CompleteEASE Manual PDF
CompleteEASE Manual PDF
CompleteEASE Manual PDF
05
CompleteEASE
Data Analysis Manual
[email protected]
[email protected]
http://www.jawoollam.com
402.477.7501
402.477.8214
Trademarks
CompleteEASE is a trademark of the J. A. Woollam Co., Inc. All other trademarks used within
are the property of their respective owners.
Copying and Distribution
You are welcome to make backup copies of the CompleteEASE software and documentation for
your own use and protection. However, you are not permitted to make copies for others or for
resale.
Contents
1.
Introduction
1.1.
1.2.
1.3.
1.4.
1.5.
1.6.
1.7.
1.8.
2.
Data Collection
2.1.
2.2.
2.3.
2.4.
2.5.
2.6.
24
Contents iii
3.
3.2.
3.3.
3.4.
3.5.
3.6.
4.
4.2.
4.3.
iv Contents
44
93
4.4.
4.5.
4.6.
5.
5.2.
5.3.
5.4.
5.5.
137
Contents v
5.6.
6.
Reference - Graph
6.1.
6.2.
7.
7.3.
7.4.
7.5.
8.2.
8.3.
8.4.
8.5.
vi Contents
198
193
8.
203
8.6.
8.7.
8.8.
8.9.
9.
9.2.
242
Contents vii
10.
10.2.
10.3.
10.4.
11.
11.2.
11.3.
viii Contents
Appendices
11.1.
253
259
1. Introduction
1.1.
Snapshots
A snapshot will compress your data file, model, and fit results into a single file with
ss extension. Thus, a single snapshot file can be stored or sent and then opened to
get back to the same result that you saved. For more details on saving or opening a
snapshot, see Section 3.6.
0BIntroduction 1
Figure 1-2. CompleteEASE can view and access two levels of sub-directories within the
Linked folder.
Multi-Processor Capability
Due to material and physics limitations the single-core processing performance of
current microprocessors is beginning to plateau. Future increases in processing
power will come by adding multiple processing cores to the system. This is already
beginning; many computers have 2 cores, and in the very near future, all computers
will likely have at least 4 (or more) cores.
CompleteEASE has been re-written to take advantage of multi-core capability. For
time-consuming operations, CompleteEASE can split the analysis across multiple
cores. While performance does not scale linearly with the # of cores, it does improve
significantly. During testing of 2 core machines the performance improvement is
1.7x. Testing of new computers with 4-core and 8-core processors showed
improvements of 3.2x and 4.8x, respectively.
Figure 1-3. CompleteEASE can be configured to use processing power from multiple cores.
Model roll-back
Make a mistake and it is now easier than ever to roll-back to an earlier point
during your analysis session. Rather than relying on a single-step reset,
CompleteEASE now keeps track of your model before and after each Fit. The Reset
list is cleared each time a new data file is opened.
Figure 1-4. Roll-back feature to reset to an earlier point during a single fit session.
2 0BIntroduction
Anisotropic Modeling
The anisotropic modeling capabilities for both Uniaxial and Biaxial materials have
been added to CompleteEASE. In fact, with right-click access to convert any layer
to its anisotropic equivalent, it is easier than ever to explore the world of anisotropic
modeling. For more details, check the Examples of Section 5.5.
0BIntroduction 3
4 0BIntroduction
1.2.
Manual Overview
The organization of the manual is briefly described below.
Chapter 1: Introduction
Chapter 1 will introduce you to ellipsometry. There is good information about basic
ellipsometry, along with the installation instructions, a list of short-cut keys, and a
list of the conventions used throughout this manual. New to this manual, we have
added a section regarding New Features of Interest as many users are upgrading
from earlier versions of CompleteEASE. In addition, new sections on File Structure
and User Setup have been added.
Chapter 2: Data Collection
Chapter 2 provides a quick overview of data collection. As the details can vary
depending on the instrument and configuration, this chapter is meant only for quick
reference and the formal information should be sought from the Hardware manual
for your instrument.
Chapter 3: Data Analysis 1 Basic
There are three chapters devoted to data analysis procedures. Each steps through
examples provided with this manual. Chapter 3 describes the basic SE data analysis
algorithms used by the CompleteEASE software. All users can benefit from
understanding how the most basic models are constructed and operate. Examples in
this chapter include:
Oxides on Silicon
Metal Substrates
0BIntroduction 5
Multi-Sample Analysis
Model Non-Idealities
Anisotropic Films
Chapter 7:
Measurement Tab
Chapter 8:
Analysis Tab
Chapter 9:
Hardware Tab
Chapter 10:
Options Tab
6 0BIntroduction
Appendix 11.1
Trouble-Shooting Guide.
Appendix 11.2
Appendix 11.3
1.3.
COMPUTER
SPECIFICATIONS
REQUIREMENTS
Operating System
Memory (RAM)
256 MB
120 MB
USB Ports
Processor
Display
Color display
at least 800x600 pixel resolution
0BIntroduction 7
Figure 1-9. CompleteEASE Setup choices. The software can be installed for use with an
ellipsometer system or for data analysis only.
The next message you will receive asks which tasks you would like the installation
software to perform. The choice ranges from creating icons to associating
CompleteEASE file types. These possibilities are shown in Figure 1-10. If you
choose, a short-cut icon (Figure 1-11) will be placed on the Windows Desktop to
start the CompleteEASE program. However, do NOT start CompleteEASE until
following the system connection instructions. The software installation process
should take less than two minutes on most computers.
8 0BIntroduction
0BIntroduction 9
1.4.
To help the user navigate this manual, the following conventions will be followed to
describe features in the CompleteEASE software.
Tabs
When the software is used with an alpha-SE system, there are typically four Tabs,
as shown across the top of the software screen: Measurement, Analysis,
Hardware, and Options. An additional In Situ tab is available for M-2000 and
RC2 systems that may be applied to real-time measurements. Throughout this
manual, Tabs will be written in bold and italics.
Screen Panels
Within each Tab, there are screen areas referred to as panels. From the
Measurement tab shown in Figure 1-1, the panels include System Status,
Measurement Controls, and Fit Results. Throughout this manual, panels will be
written in bold, blue text.
Buttons
Figure 1-13 shows the Hardware tab for CompleteEASE. Notice that there are
four panels: Hardware, Signal, Calibration, and Show Logs. Within each panel
are various buttons. Buttons will be designated with single quote marks, such as
Re-initialize and Park Z-Stage.
10 0BIntroduction
Figure 1-13.
Model Commands
Figure 1-14 shows the Analysis tab. Within the Analysis tab, there is a Model:
panel. The model includes many choices some of which are shown in bold,
underlined, red letters. Within this manual, they will also be shown in bold,
underlined, red letters. For example, Add, Delete, and Save. Model details are
further divided into expandable sections, shown in bold letters preceded by a red + or
- symbol. The symbol preceding the text shows whether the section is collapsed (+)
or expanded (-). These sections will be listed using bold text with a red + symbol:
+MODEL Options, +FIT Options, and +OTHER Options.
Mouse-selected Menus
The CompleteEASE software also utilizes both left- and right- mouse-buttons to
access additional menus and features. Throughout this manual, if you need to select
the right or left mouse button, you may see the following symbols: (R) or (L),
respectively. Figure 1-15 shows the menu when the right mouse button (R) is
pressed within the Fit: results box.
0BIntroduction 11
Figure 1-15. Right-clicking mouse (R) in Fit results section (white area) brings up a
special menu related to the Fit.
Pipelining Descriptions
The descriptions used in above sections can be pipelined to form a convenient
description of software features or menus. In this case, the description will take on a
similar form as:
Tab>Panel:>Button
For example, the Open button within the Model: Panel under the Analysis tab is
conveniently described as:
Analysis>Model:>Open
As another example, the command to right-click the mouse within the Fit: Panel
from the Analysis tab and then choose the Add to Fit Log menu choice would
read as follows:
12 0BIntroduction
1.5.
File Structure
The CompleteEASE software is installed on your hard-drive in a single directory.
However, it is helpful to understand the sub-directories used by CompleteEASE.
Figure 1-16 shows the sub-directories within the CompleteEASE folder.
The following table provides more details about each of these subdirectories and
their function within CompleteEASE.
Table 1-2. Directory Information within CompleteEASE
DIRECTORY
cnf
DAT
HTML Clipboard
jar
MAT
MOD
Recipe
FUNCTION
Hardware configuration files
Common Data files, including a sub-folder of
Examples that are shipped with the
CompleteEASE software. Any data files saved in
this directory can be viewed and opened by all
users, but are read-only. This directory is not
intended for individual user files.
Files and Graphs used by CompleteEASE to
construct reports within the HTML Clipboard.
Java Application files
Common Material Files to describe optical
constants of materials you may encounter. This
folder is further divided into: Advanced, Basic,
Dielectric, Metal, and Semiconductor folders.
Files within this directory can be viewed and
opened by all users, but are read-only.
Common Model Files to describe a sample and
the associated analysis strategy. Further divided
into Advanced, Basic, and Calibration Wafers.
Common Recipe Files describe a complete
measurement and model strategy. This is pertinent
to variable angle or variable measurement point
systems and is not used with the alpha-SE system.
0BIntroduction 13
Although the default directories will always be used by CompleteEASE, the user can
create their own directories to store Data files, Models, Materials, and Recipes. To
maintain the default files, any sub-directories created within the primary Directories
(as listed in Table 1-2) will be treated as read-only. For example, the three
subdirectories within MOD (Advanced, Basic, and Calibration Wafer) are read-only.
COMMON Location
Files can also be saved directly within the Mod directory. This location is
considered the COMMON directory for Models. Similarly, files located directly in
the \DAT or \MAT directory are considered the COMMON locations for Data files
and Material files, respectively. To demonstrate this, Figure 1-17 shows the \MOD
directory structure, along with a corresponding Analysis>Model:>Open dialog
box. Notice the same folders are shown in the dialog box, with the addition of
Recent. Most folders are shown as blue, which represents they are read-only. The
only yellow folder is Common as it maintains full access and shows the two files that
were directly within the \MOD directory.
Figure 1-17. Structure within the MOD directory along with the associated Open Model
dialog box.
