-
Principal Component Analysis and Real Data imposed on the true data and these should be removed
XPS depth profiles generate sets of spectra that are idea for before techniques such as curve synthesis are applied. In this
examination via PCA. The spectra are produced by repeat- case the spikes appear in the background and are therefore
edly performing etch cycles followed by measuring the obvious to the eye, however similar non-physical structure
count rate over an identical energy region. The resulting data that occurs on the side of a peak is less obvious and could be
set therefore varies in chemical composition with respect to missed.

Target Factor Analysis 90

 Principal Component Analysis

The first step in performing the Principal Component Anal- for most of the variation in the overall set of spectra. A more
ysis is to define a quantification region for each of the spec- even weighting between the underlying line-shapes would
tra to be used in the analysis. These regions specify the produce abstract factors that are less physically meaningful
acquisition channels that will be used in the data matrix. in appearance.
Also any shifts in the data due to charging can be removed
from the calculation by using an offset in the energy region
for those spectra affected.

Figure 60: Al 2p Abstract Factors

The only real use for the abstract factors is judging their sig-
nificance with respect to the original data. Abstract vectors
Figure 59: Al 2p Spectra that derive from noise look like noise, factors that contribute
Next, select the set of spectra in the Browser View and dis- to the description of the data contain structure. The dividing
play the data in the active tile. The processing property page line between the primary and secondary abstract factors can
labelled "PCA" offers a button labelled "PCA Apply". On sometimes be assessed based on the appearance of the ab-
pressing this button, those spectra displayed in the active tile stract factors.
are transformed into abstract factors. Figure 59 displays the Analysing the Al 2p spectra generates abstract factors and
spectra before the PCA transformation while Figure 60 eigenvalues that represent the PCA fingerprint for the data.
shows the abstract factors generated from the eigenanalysis. Table 8 is a report of the Al 2p data set generated by
CasaXPS and formatted using a spreadsheet program. Each
Note that the abstract factors are truly abstract. The first fac- row of the report is labelled by the ISO block name that con-
tor (Figure 60) looks like an Al 2p metal doublet, however tains the abstract factor corresponding to the listed eigenval-
this is because the Al 2p metal envelope dominates the data ue.
set and therefore a vector having a similar shape accounts

Target Factor Analysis 91

 Principal Component Analysis

Table 8: Report generated by a PCA for Al 2p Profile.

Factors Eigenvalue RMS RE (RSD) IE IND * 1000 Chi-sq(Calc.) Chi-sq(Expected)

Al 2p/3 4727872000 82.5236 159.3106 34.76442 398.2765 26032.99 2600

Al 2p/8 65086870 3.784555 23.78739 7.340948 65.89305 136.4193 2451

Al 2p/13 393178.7 5.875731 20.74929 7.842494 64.04102 133.7949 2304

Al 2p/18 305001.5 3.48073 17.85783 7.793798 61.79182 78.0949 2159

Al 2p/23 156694.8 3.249367 16.25038 7.929371 63.47803 59.6866 2016

Al 2p/28 98359.06 2.757161 15.2192 8.135007 67.64091 51.1035 1875

Al 2p/33 86168.29 2.48836 14.18397 8.189118 72.36718 41.82519 1736

Al 2p/38 65267.54 2.22333 13.35424 8.242415 79.01916 37.98182 1599

Al 2p/43 53613.14 2.247765 12.61316 8.257255 87.59142 42.46731 1464

Al 2p/48 43569.08 0.1744253 11.97161 8.261198 98.93895 0.1869893 1331

Al 2p/53 32387.23 1.710532 11.52946 8.344409 115.2946 29.01139 1200

Al 2p/58 28174.98 2.021671 11.12658 8.410907 137.3652 33.42906 1071

Al 2p/63 24742 1.261896 10.75487 8.461885 168.0448 15.17705 944

Al 2p/68 23980.27 0.40768 10.29759 8.407944 210.1548 1.649807 819

Al 2p/73 20710.42 1.113217 9.867347 8.33943 274.093 10.69842 696

Al 2p/78 18345.12 1.155456 9.424948 8.226769 376.9979 15.75993 575

Al 2p/83 16109.71 0.7358818 8.960655 8.062218 560.0409 5.605129 456

Al 2p/88 13003.76 0.7303461 8.600543 7.962556 955.6159 5.882052 339

Al 2p/93 12307.15 0.7049177 7.99876 7.608339 1999.69 6.193881 224

Al 2p/98 9285.948 0.000443667 7.554815 7.372745 7554.815 2.12747E-06 111

Target Factor Analysis 92

 Principal Component Analysis

Table 8: Report generated by a PCA for Al 2p Profile.

Factors Eigenvalue RMS RE (RSD) IE IND * 1000 Chi-sq(Calc.) Chi-sq(Expected)

Al 2p/103 7476.855 1.90305E-13 0 0 0 2.03675E-25 0

The chi square indicates that the data matrix can be repro- All the statistics except the Indicator Function point to two
duced to within experimental error using two abstract fac- abstract factors being sufficient to span the factor space for
tors. This is a result that is consistent with the physical the data matrix.
nature of the sample. It is also interesting (from a mathemat- It is worth examining the data set using a subset of the spec-
ical standpoint) to note that using all the abstract factors to tra and Target Testing the spectra not used in the PCA. This
reproduce the data matrix returns a chi square of zero (al- allows anomalies to be identified such as spikes in the data.
lowing for round-off errors in the computation). This should Selecting a representative subset of spectra for the PCA then
always be the case and provides an easy check to see that the target testing the remainder is particularly useful for large
calculation has been performed correctly. sets of data.
Table 9: Target Test Report for a Subset of Al 2p Data Set

Target AET REP RET SPOIL Al 2p/3 Al 2p/8

Al 2p/23 20.93032 10.38374 18.17295 1.750135 0.39033 -0.04947

Al 2p/28 23.83028 11.05117 21.11288 1.910467 0.41839 0.017257

Al 2p/33 19.83736 11.47927 16.17861 1.409376 0.42997 0.065749

Al 2p/38 19.8507 12.01348 15.80274 1.315418 0.44268 0.10609

Al 2p/43 19.9069 12.46508 15.52116 1.245171 0.4531 0.133366

Al 2p/48 57.16561 12.70691 55.73546 4.386233 0.45854 0.14688

Al 2p/53 15.37333 13.18052 7.912861 0.600345 0.46791 0.174614

Al 2p/58 21.39836 13.30379 16.76004 1.259795 0.46901 0.184805

Al 2p/63 19.92528 13.5238 14.63296 1.082016 0.47386 0.195062

Al 2p/68 27.73522 13.78354 24.06775 1.746122 0.48087 0.203826

Target Factor Analysis 93

 Principal Component Analysis

Table 9: Target Test Report for a Subset of Al 2p Data Set

Target AET REP RET SPOIL Al 2p/3 Al 2p/8

Al 2p/73 19.10189 13.88023 13.12332 0.945469 0.48192 0.210646

Al 2p/78 20.9575 13.98145 15.61204 1.116625 0.48264 0.218455

Al 2p/83 19.03813 14.15492 12.7314 0.899433 0.48483 0.229382

Al 2p/88 18.38591 14.11378 11.78317 0.83487 0.48374 0.228046

The SPOIL function and AET statistics (Table 9) show that The PCA report in Table 3 includes the spectrum labelled
Al 2p/48 differs in some respect from the other spectra in the Al 2p/48 in the data matrix. The consequence of not remov-
list tested. The spectrum in question corresponds to the trace ing the spikes is apparent in the 3-D factor space shown in
displaying the spikes seen in Figure 58 Also, another spec- Figure 61, where the abstract factor with third largest eigen-
trum that could be looked at is Al 2p/68. The AET value is value clearly contains spikes and the projection point
high compared to the other spectra. Such spectra may high- number 10 derived from the Al 2p/48 spectrum is obviously
light interfaces where either new chemical states appear (ei- a statistical outlier.
ther directly from features in the data or indirectly through
changes in the background due features outside the acquisi- PCA and CasaXPS
tion region) or energy shifts due to sample charging have al- Principal Component Analysis is offered on the "process-
tered the characteristics of the data. ing" window. The options on the property page labelled
"PCA" allow spectra to be transformed into abstract factors
according to a number of regimes. These include covariance
about the origin and correlation about the origin. Each of
these pre-processing methods may be applied with and with-
out background subtraction.
Quantification regions must be defined for each spectrum
included in the factor analysis. In addition, each spectrum
must have the same number of acquisition channels as the
others in the set of spectra to be analysed. The first step in
the calculation replaces the values in each spectrum by the
Figure 61: 3-D factor space result of interpolating the data within the defined quantifica-

Target Factor Analysis 94

 Principal Component Analysis

tion region for the spectrum. This is designed to allow ener- A button labelled "PCA Report" allows the current PCA re-
gy shifts to be removed from the data used in the factor port to be re-displayed. Care should be exercised since the
analysis. values are subject to any additional processing (including
The quantification region also provides the type of back- PCA) that may subsequently be applied to any of the spectra
ground to the spectrum. Performing the analysis on back- included in the original analysis.
ground subtracted data attempts to remove artifacts in the The PCA property page includes a button to reset the
spectrum that derive from other peaks within the vicinity of processing operations for every spectrum displayed in the
the energy region. Background contributions can be signifi- active tile. This allows a PCA calculation to be undone in
cant in PCA. Additional primary abstract factors are often one stroke. It will also undo any processing previously per-
introduced as a consequence of changes in the background formed on the data. PCA is aimed at the raw data; the chi
rather than the underlying peaks within the region of inter- square statistic is referenced to the raw data and has an un-
est. The presence of such abstract factors can be viewed as defined meaning when the data have been processed prior to
information extracted from the data, although in many cir- performing factor analysis.
cumstances they can lead to incorrect synthetic models if Target Factor Analysis in the form of target testing is also
background contributions are misunderstood. available on the PCA property page. Following a PCA, can-
A factor analysis is performed on the set of spectra displayed didates for the physically meaningful components may be
in the active tile. Although PCA is offered as a processing assessed individually or collectively. Choose an abstract
option, it is the only processing option that acts on a collec- factor from the PCA and entering this factor into the active
tion of spectra. Any other option from the processing win- tile. Then select the number of primary abstract factors for
dow would only act upon the first VAMAS block in a use in the target test procedure. A text field is offered on the
selection when that selection is displayed in a single tile. PCA property page for this purpose and is found in the sec-
The principal component analysis is performed when the tion headed "Target FA". Next, select the target test spectra
"Apply" button is pressed. Each spectrum displayed in the in the Browser view and press the button labelled "TFA Ap-
active tile is replaced by the computed abstract factors. The ply". A report detailing the statistics calculated from the
order of the VAMAS blocks containing the spectra is used TFA procedure will appear in a dialog window.
as the order for the abstract factors. The factor correspond- The TFA report may be written to file in an ASCII format
ing to the largest eigenvalue is entered first. Subsequent with TAB separated columns. When pressed, any of the but-
blocks receive the abstract factors in descending order de- tons above the columns on the report will display a file dia-
fined by the size of the corresponding eigenvalues. A report log window from which the output text-file can be specified.
showing the statistics for understanding the dimensionality This method for saving a report to file is used by the PCA
of the factor space appears in a dialog window. report (above) and the Linear Regression Report described

Target Factor Analysis 95

 Principal Component Analysis

below.
Once a set of target spectra has been identified, these spectra
can be used to reproduce the original set of spectra through
a linear regression step. Enter the set of target spectra into
the active tile; then select the original spectra in the Browser
view. Press the button labelled "Linear Regression". A re-
port shows the RMS differences between each of the origi-
nal spectra and the predicted spectra calculated from a linear
combination of the set of target spectra displayed in the ac-
tive tile. The loading used to compute the predicted spectra
Figure 62: Projection onto Abstract factors 1,2, and 4
are listed in the report. The report may be written to file us-
ing a similar procedure to the TFA report described above.
The abstract factors defining the axes are graphed together
with a list of the co-ordinate values for each of the spectra
Viewing the Data in Factor Space
projected onto the subspace spanned by the chosen abstract
CasaXPS offers an option on the "Geometry" property page factors (Figure 62) A 3-dimensional plot provides a visual
on the "Tile Display" dialog window labelled "Factor interpretation for the spectra. Patterns formed by the spectra
Space". If selected, the VAMAS blocks displayed in a tile highlight trends within the data set and the relative impor-
are used to define the axes for a subspace and the original tance of the abstract factors can be examined. A plot in
data are plotted, if possible, as a set of co-ordinates with re- which the axes are defined by unimportant factors generally
spect to these axes. The plot represents a projection of the appear random, while factors that are significant when de-
data space onto the subspace defined by a set of two or three scribing the data typically produce plots containing recog-
abstract factors. nisable structure.