Folder Links
The folders listed for Data, Models, Materials, and Recipes in CompleteEASE are
actually links to folders on your computer. The actual folders can be located
anywhere on the computer. CompleteEASE will retain a list of links that you have
made to these folders. To view the actual location of a folder, (R) on the folder
and computer location will show up at the top of the list, as shown in Figure 1-21.
NOTE: If the folders on your computer are moved, deleted, or renamed, the link
inside CompleteEASE will be broken. This can cause unstable behavior and
CompleteEASE should be closed and re-opened to re-establish links still available.
14 0BIntroduction
When you press the Add Folder Link button, a directory of your computer will
appear in the CompleteEASE Folder Browser, as shown in Figure 1-19. If you need
to make a new folder, you can press the Create New Folder button.
NOTE: Before pressing Create New Folder, you need to position your mouse on
the computer location where the new folder is to be created. After pressing the
button, verify that the location is correct from the follow-up menu before confirming.
Figure 1-18. Open Data dialog box, showing the Folder Links: to data folders on your
computer.
1. Select
location
for new
folder
2. Verify location, name
new folder and press
OK
Figure 1-19. CompleteEASE Folder Browser is used to search through folders on your
computer. To create a new folder, first select the correct location and then press Create New
Folder button. Second, type a name for the new folder. It is always wise to verify the
location where the new folder will be created on your computer.
0BIntroduction 15
Figure 1-20. Open Data list, showing the new link to Tom-2009 Data folder.
Figure 1-21. Right-click (R) on a folder to access options such as Remove from List and
Default Save Directory.
16 0BIntroduction
Viewing SubFolders
In older versions of CompleteEASE, access was only granted to individual links.
In CompleteEASE 4.0, we have expanded each folder link to view 2 sub-levels of
folders below the linked location. An example of this is shown in Figure 1-23.
Here, the only linked folder is called Measurements. CompleteEASE
automatically shows 2 levels of subdirectories within this folder.
Figure 1-23. Open Data window with a folder called Measurements linked. This allows
CompleteEASE to view and access two levels of sub-directories within the Measurements
folder.
0BIntroduction 17
CompleteEASE can link to any folder on your computer, so you can place the link
at different locations to view different levels. You will always view only the 2 sublevels below the linked folders. To demonstrate this, Figure 1-24 shows links to
three different folders: James Data-2009, Measurements, and Project A. It
turns out that all of these folders are in the same filepath and when expanded, they
view the same sub-folders. However, because the Measurements link is already 2
levels above Project A, it does not expand enough to see the third sub-levels within
Project A; namely Run 1, Run 2, and Run 3.
Figure 1-24. Three links have been created pointing to James Data-2009,
Measurements, and Project A. Even though these folders are within each other, each
will expand to show sub-directories two levels further down.
Figure 1-25. (R) on a folder name and press Default Save Directory. The location will
show up in bold and will become the default location when saving and opening files. Notice, it
shows up as bold in all different possible link locations.
18 0BIntroduction
1.6.
2.
Change Users Have Own Folders List to True, by selecting the parameter
ON check-box.
3.
NOTE: After completing these steps, you must restart CompleteEASE before the
changes take effect. However, you may also want to setup the User List before
restarting (next step).
Figure 1-26. CompleteEASE configuration dialog box. The General parameters area has
been expanded to view the User Log-In Mode, Always Add CE Folders, and Users
Have Own Folders List choices needed to setup multiple users.
After configuring CompleteEASE for multiple users, you will want to setup each
User with their own account. Press Options>Miscellaneous>Manage Users and
the User Management box will appear, as in Figure 1-27. Press Add New User and
type in the name and password for each user you wish to add.
NOTE: If the password area is left blank, no password will be required to log-in.
0BIntroduction 19
After entering each of the Users with their own account, you should close
CompleteEASE and restart. This will allow all of the changes to take effect. When
you restart CompleteEASE, it will open by asking to enter your User Name and
Password. The user names can be found in the drop-down list, as shown in Figure
1-28.
Figure 1-28. Log-in is required when multiple users have been setup. The User Names can
be selected from a drop-down box.
Any folders that are added by an individual user at this point will be shown for their
account only. However, if the folder is located within the CompleteEASE directory,
it will show up for all users. For this reason, it is preferred to create a folder outside
of the C:\CompleteEASE directory for files that individual users want to access. For
my example, a separate folder has been created at C:\Measurements. Within this
folder, each user has their own sub-directory and CompleteEASE can be configured
by each user to link to only the folders they need to view.
20 0BIntroduction
1.7.
Short-cut Guide
There are two types of short-cuts within CompleteEASE. First, like with all
software, if a letter is shown that is underlined, the short-cut for that command is to
hold ALT button and press that letter. For additional short cuts that are more global,
incorporate the CTRL, SHIFT, and ALT buttons. These additional short cuts are
listed in Table 1-3.
Table 1-3. Short-cut reference.
SHORT CUT
FUNCTION
CTRL-M
Go to Measurement Tab
CTRL-A
Go to Analysis Tab
CTRL-H
Go to Hardware Tab
CTRL-O
Go to Options Tab
CTRL-P
CTRL-D
CTRL-N
CTRL-C
CTRL-S
CTRL-1
CTRL-2
CTRL-T
CTRL-Z
CTRL-SPACE BAR
CTRL-L
CTRL-ALT-L
CTRL-R
CTRL-V
CTRL-ALT-O
CTRL-ALT-W
CTRL-ALT-S
CTRL-ALT-SHIFT-I
SHIFT-Mouse roller
CTRL-SHIFT-Mouse
roller
0BIntroduction 21
If you are working with Uniformity Mapping data, the following short-cuts can help
manipulate the data and graphs.
Table 1-4. Short-cut reference for mapping data.
SHORT CUT
FUNCTION
CTRL-Click on Point
Select/De-Select Point
CTRL-ALT-Click on Point
Delete Point
CTRL-SPACE Bar
1.8.
tan( ) ei = =
rp
rs
(1-1)
In this equation, rho () is defined as the ratio of the reflectivity for p-polarized light
(rp) divided by the reflectivity for s-polarized light (rs). is a complex number, and
the ellipsometric parameters simply report this value in polar form: tan() is the
magnitude of the reflectivity ratio, and is the phase. Ellipsometry offers three
main advantages over simple intensity-based reflection or transmission
measurements:
1.
2.
3.
The high accuracy, precision, and sensitivity of the ellipsometric measurement make
it highly suited for demanding thin film metrology applications.
22 0BIntroduction
p-plane
s-plane
s-plane
plane of incidence
In Spectroscopic Ellipsometry (SE), Psi and Delta values are acquired as a function
of wavelength. This greatly increases the information content in the data set,
enabling the simultaneous determination of multiple sample properties. However, to
extract sample parameters such as film thickness and optical constants from the
measured SE data set, an optical model must be built to fit the data. The
CompleteEASE software provides a graphical user interface for building models and
displaying measured data and model fits. The fundamentals and theory of data
analysis is explained in more detail in Section 3.1. The CompleteEASE program
also provides a simple interface to the SE hardware, making acquisition of accurate
SE data fast and easy.
For more information on ellipsometry theory, see the following references:
1.
2.
3.
4.
5.
6.
7.
0BIntroduction 23
2. Data Collection
The first step to any ellipsometry measurement is collection of data with the
ellipsometer. These operations are better discussed in the hardware manual for each
instrument. CompleteEASE is used with alpha-SE, M-2000, RC2, and AccuMap-SE
ellipsometers from the Woollam Company. The details for data collection depend on
both the system and individual ellipsometer configuration. This chapter provides a
basic overview of CompleteEASE data collection.
NOTE: Instructions in this chapter are representative of various systems. For more
complete instructions, including details on system alignment and calibration, please
consult your corresponding Hardware manual.
2.1.
alpha-SE Systems
CompleteEASE uses a simplified screen for data collection with the alpha-SE, as
there is only one angle of incidence and no mapping to consider. The Measurement
tab for an alpha-SE will appear as in Figure 2-1.
There are four choices to make before collecting data. These are the measurement
Mode, the type of Sample Alignment, the Model to use for data analysis, and
whether to Save Data after Measurement. The first three options provide a dropdown menu, as shown in Figure 2-2. Table 2-1 describes each choice in more detail.
24 1BData Collection
Figure 2-2. Options for measurement Mode, Sample Alignment, and Model.
Option
Description
Mode
Standard
Fast
Sample
Alignment
Long
Transmission
M.M.
M.M. Long
None
Standard
Robust
A thorough alignment of the light beam that searches over the full
translation range. Use for glass substrates with multiple reflected beams.
Fixed Height
Prompt Height
Allows user to enter the sample stage height for measurement. It is often
helpful to first perform a z-stage scan from the Hardware tab.
Press Choose from File Dialog.
Model
Common Models
For convenience, you can move Models that are commonly encountered into the
Common folder. To move into the Common folder location, go to the Analysis
tab and press Open from the Model: section. Search for the model of interest and
then drag the model name (right side of Figure 2-3) with mouse to the Common
folder (left of Figure 2-3). You can do this with multiple models.
1BData Collection 25
Figure 2-3. From the Analysis tab, press Open Model and then you can drag-and-drop
models from any folder (listed on right) into the Common folder (shown on left).
All models that you move to the Common folder will now appear in the drop-down
menu from the Measurement>Measurement Controls>Model drop-down menu.
This is shown in Figure 2-4, where the Common folder (left) has two models and
these same models appear in the Model drop-down menu (right).
Figure 2-4. The models moved to the Common folder (shown on left) will be quickly
accessible from the Measurement tab pull-down Model menu (seen on right).
Measuring a Sample
Once the Mode, Sample Alignment, and Model are selected, you can press the
Measure button to begin the measurement, as seen in Figure 2-5. As the alpha-SE
has automated sample alignment, the system will align to the sample (using the
selected Sample Alignment) and then collect measured data (using the selected
Mode). Finally, if a Model was selected, the data will be analyzed to determine the
resulting film properties.
26 1BData Collection
Figure 2-5. Pressing Measure will start the data collection procedure. The System Status
will be updated to show the current hardware operation.