Target Factor Analysis 96

Monte Carlo Methods
- Uncertainties in Intensity Calculations CHAPTER 11

Optimization routines are completely deterministic, in that ess, and these true values will lie inside the region of the N-
the same set of parameters, used in the same functional dimensional parameter space defined by the set of outcomes
forms, applied to the same set of data with the same initial to this sequence of experiments. Obviously, if the synthetic
and final conditions will result in the same set of parameters model does not describe a set of parameters in tune with na-
on termination. Vary any of the above conditions and the re- ture, the results may be in more doubt than the measured dis-
sult from the optimization routine will change in some re- tribution might suggest. However, given that all is well then
spect. One method for assessing the uncertainty in the the task is to offer a means of understanding the uncertain-
parameters for a peak model is to vary these optimization ties in the peak parameters within the context of these pa-
conditions by repeating an experiment on, what are hoped to rameter distributions.
be, identical samples. Then for each set of data apply the
same optimization routine to the same synthetic model and Peak identification in XPS spectra represents a challenge
so determine a distribution for the set of parameters used to since synthetic models more often than not involve overlap-
quantify a sample. Such a procedure will vary almost every ping line-shapes (Figure 63), the consequence of which is
aspect of the measurement process and so result in a distri- correlated optimization parameters. That is to say, if a single
bution for the parameters that truly represent the nature of data envelope results from two overlapping peaks and if one
the experiment. of these underlying peaks is reduced in intensity then in or-
der to describe the same data envelope the other must in-
The basis for such an approach as described above lies in the crease in intensity. Adjustments to peak parameters of this
assumption that there exists a set of parameters (only known nature are inherent to any optimization procedure and the
to nature) that does not depend on any optimization routine choice between the possible combinations of these peak in-
nor any other errors introduced into the measurement proc- tensities is made based upon a chi square or root mean

97
 Monte Carlo Methods - Uncertainties in Intensity Calculations

square metric. The problem is therefore to identify the point artificial data. The results represent a distribution for each of
at which these goodness-of-fit metrics fail to produce results the parameters from which an error matrix and tabulated pa-
that can be believed, and provide some means of illustrating rameter distributions can be extracted.
the degree of uncertainty.
Error Matrix
Given a distribution for each of the parameters, the best way The result of a Monte Carlo simulation for a peak fit takes
to describe the nature of the uncertainties is to offer an error the form of m optimization parameters from each of the n
matrix as well as a tabulated form for the distributions. The simulated data sets. An error matrix derived from these pa-
error matrix provides numerical values from which the de- rameter distributions is defined to be an m x m matrix as fol-
gree of correlation can be assessed while scatter plots taken lows:
from some subset of these distributions allows visual inspec-
tion for the same information. Ideally a single number for n

∑ ( xsik – xˆ i ) ( xsjk – xˆ i )
1
each parameter would provide the simplest means of stating e ij = -------------
n–m
k=1
the uncertainties, but as the old adage goes “To every com-
plex problem there is a simple solution that is wrong”, and
so it is for correlated parameters. The unfortunate fact is that Where the xsik are parameter values calculated for each of
if the peaks weren’t correlated then synthetic models would the simulation steps and each distribution is centered with
be unnecessary. respect to the mean rather than the initial parameter value.

The standard error in each parameter is given by eii and the

e ij
Monte Carlo Data Sets correlation between parameters i and j is given by ---------------
-
e ii e jj

Repeating an experiment many times is not always practical An alternative method for estimating uncertainties in the
and for most XPS spectra peak models are developed based peak parameters is to quote the inverse of the Hessian matrix
upon a single acquisition sequence. Estimates for uncertain- used in the Marquardt Levenberg optimization routine. So
ties in the peak parameters must be made using assumptions the question is, why bother with Monte Carlo simulation
about the noise typical of XPS spectra. The essence of Mon- when this very information desired is offered by the optimi-
te Carlo methods for estimating uncertainties is to take a zation procedure itself? The answer comes in two forms.
data set, remove the noise from the data, then repeatedly add Firstly, the error matrix derived from Monte Carlo is not
noise to the synthesized data to generate a set of simulated limited to the problem of just fitting the peaks. The simula-
experimental results and apply the set of operations to that tion procedure can also take account of the background as

Monte Carlo Data Sets 98

 Monte Carlo Methods - Uncertainties in Intensity Calculations

well as the peaks, and so determine a global error matrix for

the calculation rather than one that focuses on the stability of Monte Carlo: A Simple Example
the peak parameters alone. Secondly, the Hessian matrix has
a tendency to be near singular when the Marquardt algo- The best way to introduce the ideas behind Monte Carlo
rithm terminates. Indeed the optimization routine follow a methods and how the results can be interpreted is to look at a
sequence of steps that switch between a direct solution of the simple example. (Note that the “Monte Carlo” button may
be found on the Components tab of the Quantification mod-
Hessian matrix and a solution of a regularized form of the
ule)
matrix. The regularization step occurs when the direct Hes- Consider the data envelope in Figure 63. The spectrum is a
sian inversion fails to produce an improvement in the fitting synthetic envelope created from two GL(50) line-shapes
parameters and a common reason for this to occur is that the without any background and where the peaks are separated
Hessian is ill conditioned. Under these circumstances the co- by an energy gap consistent with C 1s lines in a PVC spec-
variance matrix derived from the Hessian should not be trum (this is actually the PVC data available from the “Test”
trusted. tab of the Processing module). In the absence of noise and
experimental error the optimization routine always returns
the peak parameters tabulated on the spectrum in Figure 63.
The next step in the simulation is to introduce noise onto the
data envelope that is consistent with noise found in experi-
mental XPS spectra, i.e. variations about the channel inten-
sity of magnitude related to the square root of the counts.
Figure 64 shows the data envelope from Figure 63 after
noise has been added and the peak parameters refitted.
This procedure yields the first set of simulation results. If re-
peated many times, the output is six distributions, one for
each of the peak parameters involved and these can be used
to assess the influence of experimental noise on these quan-
tities. Note that this differs in some respects from adopting a
purely experimentally determined parameter distribution.
The initial stating point for the peak parameters will not be
Figure 63: Simulated PVC C 1s Data Envelope. identical for an experimental data set since the experimental
data may be subject to sample charging, and any errors in the

Monte Carlo: A Simple Example 99

 Monte Carlo Methods - Uncertainties in Intensity Calculations

measurement procedure that can not simply be described by

the random nature of the counting system will be omitted. A
Monte Carlo simulated data set only tests the stability of a
model with respect random noise and therefore may neglect
other factors present in an experimental data set.
Table 10 is the error matrix that results from a Monte Carlo
simulation for the spectrum and synthetic model in
Figure 63.
If viewed as uncorrelated error distributions (which is cer-
tainly not true) then one standard deviation in each of the pa-
rameters is given by the square root of the diagonal elements
in the matrix.

Figure 64: Simulated C 1s data where noise has been introduced

to the system.

Table 10: Error Matrix for the C 1s spectrum in Figure 63.

1:Area 1:Pos’n 1:FWHM 2:Area 2:Pos’n 2:FWHM

1:Area 39.230 0.122 0.196 -31.512 0.128 -0.186

1:Pos’n 0.122 0.001 0.001 -0.120 0.000 -0.001

1:FWHM 0.196 0.001 0.002 -0.185 0.001 -0.001

2:Area -31.512 -0.120 -0.185 38.330 -0.125 0.202

2:Pos’n 0.128 0.000 0.001 -0.125 0.001 -0.001

2:FWHM -0.186 -0.001 -0.001 0.202 -0.001 0.002

Monte Carlo: A Simple Example 100

 Monte Carlo Methods - Uncertainties in Intensity Calculations

It may be seen that the areas of the two peaks are anti-corre-
lated, where the quantity –31.512 / √(39.230 x 38.330) char-
acterizes the degree of interaction between the two
parameters. The minus sign indicates the parameters are
anti-correlated. A scatter plot constructed from the two dis-
tributions for the peak areas is shown in Figure 65, where
the anti-correlation is obvious from the bias in the scatter
along a direction 135° to the positive x axis. An alternative
scatter plot (Figure 66) for the area and FWHM parameters
taken from peak C 1s 1 (Figure 63) shows that these two pa-
rameters exhibit positive correlation.
The uncertainty for correlated parameters can be estimated
from plots such as those shown in Figure 65 and Figure 66.
By projecting horizontal or vertical lines onto the parameter
axes in such a way that the appropriate number of points lie Figure 65: Scatter plot showing the anti-correlation between the
between the projection lines for a given confidence limit, the peak area parameter distributions.
uncertainty for a parameter can be assessed in the context of
others. Again, the uncertainty for a parameter should be All the parameter distributions are reported relative to the in-
viewed in the context of all other correlated distributions, itial parameters used to define the data envelope. As a result
and yet the ellipsoid in 6-dimensional parameter space is dif- the relative parameter distributions may not be symmetrical
ficult to quantify; the procedure based on scatter plots about the initial value and asymmetric confidence intervals
should be seen as merely a step in the right direction rather are possible. Note that the error matrix is calculated from the
than arriving at the ultimate destination. In the case of the distributions centered on the mean parameter value for the
peak area shown in both Figure 65 and Figure 66, the esti- distribution, not the initial values.
mate taken from the diagonal element of the error matrix
would seem to be reasonable. Both scatter plots show that One of the real advantages of using Monte Carlo error anal-
about 65% of the points lie inside projection lines positioned ysis is that it highlights when a quantification parameter is
at about 0.95 and 1.05. This interval represents about ±5% poorly determined by the combination of model and optimi-
of 114.2 CPSeV (see Figure 63) and is not too different from zation procedure. It also allows the influence of constraints
the uncertainty taken from the error matrix ±6.2 CSPeV. within a model to be evaluated. Adding information about
chemical shifts, relative peaks widths and/or peak areas can

Monte Carlo: A Simple Example 101

 Monte Carlo Methods - Uncertainties in Intensity Calculations

using combinations of intensities from integration regions

and synthetic components. The purpose of this section is to
point out some of the less obvious consequences that result
from the specification of the background associated with the
recorded peaks.

Monte Carlo End Point Determination

Consider an all too common situation where a pair of unre-
solved peaks is the subject of interest and the total intensity
is the value that is calculated from the data.
The data in this example (Figure 67) are purely synthetic;
two Gaussian peaks of equal intensity and equal FWHM of
Figure 66: Scatter plot between Area and FWHM parameters for 2eV are offset by 1eV on a linear background. This provides
Peak C 1s 1 in Figure 65
an envelope open to analysis by Monte Carlo simulation that
alter the manner in which noise adjusts the parameters from will provide visual insight into the errors associated with
their initial value. Monte Carlo derived scatter plots can of- noise in the data.
ten help to understand how rigid models based upon chemi-
cal knowledge can reduce the range of outcomes for a given Integration Region Limits
set of parameters.
The purpose of this analysis is to illustrate the consequences
of choosing inappropriate end points for the integration re-
Quantification gion. Note that the peak-fitting procedure is not at issue
since only the data are used to calculate the intensities for the
Quantification of AES/XPS spectra is routinely performed combined peaks.

Quantification 102
 Monte Carlo Methods - Uncertainties in Intensity Calculations

. Each calculation includes determining a new linear back-

Figure 67: Two “unresolved” peaks Figure 68: Peak areas from one end point

ground from the information in the simulated noise. In

Figure 66 shows a scatter plot where each point is a normal-
ized area plotted against simulation index. These areas are Figure 68 the background is computed using only one point
calculated from the original data envelope after the introduc- at each end of the integration region while Figure 68 is the
tion of pseudorandom data and the computed area is refer- equivalent experiment but in this case an average of 21
enced to the initial area before noise was superimposed on points defines the background limits.
the spectrum
It is clear from the two scatter plots that the intensity calcu-
lation using just one data point at each end of the integration
region is much more sensitive to noise. The introduction of
more data channels when determining the background im-
proves the situation.

Quantification 103
 Monte Carlo Methods - Uncertainties in Intensity Calculations

Intensities Determined by Peak-Fitting

Two peaks of equal width and area but separated by 1eV

represent an intractable problem for a peak-fitting algorithm
if the intensity of the individual peaks is required and no fur-
ther information is available that can be used to provide con-
straints for the peak fitting parameters.
Figure 70 shows a Monte Carlo simulation for the peaks
shown in Figure 67. Each Monte Carlo simulation step takes
the envelope for the two peaks plus the background then su-
perimposes a normal deviate of mean zero and standard de-
viation one scaled by the square root of the counts in each
data channel. The envelope constructed in this way simu-

Figure 69: Peak areas - end points averaged

Not all peaks are isolated and so under these situations it

falls to the acquisition phase to improve the statistics for the
data channels that determine the background. An acquisition
system that accurately determines the points used to define
the background will yield intensities with significantly more
precision than those that treat the end point as equal to any
other in the acquisition region15.

15. K. Harrison and L.B. Hazell, Surf. Interface Anal. 18, 368
Figure 70: Linear background using single end points..
(1992)

Intensities Determined by Peak-Fitting 104

 Monte Carlo Methods - Uncertainties in Intensity Calculations

lates a sequence of experiments performed on the same sam- In both cases the background used for the Monte Carlo sim-
ple, so that the only variation in the data is the random noise ulations shown in Figure 68 and Figure 71 are determined
inherent in the acquisition process. For each Monte Carlo using one end point for the integration region. The asymme-
envelope the peak parameters are refitted. The points used in try in the distribution derives from the errors introduced by
the scatter plot are determined from the intensities for the the background calculation that occurs prior to performing
two peaks after automatically fitting the peak parameters the peak fitting procedure. A comparison of the distributions
suggests that the intensity determined form the peak fitting
(Area, FWHM and Position) to the new data envelope; these
procedure will be more accurate than the integration region
intensities are normalized with respect to the initial values
approach although there is a tendency to over estimate the
for the peak areas. It is apparent from Figure 70 that the un- total area when the sum of two fitted peak intensities is used
certainty in the individual peak intensities may be in error by and the background calculation is sensitive to noise. Again,
as much as 100% although the total area is determined much the use of more acquisition channels to define the integration
more accurately (Figure 71). The scatter in the computed in- region end points improves the precision in the results for in-
tensities in Figure 71 should be compared with that for the tensities calculated from the sum of the fitted peak areas
intensities shown in Figure 68. (Figure 72).