If the Save Data After Measurement check-box was selected, a dialog box will
appear, as shown in Figure 2-6. Find the location you wish to save the file within the
Folder Links: area and type in a File Name and Comment. If the folder link does
not exist, you can create a new folder or add a new link. These instructions were
detailed in Section 1.5.
NOTE: CompleteEASE data files are encrypted & saved with .SE or .iSE
extensions. These files are only accessible by the CompleteEASE program.
Figure 2-6. Save Data dialog box. Type a File Name and comment and then press Save.
After SE data is acquired, the model analysis will be performed and the fit
parameters displayed in the Fit Results panel. The model fit to the measured data is
also displayed in the graph: the black dashed Model curves should lie essentially
on top of the colored Psi and Delta measured data curves if the model fits the data
well. If this is not the case, the wrong model was selected to analyze the data. The
final result for a thin oxide on Silicon is shown in Figure 2-7. The model parameters
will be reported in the Measurement>Fit Results section, while a graph of Psi
and Delta versus wavelength are displayed at the bottom of CompleteEASE along
with corresponding Fit curves from the model.
1BData Collection 27
Figure 2-7. Successful measurement of thin dielectric film using an alpha-SE system.
28 1BData Collection
2.2.
Figure 2-8. Measurement tab for an M-2000 system with automated angle of incidence,
automated sample alignment, automated sample translation, focusing optics, and an
integrated camera.
Measurement Recipe
The measurement controls for this type of system are integrated into a Recipe. Each
Recipe describes the three basic components of data collection: 1) Data Acquisition,
2) Mapping Scan settings, and 3) Modeling.
The drop-down box from Measurement>Measurement Controls>Recipe: allow
the following options: 1) Prompt for Acquisition Parameters, 2) Prompt for Recipe
Components, 3) list of all Common recipes, 4) Choose From File Dialog, and 5)
Create/Edit Recipe. These options are shown in Figure 2-9.
1BData Collection 29
<Select a Recipe>
This listing is not an option, but rather a note to the user that they need to Select a
Recipe or change this setting before they can proceed.
30 1BData Collection
Common Recipes
All recipes that are saved in the Common Folder (C:\CompleteEASE\recipes) will
appear in the drop-down list as shown in Figure 2-9. In this example, the Recipe
called 55, 65, and 75 3sec, 3mm thick quick was created and moved to the
Common Folder. To move a recipe to the Common Folder, select the Choose From
File Dialog option that is described next. This will open the Recipe Dialog box, as
shown in Figure 2-12. Navigate through the folder links on left to find the recipe
you would like to add to Common folder. Then, drag-and-drop the recipe listed
under Files: to the Common folder listed under Folder Links:.
1BData Collection 31
Create/Edit Recipe
The final button allows you to create or edit a recipe. This selection opens the same
box as shown in Figure 2-11. Each of the options for a recipe will be described later
in this chapter.
The three primary components of any recipe are the Acquisition Parameters, Scan
Pattern, and Model; as shown in Figure 2-13. Each of these components is discussed
in further detail.
32 1BData Collection
2.3.
Acquisition Parameters
The Acquisition Parameters describe how the data will be collected at each point of a
Measurement Recipe (whether single-point or scan). A common box for Acquisition
Parameter choices is shown in Figure 2-14.
Figure 2-14. Acquisition Parameters Setup for M-2000 and RC2 systems. Choices listed
above can vary from system-to-system.
Data Type
The Data Type specifies the type of data to collect during the measurement, as
shown in Figure 2-15.
1BData Collection 33
Standard
Most applications require a Standard measurement. This is simply a standard
Spectroscopic Ellipsometry measurement. The data collected will include Psi and
Delta versus wavelength and angle of incidence. In addition, the depolarization and
intensity reflected will be measured. However, the measurement accuracy of
reflected intensity and depolarization depends on the accuracy of recent intensity
baselines and DC calibrations, respectively. For more details, please consult your
Hardware manual. Most models will ignore the depolarization and Intensity
measurements, unless instructed to consider this extra data.
Generalized Ellipsometry
Measurement of the complete Jones matrix of a sample, including the off-diagonal
elements (which are assumed equal to zero for Standard SE measurements). This
type of measurement is useful for anisotropic samples where the optical axis is not
aligned parallel or perpendicular to the plane of incidence. Generalized Ellipsometry
measurements are not compatible with measurements that exhibit depolarization, as
it is based on a Jones matrix description of the interaction between measurement
beam and sample.
Generalized Ellipsometry measurements consist of three ratios, so you will end up
with three Psi curves and three Delta curves for each angle of incidence that is
measured.
Mueller Matrix
Measurement of the polarization change using Stokes vectors to describe the light
beam. The Mueller-matrix is a 4 X 4 matrix, but not all elements can be measured
with each ellipsometer. In fact, the only ellipsometer to measure all 16 elements of
the Mueller-matrix is the RC2 system. For most samples, there is significant
redundancy in the Mueller-matrix and no need to measure all 16 elements. For
isotropic samples (with or without depolarization), the complete interaction can be
described by measuring 4 elements (M11, M21 or M12, M33 or M44, and M34 or
M43). If the sample also contains anisotropy, the number of necessary elements
climbs to 7 or 8. Only for the most advanced samples are more elements required for
complete characterization. Thus, less than 1% of all Ellipsometry Applications
require Mueller-matrix measurements.
Transmission Intensity
Measure Intensity of transmitted light through the sample.
Reflection Intensity
Measure Intensity of reflected light .
Sample Alignment
Sample Alignment options depend on the type of system and configuration. In
general, there are two basic steps to sample alignment: Tilt alignment and Height
alignment. Common selection for each are described in this section.
34 1BData Collection
Tilt Alignment
Tilt alignment refers to the tip and tilt of sample relative to the incoming light beam
such that the reflected beam is aligned on the receiver unit at the correct angle of
incidence. Tilt alignment options include Skip, Manual, or Automated; as shown in
Figure 2-16.
The Tilt alignment is often Skipped when using focusing probes, as the tip-tilt
alignment becomes less sensitive. It can also be skipped for applications where
common samples are measured that remain flat. For example, if a system is
continuously used to measure 200mm silicon wafers, the tip-tilt alignment of each
wafer will probably be very similar from sample-to-sample.
For Manual or Automatic Alignment, the system will study the reflection of a light
beam off the sample that is incident on a 4-quadrant detector. This detector allows
the beam to be centered onto the 4-detector elements by adjusting tip and tilt until all
4 detectors have the same intensity. This method allows very repeatable alignment
of the beam from sample-to-sample. An example of the Tilt Alignment step for an
Automatic System is shown in Figure 2-17.
1BData Collection 35
36 1BData Collection
Figure 2-20. When opening Acquisition Parameters, you can find files that you would like to
see immediately from the Recipe drop-down menu and drag them into the COMMON folder
shown on the left of this box.
1BData Collection 37
2.4.
Scan Pattern
The Scan Pattern is used to describe where each measurement should be performed
across a sample. This option is only needed for ellipsometers with automated sample
translation. To make a new Scan Pattern, press Edit/Create button to the right of
the Scan Pattern selection of the Recipe Components box, as shown in Figure 2-21.
Figure 2-21. Choose Edit/Create Scan Pattern from the Recipe Components dialog box.
The Scan Patter Editor, shown in Figure 2-22, allows you to describe both Circular
and Rectangular samples. There are many options to automatically fill-in the points
to measure, including Grid Fill, R-T Grid Fill, and Line Fill. Most of these
selections are intuitive and best learned by testing each button a few times.
38 1BData Collection
As an example, if you press the Grid Fill button, you will get the options shown in
Figure 2-23. The Scan Pattern also describes the overall Alignment methodology.
For example, you can choose to Align at a single point only. You can also choose to
do a full alignment at the first point and then z-alignment at all other points.
Figure 2-24. Open Scan Pattern Editor. To move routine Scan Pattern files from any
location, simply drag them with the mouse from the list on right to the COMMON folder listed
to the left of this screen.
1BData Collection 39
2.5.
Model
A model is a description of how the data will be fit by CompleteEASE to
determine film thickness and optical constants (or other material properties). Models
will be described in significant detail in the upcoming chapters, with examples of
how to work with many common models. You can select any model of interest to
analyze the data during the recipe. For scans of multiple points, the model will be
used to analyze data from each point.
Common Models
For convenience, you can move Models that are commonly encountered into the
Common folder. To move into the Common folder location, go to the Analysis
tab and press Open from the Model: section. Search for the model of interest and
then drag the model name (right side of Figure 2-25) with mouse to the Common
folder (left of Figure 2-25). You can do this with multiple models.
Figure 2-25. From the Analysis tab, press Open Model and then you can drag-and-drop
models from any folder (listed on right) into the Common folder (shown on left).
All models that you move to the Common folder will now appear in the drop-down
menu from the Recipe Components box.
40 1BData Collection
2.6.
For maps of multiple points, the Graph can be toggled to Graph the parameters for
all points measured up to the current point. To do this, press the Graph Parms
button within the Fit Results section. The parameter to graph can be selected from a
drop-down list above the Graph, as shown in Figure 2-27.
1BData Collection 41
When the Scan is completed, a table of results will be presented in the Fit Results
panel, as shown in Figure 2-28. The results of any model that are run during recipe
are saved with the measurement data and can be opened at a later time by pressing
the View Previous Results button from the Fit Results pane. This opens the
window shown in Figure 2-29.
42 1BData Collection
1BData Collection 43
We have found that the best way to learn software features is through a series of realworld examples, where the user can follow along. To this end, we have developed
examples that spread over multiple chapters to teach the common CompleteEASE
data analysis procedures. Following this tutorial is instructive for both beginning
and expert users. Chapter 3 begins with basic data analysis examples. Here, we
concentrate on thin films that are transparent. Chapter 4 continues with intermediate
concepts, concentrating on absorbing materials and the extra difficulties involved
with their characterization. Finally, Chapter 5 introduces a few advanced topics.
The section examples from this chapter are listed below, along with the primary
CompleteEASE features that are discussed.