Figure 72: Linear background with twenty-one end points.

Figure 71: Total intensity, linear background from one end point. In addition, when the uncertainty in the background is re-
duced, asymmetry in the distribution of area ratios is im-

Intensities Determined by Peak-Fitting 105

 Monte Carlo Methods - Uncertainties in Intensity Calculations

proved dramatically (compare Figure 71 and Figure 72).

Summary
Introducing additional information into the background cal-
culation does not improve the uncertainty in the intensities
for the individual peaks determined by a peak fitting proce- Monte Carlo simulation provides a valuable tool for under-
dure, but it does remove some of the correlations in the peak standing the uncertainties associated with data reduction for
intensities due to the background. These correlations are vis- AES/XPS spectra. The results presented here are available
ible in Figure 70 at the extremes of the distribution but are in other forms [15] [16] [17] however the visual feedback of-
not present when the same experiment is performed but with fered by Monte Carlo simulations[17] provides an insight for
analysts who feel less comfortable with a mathematical de-
an improved background specification (Figure 73).
scription of the same concepts. By using Monte Carlo simu-
lation, an understanding of the errors involved in extracting
information from AES/XPS spectra is hopefully opened to a
wider audience.

16. P.J. Cumpson and M.P.Seah, “Random Uncertainties in AES

Figure 73: Individual intensities - Linear background with twenty- and XPS: Peak Energies, Areas and Quantification.” NPL
one end points. Report DMM(A) 26’,May 1991; Surf. Interface Anal. 18,
345 (1992) and 18, 361 (1992).
17. S. Evans, Surf. Interface Anal. 18, 323 (1992).

Summary 106
Analytical Applications CHAPTER 12

The following sections provide examples of the treatment of 74) includes structure that offers chemical information
particular analytical problems in greater depth, emphasising about a sample, but without some initial starting point it is
the special tools and procedures available within CasaXPS difficult to construct an appropriate model for the data enve-
as appropriate. It is intended that these examples should pro- lope.
vide a greater insight into some of the powerful features of
The high-resolution spectrum in Figure 74 derives from poly
the system and an indication of a possible analytical ap-
(acrylic acid) PAA reacted with inorganic material (or par-
proach which may be applied effectively to certain prob-
tially reacted) in an acid-base reaction. If Gaussian/Lorentz-
lems. It is not suggested that these are the only aproaches
possible or available - even within CasaXPS - but merely ian (GL) line-shapes are added in an arbitrary way, the curve
that these indicate known, good methodology. fit yields little information about the sample other than to
say that it deviates from the published data for PAA and
therefore demonstrates the presence of additional chemistry
Organic Polymers and Curve Fitting at the surface. Figure 75 shows a synthetic model for this
data envelope where three GL peaks have been added, then
fitted using a Marquardt-Levenberg20 optimization algo-
The availability of data from well-characterized samples rithm. No constraints have been applied and the result is a
such as those offered by Beamson and Briggs18 owe much reasonable fit to the experimental data but the fitting param-
to the popularity of XPS as a tool for understanding the
chemistry of polymers. A typical C 1s envelope19 (Figure 19. Jones F. et al, “Fluoride uptake by glass ionomer cements: a
surface analysis approach”, to be published
18. Beamson G. and Briggs D., “The XPS of Polymers Database 20. Press W.H. et al, “Numerical Recipes in C”, Cambridge Uni-
- CD-ROM”, Surface Spectra Ltd (2000) versity Press (1988)

107
 Analytical Applications

Figure 74: C1s high-resolution spectrum taken from a polymer- Figure 75: An initial fit to the C 1s spectra shown in Figure 74
based sample using an ESCALAB 220i at University College, London.
PAA. The PAA stoichiometry is still doubtful but the essen-
eters are not readily open to chemical interpretation. For ex-
tial positions for a pure PAA envelope (as indicated by
ample, the FWHM are much bigger than would typically be
expected from C 1s profiles the given instrumental resolu- Beamson and Briggs) have more or less appeared. Further
tion. The three-peak model suggests further deconvolution input is required to make sense of the new synthetic model.
is required before the sample can be fully understood.
To constrain a parameter so that it does not adjust during an op-
If it is assumed that the FWHM for a C 1s photoelectron timization step, set the constraint interval to have the same value
peak is 1.1 eV (only a guess), then applying peaks with said as the fixed value: e.g., if it is required to set the FWHM value to
constraint results in a peak fit shown in Figure 76. Two ad- 1.1, the constraint interval should be entered with the value
ditional peaks appear in the model and further more three of “1.1,1.1”. Actually, it is sufficient to set the parameter value out-
the peaks look like they may have something to do with pure side the constraint range currently defined for the parameter.

Organic Polymers and Curve Fitting 108

 Analytical Applications

Spectra contained in the Beamson and Briggs polymer data- To link a component parameter the constraints must be adjusted
base offer the opportunity to examine more complex poly- as follows. Each synthetic component defined on the Quantifica-
tion Parameter dialog, Components Property Page, appears as a
mer data in the context of known synthetic models. Figure column of parameters in the scrolled list shown in Figure 76.
77 is an example of such data where a set of three line- These columns are headed “A”, “B”, “C” and so on. To con-
shapes has been used to model the clean PAA C 1s data en- strain the area of the component in column B to be half of the
velope. The important feature is that the stoichiometry and area of the component in column A, the area constraint in column
chemical shifts for the C 1s lines are incorporated into the B should be set to “A * 0.5”. Similarly, to offset a component in
column C by 0.2 from the component in column B, enter “B+0.2”
model and this information is then transferable to other pol- in the position constraint field in column C.
ymer spectra. The peaks in Figure 77 are linked in area, but
only the position of peak “C 1s b” and “C 1s a” are con-
strained by an offset.

Figure 77: C 1s envelope from clean PAA acquired on a

Scienta-300, RUSTI, Daresbury Laboratory UK.
Figure 76: Same C 1s envelope as Figure 75 but the synthetic peaks
are all constrained to have FWHM equal to 1.1 eV.

Organic Polymers and Curve Fitting 109

 Analytical Applications

The peak shapes from a Scienta ESCA-300 may differ from

a VG ESCALAB 220i (the source of the real data in Figure
74) or any other manufacturer’s instrument, but the essential
structure should be suitable as a basis for the new model.
Copying the Beamson and Briggs pure PAA model into the
data in Figure 76 leaves a residual that requires an adjust-
ment for the two non-PAA peaks together with the introduc-
tion of a third peak. The new peak in Figure 78 is
constrained to be the same width and position as the saturat-
ed PAA C 1s peak located at 285 eV (BE). The area of this
new peak is allowed to adjust at will and accounts for carbon
with the same characteristics as the PAA peak at that posi-
tion. The consequence of introducing the new peak is that
the PAA synthetic model can adjust without breaking the
stoichiometric relationships for pure PAA, while differences
in the intensity of the saturated peak from the PAA structure
are allowed for by this additional constrained component
(Figure 78).
Synthetic models such as the one in Figure 78 can be tested
using Monte Carlo simulation techniques21 to assess the sta- Figure 78: Final form for the synthetic model. The Glass Ionomer
bility of the peak parameters with respect to noise in the da- Cement (GIC) [2] C 1s envelope containing three peaks from PAA plus
three additional peaks not seen in a clean PAA spectrum.
ta. Once a set of peaks and constraints has been developed,
the Monte Carlo procedure simulates repeated identical ex-
highlight the way noise influences the data model. Figure 79
periments on the same sample, and for each simulation a
is a scatter plot for the normalized peak areas of the non-
new fit is determined for the peak parameters. The result of
PAA peaks against the saturated C 1s peak from the pure
this procedure is a set of distributions for the individual pa-
PAA model. The peaks at 288.4eV and 285eV have areas
rameters from which scatter plots may be constructed that
that are anti-correlated with the pure PAA model (all the
21. Cumpson P. J. and Seah M. P., “Random Uncertainties in peak areas within the PAA model are constrained to one an-
AES and XPS”, Surface and Interface Analysis, 18 361 other,) but the synthetic component at 286.54eV shows a de-
(1992) gree of correlation with the intensity of the pure-PAA area.

Organic Polymers and Curve Fitting 110

 Analytical Applications

At first glance the behavior of the GIC 2 peak at 286.54eV experimentalist and may need changing in view of other in-
is unexpected since one might think that two peaks next to put, however such a model is only possible when peak fitting
one another should produce anti-correlated area distribu- routines offer mechanisms for fixing parameters with re-
tions. The constraints have altered the concept of “next to” spect to one another. The role played by the pure PAA syn-
since the PAA sub-model spreads across the entire envelope thetic model is that of a foundation shape from which
and it becomes difficult to judge by eye what the influence differences in the unknown polymer can be assessed. These
of noise might be on the final result. This type of insight can additional peaks may still require further interpretation, but
only help to understand what constraints do to a fitting pro- with the aid of chemical knowledge and supporting evidence
cedure as well as provide a rule-of-thumb estimate for error a meaningful model can emerge from seemingly intractable
bars (multi-dimensional error distributions can seldom be data.
described by a single number.)

Quantification using “Tagged” Regions

Introduction
Routine measurements using XPS often involve acquiring a
wide scan spectrum to determine the general composition of
a sample. The survey spectrum is then used to select narrow
energy regions where detailed structures are present. These
narrow scan spectra are typically needed when overlapping
peaks are responsible for the data envelope such as is often
seen in the case of C 1s spectra. Synthetic line-shapes must
be used to extract the chemical-state information in the data,
but intensities determined from these models are not always
Figure 79: Monte Carlo simulation results for normalized peak areas.
The three peaks associated with GIC are plotted against the saturated C 1s comparable with intensities determined from integration
peak area from the pure PAA model (wide scan) regions. The most notable case is when asym-
metric line-shapes are used to model the data. Moreover, the
transmission characteristics of an instrument require that
The exact meaning for the model in Figure 78 is left to the each operating mode must be characterized by a transmis-

Quantification using “Tagged” Regions 111

 Analytical Applications

sion function and although it is possible to correct the data

for such differences, problems may arise due to changes in
the system from aging or tweaks to lens functions by the op-
erator. Both require recalibration of the instrument, without
which peak areas from a wide scan acquired using one pass
energy cannot be used with intensities measured via other
modes.
If a spectrometer is characterized by a set of relative sensi-
tivity factors (RSF’s) applicable only to the survey mode of
an instrument, but the resolution of the survey mode is insuf-
ficient to provide the desired chemical state information,
then one way to provided a more detailed quantification is to
reference the intensities from the to high-resolution spectra
to this survey mode. Peak areas determined from the high-
resolution data provide the relative proportions for the
chemical-state intensities and therefore the corresponding
concentration from the survey data can be subdivided using Figure 80: C 1s high-resolution data used to proportion the quantification
these proportions, thus allowing detailed quantifications for C 1s in the survey spectrum.
without the need for transmission correction. Such a proce-
dure effectively takes out the transmission function from the information lost by the limited resolution used for the survey
quantification step and lessens the need for the time consum- spectrum. In the case of relatively pure samples, such as the
ing calibration of each operating mode. Indeed, for many in- one in Figure 80, it might be possible to run a set of high-
struments transmission function correction is not available resolution spectra for each of the peaks identified from the
and so this procedure represents the only way to combine survey data, but many samples include so many elements
data from different operating modes, which would otherwise that full quantification via high-resolution spectra would be
require the maintaining of multiple sets of RSF values. both time consuming and costly. Degradation of a sample
from exposure to X-rays and financial considerations can
“Tagging” regions in a survey scan make the use of high-resolution spectra unattractive as a rou-
tine analysis regime.
Figure 80 shows a typical situation from an XPS measure-
ment where a survey spectrum provides the overall elemen- The data in Figure 80 are quantified via a table that shows
tal quantification but a narrow C 1s scan offers a wealth of the elemental concentrations plus a further breakdown of the

Quantification using “Tagged” Regions 112

 Analytical Applications

C 1s elemental composition into chemical state concentra- tween the components based upon tags, nor do any of the
tions. The proportions determined from the peak model for annotation options involving quantification tables unless the
the high-resolution envelope are used to show how the ele- “Use Tag Field” checkbox is ticked. The tag mechanism is,
mental concentration is subdivided into chemical-state in- however, always used when the Combined button on the Re-
tensities. This is achieved by assigning a “tag” to each of the port Spec property page from the Quantification Parameters
quantification items used in the analysis. That is to say, in dialog window is selected. No other quantification reporting
addition to the name field in each integration region or syn- options use tags.
thetic component there is a Tag entry (in the form of a
string), which is used to link intensities from synthetic peaks “Tagging” for different line shapes
to a specific integration region. Although the primary reason for using tagged reports is to
If the information is available, CasaXPS will enter into the reference results from peak fits to survey spectra, it is possi-
Tag field the element and transition associated with the ac- ble to use the same mechanism to remove some of the ambi-
quisition, and only those items with tags are included in the guities from comparisons between intensities derived from
quantification table. To remove an item from the quantifica- peak fits that employ different line-shapes. Any line-shape
tion step delete the string from the Tag field and press return. involves a functional form that may extend beyond the ac-
The system will enter the key word NoTag when return is quisition regions to which it is applied, especially when
pressed to indicate that the item is not to be used in the quan- asymmetric peaks are involved. The tag mechanism allows
tification. Note that the NoTag entry is necessary to exclude the peak fitted results to be referenced to the underlying in-
from the elemental concentration table the integration region tegration region and so intensities for the chemical states are
defined for the high-resolution scan seen in Figure 80. The calculated based upon the data rather than the implementa-
background to the peak fit for the C 1s spectrum requires a tion of the synthetic components.
region to be defined, but the intensity from this region must Figure 81 shows a set of quantification tables. These tables
not be included in the results from the survey spectrum. In derive from components and integration regions taken from
this example the tag field in the region used to define the the two high-resolution scans below the survey spectrum.
background to the peak fit must be set to NoTag, while each Note that the Al 2p doublet is fitted using a Doniach-Sunjic
of the synthetic components are tagged with the same name asymmetric line-shape, while the O 1s spectrum is modeled
as the corresponding region in the survey quantification, using symmetric Voigt approximations. The table headed
namely, “C 1s”. “Components Only” shows the atomic concentrations where
Not all forms of quantification tables use tags. For example, the intensities are calculated from the line-shapes. These
the Component annotation option used to display the com- concentrations do not agree with the elemental results calcu-
ponent table in the upper spectrum does not discriminate be- lated from the integration regions alone, where the Al 2p

Quantification using “Tagged” Regions 113

 Analytical Applications

models.

Trend Analysis for Metal Oxidation

Introduction
Many XPS/AES experiments are performed to monitor how
the chemical composition of a sample varies with one or
more variables. These variables may be time, tilt angle of a
sample (depth), ion gun etch-time or any other quantity that
characterises the state of an experiment. The challenge is to
devise methods for identifying changes in the spectra and
therefore provide a way of following these changes as a
function of the experimental variable.
The usual tools available for reducing XPS/AES spectra are
simple quantification regions and/or synthetic spectra deter-
mined by optimisation procedures with various forms of pa-
Figure 81: Tag mechanism applied to differing line-shapes. rameter constraints. Implicit in both of these data reduction
The Al 2p has been modeled using Doniach-Sunjic profiles, while the methods is the introduction of a background approximation
O 1s spectrum is fitted with Gaussian-Lorentzian line-shapes.
and in the latter case rigid models for the synthetic line-
concentration is computed to be 85.38% while components shapes. The consequence of the assumptions involved in
only reports 88.24% for the same quantity. The differences these data reduction techniques is hard to assess with respect
to the results for individual spectra. Techniques for estimat-
are due to the cutoff criterion used to limit the infinite area
ing the errors due to noise in the data are available for both
under a Doniach-Sunjic profile. Using the tag mechanism peak-fitted parameters and quantification region values, but
the peak fit can be used to subdivide the concentrations from there is still the question regarding what influence these as-
the integration regions using the individual components sumptions have on the trend itself.
from the peak models. The concentrations for the synthetic An alternative means of following a trend through a data set
components are now consistent with the results of the inte- is to perform a Principal Component Analysis (PCA). This
gration regions but are in the proportions determined by the technique is a linear least square procedure that transforms a

Trend Analysis for Metal Oxidation 114

 Analytical Applications

set of spectra (when viewed as vectors) into a set of orthog-

onal basis vectors (abstract factors) that span the same sub-
space as the original data set. Trends within the original
spectra can be assess by examining the so called abstract
factors and plotting the co-ordinates of the spectra with re-
spect to the principal abstract factors as a function of the ex-
perimental variable involved. Here the term principal
abstract factor means those abstract factors that contain sig-
nificant structure and are necessary for providing an ade-
quate description for the entire set of spectra. Plotting the co-
ordinates for the spectra with respect to the subspace
spanned by the principal components shows how each of
these designated abstract factors contributes to the descrip-
tion of the spectra and depending on the nature of the ab-
stract factor trends can be identified. No a priori model is
required by PCA although pre-processing the spectra can be
used to assess the consequences of including, for example, a
particular background approximation.
Principal Component Analysis is not a substitute for the con-
ventional quantification techniques and requires care before Figure 82: Oxidation Sequence
the results can be used to assess trends. There have been at-
tempts to place physical meaning on the abstract factors but data set may have a dominant trend and therefore the ab-
it should always be remembered that the abstract factors are stract factors apparently have chemical meaning in their
truly abstract and owe their nature and size to a mathemati- own right. The temptation to interpret these abstract factors
cal procedure for generating an orthogonal basis set for an n- as chemically meaningful quantities should be avoided nor
dimensional vector space. The first abstract factor is chosen should the number of significant factors be slavishly used to
to be in the direction of the maximum variation in the entire imply the number of synthetic component that are required
set of spectra in the linear least square sense. Subsequent to model a data envelope. However PCA can produce some
factors are generated with the same criterion as the first but interesting insights into the data and with the appropriate
subject to the constraint that each new vector is orthogonal care provides supporting evidence for a synthetic model.
to all those that have already been constructed. Sometimes a A trend analysis for the oxidation of a piece of aluminium

Trend Analysis for Metal Oxidation 115

 Analytical Applications

provides a good example of how PCA can assist in the inter-

pretation of a data set. The essential problem is to construct
a synthetic model for the Al 2p or Al 2s photoelectric lines
measured periodically as an Aluminium sample oxidises in
a vacuum chamber. The relative sizes for the individual syn-
thetic lines provide a quantitative measure for the trends
within the data. The problem lies in knowing what chemical
states are present and therefore what features the model must
accommodate.

PCA for Aluminium Oxidation Sequence

The spectra shown in Figure 82 were recorded on a Scienta
ESCA300 XPS spectrometer at RUSTI and represent a trend
observed as a clean Aluminium sample, created by argon-
ion etching, oxidised in a vacuum chamber as a result of
combination with low level residual water and oxygen (1 x
10–9 mbar). These are particularly good data for a PCA since Figure 83: First three abstract factors generated from the spec-
the spectra have been recorded over the Al 2s as well as the tra shown in Figure 82.
Al 2p photoelectric lines and both structures should vary to-
gether as a unit over the course of the experiment.
A PCA for the data set shown in Figure 82 using correlation principal components when reproducing the original spec-
about the origin produces the abstract factors and associated tra. The trends within the data set can therefore be assessed
eigenvalues displayed in Figure 83. The first abstract factor by plotting the co-ordinates for the spectra when projected
is clearly a vector in the direction of the maximum variation onto the two dimensional subspace spanned by the first two
in the overall data set. There are shapes within this first ab- abstract factors.
stract factor that belong to what can be identified as alumin- At this point it would be easy to conclude that any synthetic
ium bonded with oxygen. The second abstract factor model for the data envelopes requires two fundamental line
exhibits structure associated with each of the aluminium shapes. The structure associated with the aluminium peaks
peaks and what is more the structure is of similar form. is positive for the region where the oxide peak should occur
The third abstract factor contains almost no significant and is negative in the region of the main metal line shapes.
structure and all the PCA statistics point to the need for two The second abstract factor represents an adjustment to the

Trend Analysis for Metal Oxidation 116

 Analytical Applications

average shape manifested in the first abstract factor that is loading represents a removal of the oxide shape from the av-
required to describe each spectrum in the data set to within erage description represented by the first abstract factor,
experimental error. The loading for each of the abstract fac- while the metal line shape is enhanced by virtue of the neg-
tors demonstrates that the average form is almost constant ative structure beneath that part of the spectrum. The adjust-
across the data set but that the adjustment smoothly moves ment to the average shape reverses as the oxidation reaction
from a negative contribution for the second abstract factor to progresses. The PCA has managed to describe the trend us-
a positive loading as the oxidation proceeds. An interpreta- ing essentially a single loading and a careful choice of the
tion for this trend might be the consistent conversion of one basis vectors. An interpretation for this trend could be a mi-
form of Aluminium to another. gration of the metal form of aluminium to aluminium bond-
ed with oxygen and/or the correlated attenuation of the metal
signal due to the build up of an oxide over-layer. A strong
conclusion from the PCA is that, whatever processes are in-
volved, these processes occur in a linearly related fashion.
This statement does not exclude a hydride moving to an ox-
ide state with the same speed as the metal making the same
adjustment. It does, however, show that metal and oxide
line-shapes are sufficient to describe the oxidation process
measured by the XPS spectra.

The next step is to construct a synthetic model that produces

the same trend as identified by the PCA.

Figure 84: PCA loading for first two abstract factors Synthetic Model for an Oxidation Sequence
Photoelectric peaks for metals require synthetic models ca-
The graph in Figure 84. shows a plot for the loading associ- pable of describing asymmetry in the observed line shapes.
ated with each of the first two abstract factors. The loading Various approximations have been introduced to account for
factor for the overall shape seen in the first abstract factor is the asymmetry seen in the aluminium lines, ranging from ad
virtually constant across the set of spectra. All the variation hoc functions to theoretically based shapes. As always, re-
within the data set is accounted for by the second abstract gardless of the asymmetry function a background approxi-
factor. This is an interesting result in the context of the struc- mation must be introduced to facilitate the optimisation
ture seen in the second abstract factor. The initial negative procedures that ultimately determine the peak parameters.

Trend Analysis for Metal Oxidation 117

 Analytical Applications

The Al 2s line is an interesting one to study since the data intensity will not be consistent with the direct integration
envelopes should result from only two photoelectric lines over the data region. If the data region is used to determine
namely the metal and the oxide Al 2s resonance features (al- the limits for the theoretical line shape then the results for
though a small plasmon structure from the Al 2p line should the intensity will vary with the acquisition region. Alterna-
also appear at the same BE as the Al 2s lines.) This is in con- tively, a more extensive but arbitrary limit will result in
trast to the Al 2p data envelope since in this case the data en- more repeatable results for intensity values however these
velope is constructed from doublet structures that therefore intensities will not compare to quantification values derived
complicate the nature of the model. The data set under study from different synthetic models.
offers a near pure Al 2s Metal spectrum from which the The Voigt type synthetic line shapes are more ad hoc in na-
asymmetry can be assessed and then applied to the subse- ture than the Doniach Sunjic approximation but have the
quent spectra in the oxidation sequence. merit that they make no attempt to model the background re-
The theoretical line shape for a metal such as the aluminium sulting from the primary peak. The form for the background
has been shown to have an asymmetric line shape given by must derive from one of the standard approximations nor-
Doniach and Sunjic22, however from a practical perspective mally applied (Linear, Shirley, Tougaard etc). Modifying
the Doniach Sunjic line shape presents a number of prob- the Voigt function by an exponential tail revisits the problem
lems. The principal problem is the question of area under the associated with the meaning of the intensity when compared
curve and how it relates to the intensity of the photoelectric to the actual acquisition region (i.e infinite extent vs defined
line. The Doniach Sunjic line provides a very good fit for the region). The grey area introduced by modelling data with
observed data provided the background does not attempt to functional forms that blur the boundaries between peak and
background are also present when line shapes are extracted
model the background contribution from the Al 2s line it-
from experimental data as well as issues associated with ar-
self. The functional form for the Doniach Sunjic line typical-
bitrarily truncating the model data. Good statistics for a
ly extends beyond the data region and therefore integrates to
model do not necessarily translate into good quantification
a larger intensity than is calculated from the data region in
results for practical reasons.
question. Truncating the function to the integration region
destroys stoichiometric relationships between peaks or rela-
Peak Fit Using Doniach Sunjic Line Shapes
tive intensities for multiple-peak models. The underlying
line shapes must extend far enough to reduce the influence The preceding discussion attempts to highlight the dangers
of offset peaks within the same region but whatever choice of modelling spectroscopic data using synthetic line shapes.
is made for the extent of the Doniach Sunjic line shape the Nevertheless trends within data such as the set shown in Fig-
ure 85 must be assessed by some means and an approxima-
22. tion chosen. In the case of the Al 2s peaks the Doniach

Trend Analysis for Metal Oxidation 118

 Analytical Applications

component is introduced to account for the oxide structure

seen on the other spectra. The position and width for the ox-
ide component is determined from the last spectrum in the
data set (Figure 86) and then fitted across all the spectra in
the experiment.

Figure 85: Al 2s metal data fitting using a Doniach Sunjic line

shape.