Section 3.1 General Data Analysis Theory
Section 3.2 Samples with Known N,K: SiO2 on Si
FEATURES IN THIS EXAMPLE
Opening model and data files
Fitting data
Thickness Pre-Fitting
Global fits
Expanding layers
Depolarization data
Grade Layer
HTML clipboard
Creating a report
Fit log
Open/Save Snapshot
3.1.
2.
A layered optical model is built which represents the nominal structure of the
sample. This model is used to generate SE data.
3.
Model fit parameters are defined, and then automatically adjusted by the
software to improve the agreement between the measured and model-generated
SE data. This is known as fitting the data.
4.
The results of the fit are evaluated. If the results are not acceptable, the optical
model and/or defined fit parameters are modified and the data is fit again.
The above steps will be described in more detail in subsequent sections. Keep in
mind that while the basic SE data analysis approach is straightforward, real-world
samples can often be difficult to analyze. The CompleteEASE software is designed
to simplify common SE data analysis procedures which work for many types of
samples, but there is no substitute for experience when dealing with complex
samples. Please contact your J.A. Woollam Co., Inc. representative if you require
assistance.
= ( n +i k
1
= sin( ) 2 1 + tan( ) 2
1
(2-1)
t1
n,k (film 2)
t2
n,k (film 1)
n,k (substrate)
Figure 3-2. Schematic representation of a layered optical model with 2 films, parameterized
by thicknesses t1 and t2, and optical constants n & k.
(a)
(b)
Figure 3-3. Layered optical model in CompleteEASE, with model generated data: (a) the
model generated data (black dashed lines) are not in agreement with the measured SE data
(colored curves), (b) when the top layer thickness is adjusted appropriately, the model
generated data lies directly on the measured data.
Sometimes an ideal layered model does not adequately describe the optical behavior
of the actual sample. The CompleteEASE software can also model two common
non-idealities: surface roughness and index gradients. Surface roughness is
shown in Figure 3-4: to model the actual sample which may have a non-abrupt
rough surface, an effective roughness layer is added to the model. The optical
constants of the effective roughness layer are derived by mixing the optical
constants of the underlying material with the optical constants of void (which has
optical constants of n=1, k=0). The Bruggeman Effective Medium Approximation
(EMA) is used to calculate the optical constants of this mixed layer assuming 50%
void content. While the effective roughness layer approach is certainly an
approximation to the actual sample, this approach works extremely well for
modeling SE data when the size of the surface roughness is much less than the
wavelength of light used to measure the sample. For most SE systems, this implies
that surface roughness features must be less than ~40nm.
Figure 3-4. Optical model for surface roughness: (a) actual sample with non-abrupt
rough surface, (b) optical model with effective roughness layer.
Sometimes the optical constants of a layer are not constant throughout the layer.
This may be caused by process variations during the film deposition. Figure 3-5
shows how CompleteEASE can model a graded film: the layer is divided into sublayers with small thicknesses, and each sub-layer has slightly different optical
properties. In CompleteEASE, a linear variation in the film index n is assumed for
the graded layer. This simplified approach, which is automated in CompleteEASE,
works well for modeling many types of samples.
Graded Film
Substrate
Substrate
Figure 3-5. (a) Continuous variation in the optical properties of the graded film, (b)
CompleteEASE approximation of the graded film with discrete layers.
[(
n
1
N Ei N Gi
3n m i =1
) + (C
2
Ei
CGi
) + (S
2
Ei
S Gi
) ]1000
2
(2-2)
2.
The N, C, & S parameters for the current data set in CompleteEASE can be plotted
by selecting the N, C, & S entry in the Graph Type: field on the Options tab.
The lower the MSE value, the better the fit or agreement between the measured
and model generated SE data. As the typical precision and accuracy of the measured
ellipsometric data in terms of N, C, & S is ~0.001, a multiplicative factor of 1000
is included in the MSE definition of (4-2). This implies that an ideal model fit
should have an MSE of ~1. While this is achieved on certain types of samples
(single thin films on Si substrates, for example), the best model fits for more
complex samples (with thick and/or multiple layers) may exhibit much larger MSEs
(>10) and still be considered acceptable. It is also possible to get an MSE smaller
than 1, but this implies the match to data is better than accounted for with random
measurement error. In other words, the results likely include some of the noise
present from data set.
The Try Alternate Models command automatically fits the data using the
ideal layered optical model, and then with surface roughness and index
grading non-idealities added to the model. The results of the various model
fits are summarized in tabulated and graphical format, and the user can then
decide which model is most appropriate for the sample.
The Thickness Pre-Fit and Global Fit options perform a large number
of trial fits to the data set, using a wide range of initial starting values for
selected fit parameters. The best fit model is then reported.
1.
2.
3.
If the model fit is deemed unacceptable, try modifying the optical model and/or
defined fit parameters, and repeat the fit procedure until an acceptable fit is achieved.
Finding an acceptable balance between the quality of the fit and the model
complexity requires patience and experience.
A.
Figure 3-6. Index must increase toward shorter wavelengths if k=0. These optical
constants are physical as they obey this rule.
Figure 3-7. Index must increase toward shorter wavelengths if k=0. These optical
constants are not physical because n turns-over (decreasing toward shorter
wavelengths) below 600nm without any absorption (k=0).
B.
Figure 3-8. Index will turn-over and decrease toward shorter wavelengths when
material becomes absorbing (k increasing toward shorter wavelengths). These
optical constants are physical as they obey this rule.
C.
D.
Figure 3-10. The graphed optical constants are unphysical because the
extinction coefficient, k, goes negative.
3.2.
Fitting data
Thickness Pre-Fitting
The simplest types of samples to analyze are those in which the optical constants of
each material are known. For this case, only layer thickness needs to be determined
in the optical model fit. Unfortunately, most samples do not fall into this category,
as the optical constants for most materials are rarely constant. Instead, they often
depend on the material deposition and processing conditions. The analysis examples
in the following sections demonstrate how to determine both layer thickness and
optical constants from SE data. However, in this section we show the classic
ellipsometry example which assumes known optical constants for the materials:
thermally-grown silicon dioxide films on silicon wafer substrates (SiO2 on Si).
For high quality thermally grown oxide films on semiconductor-grade Si wafers, the
optical constants of both the Si wafer substrate and the SiO2 film are known to be
constant. The CompleteEASE software has a standard model for analyzing this type
of sample. To open this model:
1.
2.
3.
Highlight (by clicking) the Basic folder in the Location list on the left
side of the screen.
4.
2.
3.
4.
After opening the specified model and data files, the CompleteEASE screen should
appear as shown in Figure 3-11. Note the experimentally measured SE data (the
ellipsometric Psi and Delta parameters versus wavelength) appear in the graph panel
as red and green curves, and the optical model contains a substrate (SI_JAW) and
two layers (INTR_JAW and SIO2_JAW).
Figure 3-11. CompleteEASE screen after opening the Oxide on Si model and the 25nm
Oxide on Si example data.
As an exercise, use the Si with Thermal Oxide model to determine the total film
thickness for each oxide on Si example file: 60nm Oxide on Si, 120nm Oxide on
Si, 300nm Oxide on Si, and 1600nm Oxide on Si. Open each data file
(Analysis>Data >Open) and click Fit; the results should agree with the values
shown in Table 3-1. There are a few important trends to note in this table:
1.
The MSE increases with thicker films (from ~1 to >10), though all data fits
are very good to the eye (i.e. black dashed Model curves lie on colored
experimental curves).
2.
Example File
MSE
25nm Oxide on Si
0.943
0.001 0.0023
27.42 0.007 nm
60nm Oxide on Si
0.687
0.037 0.0014
58.54 0.006 nm
120nm Oxide on Si
2.510
0.071 0.0044
124.61 0.007 nm
300nm Oxide on Si
3.673
-0.051 0.0070
273.02 0.014 nm
1600nm Oxide on Si
13.885
-0.094 0.0261
1640.58 0.195 nm
To simplify the model and provide more unique fit results, the interface layer
(Thickness #1) is fixed at 1nm. This value is typical for thinner oxide films
(thickness < 100nm); while for thicker oxides, a slight fit improvement can be
obtained by also fitting the interface layer thickness. For example, open the
1600nm Oxide on Si file and fit the data with the Si with Thermal Oxide
model. Now add the interface layer thickness as a fit parameter by right-clicking
the mouse on the Thickness #1 value (that is, (R) on 1 nm). The value is
now shown in a bold font followed by the word (fit), indicating that this value is
now defined as a fit parameter. (To turn off a fit parameter, simply right-click
(R) again on the parameter value.) Click on the Fit button, and the results box
shown in Figure 3-13 should appear. Note the fit value for Thickness #1 is 3.48 (an
increase from the nominal value of 1nm) and the MSE is reduced slightly from
13.885 to 10.316. Unless the explicit goal of the sample analysis is to characterize
the interface thickness, it is best to fix Thickness #1 at 1nm nominal value.
Figure 3-13. Results from Si with Thermal Oxide model fit to 1600nm Oxide on Si data
with interface layer (Thickness #1) added as a fit parameter.
Derived Parameters
Even though the interface thickness increased in the preceding fit, the Total
Thickness remained essentially the same (1640.58 vs. 1640.52 nm) for both fits. The
Total Thickness is not an actual fit parameter, as the value is derived by
summing all layer thicknesses in the model. Derived parameters can be added to the
results by turning ON the Include Derived Parameters option within Model
>+FIT Options (Figure 3-14). In addition to Total Thickness, additional derived
parameters include n and k values for a specified layer and wavelength, optical
thickness, total optical thickness and many other possibilities. These will be
addressed further in a later example.
1
C.M. Herzinger, B. Johs, W.A. McGahan, J.A. Woollam, and W. Paulson, Ellipsometric determination of optical constants for
silicon and thermally grown silicon dioxide via a multi-sample, multi-wavelength, multi-angle investigation, J. Appl. Phys. 83 (1998)
3323.
If you wish to delete a Derived Parameters, simply right-click on the number (shown
in blue) next to the parameter you wish to delete.
Figure 3-15. Fit parameter dialogs, showing how to (a) Hide or (b) Rename fit parameters.
High and Low specs for the fit parameter are also set in (b).
Check the Rename box, type in a different Name: for the parameter
(e.g., Top SiO2)
Type in values for Low Spec. (1200) and High Spec. (1400).