Sunjic line shape provides the best form for modelling the
observed data. Given these doubts it is therefore wise to test
the results against other techniques in order to build confi- Figure 86: Oxidised Aluminium data envelope. Note that the ox-
ide line-shape has been chosen to be the same as the metal even
dence in the conclusions. In this case PCA is used to “cross though the asymmetry of the metal should not exist in the oxide.
check" This has been done to so that the intensity measued by the oxide syn-
Figure 85 shows the initial spectrum in the data set fitted us- thetic line-shape is, in some sense, comparible to the metal.
ing a single Doniach Sunjic line shape. Note how the back-
ground approximation is a simple linear form offset from the A trend for the oxide peak intensity can be identified but the
spectrum. true intensities involved are difficult to assess for the reasons
The model from the first spectrum in the sequence is used as given above. Figure 87 shows a comparison between the re-
the basis for modelling the remaining spectra. An additional sults from modelling the spectra using two synthetic compo-

Trend Analysis for Metal Oxidation 119

 Analytical Applications

nents and the principal component analysis. The two PCA tion of the chemical processes involved The trend analysis
trends correspond to covariance about the origin with (PCA based on PCA is modelling changes in both the background
BG) and without (PCA) background subtraction. Note how and the intensity of the photoelectric lines, but does show
the trend PCA BG contains elements from both the PCA and that these adjustments throughout the data set are well mod-
the normalised peak-fitted (Normalised PF) results. Clearly elled by two underlying shapes in the form of abstract fac-
the background approximation influences the nature of the tors.
trend computed using PCA and therefore similar influences
must also be present when peak fitting is used to extract the
trend. Adjusting a Custom Quantification Report

The standard quantification reports treat each quantification

item as a separate entity and therefore a percentage concen-
tration is reported for each item used to quantify a sample.
Custom reports, on the other-hand, sum the intensities for
any quantification items defined with the same name. It is
therefore important when using the Custom report option to
label synthetic peaks with a different name from the region
used to define the background for those peaks.
Consider the following example:
Figure 88 shows three high-resolution spectra used to quan-
tify a sample. The O 1s region is fitted using two synthetic
components, whilst the C 1s and the Al 2p spectra are quan-
Figure 87: Comparison of trends identified using Curve-fitting tified using integration regions only.
and PCA. Each spectrum has an integration region defined and the
names for these regions are O 1s, Al 2p and C 1s. In addition
The PCA without background subtraction shows the to the integration regions, the O 1s spectrum is fitted using
smoothest trend in Figure 87 and the limited input assump- two synthetic components and both of these components are
tions suggests this curve is the best description of the trend named using the same name as the integration region for the
within the data set, although not necessarily the best descrip- O 1s spectrum. A Standard report generated from these inte-

Adjusting a Custom Quantification Report 120

 Analytical Applications

A custom report using the same names as those shown in

Table 11 is as follows:
Table 12: Custom report using the same names as those used in
Table 11.

O 1s Al 2p C 1s
O 1s % Al 2p % C 1s %
CPSeV CPSeV CPSeV

34075.5 11360.8 2860.9 70.5538 23.5227 5.92353

Note that the O 1s percentage concentration is larger than

the equivalent value reported by the Standard report. The

Figure 88: Three High-resolution Spectral Regions

gration regions is as follows:.

Table 11: Standard report generated from the quantification

regions shown in Figure 88

%
Name Energy FWHM R.S.F. Area
Conc

O 1s 532.35 2.38589 0.78 13299.4 54.52

Al 2p 72.7 0.740902 0.193 2192.6 36.33
Figure 89: Synthetic peaks for the O 1s spectrum are now named
C 1s 286.3 1.39428 0.278 795.3 9.148 differently from the integration region.

Adjusting a Custom Quantification Report 121

 Analytical Applications

problem lies with the same name being used for more than .
one quantification item. The integration region for the O 1s
spectrum has the same name as both the synthetic compo- Table 13: Custom report after the synthetic components have
nents and therefore the three quantities associated with the been renamed
O 1s spectrum are summed to provide the intensity used in
the custom report. O 1s Al 2p C 1s
O 1s % Al 2p % C 1s %
CPSeV CPSeV CPSeV
This was not the intention. To prevent this type of ambiguity 17050.5 11360.8 2860.9 54.523 36.3287 9.14836
it is best to use a different name for the integration region
from that used for the synthetic components. Figure 89
shows the same data where the synthetic peaks have been as-
The value of the Custom report lies in the ability to combine
signed a common name, which is different from the integra-
intensities, such as those from the synthetic components
tion region.
used to model the O 1s region, with the intensities from the
integration regions for the C 1s and the Al 2p regions.
A Custom report generated from this newly named set of
Table 14 has been calculated from a combination of the
quantification items is shown in Table 13 and can be seen to
“Peak O s” intensities and the previously used Al 2p and
be identical to the report in Table 11
C 1s regions. The real value comes when the O 1s region,
say, has an interference peak which requires excluding from
the final results by fitting multiple peaks to the data but only
selecting a subset of these peaks for the purposes of the
quantification report

Table 14: Custom report using the Synthetic Peaks for the O 1s
intensity but the integration regions for the C 1s and Al 2p
intensities

Peak
Al 2p C 1s Peak
O 1s Al 2p % C 1s %
CPSeV CPSeV O 1s %
CPSeV

17024.9 11360.8 2860.9 54.4856 36.3585 9.15586

Adjusting a Custom Quantification Report 122

Using Different File Formats CHAPTER 13

In order to further enhance the “cross platform” approach in- dow where the “Files of type” filter allows only “.mpa” files
herent in ISO 14976, CasaXPS provides several file conver- to appear in the listing. Simply select a DS800 file and press
sion routines for the “native” file format of systems which Open or double-click the desired file. CasaXPS will convert
either do not have any “VAMAS” output capability or which the binary file into the ISO 14976 file format, where the new
do not implement the specification fully. brief descriptions file will have the same name as the original .mpa file but
of procedures for these conversions are given below. with a .vms extension added.

Kratos DS800 Binary Files

Kratos DS800 binary files can be directly converted through

CasaXPS.
The first step to creating a new VAMAS file is to enable the
Convert toolbar button. The Convert option is disabled un-
less the selected sub-frame window representing a file con-
tains no VAMAS regions. If no empty sub-frames are
amongst the list of windows then the New toolbar button
must be pressed to create an empty sub-frame. Figure 90: Convert to VAMAS file showing DS800
files.
Figure 90 shows the “Convert to VAMAS file” dialog win-

123
 Using Different File Formats

less the selected sub-frame window representing a file con-

VG Eclipse Files tains no VAMAS regions. If no empty Experiment Frame is
amongst the list of windows then the New toolbar button
VG Eclipse files are binary files, one file per spectral region, must be pressed to create an empty Frame.
organised in directories where files with common acquisi- There are three ways to convert VG Eclipse data.
tion characteristics appear in the same directory or inside • If an individual file is required then selecting the .dts file
sub-directories within that directory. CasaXPS will convert via the Covert to VAMAS file dialog window will result
these directory structures into a single ISO 14976 file where in the creation of a VAMAS file that contains the spec-
experimental information and relationships are maintained. trum in that .dts and no others.
Quantification in CasaXPS can be performed in an identical • If, on the other hand, all the .dts files are require in the
way to the Eclipse data system; transmission correction is same VAMAS file, the Covert to VAMAS file dialog
accounted for by CasaXPS whenever the information is in- window should be used to find the directory containing
cluded in the original files. the spectral regions. The name of the new file entered in
Individual spectral regions are stored in files with a .dts ex- the dialog window must be specified in the text-field, but
tension. Sets of regions are stored in individual files all lo- with a .col extension. It is the extension given to the file
cated in the same sub-directory. Older versions of Eclipse name that flags the type of conversion required, however
may have .col extensions associated with these data directo- the new file created by CasaXPS will replace the entered
ries, but later versions of Eclipse generate such directories extension by .vms.
without this naming convention. Similarly, depth profiles or • The third method converts depth profile type file struc-
angle resolved experiments appear as sets of directories, one tures. Using the Covert to VAMAS file dialog window,
per acquisition region, where each sub-directory contains move to the directory that contains the sub-directories
the .dts files for one of the acquisition regions involved in containing the spectral regions. Whilst at the level of the
the experiment. Again, older versions of Eclipse use a nam- sub-directories enter the name of the new file name and
ing convention where each sub-directory was given a .mle add a .mle extension. The .mle flags that CasaXPS
extension. should read each sub-directory within the current direc-
The first step to creating a new VAMAS file is to enable the tory and convert all the .dts files found within the sub-
Convert toolbar button. The Convert option is disabled un- directories. The new file name will replace the .mle

VG Eclipse Files 124

 Using Different File Formats

extension with .vms. Figure 91 shows the dialog window these directories is shown in Figure 92.

Figure 91: Convert to VAMAS File Dialog Window

where the visible directories are about to be converted

using the .mle flag. The file name profile .mle indicates
that each of the directories form part of a depth profile Figure 92: VAMAS File Block Structure from a VG Eclipse
and should be interpreted as the names for the regions Depth Profile
found inside these directories. The result of converting

VGX900 (Ron Unwin)

Ron Unwin developed the VGX900 system and so all the

various forms of Ron Unwin’s files can be converted in the
same way.

The first step to creating a new VAMAS file is to enable the

Convert toolbar button. The Convert option is disabled un-
less the selected sub-frame window representing a file con-
tains no VAMAS regions. If no empty Experiment Frame is

VGX900 (Ron Unwin) 125

 Using Different File Formats

amongst the list of windows then the New toolbar button name at conversion time can alter these two options:
must be pressed to create an empty Frame. -anode Mg specifies Magnesium anode.
Files generated by the VGX900 system are ACSII files and
-energy KE specifies scans with increasing KE sweep.
each file may contain one or more spectral regions. The
method for loading these files into CasaXPS is to collect a
set of the files in a sub-directory and then enter a new file
name in the “Convert to VAMAS file” dialog window, but
add an extension of .unw to the specified name (Figure 93).
All files in the directory will be read and appear in a single
VAMAS file. Since the files have no characteristic file type
it is essential that only files generated from the VGX900
system are present in the sub-directory and no others.
Some formats of the VGX900 system do not include infor-
mation necessary for a proper description of the data (in the
VAMAS sense), but the system does allow the writing of
VAMAS files based on a user specified template. This al-
Figure 93: Convert to VAMAS file dialog window VGX-
lows data to written in a form directly readable by CasaXPS, 900.
but the user must ensure that the correct fields have been Note the .unw extension used to indicate that the set of files con-
used to describe the spectra. tain spectra acquired using Ron Unwin's data system. The –an-
Alternatively, the conversion option in CasaXPS can specify ode Mg is only used to force the X-ray source to be Magnesium
some of the missing information using flags. The X-ray an- instead of the default setting. The latest release of Ron Unwin’s
system includes this information.
ode defaults to Aluminium and the energy-scale defaults to
Binding Energy. Appending the following strings to the file

VGX900 (Ron Unwin) 126

 Using Different File Formats

The first step to creating a new VAMAS file is to enable the

Convert toolbar button. The Convert option is disabled un-
less the selected sub-frame window representing a file con-
tains no VAMAS regions. If no empty Experiment Frame is
amongst the list of windows then the New toolbar button
must be pressed to create an empty Frame.
There are two methods for loading Bristol files.
• The simplest method is to convert files that are listed in a
.seq file. All that is required is for the .seq file and all the
.SP_ files listed within that file are located in the same
sub-directory, then the .seq file is selected through the
“Convert to VAMAS file” dialog window. If for any rea-
son the set of .SP_ files does not match the set listed in
the .seq file, then the .seq will need to be edited to
remove or add entries as necessary. The VAMAS file that
results will include an experimental variable for each
spectrum loaded.
Figure 94: The set of spectra contained in the sub-directory • The alternative method for loading the .SP_ files is to
shown in Figure 93 are offered in CasaXPS using the browser and
spectrum display frames.
collect a set of these files in a sub-directory and then
enter a new file name in the “Convert to VAMAS file”
dialog window, but add an extension of .bri to the speci-
fied name (Figure 95). All .SP_ files in the directory will
be read and appear in a single VAMAS file. The default
“Dayta” System Files (Bristol IAC system) action is that no experimental variable it defined for each
spectrum. As a consequence spectra read will all appear
Spectra acquired by the Bristol “Dayta” system are stored as in the same row in CasaXPS (i.e. in a format suitable for
sets of ACSII files with .SP_ file extensions. These files may quantification of high resolution spectra). If on the other
be connected to a particular experiment through an associat- hand it would be more convenient to separate spectra by
ed .seq file. Sets of .SP_ spectrum files are listed in the .seq an experimental variable then the new file name entered
file together with the any experimental information (e.g. in the dialog window should be followed by a –col string
etch times). e.g. “newfile.bri –col” (omit the quotation marks). The

“Dayta” System Files (Bristol IAC system) 127

 Using Different File Formats

experimental variable is the index number determined alent to the experimental variable fields used in the
form the order in which the files are read. ISO 14976 standard and the Vision Objects are equivalent to
the VAMAS blocks.
The principal problem with converting the Vision 1.x/2.x
data sets is the internal data-structures used to maintain the
experimental information. The file structure is a link-list of
file records that hold a hierarchical description of the data.
Fortunately there is a utility called dump_dataset (available
from Kratos), which converts the binary format to an ASCII
version and it is the ASCII version that CasaXPS converts to
the ISO 14976 standard.
To create an ASCII version of a Kratos Vision data set fol-
low this procedure:
Figure 95: Convert to VAMAS file dialog window for Dayta
A.bri file extension has been specified. The result is a file called
at the command line level on the Unix workstation, move to
newfile.vms containing all the .SP_ data files seen in the window the data directory and type
$dump_datasetfilenameall>new_filename.kal
(where  represents space and $ the command line prompt)
The extension .kal is recognized by CasaXPS and tells the
conversion routine to parse the data using the Kratos Vision
Kratos Vision 1.x/2.x ASCII files option.
The first step to creating a new VAMAS file is to enable the
The Kratos Vision data sets are probably the closest in logi- Convert toolbar button. The Convert option is disabled un-
cal design to the ISO 14976 file format. A single binary file less the selected sub-frame window representing a file con-
maintains the context for an experiment, where an experi- tains no VAMAS regions. If no empty Experiment Frame is
ment may include multiple acquisition regions characterized among the list of windows then the “New” toolbar button
by, so-called, state changes. These state-changes are equiv- must be pressed to create an empty Frame.

Kratos Vision 1.x/2.x ASCII files 128

 Using Different File Formats

SSI M-Probe Files

SSI M-Probe files may exist in an ASCII format where mul-

tiple regions appear in the same file or alternatively, the data
is described by a control file with the spectra stored in sepa-
rate .reg files in a sub-directory where the sub-directory has
the same name as the base-name of the control file.

The first step to creating a new VAMAS file is to enable the

Convert toolbar button. The Convert option is disabled un-
less the selected sub-frame window representing a file con-
Figure 96: Convert to VAMAS file Dialog Window. The tains no VAMAS regions. If no empty Experiment Frame is
file display filter is set to show only .kal files amongst the list of windows then the New toolbar button
must be pressed to create an empty Frame.
Selecting a .kal file via the “Convert to VAMAS file” dialog
There are two methods for loading M-probe files:
window causes an ISO 14976 file to be generated with the
same name, but with a .vms extension added. The spectra
• ASCII format files must have an .mrs extension and
will be converted and other experimental information ex-
may be opened directly through the Convert Dialog win-
tracted so that the ISO 14976 file may be quantified to pro-
dow.
duce identical results to those from the Vision systems. That
is, transmission correction is included as part of the
ISO 14976 file. • The binary format consists of an .mrs file together with a
directory of the same name as the base-name of the .mrs
file. To load the information held in this binary format
Element library files from the Vision 1.x system can also be the base-name of the .mrs file must be specified but with
converted to CasaXPS format and Vision 2.x element librar- an .ssi extension replacing the .mrs extension
ies can be converted on request. (Figure 97.)

SSI M-Probe Files 129

 Using Different File Formats

amongst the list of windows then the New toolbar button

must be pressed to create an empty Frame.
The ASCII files may contain single regions, so called mul-
tiplexed regions, or depth profile data. To convert a set of
files in any of these formats into a single VAMAS file, the
.asc files must be collected into a sub-directory that contains
only valid .asc data files and then a new filename must be
entered with an extension of .qua (Figure 98) A VAMAS
file containing all the regions from the files in the directory
with an .asc extension will be created.

Figure 97: The binary data held in the file/directory pair named Al-
foil will be converted to ISO format.

PHI MultiPak ASCII files

PHI MultiPak writes an ASCII format. Data in this format

can be converted to ISO format through CasaXPS.

The first step to creating a new VAMAS file is to enable the Figure 98: MultiPak ASCII files are converted by entering a filena-
me with an extension of .qua.
Convert toolbar button. The Convert option is disabled un-
less the selected sub-frame window representing a file con-
tains no VAMAS regions. If no empty Experiment Frame is

PHI MultiPak ASCII files 130

Command Summary CHAPTER 14

The CasaXPS system has three “levels” of control - the fa- (top) tile.
miliar Windows style “Menu bar”, a system specific “Tool- Convert (“greyed out” (disabled) unless the current
bar” and “Options bar” (providing commands or command Expeiment Frame is empty ) brings up the “Convert” file
windows accessed by clicking “button” icons) and keyboard browser (see “Using Different File Formats” on
shortcuts. This chapter provides brief “one line” descrip- page 123)
tions indicating where a command is - how it is accessed, Close dismisses (closes) the current Experiment Frame,
what the command does, and the part of the system or data providing the file it represents is saved.
to which it applies. Save As... enables changing a file name and/or saving to
a different directory
Save Picture... brings up a “save” file browser enabling
Main menubar the displayed graphic to be saved as a “Windows
Enhanced MetaFile” (.EMF)
Print opens the standard Windows print dialog box for
The main system window Menu bar contains five menus,
the system default printer.
File, View, Window, Options, and Help .
Print Setup invokes the standard Windows print set up
box for the system default printer.
File
New opens a fresh (blank) Experiment Frame inside the (recent files) the system indicates the four most recently
CasaXPS main window (Programme Frame). opened files (a “push down” list). Clicking on one of
Open... brings up a “file dialog” window so that a cho- these opens a new Experiment Frame containing the file.
sen (existing) file is opened and displayed in the current Exit quits CasaXPS entirely.

131
 Command Summary

View tidily, so that the top and left edges are visible “behind”
Toolbar toggles (sets on or off) visibility of the Toolbar the front one.
Options Bar toggles visibility of the Options bar Tile fills the work area uniformly with the open windows
so that all their contents are visible. The usefulness of this
Status Bar toggles visibility of the Status bar command depends on the screen area available, but it works best
with a even number and is seldom effective with more than four
Toolbar tiles, on an average (800x600 pixel) screen
Options Bar Arrange Icons tidies any minimised processing win-
dows, “docking” them in rows along the lower edge of
the work area.

(File List) provides a numbered list of the file names for

the open windows

Options
Launches one of six dialog windows for the major com-
mand sets and processing activities:-
Page Layout... deals with the way in which the graphi-
cal display area of the active tile is arranged (in rows and
columns).
Tile Display... deals with the way that parameters
(e.g.energy scale, intensity) are displayed within a tile
and sets attributes such as colour.
Quantify... provides access to the Regions and Compo-
Status Bar
nents specifications, enables definition of Report format
(standard or custom, with defined names and tags) and
also provides a numerical Data Editor (for spike
Window removal)
New Window launches a fresh (blank )processing win- Elements... controls the setting and display of the basic
dow (a tile) within the CasaXPS work area line position database for the system, accessing Element
Cascade overlaps a number of processing windows Table, Periodic Table, and Load File dialog

Main menubar 132

 Command Summary

Processing... provides access to the major processing

routines of the system: Smoothing, Integration, Differ- Toolbar
entiation, and Principlal Component Analysis (PCA), as
well as Test Data (including spectrum synthesis),
(Energy) Calibration, Intensity Calibration, a spec- Export Tab ASCII
trum Calculator (e.g. for addition and subtraction of Export MetaFile
whole spectra) and the Processing History for the Copy Display MetaFile *
selected tile.
Annotation... provides Peak Labels, and (general) Text
Main File Access Bar
annotation for the display, control for Regions, Compo-
nents, and Quantification (Table) storage and presenta-
tion and access to the Annotation History for the
selected tile.

Help Copy Display bitmap*

About launches a window giving the system version Save File to Disk
number, license status, and a dialog enabling change or Convert File
update of license. * to Clipboard
Open File
Help launches on screen (html file based) help.
New File (Experiment Frame)

The Button bars File Access buttons File Processing buttons

Variable/Calibration controls

Display Options buttons Display Properties buttons Block comment/info controls

Display Scaling buttons Display Modifier buttons

Toolbar 133
 Command Summary

Options Bar

Launch Library window

Launch Annotate window
Display Options buttons
Launch Processing window

(see page 135)

Insert many blocks into current scale
File Processing buttons
Insert one block into current display scale

Launch Quantification window Display all selected blocks in a tile

Launch Tile Display window Display one block per tile
Launch Page Layout window

The Toolbar (upper) buttons in general provide access to menus or dialog boxes.

The Options bar (lower) buttons in general execute actions immediately or (reversibly) Printing and Help
change the format of the display.

“Hovering” over a button (placing the mouse screen pointer on an icon without clicking)
Print Current (spectrum) Display
produces a descriptive label for that button (a “Tool Tip”), and a slightly longer description in
the status bar Screen Preview for Display Print
Help / About Casa XPS
Function Buttons F5 - F10 mirror the Processing buttons (Page Layout - Element Library)

Options Bar 134

 Command Summary

.
Decrease intensity scale maximum
Toggle normalised display
Increase intensity scale maximum
Toggle subtracted display
Reset Intensity scale to original maximum
Increase Energy scale range*†
Display Modifier buttons
Step zoomed energy scale left*

Display Scaling buttons Toggle background

Toggle residuals display
Step zoomed energy scale right* Toggle components display
Reset to original scale
Toggle (shaded) region display
Zoom out (step back round history)
Zoom in (requires selection box)
NB. Modifiers require prior function definition
* rescales intensity to suit
† adds 20% left and right

Display Properties buttons Comment, Information & Variable control

Edit Block Comment

For Display
Edit Block Information

† “tri-state” buttton

For Browser
Toggle Offset for multiple traces †
Set Experiment variable (linear)
Toggle 2D, 3D, and Factor Space Display †
Edit Species/Transition parameters
Toggle Counts and CPS Intensity scales
Edit Source/ Analyser parameters
Toggle Binding and Kinetic Energy scales

Options Bar 135

 Command Summary

typeface (“Font”).
Processing Dialog windows

Page Layout
(see “Tiles of Spectra” on page 26)

Min and max enable setting the span of the display (which
may be greater or less than the data range for the selected
Sixteen predefined layouts are provided, corresponding to tile). Type in decimal values: values relate to the scale se-
arranging spectra in symmetrical rows up to the maximum lected above, so take care with max and min values for BE!)
(16) allowed. All predefined formats may be changed as re- Set Scale enables a linear remapping of the x-axis for use in
quired, for example to provide a project specific layout set, e.g. depth profiles, Most usefully employed when the output
the three controls in each layout functioning as “radio but- report is saved to a fresh file with “.vms” extension, then re-
tons”, allowing selection of only one alternative or combina- opened in the system and the scale changed.
tion for each case.
Display Minor Tick-Marks is a simple “on - off” control
Tile Display Parameters Display Axis Scale similarly turns off the x-axis scales and
(see “Tile Display” on page 27) labels
The six “tabs” give access to tile (individual spectra) param- Rescale Axis changes the scale to KE and also changes the
eters in a straightforward way. Settings may be applied to in- label to whatever is typed in the adjacent box
dividual tiles or all tiles in an Experiment (via “Global” tab).
Y-Axis
X-Axis
Dependent variable provides a “radio button” choice be-
Independent variable provides a “radio button” choice be- tween “counts per second” and “counts” (“counts per bin”
tween BE and KE for x-axis display, and a means of label- i.e. counts accumulated in each energy interval acquired) for
ling the axis as desired (the type-in boxes) and setting the y-axis display, a means of labelling the axis as desired (the

Processing Dialog windows 136

 Command Summary

type-in boxes) and of setting the typeface (“Font”) for this plays only. 2D has no further options.
label.
Reverse Display mirrors the order of plotting traces in a 2D
Min and max enable setting the span of the display (which or 3D display (“top” to bottom” or “front” to “back”) but not
may be greater or less than the data range for the selected the traces themselves (spectra are always plotted with in-
tile). Type in decimal values, which relate to the scale select- creasing KE left to right).
ed above.
Back Plane Parameters provide scaling and offset (and
Offset spectra provides an automatically calculated shift in thus “perspective”) for the “furthest” (“back”) trace of a se-
the y-axis for each overlaid spectrum. The “%” value sets ries of traces visualised as slices of a cube, subject to plot-
the size of the y-axis scale in relation to the available area ting within the tile area. Values are percentages. Offsets may
(not to the offset). The offset is then calculated to divide the be positive or negative and relate to the leftmost or rightmost
remaining space equally. point in the dispay subject to the constraint of plotting within
Display Minor Tick-Marks is a simple “on - off” control the viewable area in the tile.

Display Axis Scale similarly turns off the y-axis scales and Front Plane Parameters are similar except that y=0 is al-
labels ways the front baseline (no offset). Colours for 3D back-
ground and fill are set in the “Colours” tab
Normalise Display sets y values equal at indicated point on
the x-axis. To move normalisation point, use “shift-left- Scatter Parameters provide for rotation (in degrees, posi-
click” in the display tile (uses the foreground trace scale in a tive or negative) about the three axes (x, y, and z), and shift
multi-trace display). Control mirrors the “normalise” button. (percentage of diplay area) in x and y, constrained as above.
Display Exp. Variable Left/Right labels each trace at its Display
right or left extremity with the corresponding value of the
Display provides simple “checkbox” (on - off) controls for
experimental variable (e.g. etch time) Non - exclusive.
items to be included in the display. Traces are normally dis-
Display Background Subtracted includes or excludes played as “lines” (continuous curves) unless “Draw Points”
background from display. Greyed when not available is checked, when actual point-by-point values are potted
with no joining lines (useful, e.g., for curve fitting, overlay-
Geometry
ing envelopes (lines) and points). Typefaces for the Header,
Tile Display Geometry selects either 2D, 3D, or Factor Title, and Axes labels are controlled here (a fresh header
Space. The parameters immediately below apply to 3D may be added in the input box here) and the line width (in
(Back Plane and Front Plane) and Factor Space (Scatter) dis- pixels) of the display defined (e.g. thicker lines for clearer

Processing Dialog windows 137

 Command Summary

AV presentation or graphical capture and scaling) same logical level on the currently operational system disk
as that at which the file “casaxps.exe” is stored. CasaXPS
Colours
will then create and store files for all the default parameters
provides a means of setting colour for Spectra, Spectrum for acquisition, colours and so on here.
Background, Tile Background, 3D Background, Region
Background, 3D Fill, Residual Trace, and Synthetic Compo- Quantification
nents.
Clicking any of these items brings up a Standard + Custom (see “Quantification” on page 42
Colour swatch for selection and a further click on “Add to
Custom Colours” extends this to a Colour Picker with HLS,
RGB and visual presentation

When more than one item of the same type is present in a

display (e.g. spectrum traces), then custom colours are cy-
cled through in order Regions
Global This tab provides access to the mostly self explanatory pa-
This tab provides a means of applying the changes made rameters for any new or existing region for the displayed
with the other tabs to all the tiles in the current Experiment block or blocks.
Frame, including (or otherwise) x and y axis ranges and the Note that all the parameters in the region list are editable. A
title. the Tabs also include buttons to enable saving the de- selected region as a whole (click on the header letter) has a
fault parameters for colours, settings, and fonts (typefaces). blue background: an individual parameter selected by a
A directory called “casaxps.def” should be created at the mouse left click has a plain white background and the exist-

Processing Dialog windows 138

 Command Summary

ing value is left justified. Changing an entry requires a “re- (note that the original data are not affected by this correc-
turn” (enter) key at the end of the entry to validate it: merely tion, only the quantification). The manual entry (type-in)
clicking on a new parameter restores the old value. Note also box in general should only be used when automatic correc-
that correct spelling is required for background type (linear, tion is not available. The Update button applies a new man-
Shirley, Tougaard - though case is insignificant). ual value or the automatic setting after any change in the
status of the checkbox.
Create brings up a new region with (as far as possible) the Note that the width of the region list may be altered in the
parameters entered for the displayed peak. usual fashion by dragging on the edge of the divider in the
header bar.
Create From Labels provides a region whose parameters
are taken from the library entry with the same label as the Components
displayed peak (see “Step 5: Create quantification regions”
on page 11) (see “Tile Display” on page 27, and also see “Line Shapes
and Backgrounds” on page 55)
Delete deletes the selected region and its parameters from This tab defines the parameters for each synthetic compo-
the Quantification window (and thus from calculations and nent for a particular region and also controls peak fitting.
display).