Click Ok to close the dialog box, and fit the data again (with the 1600nm Oxide
on Si data file loaded). Fit results will appear as shown in Figure 3-16. Since the
Top SiO2 value (1639.58) is outside the Low and High parameter Spec (1200
1400), an (Out Of Spec.) message appears after the result.
Figure 3-16. Fit results with parameter renamed and Top SiO2 parameter Out-of-Spec.
Thickness Pre-fitting
One further concept will be demonstrated using Si with Thermal Oxide model.
For this example, reload both the Si with Thermal Oxide model and the 1600nm
Oxide on Si example data file. From the Model>+FIT Options section, turn the
Perform Thickness Pre-Fit option to OFF by clicking on the underlined blue
value. Then press Fit>Generate. The screen should appear as shown in Figure
3-17. Note that the data generated by the model (black dashed curves) using the
default thickness (100nm) does not match the experimental data. Now click the Fit
button: the data fit is still very poor, and the reported MSE is very high (677). This
illustrates what happens if you start the fit with parameter values that are not
reasonably close to their best fit values.
Figure 3-17. Data generated from Si with Thermal Oxide model, using default parameter
values. The Perform Thickness Pre-Fit option is turned OFF.
Now set Thickness #2 to 1650 and Angle Offset to 0 (by clicking on the
respective parameter values), and click Generate. The screen should appear as in
Figure 3-18. While the model and experimental data curves do not line up exactly
(slight shift), at least the basic structure and features in the data sets appear similar.
Click the Fit button, and observe that a good quality fit is obtained.
Figure 3-18. 1600nm Oxide on Si data set, with model data generated using a closer
(1600nm) starting thickness value.
A good starting value for Thickness #2 is critical to achieve the best data fit.
Without the Thickness PreFit, the software tries to adjust the fit parameter
(thickness) to lower the MSE. However, the MSE profile is not a single minimum at
the correct thickness. It also contains many other minima, which are called local
minima. These local minima can trap the normal fit process. Each of these local
minima are at a much larger MSE value, as shown in Figure 3-19., so it is easy to
distinguish the correct answer from a thickness result that does NOT match the
Experimental curves. However, it is tedious to guess starting film thickness values
for all samples.
Figure 3-19. MSE profile for 1600nm SiO2 on Silicon data using the Thermal Oxide on
Silicon model. The only answer occurs when the SiO2 thickness is near 1640nm. However, a
normal ftt will only find this answer if the starting point is near the final answer. To overcome
this problem, our patented Thickness PreFit automatically searches for a good starting
thickness near the final answer.
Figure 3-20. Thickness Prefit uses a proprietary algorithm to very quickly estimate the
thickness of a film based on oscillations in the data.
Hold down the Shift key as you roll the mouse wheel up and down. The
model parameter is increased or decreased by a set increment and data are
re-generated with the new model values.
3.3.
Global Fits
Expanding Layers
The next set of examples will demonstrate how to determine the film thickness and
index of refraction n for transparent films deposited on substrates with known
optical constants. While there are certainly cases in which it is not known a priori if
the film is transparent (films which are absorbing are covered in the following
chapter), there are also many applications where the film can be assumed
transparent, at least in the visible spectral range: optical coating materials (SiO2,
TiO2, Ta2O5, MgF2 etc.), some nitrides (Si3N4, AlN), many organic films
(photoresists, PMMA, spin-on polymers, etc.), and so on. In addition, the presence
of periodic data oscillations suggests interference between top and bottom of a semitransparent film (as shown in Figure 3-21).
The index of refraction for transparent films in the visible spectral range is often
specified using the Cauchy dispersion relation given by Eqn. 4-3. The An
parameter relates to the approximate amplitude for the material index, while Bn
and Cn parameters provide the shape or curvature of the index versus wavelength.
Figure 3-22 shows index dispersion curves for three common transparent materials.
For most transparent materials in the visible spectral range, the index curves slightly
upward at shorter wavelengths, a behavior which is well described by the Cauchy
dispersion equation.
n( ) = An +
Bn
Cn
(4-3)
Figure 3-22. Index of refraction n for common transparent materials, described by the
Cauchy dispersion equation.
To determine the index and thickness of the Flow-on Glass film, open the Si with
Transparent Film model (which is also found under the Basic folder in the Open
Model dialog box), and click the Fit button. A nice data fit is obtained (though the
fit takes longer than the Si with Thermal Oxide data fit) with the fit results shown
in Figure 3-24. The reported fit parameters are Thickness #1 (which is the film
thickness) and the films Cauchy parameters: A, B, and C. The film index of
refraction at a specified wavelength (632.8nm) is also reported as a derived
parameter.
Figure 3-24. Fit results for Flow-on Glass on Si example file using the Si with
Transparent Film model.
To display the films index of refraction versus wavelength, right-click (R) on the
material name Cauchy Film and choose the option Graph Layer Optical
Constants (Figure 3-25). The optical constants will be displayed in the graph
window as shown in Figure 3-26.
To display and edit the film Cauchy parameters, click (L) the red + symbol to
the left of Layer #1. The layer expands (and the + turns to a -, which can be
clicked to collapse the layer), and the layer fit parameters are now accessible, as
shown in Figure 3-26. The k Amplitude, Exponent, and Band Edge
parameters describe an exponential Urbach absorption tail - not required for this
example.
NOTE: The Cauchy layer is described in further detail in Section 8.5
Figure 3-25. To view a layer optical constants, (R) over the layer name and choose to
Graph Layer Optical Constants.
Figure 3-26. Optical constants displayed versus wavelength. Also, the Cauchy layer is
expanded, displaying its available fit parameters.
Global Fits
Like the Si with Thermal Oxide model, the Si with Transparent Film model uses
the Thickness Pre-Fit option to provide a good starting value for the layer
thickness. To provide starting values for the film index, a Global fit is performed.
In a Global fit, a wide range of parameter starting values are attempted and the
values which result in the lowest MSE are then used as starting values for the final
fit. The Use Global Fit option is found within Model>+FIT Options. Turn this
option to ON to display additional settings which are used to customize the Global
fit procedure (Figure 3-27).
In the Si with Transparent Film model, the Cauchy A parameter for the film
(which is basically the approximate index of the film) is defined as a Global fitting
parameter. Up to 3 fit parameters can be used in the Global fit. The Min. and
Max. values for A are 1.3 and 3.0, and the # of Guesses is 30. When the Fit
button is pressed, the Global fit procedure performs 30 trial fits with different
starting values for A, distributed in equal increments between 1.3 and 3.0. At each
trial fit (i.e. for each value of A - which corresponds to the film index), a thickness
pre-fit is also performed such that the entire array of film thicknesses and indices are
attempted. A final fit is performed using starting parameter values from the trial fit
which had the lowest MSE.
Figure 3-27. Settings for the Si with Transparent Film model; note the Use Global Fit
and Include Derived Parameters sections.
Since the Global fit process is computation intensive, a status bar with a Stop
button is displayed during the Global fit (Figure 3-28). To speed up the Global fit,
only a subset of the experimental data points are used during trial fits based on the
Number of Data Points to Use option. The number of iterations used in the trial
fits can also be limited. The default values for these options (20 and 5) are adequate
for most samples, but they can be increased if necessary: larger values may improve
the robustness of the global fit, but with the downside of increased analysis time.
Likewise, the Min., Max., and # Guesses for the parameter(s) may also need to be
tuned to handle specific types of samples.
Figure 3-28. Status box and Stop button displayed during a Global Fit.
3.4.
Transparent Substrates
NEW FEATURES IN THIS EXAMPLE
Graphing Multiple Data Sets
In the previous example, we introduced the Cauchy dispersion relation to model the
index of refraction for transparent films. In addition, the Cauchy can also be applied
to transparent substrates. This is the most common approach for substrates of glass
or plastic and the Glass Substrate model will use a Cauchy to determine the index
for glass substrates. This example will demonstrate the modeling of transparent
substrates.
With transparent substrates, we must consider a possible complication the light
may reflect from the back surface of substrate and reach the detector. The backside
reflections will join the surface reflection, but these two collected beams will be
incoherent. This is an optical description for light beams that have lost phase
information relative to each other. In this case, the backside reflection will only
adjust the overall amplitude of the signal and not produce oscillations like we see
for thin films. This amplitude offset from backside reflections can lead to an error
in the reported substrate index if not handled correctly. Fortunately, the backside
reflections can be corrected in the software (described below).
If the backside reflections can not be avoided, then they need to be included in the
software calculation. This will be demonstrated during the example. Fortunately,
most Woollam SE systems collect depolarization data, which provides evidence of
whether backside reflections are present in the collected data. If the depolarization is
essentially zero, then no backside reflections reached the detector. This is the
preferred condition. If backside reflections reach the detector, the CompleteEASE
software can incorporate incoherent interference modeling to correct for the
situation. Lets consider data from a glass substrate to better understand this
situation.
2
R.A. Synowicki, Suppression of Backside Reflections from Transparent Substrates, Phys. Stat. Sol. (c) 5, No. 5 (2008) 1085-1088.
Figure 3-29. (R) on Open command to access the Append Data command.
When multiple data sets are appended, CompleteEASE allows all three to be
accessed in the Graph window (either individually or the complete set). Select the
Graph Type command at the top-left of the graph panel and choose to compare Psi.
From this same location, also turn off the Double Y Axis. Finally, choose the
Graph All Data Sets at the top of the Graph panel. This will allow a comparison of
the Psi curves collected from all three data sets as shown in Figure 3-30.
Figure 3-30. When multiple data sets are appended, there are new choices at the top of the
Graph Panel to go to any specific data set. Selecting the Graph All Data Sets check-box,
allows comparisons with all data sets graphed.
Figure 3-31. Right-click (R) within the graph area to access the Graph Scratchpad.
Figure 3-32. CompleteEASE Graph Scratchpad that allows further customization to the
graph.
Figure 3-33. The Graph ScratchPad is used to compare data from glass substrate where the
backside is roughened or taped to suppress backside reflections.
Before proceeding, clear the contents in the Graph ScratchPad by viewing the
ScratchPad and pressing Delete All Curves.
In a later example, we will learn how to fit multiple data sets simultaneously.
However, for this example, we will focus on fitting each individual data set. Start by
opening the glass substrate_rough data file. Next, open the Glass Substrate
model from the Basic folder. Press Fit and you should get the result as shown in
Figure 3-34. The data appear noisy, but this is primarily due to the graph scale.
Figure 3-35. Index of refraction from Cauchy model fit to bare glass substrate.
Graphing Depolarization
Ellipsometry measurements use polarized light. If the sample is isotropic and ideal,
this measurement can be described by two values: and . However, if a sample is
partially depolarizing, two parameters are insufficient to fully describe the change in
polarization. The patented Rotating Compensator technology used in the alpha-SE,
M-2000, and RC2 systems permits measurements of three reported quantities: , ,
and %depolarization. The %depolarization should remain zero for ideal sample
conditions.
HINT: Some ellipsometers require DC measurement to acquire accurate
depolarization data. Please consult your hardware manual (or contact JAWCo) to
best measure the depolarization.
Shortcut:
CTRL-Z to graph
Depolarization
Before proceeding, press on Graph Type at the top-left of the Graph. Change
from Psi & Delta to Depolarization. Notice, the Depolarization graph in
Figure 3-36 is near zero. This indicates that little light is reflected from the
backside into the detector.
Figure 3-36. Depolarization from uncoated glass slide (no backside reflections).
Shortcut:
CTRL-P to graph
Psi and Delta.
Next, open the glass substrate_smooth data file. The graph of Depolarization is
now centered at almost 2%. This indicates the backside was smooth and
incoherent reflections from the smooth backsurface reached the detector unit.
Change the Graph Data back to Psi. Press Fit:>Generate. The model no
longer matches the experimental measurement even though this data is from the
same glass substrate (only with backside reflections). The backside reflections do
more than introduce depolarization they also shift the ellipsometry values. Fit
the data with the same model and you should get results as shown in Figure 3-37.
Figure 3-37. Fit results and optical constants for glass substrate_smooth data modeled
without backside reflections.
Before proceeding, view the optical constants and add these new results to the Graph
ScratchPad - allowing comparison of the two results in the ScratchPad as shown in
Figure 3-38. The first fit result should be correct, as the glass did not include
backside reflections. The second fit, with a much lower index of refraction is
incorrect because the backside reflections were present in the data, but were not
considered in the model. Fortunately, there is a method that can account for
backside reflections in measured data and help get correct results. This is considered
in the following section as it is a common occurrence with transparent substrates.
Figure 3-38. Comparison of fit results for refractive index from within the CompleteEASE
Graph ScratchPad.
Figure 3-39. Reflections from the backsurface of transparent substrate are incoherent. This
light can be avoided by roughening the backsurface to scatter the reflected light.
Figure 3-40. Model Options expand to allow Include Substrate Backside Correction. The
number of Back Reflections can be fit to best match the experimental data.
Fitting Depolarization
There is correlation between the substrate index and the amount of backside light
collected when modeling ellipsometry data at one angle of incidence. However,
depolarization data is only influenced by the amount of backside light collected. The
default setting is to ignore the depolarization data and only fit Psi and Delta. In fact,
the depolarization fitting selection is a hidden feature of the model. To show this
feature, go to the bottom of the Model: pane and press Configure Options. This
will bring up the dialog box shown in Figure 3-41. Check the box next to Include
Depolarization Data in the Fit Options section and then Ok. With this option
shown, choose Model:>+Fit Options and turn the Include Depolarization Data to
ON, as shown in Figure 3-42.
Figure 3-41. Press Configure Options to show additional Model options. In this example, we
will add Include Depolarization Data as shown.
Figure 3-42. The Use Depol. Data in Fit selection within the Fit Options section of model.
Press Fit again for the Glass Substrate_smooth data including the backside
reflections (with fit to # of Back Reflections collected) and using the Depolarization
data in fit. You will probably have to reduce the # of Backside Reflections down to
1 (from default of 5) to get the depolarization data to fit. When done correctly, the
result should appear as in Figure 3-43. The measurement of depolarization provides
sensitivity to the amount of backside reflected light that is reaching the detector. In
this case, the best fit is with 1.5 backside reflections, as shown in Figure 3-44 (where
the number of backside reflections is varied to see whether the number of backside
reflections is a unique result). This means the substrate is thin enough that the first
reflection from backside is completely collected in the detector, while the second
reflection is spatially shifted enough to only collect about half of this second
reflection. Of course, as the number of reflections increases, the amount of
additional light reaching the detector is much lower and the sensitivity goes down.
Thus, there is little difference in the calculation between 3, 4, and 5 total reflections
from the backside.
Before proceeding, lets view the optical constants from this result and copy the
graph to the Graph ScratchPad for further comparison with our previous results. The
final comparison of optical constants is shown in Figure 3-45. The index for
backside roughened glass is similar (but not exactly equal) to the results for smooth
backside glass when the fit is corrected including depolarization. If the backside
effects are not considered, the fit results significantly affect the index of refraction.
There are two possibilities as to the difference in final results when corrected:
Figure 3-43. Fit results including backside reflections and depolarization data.
Figure 3-44. Uniqueness Test for the # of Backside Reflections, showing that the best fit
results from about 1.5 backside reflections. The depolarization measurement helps to quantify
this result.
3.5.
Transparent Films - 2
FEATURES IN THIS EXAMPLE
Surface Roughness
Grade Layer
Figure 3-46. Fit results from analysis of ZrO2 on Glass example file, using the Glass with
Transparent Film model.
Figure 3-47. ZrO2 optical constants from fit using Cauchy dispersion layer.
The main difference between the Glass with Transparent Film and the Si with
Transparent Film model which was used in previous examples is the material file
used for the substrate: 7059_Cauchy instead of SI_JAW. As the optical
constants for 7059 glass are representative of many glass substrates, this model can
be used as a starting point for many transparent films on glass. More accurate results
could be achieved by replacing the 7059_Cauchy material with optical constants
measured on your bare glass substrate (see section 3.4). In addition, because the
substrate is transparent, backside reflections may be an issue. To test whether
backside reflections are present in this data, press CTRL-Z to view the
Depolarization data as shown in Figure 3-48. As the depolarization is centered at
zero, it is likely that the backside reflections were suppressed to avoid collection at
the detector. These issues are discussed in more detail in Section 3.4.
Figure 3-48. Depolarization data is near zero percent, which indicates that backside
reflections were not collected even though substrate was transparent.
Surface Roughness
An interesting feature to the current model is the inclusion of surface roughness (the
Include Surface Roughness option is set to ON). Actually, the best fit surface
roughness value for this sample is rather large: 11.17 nm. To test the importance of
roughness in this model, try fitting the data without surface roughness: toggle the
Include Surface Roughness = value to OFF by clicking (L) on the blue
underlined value. Since we already know the nominal thickness and index, it is also
wise to turn OFF both the Use Global Fit and Thickness Prefit. Go ahead and
click the Fit button, which will fit the data without roughness resulting in a poor
fit and higher MSE, as displayed in Figure 3-49.
Figure 3-49. Fit to ZrO2 on Glass data with surface roughness OFF.
Graded Index
Surface roughness, which is one of the model non-idealities discussed in section
3.1, is very important to achieve a good model fit on this sample. Another nonideality described in section 3.1 is index grading. With the surface roughness turned
off, try grading the film as described below:
(R) on the name Cauchy Film of Layer #1. You should see a dropdown menu, as shown in Figure 3-50. Choose Grade Layer.
Figure 3-50. Right-click on Cauchy Film to access drop-down menu, then choose to Grade
Layer.
NOTE: Surface roughness and index grading often have the same effect on modeled
data, as they can both reduce the optical density of a material toward the surface. If
both give similar MSE values, it is up to the user to judge which is best for their
sample.
While the fit appears quite good to the eye, the MSE is slightly higher than for the
model which included surface roughness (2.1 vs. 1.175).
Figure 3-51. Fit to ZrO2 on Glass data with index Grading ON.
It is also possible to combine surface roughness and index Grading in the model:
turn the surface roughness back ON and press Fit (leaving the %
Inhomogeneity defined as a fit parameter). The results shown in Figure 3-52
indicate this model yields the lowest MSE yet (0.78).
Figure 3-52. Fit results for ZrO2 on Glass data fit with model that includes both surface
roughness and grading.
Figure 3-53. Remove Grading by right-clicking on the Graded Layer name and then select
Remove Grading.
The process of trying optical models with and without surface roughness and index
grading is so common in SE data analysis that CompleteEASE has a command to
automate the procedure. Once a good starting model is established, click
Model>+OTHER Options>Try Alternate Models, as shown in Figure 3-54. Fits
will automatically perform with the ideal optical model, with surface roughness, with
index grading, and with roughness & grading. Fit results are summarized in
tabulated and graphical formats, as shown Figure 3-55. The highlighted green
column is a suggestion of the best optical model, based on 25% improvement in
MSE for each additional fit parameter.
Figure 3-54. Model panel with the +OTHER Options section expanded to show the Try
Alternate Models option.
3.6.
HTML Clipboard
Creating a Report
Fit Log
Save/Open Snapshot
Figure 3-56. Fit to SiNx on Si-1 data file using Si with Transparent Film model.
Copy to Clipboard
Each pane (white area) in the Analysis tab of CompleteEASE offers right-click
(R) menu choices. Examples of these right-click menus are shown in Figure 3-57.
In addition, the Graph also features a right-click menu, as shown in Figure 3-58. To
assist in creating reports, the model, fit results, and various graphs can be copied to
the clipboard and then pasted in various programs such as Word, Excel, or
PowerPoint. The Fit results can be copied to the clipboard using Formatted or Table
choices, where the former is best for pasting into a word processor and the latter is
preferred for pasting into a spreadsheet. The Model can be copied in its entirety or
as the layers only.
Figure 3-57. Right-click (R) menu choices from the Fit: and Model: panes.