Calculate Error Bars completes the standard deviation Note that the Cross Section is the Tougaard “3-parameter”
calculation and related procedures. constant set (consisting of four (!) comma separated values)
(see “U 4 Tougaard (short form U 4) :” on page 67).
Save Region As VAMAS File records the region parame- Offset provides a numerical adjust for spectrum overlay in
ters in the block header (see page 145) display after background subtraction. the value entered is a
percntage (of the largest peak to be displayed) and may only
Intensity Calibration provides either automatic incorpora-
be positive. Positive values result in movement of the region
tion of the transmission function data held in the (Kratos ‘downwards”, offset towards the baseline.
style) VAMAS file (see also page 145) or manual entry of a
specific factor (transmision exponent) in the type-in box. If Max Height and Min Height are derived from the region
CasaXPS detects the presence of transmission function cor- under consideration (signal values above / below back-
rection values in the data block (an appropriate ‘correspond- ground) and used e.g. with AES signals (or other deriva-
ing variable’) then the ‘automatic’ checkbox will be ticked tives) in determining peak to peak concentration.
and the quantification will proceed using the correction See page 11 for notes on other button use.

Processing Dialog windows 139

 Command Summary

The Report Spec tab enables customised output of quanti- different and specialised procedure.The two Calibration
procedures are straightforward (see “Energy Calibration” on
page 35,“Intensity Calibration” on page 36) as is the Calcu-
lator. The Test Data tab provides not only standardised
files for benchmarks, but also access to the Monte Carlo pro-
cedures (see “Monte Carlo Methods - Uncertainties in Inten-
sity Calculations” on page 97).
Processing History provides a complete or selective “un-
do” function as well as a record of applied methods.

fied values, with the ability to apportion and aggregate areas

and set up mathematical relationshipships, as well as pro-
duce ‘standard’ reports. The Data Editor provides access to
individual channel counts to enable, for example, spike re-
moval. Note that this is a potentially dangerous procedure
which can modify original data.
Reset removes all processing from the selcted block and re-
turns it to the original sate.
Processing
Apply selection resets the displayed block then reapplies
(see “Processing” on page 34)
only the selected items from the existing history. Use ‘con-
The Smoothing, Integration, and Differentiation tabs trol’ and ‘shift’ keys with the mouse to select non-contigu-
have similar controls (and use related procedures). Zero ad- ous or continuous items, respectively.
just (in Integration) provides for a shift of the minimum or-
dinate value after integration to the axis zero. PCA (see Propagate flag sets the ‘propagate’ flag to apply the current
“Principal Component Analysis” on page 80) is an entirely selection from the history to a series of (selected) blocks.

Processing Dialog windows 140

 Command Summary

Annotation Library
(see “Graph Annotation” on page 31) (see “Element Library” on page 38)

Controls placement and style of text for annotation (Text),

Peak Labels, Regions and Component summaries, and
In order to function correctly, CasaXPS should have an ele-
Quantification tables in the display tile and printed output.
ment library (table of defined line positions, widths, shapes,
An Annotation History panel functions in a similar fashion
senisitivity factors, excitation source and labels) present in
to the Processing History, enabling removal of unwanted
the system. This table may list any (single or composite)
items or selective application. Each tab has provision for
transition appropriate for the analysis undertaken (it is not
changing of typeface (font, size and colour). Position for an-
limited to chemical elements) and may be changed at any
notation items may be referenced to the display tile frame or
time (so that several files may be held on disk), but only one
the data itself (for convenient placement when multiple
library (of unlimited size) may be active at any time.
blocks are displayed) and labels may be orientated vertically
Input File (above) enables loading, changing or combining
or horizontally. Positioning targets (small squares with cen-
the line position data file which forms the basis for feature
tral dots) may be set with the mouse anywhere within the
identification. File selection (of either CasaXPS or Kratos
display tile and disappear when the annotation window is
basic type) may be direct, typing ifn a file name or using a
dismissed.
file browser, and the selected file may be merged with or
may replace the existing file (if any).
The Element Table window provides a scrolled list of the
contents of the Library File: clicking on a name (toggle)
transfers a labelled position marker to the display for the se-
lected item Position markers may be removed from the dis-

Processing Dialog windows 141

 Command Summary

play with the Clear All Elements button. Find Peaks adds
coloured markers to the display for all major peaks in the
displayed block, according to a defined algorithm (see
XREF xxx): the Clear Markers button removes these. Keyboard & Mouse Shortcuts

Function Shortcut*

New Experiment Frame Ctrl+ N

(Browser)

Open ISO File Ctrl+ O

Print (graphic) Ctrl+ P

Copy (graphic to clip- Ctr+ C

board)

Set ‘Normalise’ point Shift+ (left click)

The Periodic Table window provides a simpler means of Zoom out (cycles) Ctrl+ (right click)
selecting elemental transitions (from the current element li- Context menu (right click)
bary, if any) for display labelling. Each element symbol is a
toggle (click on - click off) button). Edit a data point Ctrl+ Shift+ (left
click)

Display (spectrum) F1

Overlay (spectra) F2

Zoom out F3

Reset Zoom F4

* F1 - F4 are single press “Function“ keys: for the other commands,

hold down the “control” (Ctrl) and/or “shift” key and press the
listed key or mouse button.
“Drag and Drop” - you may load ISO 14976 files by dragging them
from Windows™ Explorer and dropping onto an open CasaXPS
Programme Frame.

Processing Dialog windows 142

Appendix 1: ISO 14976 format files annotated
1. Simple single block XPS (region) spectrum

Item Description Line ISO 14976 Format Item (in PET-C1s.vms file)

format identifier (1) VAMAS Surface Chemical Analysis Standard Data Transfer Format 1988 May 4
institution identifier (2) Acolyte
instrument model i.d. (3) Kratos AXIS-HS
operator i.d. (4) AC
experiment i.d. (5) PET C1s Test Spectrum
# comment lines (6) 1
(comment lines) (7) Acolyte copyright 1999
experiment mode (8) NORM
scan mode (9) REGULAR
# regions (10) 1
# experiment variables (11) 0
# parameter exclusion entries (12) 0
# manual items (13) 0
# future experiment items (14) 0
# future block entries (15) 0
# blocks (16) 1

block i.d. (17) C1s 1

sample i.d. (18) PET
year (4 digits) (19) 1999
month (20) 06
day (21) 24
hour (22) 14
minute (23) 45
second (24) 00
# hours + GMT (25) 1
# lines in block comment (26) 0
technique (27) XPS
source (28) Al
source energy (29) 1486.6
source strength (30) 300

143
 source width x, (mu) (31) 500
source width y, (mu) (32) 500
source polar angle (33) 54
source azimuth (34) 180
analyser mode (35) FAT
analyser resolution characteristic (36) 20
magnification of analyser (37) 1
analyser work function (38) 4.5
target (sample) bias (39) 0
analysis width x, (mu) (40) 200
analysis width y, (mu) (41) 200
analyser axis polar angle (42) 0
analyser azimuth (43) 180
species (44) C
transition state (45) 1s
charge of analysed particle (46) -1
abscissa label (47) kinetic energy
abscissa units (48) eV
abscissa start (49) 1191.6
abscissa increment (50) 0.05
# corresponding variables (51) 1
corresponding variable label (52) counts per channel
corresponding variable units (d= none) (53) d
signal mode (54) pulse counting
signal collection time/channel (55) 0.5
# scans for this block (56) 1
signal time correction (57) 400E-9
sample normal tilt (58) 0
sample normal azimuth (59) 0
sample rotation angle (60) 0
# additional params (61) 0
# ordinate values (62) 401
min. y (63) 42
max. y (64) 2250
(first data point) (65) 179
(399 data points) ( ) ( )
(last data point) (465) 59
terminator (466) end of experiment

144
2. Multiple block depth profile composed of XPS (wide scan) spectra

Note: the line numbers for the “experiment header” and “block 1” have been deliberately left the same as in example 1, so that
differences stand out clearly. Extra lines (variables and values) in this example are denoted with a + sign.

Item Description Line ISO 14976 Format Item (in DATA2.vms file)

format identifier (1) VAMAS Surface Chemical Analysis Standard Data Transfer Format 1988 May 4
institution identifier (2) Not Specified
instrument model i.d. (3) AXIS-165 (SAC)
operator i.d. (4) kratos (VISION User)
experiment i.d. (5) /export/home/kratos/data/2043it1.dset
# comment lines (6) 0
experiment mode (8) NORM
scan mode (9) REGULAR
# regions (10) 1
# experiment variables (11) 1
experiment variable label + Etch Time
experiment variable units + s
# parameter exclusion entries (12) 0
# manual items (13) 0
# future experiment items (14) 0
# future block entries (15) 0
# blocks (16) 117

block i.d. (17) OSnIn a a/2

sample i.d. (18) OSnIn/2
year (4 digits) (19) 1999
month (20) 12
day (21) 22
hour (22) 17
minute (23) 40
second (24) 2
# hours + GMT (25) 0
# lines in block comment (26) 3
“comment” lines used internally + Lens Mode:Magnetic Resolution:Pass energy 80 Anode:Al(180 W)

145
 by Kratos Vision system + Step(meV): 300.0 Dwell(ms): 136 Sweeps: 1
CasaXPS adds “history” here + Acquired On:Tue Dec 22 17:40:02 1998
technique (27) XPS
experiment variable value + 0
source (28) Al
source energy (29) 1486.6
source strength (30) 180
source width x, (mu) (31) 1E+37
source width y, (mu) (32) 1E+37
source polar angle (33) 1E+37
source azimuth (34) 1E+37
analyser mode (35) FAT
analyser resolution characteristic (36) 80
magnification of analyser (37) 1E+37
analyser work function (38) 1E+37
target (sample) bias (39) 0
analysis width x, (mu) (40) 1E+37
analysis width y, (mu) (41) 1E+37
analyser axis polar angle (42) 1E+37
analyser azimuth (43) 1E+37
species (44) Wide
transition state (45) none
charge of analysed particle (46) -1
abscissa label (47) kinetic energy
abscissa units (48) eV
abscissa start (49) 926.7
abscissa increment (50) 0.3
# corresponding variables (51) 2
corresponding variable label (52) Intensity
corresponding variable units (d= none) (53) d
+ Transmission
+ d
signal mode (54) pulse counting
signal collection time/channel (55) 0.136
# scans for this block (56) 1
signal time correction (57) 0
sample normal tilt (58) 1E+37

146
 sample normal azimuth (59) 1E+37
sample rotation angle (60) 1E+37
# additional params (61) 0
# ordinate values (62) 878
min. y (63) 427
max y (64) 15744
(first data point) (65) 2967
(first transmission value) + 1
(second data point) (66) 2954
(second transmission value) + 1
(875 total data & transmission points) (+ ) ( )
(last data point) (950) 458
(last transmission value) (951) 1
block i.d. (952) OSnIn a a/6
sample i.d. (953) OSnIn/6
year (4 digits) (954) 1999
month (955) 12
day (956) 22
hour (957) 17
minute (958) 43
second (959) 27
# hours + GMT (960) 0
# lines in block comment (961) 3
+ Lens Mode:Magnetic Resolution:Pass energy 80 Anode:Al(180 W)
+ Step(meV): 300.0 Dwell(ms): 136 Sweeps: 1
+ Acquired On:Tue Dec 22 17:43:27 1998
technique (965) XPS
experiment variable value + 30
source (967) Al
source energy (968) 1486.6
source strength (969) 180
.
and so on, as above for this block (block 2)
.
(last data point) (1884) 504
(last transmission value) (1885) 1
block i.d. (952) OSnIn a a/11

147
 sample i.d. (953) OSnIn/11
.
and so on, as above for this block (block 3)
.
and also for another 114 blocks,
.
.
until
.
(last data point) (15894) 504
(last transmission value) (15895) 1
terminator (15896) end of experiment

Appendix 2: ISO 14976 and the World Wide Web

The ISO 14976 file format is designed for simple, unambiguous communication of information derived from surface chemical
analysis experiments. The World Wide Web is one of the most important communications links for experimentalists, and so it is
appropriate that means are provided for transport and display of ISO data over the web.
Standard methods of text file compression and transfer work well with the format because it is text based. Display of spectra in a
web environment - on a browser page for example, is a (non trivial) matter which is of interest in many contexts, and there exists
(at least one) simple Java applet which enables display of XPS spectra “live” on a web page. The applet is also designed to
provide a basic practical mechanism for “platform independent” (that is “universal”) transfer and display of XPS spectra across
the world wide web.