HTML Clipboard
In addition to the standard clipboard, there is a choice from Model, Fit Results and
Graph to use the HTML clipboard. This is a special area that can be used to create a
report and then later copy in its entirety to a separate document. In addition,
CompleteEASE has a quick report that can be automatically created from the Fit:
panel. After you have finished analyzing a sample, such as the SiNx film that we are
currently working with, press the (R) from the Fit: panel and choose Add Report
to HTML Clipboard. Alternatively, you can press CTRL-R to achieve the same
function. After copying the Report into the HTML clipboard, press CTRL-V to
view the HTML report (or access the right-click menu again and choose View
HTML Clipboard). The clipboard will contain fit results, model and graph in a selfcontained report, as in Figure 3-59.
Fit Log
Now, work through all five SiNx on Si samples in the Examples folder. To record
the final fit results for each film, right-click (R) on the Fit: pane and Add to Fit
Log after the fit is finished. This will save the current fit results. Type in an
appropriate name for each fit result.
The short-cut key to add results to the Fit Log is CTRL-L.
After all five data files have been fit and added to the Fit Log; choose to View Fit
Log from the right-click (R) menu of the Fit: pane. This should bring up a Fit
Log with each of the five entries, as in Figure 3-60. You can press on any of the
individual log entries to see the results from that sample.
The short-cut key to View the Fit Log is CTRL-ALT-L.
Figure 3-60. Fit Log showing results from all five SiNx on Si data files.
Comparing Results
Now, select all five Log Entries by selecting the first entry with your mouse and then
holding the shift-key down while you press the last Log entry. With multiple Log
entries selected, press the Compare button. This will create a Table of results for
each sample, as in Figure 3-61. The columns/rows can be reversed by a simple
button click. Also, the statistics can be added with a click. This table can be
conveniently copied to the clipboard.
Figure 3-61. Comparison Table showing results from all five SiNx films.
Figure 3-63. Choose the layers to compare and the type of graph.
Figure 3-64. ReAnalyze Data window opened from within the Fit Log.
Generate Report
Now, use another one-step approach to create a report for all the results. View the
Fit Log (CTRL-ALT-L) and select all five entries. Next, press the Generate
Reports button, as shown in Figure 3-65. This copies the results of each log entry
into the HTML Clipboard. When finished, press the View Reports button to view
the HTML clipboard, as seen in Figure 3-66.
Figure 3-65. All log entries are selected and the Generate Reports button is pressed to copy
all results into the HTML Clipboard.
Figure 3-66. HTML Clipboard after a report has been generated for all entries in the Fit Log.
Snapshot
A better way to send single fits to a colleague, or save them for future use, is to save
the Snapshot of your fit result. A snapshot will compress your data file, model, and
fit results into a single file with SS extension. Thus, a single snapshot file can be
stored or sent and then opened to get back to the same result that you saved. To save
a Snapshot after performing a Fit on a SiNx data example, simply right-click your
mouse over the Model:> Save button, as shown in Figure 3-67. To open the
Snapshot, simply right-click your mouse over the Open button above the model.
Figure 3-67. Save a Snapshot of your result by right-clicking mouse button over the Save
button in the Model: panel.
Save Model
Building a Model
Starting Mat =
Saving a Model
Zoom All
Gen-Osc Layer
Parameterize Layer
Tauc-Lorentz
Cody-Lorentz
Parameter Uniqueness
Graph ScratchPad
4.1.
Metal Substrates
FEATURES IN THIS EXAMPLE
Save Optical Constants
Save Model
Building a Model
Substrate optical constants are an important consideration when modeling thin films.
Although certain materials, such as crystalline silicon are well understood and will
remain constant, others substrates such as glass and metals are more variable. In
these cases, it is best to measure a bare substrate to determine optical constants.
Glass substrates were considered in Section 3.4. We will consider a metal substrate
for this example. Although the substrate is called a metal, it is actually a glass or
silicon substrate that is coated with an optically opaque metal film. The
measurement beam will not penetrate through the metal film, so we can consider it a
substrate.
Open the Au_Substrate data file. Substrates offer a special case, as the
measurement involves only a single-interface (reflection from surface). The
experimental measurement of Psi and Delta can be directly inverted to determine n,k
using the pseudo-substrate approximation. To keep the optical functions smooth, a
B-Spline layer can be used for the model. While this example will use the B-Spline,
we will save the intimate details for a later example. In this exercise, we will build a
model to fit any metal substrate and then save this model for future use.
To start, open the Multilayer (empty) model from the ADVANCED folder, and
your screen should appear as in Figure 4-1. Left-click (L) on the Substrate =
none. This will bring up the Open Material dialog box. Choose the B-SPLINE
material file from the Basic folder and press Open. Expand the B-Spline layer such
that it can be edited.
The resolution defines how closely spaced to apply nodes when allowing the
optical constants to vary. The nodes are matched to the starting material file to
provide an estimate before fitting. It is not necessary for the starting material to be
the same as the material you have measured this is the reason for choosing Ag in
this example.
Figure 4-1. Starting the example for Au_substrate with the Multi-Layer (empty) Model.
Generate data and then attempt the fit. It should quickly converge to the answer as
shown in Figure 4-2.
To view the optical constants as shown in Figure 4-3, right-click on Substrate = BSPLINE and choose Graph Layer Optical Constants. To view the spline points as
in Figure 4-4, press on Draw Node Graph within the B-Spline layer. The red and
green curves are the smooth optical constants (e1 and e2) that are derived from the
B-Spline fit, with the open circles showing the individual spline nodes.
As an exercise, change the Spline resolution to larger and smaller values and repeat
the fit. Does the fit quality change?
Figure 4-4. Spline points at 0.1eV resolution fit to the Au_Substrate data.
Save
When happy with results, press Save right above the model to save the gold optical
constants. This opens the Save Layer Optical Constants dialog box as shown in
Figure 4-5. Choose the Tabulated choice from the right. Next, press mouse on the
substrate layer itself. After entering the file name and comment, the tabulated n,k at
each measured wavelength will be saved for future use.
Figure 4-6. Right-click over a layer name to choose Save Layer Optical Constants from a
drop-down menu.
Figure 4-7. Choose whether to save the Parameterized optical constants (values used by
equation) or a tabulated list of optical constants at each wavelength.
Au Substrate Model
After optimizing the analysis for a new sample, the model can be saved for future
use. Lets use this Au substrate as an example. If additional Au substrates will be
measured in the future, a customized model would help reduce analysis time. For
this example, modify two additional settings:
Replace the Ag reference in the B-SPLINE layer with the metal optical
constants saved in this example (previous step).
Include Derived Parameters for both n and k of the substrate (make sure
to change layer # to 0).
Press the Save button at the top of Model: pane to save the final model as metal
substrate example as in Figure 4-8.
NOTE: If a model uses a new material file, both model and all material files must be
located in the same folder.
4.2.
This example uses the Au substrate result from Section 4.1. Thus, you will need to
complete 4.1 before proceeding to this example. To insure that the Au substrate
model works before proceeding, open the Au_substrate data file again and press
fit. If the fit does not work, then repeat Section 4.1 before proceeding.
Self-Assembled Monolayers
Self-Assembled Monolayers (SAMs) are very thin organic layers that consist of a
single monolayer. Thus, they are generally less than 10nm thick and often closer to
1-2nm. Spectroscopic Ellipsometry is very sensitive to layers of this thickness, due
to the phase information available from the measured Delta parameter. However, the
index of refraction from such thin layers is not typically attainable. Instead, the
nominal index of refraction must be assumed in order to fit the thickness. As layers
become thicker than 10-20nm, spectroscopic ellipsometry gains sensitivity to index
of refraction in addition to thickness. A common method for thinner layers is to
measure a similar material as a thicker film. From the thicker film, the index can be
measured and then fixed when measuring thinner layers. Because this
approximation may not always be true for such thin layers, the measurement of
SAMs is often qualitative and best used for comparison between multiple samples.
Open the Organic on Au data file and press Generate. The graph should appear
as in Figure 4-9. The Model dashed curves will represent a bare Au surface, as they
come from our model with only Au. Notice a shift in Delta toward lower values (at
all wavelengths). This indicates the presence of a thin surface coating. A larger shift
indicates a thicker film on the Au surface.
Figure 4-9. Data from Organic on Au compared to Generated Data from the bare Au
surface. Notice the shift in Delta toward lower values for the coated sample, which indicates
a surface coating.
Add a new layer to the Model by pressing Model:>Add. Position the blue bar above
the Au substrate and click to add the layer. Choose Cauchy layer from the Basic
folder and press Open. The default values for Cauchy layer will be close to the
values needed for generic organic films. Thus, we will fix the optical constants for
the Cauchy layer and only turn on the thickness as a fit parameter.
BEFORE proceeding, it is important to FIX the Au optical constants. If we fit at this
time, the Au optical constants would also move and confuse our overall result.
To prevent the Au substrate optical constants from fitting, make sure to do the
following:
Turn OFF the Fit Opt. Const. command inside the B-Spline layer, as
shown in Figure 4-10.
Figure 4-10. Turn OFF the Fit Opt. Const. to insure the Au optical constants stay fixed at
the values fit from the bare substrate data.
Turn on the Thickness of the Cauchy layer as a fit parameter (by clicking it with the
right mouse button). Press Fit and the resulting thickness of 7.48nm should be
reported. If you get a different thickness, make sure the Au optical constants did not
adjust from the previous fit.
Parameter Uniqueness
We will use this result to demonstrate a feature in CompleteEASE that helps to
visualize the model uniqueness. Go to the bottom of the Model: Panel and expand
the -OTHER Options. In this section, press the Parameter Uniqueness command.
You will be asked which of the fit parameters you want to select for testing. In this
case, we have only fit the thickness, so choose Thickness #1, as shown in Figure
4-11.
Next, you need to enter the Minimum value, Maximum value, and # of fit
points. These will define the test conditions. The parameter space between the
minimum and maximum will be divided into equal spacing to match the total number
of points entered. Then, the fit will proceed with the test parameter (Thickness #1 in
this case) fixed at each individual value. The MSE is captured from each fit for
graphing.
For this example, choose a Minimum Value of 2.5nm, Maximum Value of 15nm,
and 100 total points, as shown in Figure 4-12. You will get a warning that no other
fit parameters are selected, as shown in Figure 4-13. This is okay, as the test will
still proceed to show the MSE profile as the Cauchy thickness is varied, as shown in
Figure 4-14.