See, for example, “http://www.acolyte.co.uk/JISO”

“MIME” type. MIME types (Multimedia Internet Mail Extensions) are the designators which enable applications such as web browsers and their
“helpers” to decide how to handle (interpret) an otherwise “unknown” file type. The “.vms” file extension is used for the
ISO 14976 spectra and is in process of submission through the IETF as a “Chemical MIME” type. The present status is
“experimental” (i.e. the type/sub-type) “chemical/x-vamas-iso14976” should be set. Since the format is solely text based,
however, http systems (and browsers communicating with them) will in general react correctly if configured to serve (or read)
unknown MIME types as “text/plain”. In most cases this will be the default action

148
Appendix 3: Definitions and Formulae

Glossary of terms

Block (ISO). A sub unit of an Experiment (file) (q.v.) which consists of a header - a set of parameters that only apply to that
block - followed by a series of ordinate values which may represent a curve (e.g. a depth profile), a spectrum, or a map
Experiment (ISO). A complete ISO 14976 file which defines the context of and data acquired during a surface chemical anal-
ysis determination. Consists of a header containing of a series of parameters which apply to the measurement procedure as a
whole, followed by one or more blocks (q.v) of data
VAMAS. the Versailles project on Advanced Materials And Standards

Formulae: Peak shapes

Gaussian/Lorentzian Product Form

 (x – E) 
2
exp  ( – 4 ln 2 ) ( 1 – m ) ------------------ -
 F
2

GL ( x, F, E, m ) = --------------------------------------------------------------------------
2
(x – E)
1 + 4m ------------------ 2
-
F

Gaussian/Lorentzian Sum Form

 (x – E) 
2
m
- ( 1 – m ) + --------------------------------
SGL ( x, F, E, m ) = exp  – 4 ln 2 ------------------
 F
2
 (x – E)
2
1 + 4 ------------------2
-
F

149
 Doniach Sunjic:

πα
cos  ------- + ( 1 – α ) atan  ------------ 
x–E
 2  F 
DS ( x, α, F, E ) = ---------------------------------------------------------------------------
2 2 (1 – α) ⁄ 2
(F + (x – E) )

Appendix 4: References and other Resources

World wide web links:

CasaXPS http://www.casaxps.com

UK Surface Analysis Forum http://www.uksaf.org

Kratos Analytical http://www.kratos.com

Physical Electronics http://www.phi.com
Thermo-VG http://vacuumgenerators.com

American Vacuum Society (AVS) http://www.vauum.org

ECASIA http://dmxwww.epfl.ch/ecasia/
QSA http://www.surrey.ac.uk/MME/QSA/
Queen Mary/Westfield College http://www.chem.qmw.ac.uk/surfaces/
Institute of Physics(UK) http://www.iop.org
Royal Society of Chemistry(UK) http://www.rsc.org
American Chemical Society (ACS) http://www.acs.org
American Institute of Physics (AIP) http://www.aip.org
IUVSTA http://www.vacuum.org/iuvsta/default.asp
Materials Research Society (USA) http://www.mrs.org/

150
 References from the Text Footnotes:
4. Tanaka A., J. Surf. Analysis, 1 189 (1995)
5. M.P. Seah and M.T. Brown, J. Elec. Spec., 95 (1998) 71-93
6. Shirley D. A., Phys. Rev., 55, 4709 (1972)
7. Doniach S. and Sunjic M., J. Phys. 4C31, 285 (1970)
8. Evans S., Surf. Interface Anal., 17, 85 (1991)
9. Castle J. E. et al., J. Electr. Spectr. Related Phenom,106, 65 (2000)
10. G. Wertheim, J. Electron Spectrosc. 6, 239 (1975)
11. Tougaard S., Surf. Interface Anal., 25 137 (1997)
12. Tougaard S., Surf. Interface Anal., 11 453 (1988)
13. Jo M., Surface Science, 320 191 (1994)
14. Fiedor J.N., Proctor A., Houalla M. and Hercules D.M., Surf. Interface Anal., 20 1 (1993)
15. Do T., McIntyre N.S., Harshman R.A., Lundy M.E. and Splinter S.J., Surf. Interface Anal., 27, 618 (1999)
16. (to be added)
17. K. Harrison and L.B. Hazell, Surf. Interface Anal., 18, 368 (1992)
18. P.J. Cumpson and M.P.Seah, Random Uncertainties in AES and XPS: Peak Energies, Areas and Quantification. NPL
Report DMM(A) 26 ,May 1991; Surf. Interface Anal., 18, 345 (1992) and 18, 361 (1992).
19. S. Evans, Surf. Interface Anal., 18, 323 (1992).
20. Beamson G. and Briggs D., The XPS of Polymers Database - CD-ROM , Surface Spectra Ltd (2000)
21. Jones F. et al, Fluoride uptake by glass ionomer cements: a surface analysis approach , submitted to Biomaterials.
22. Press W.H. et al, Numerical Recipes in C , Cambridge University Press (1988)
23. Cumpson˚ P. J. and Seah M. P., Random Uncertainties in AES and XPS , Surface and Interface Anal., 18 361 (1992)

Appendix 5: Quick Reference Card

(following two pages - for printing)

151
 Keyboard & Mouse Shortcuts

Function Shortcut*

New Experiment Frame Ctrl+ N

(Browser)

Open ISO File Ctrl+ O

Print (graphic) Ctrl+ P

Copy (graphic to clip- Ctr+ C

board)

Set ‘Normalise’ point Shift+ (left click) To Contact Us

Zoom out (cycles) Ctrl+ (right click)
Phone:
Context menu (right click)
+44 (0) 1625 535346
Version 2.0
Edit a data point Ctrl+ Shift+ (left
click) E-Mail:
Display (spectrum) F1 [email protected]
QUICK REFERENCE CARD
Overlay (spectra) F2 World Wide Web:
Zoom out F3 http://www.casaxps.com Export Tab ASCII
Reset Zoom F4 Export MetaFile
Copy Display MetaFile *
* F1 - F4 are single press “Function“ keys: for the other
commands, hold down the “control” (Ctrl) and/or “shift” key
and press the listed key or mouse button. Main File Access Bar
“Drag and Drop” - you may load ISO 14976 files by dragging
them from Windows™ Explorer and dropping onto an open
CasaXPS Programme Frame.

Casa Software Ltd.,

Printing and Help 26 Burford Crescent
Wilmslow Copy Display bitmap*
Cheshire SK9 6BN Save File to Disk
Print Current (spectrum) Display United Kingdom
Convert File
Screen Preview for Display Print
Open File * to Clipboard
Help / About Casa XPS
New File (Experiment Frame)
 The Toolbar (upper) buttons in general provide ac-
The Button bars File Access buttons File Processing buttons cess to menus or dialog boxes.
Variable/Calibration controls The Options bar (lower) buttons in general execute
actions immediately or (reversibly) change the for-
mat of the display.

“Hovering” over a button (placing the mouse screen

pointer on an icon without clicking) produces a
descriptive label for that button (a “Tool Tip”), and a
Display Options buttons Display Properties buttons Block comment/info controls slightly longer description in the status bar
Display Scaling buttons Display Modifier buttons
Function Buttons F5 - F10 mirror the Processing
buttons (Page Layout - Element Library)
Decrease intensity scale maximum
Launch Library window Increase intensity scale maximum Toggle normalised display
Reset Intensity scale to original maximum Toggle subtracted display
Launch Annotate window
Increase Energy scale range*†
Launch Processing window
Step zoomed energy scale left* Display Modifier buttons

File Processing buttons Display Scaling buttons

Toggle background
Step zoomed energy scale right* Toggle residuals display
Launch Quantification window
Reset to original scale Toggle components display
Launch Tile Display window
Zoom out (step back round history)
Launch Page Layout window Toggle (shaded) region display
Zoom in (requires selection box)
* rescales intensity to suit NB. Modifiers require prior function definition
† adds 20% left and right

Display Options buttons Display Properties buttons Comment, Information & Variable control

Edit Block Comment

For Display
Edit Block Information

† “tri-state” buttton

For Browser
Insert many blocks into current scale Toggle Offset for multiple traces †
Set Experiment variable (linear)
Insert one block into current Display scale Toggle 2D, 3D, and Factor Space Display †
Edit Species/Transition parameters
Display all selected blocks in a tile Toggle Counts and CPS Intensity scales
Edit Source/ Analyser parameters
Display one block per tile Toggle Binding and Kinetic Energy scales
 Index

Symbols Browser 15, 16

~GL(50) 99 browser 7
“Dayta” System Files 127
’GL(p)’ 60 C
’GL(p)K(b0,b1)’ 60 CasaXPS 5
’GL(p)T(k)’ 60 installing 6
’SGL(p)’ 60 starting 6
’SGL(p)K(b0,b1)’ 60 terminology 6, 7
’SGL(p)T(k)’ 60 Colours 28, 138

Numerics D
1E37 21 data
Zooming 26
A Data Display 15
AET Data Editor 49
apparent error in the test vector 84 Define Custom Colors 28
Analyser Derivatives 50
response function 55 Differential charging 55
annotation display
history 32 3-D plot 28
moving 32 Geometry 27
repositioning 15 window 7
annotation dialog 11 Display Parameters 29
artificial peaks Doniach Sunjic 56, 58, 59, 61, 150
test data 47 DS800 Binary Files 123
asymmetric line shapes,alternative 58
Auger lines 39 E
Eclipse Files 124
B Element Library 16
Background Subtraction 65 loading 40
Backgrounds 43 Element Library dialog 9
Beamson and Briggs 107 Enhanced Metafiles 31
block Excel 54
window 7 experiment file 8
Briggs and Seah 61 Experiment Frame 9
 Index

experimental variable 22 line separator 20

parameter exclusion list 20
F Partially Encoded 19
F line shape 64 Selecting 26
F.W.H.M 40 spectrometer geometry 20
File Formats 123 experiment. 20
File formats Experimental Variable 22
DS800 Binary Files 123 File Format 19
fonts 29 File Structure 20
Quantification 24
G Transmission Functions 23
Gaussian/Lorentzian
Product 57 K
Sum 57 Kappa 59
Gelius, Ulrich 58, 61
Geometry 137 L
Graph Annotation 17
Line shapes 44
Asymmetric Blend 58
H
Doniach Sunjic 58
H line shape 63 F profile 64
Gaussian/Lorentzian 57
I H form 63
IBM 5 line shapes, including background 59
Identifying Peaks 41
IE 86 M
IND 86 Macintosh computer 20
Installing CasaXPS 6 Microsoft Foundation Class 5
integration regions 13 Microsoft Windows 95 5
intensity calibration 23 Multiple Document Interface 8, 25
ISO 14976 5
Binding vs Kinetic Energy 22 N
blocks 16, 20
Name/Formula List. 53
Data Blocks 21
NPL 20
Experiment header 21
 Index

O RSF 112
offset spectra 27 RSF values 24
RUSTI 69
P
Page Layout dialog 15 S
peak labels 32 Shirley 42, 43, 59
Peak Parameters Shirley background 74
optimisation 47 Simulating Spectra 69
PHI MultiPak ASCII files 130 splitter 25
Phonon broadening 55 bar 7, 25
polymer database 109 window 25
Processing 17 SSI M-Probe Files 129
Starting CasaXPS 6
Q Synthetic Component 44
Quantification 17
Calculation 45 T
Custom Report 46 Tables
Propagating 49 quantification 17
Report 45 Tag field 113
Quantification Parameters dialog 11, 13 test data 47
quantification regions 31 text annotation 33
Tile Display 27
R Tiles 26
R.S.F 40 tiles
REELS 66 preferred layouts 27
Regions 42 reducing scrolled list 27
name 42 rows & columns 27
Regions Property Page 12, 13 Tougaard 42, 43, 66
REGULAR scan 22 transition
Relative Sensitivity Factors 22 name 39
REP transmission correction 24
real error in the predicted vector 84 transmission encoding 146
RET trend analysis 114
real error in the target vector 84 typefaces 29
 Index

U W
universal cross-section Windows 98 7
adjusting 68
X
V X-Axis 136
VAMAS x-ray line shape 55
definition 18
VAMAS Files 17 Y
VGX900 files 125 Y-Axis 136
Vision 1.x/2.x ASCII files 128
Voigt 57, 59, 61 Z
Zoom Reset 14
zoom states 26
© 2001 Casa Software Ltd., 26 Burford Crescent, Wilmslow, Cheshire SK9 6BN, United Kingdom

Tel: +44 (0) 1625 535346 E-Mail: [email protected] World Wide Web:http://www.casaxps.com