Figure 4-13. Warning message shown when the only fit parameter turned on is the test
parameter. It is okay to proceed past this warning to see the MSE profile as that single fit
parameter is varied.
Figure 4-14. MSE profile from Uniqueness test of the thickness for thin Self-Assembled
monolayer on Au when the optical constants are fixed.
Right-click on the graph area and copy the MSE profile data to the Graph
ScratchPad. This will allow us to compare the MSE curve for this fit with our next
test.
Next, allow the A, B, and C parameters in the Cauchy layer to fit. Repeat the
Parameter Uniqueness test with the same settings for Thickness #1. This time, the
result should appear as seen in Figure 4-15. Add this result to the Graph ScratchPad
and then view the ScratchPad to allow comparison of the two tests as shown in
Figure 4-16.
Figure 4-15. MSE profile from Uniqueness test of the thickness when the Cauchy parameters
for thin layer are allowed to vary at each point.
Figure 4-16. Comparison of MSE profiles from both tests, as seen in the Graph ScratchPad.
4.3.
Starting Mat =
Saving a Model
Zoom All
SIC
ZnS, ZnSe
TiO2
Figure 4-17. Data from SiC film on Silicon. The interference oscillations in data suggest the
film is transparent at longer wavelengths, while the decrease in amplitude of these oscillations
at short wavelengths indicates UV absorption.
Unlike the transparent films we have witnessed, the interference oscillations decrease
in amplitude toward shorter wavelengths. This is a strong indication of absorption in
the ultraviolet spectrum. The oscillations are dampened, as less light is able to
penetrate the film and return to the surface to interfere with the top-reflected light.
This concept is illustrated in Figure 4-18. To model the SiC film on Silicon will
require a material file that can describe both transparent and absorbing regions.
Figure 4-18. Interference oscillations occur when the light can travel through a thin film and
return from the bottom interface (red). If the light is absorbed, the light only comes from the
top-surface (blue) and will not produce interference.
Figure 4-19. Data fit to SiC on Si using the Si with Absorbing Film model.
Figure 4-20. Optical constants for SiC film using the B-Spline layer.
Figure 4-21. Settings for the Si with Absorbing Film optical model.
Clicking on Draw Node Graph displays the graph shown in Figure 4-22. This
graph shows the location of the spline control points. The B-Spline layer always
works in terms of the complex dielectric function = 1 + i2 = (n + ik)2, which is
often written as e1 & e2. One set of control points (and associated spline curve) is
used to specify the dispersion in e1 and a second set of control points is used to
describe e2. Using a smaller Resolution value adds more control points,
resulting in more flexibility to accommodate sharper features in the dielectric
constant spectra. However, setting the Resolution too low can result in unwanted
noise or unphysical artifacts in the dielectric function spectra.
Figure 4-22. B-Spline dielectric function e1 & e2, showing the location of the control
points (red and green dots), which are equally spaced in eV.
There are two ways to initialize the B-Spline layer control point values:
1.
Click on the n = and/or k = fields which sets all control points to this
new value over the entire spectral range.
2.
Click on Starting Mat = which chooses control points to match the optical
constant values from the specified material file.
The Si with Absorbing Film model uses the first method combined with the
Global Fit option to search a range of n and k values for the film. The Use Global
Fit section in Figure 4-21, specifies twenty n values ranging from 1.3 5 and three
k values ranging from 0 0.5.
Note: The Global Fit is a slow procedure and should be turned off if you have a
good starting guess for the optical constants. In this manner, you can save
significant time by avoiding the automated recipe especially when dealing with
multi-angle data sets.
The Fit Opt. Const. button allows the user to turn on and off the fitting of spline
points from within the B-Spline. Thus, an absorbing material could be fit using the
B-Spline and then the Fit Opt. Const. button can be toggled to OFF to fix the
optical constants of this layer in place for further analysis.
Note: To FIX optical constants, the Global Fit also needs to be turned off
In the B-Spline layer, the Use KK mode is an advanced calculation that maintains
Kramers-Kronig consistency between the resulting e1 and e2 spline curves. When
this feature is ON, the optical constants will be constrained to a physical shape thus
it should provide further assurance that the resulting optical constants are correct.
This feature will also reduce the total number of fit parameters, as only the e2 spectra
is described with a series of spline points while the e1 spectra is calculated via the
Kramers-Kronig transformation. Additional Advanced Options in the B-Spline
layer are further documented in Section 8.6.
108 3BData Analysis 2 - Intermediate
Range-Selecting Wavelengths
To manually perform a WVL-EXPAND fit, we first need to learn how to select a
subset of data over a specific wavelength range. This is simple in the
CompleteEASE program. First, choose a wavelength that will serve as an end-point
for the new data subset. Left-click (L) the mouse on this wavelength and hold the
mouse button down while dragging to the opposite end-point wavelength. If the
mouse drags off the side of a graph, all wavelengths in that direction will be included
(even expanding the graph to include wavelengths not shown).
As an exercise, open any data file and try the following:
1.
2.
Range select wavelengths of 600nm and above. For this, start the mouse-button
just inside the current graph (near 600nm) and then drag beyond the right graph
boundary to incorporate all longer wavelengths (Figure 4-24).
3.
Figure 4-25. Press Zoom All to select the entire wavelength range.
Now, lets return our attention to the WVL EXPAND fit. This fit works by taking a
limited wavelength range (as selected by the user), applying the current fit, and then
gradually increasing the range to both shorter and longer wavelengths until the entire
data set is included in the fit. The fit is performed at each new increment of
wavelengths.
To manually perform a WVL EXPAND fit, you choose the desired starting
wavelength region, and press +OTHER Options>WVL. Range Expansion Fit at
the bottom of the Model pane.
Figure 4-26. Global Fit and Wavelength Range Expansion Fit regions of the model for
absorbing film using B-Spline layer.
Figure 4-26 shows the additional options available when a WVL-EXPAND fit is
turned ON:
Adjust the Spline Layer n and Spline Layer k ranges to values which
closely bracket the nominal optical constants.
For thicker films (i.e., data sets with many oscillations) it may be necessary
to increase the # Guesses; thinner films can use fewer # Guesses, which
will reduce the data analysis time.
For thin films (<100 nm) the WVL-EXPAND Fit may not be required.
For very thick films (>1 2 microns), it may be necessary to specify a more
limited spectral range (it is preferable that the film is transparent in the
limited spectral range).
Unless the nominal layer thickness is accurately known (and set in model),
Perform Thickness Pre-Fit option should be ON.
Before beginning the global fit, make sure all other model parameter values
are set appropriately (for example, if Surface Roughness is included in the
fit, its starting value should probably be zero).
4.4.
Parameterize Layer
Tauc-Lorentz
Cody-Lorentz
In the previous example, we introduced the analysis of absorbing thin films via the
B-Spline layer. Another method for absorbing films that can be used independently
or after initial B-Spline modeling is the Gen-Osc layer. The Gen-Osc models the
absorbing features by describing the shape of the absorption. This example
demonstrates the common application of the Gen-Osc layer.
Parameterize Layer
The resulting B-Spline optical constants from the Si with Absorbing Film model
fit may be completely adequate for many applications. While the B-Spline layer
guarantees the optical constants versus wavelength will be smooth, it does not
always guarantee they are physical (that is, Kramers-Kronig consistent).
Furthermore, the B-Spline layer requires many fit parameters (two times the number
of control points), which may lead to correlation in the data fit. Once the general
shape of the optical constants are determined using the B-Spline layer, it is often
desirable to parameterize the optical constants with a simple analytic expression.
The Parameterize Layer command can conveniently perform this procedure.
NOTE: The Goal of Layer Parameterization is to match the shape of n,k from the BSPLINE fit using a dispersion model.
To parameterize the B-Spline optical constants for the SiC film in the previous
example, right-click on the B-spline layer and choose Parameterize Layer from the
drop-down menu, as shown in Figure 4-27. This opens Parameterize Layer dialog
box as shown in Figure 4-28, with the B-Spline layer optical constants plotted in the
graph. The default model (Gen-Osc) will automatically open in this window with a
starting oscillator (in this case, Cody-Lorentz).
Figure 4-27. Right-click on a layer to drop-down menu, then press Parameterize Layer.
Figure 4-28. Parameterize Layer dialog box, showing starting optical constants of the layer
along with the default Cody-Lorentz oscillator.
The Gen-Osc material file (which stands for General Oscillator) allows a
combination of various oscillator lineshapes. We want to match the shape of SiC
optical constants with an oscillator. There are many choices which are detailed in
Section 8.6. For this example, left-click on the Cody-Lorentz name and choose
Tauc-Lorentz from the Input box, as shown in Figure 4-29.
The basic steps for matching optical constants with a new oscillator model are as
follows:
1.
2.
3.
Each of these three steps will be demonstrated for the SiC example, where we match
the SiC B-Spline results with the Tauc-Lorentz oscillator parameters.
Figure 4-30. Left-clicking on the number next to an oscillator selects that oscillator for
manipulation.
Figure 4-31. Press the Expand X-Axis button to view all of the control points that
manipulate the oscillator, even if they are outside the measured spectral range.
Before we adjust any of the control points, it is easier to visualize the oscillator if we
change the OC View/Fit Type to Imaginary Part Only. This will show the green
curve of e2 (or k) from our B-Spline fit along with the dashed gray curve from the
oscillator we are adjusting. The goal is to match these two curves. If you position
your mouse over any of the gray control boxes, your pointer should change to an XY
arrow to show that you can now adjust that control box. This is shown in Figure
4-32, where the top control box is selected. This top control box can move both the
amplitude and the center energy. Practice moving the different control boxes to
better match the dashed curve to the green B-Spline data.
Figure 4-32. Cursor positioned above the top control point turns to an XY Arrow to show that
this control point can be adjusted to change the shape of the selected oscillator.
After the oscillator parameters have been adjusted to better match the reference e2
shape, right-click (R) on the Amp., Br, Eo, and Eg parameters. This will
turn them ON as fit parameters. Click the Fit button. In this case, we are not
fitting the experimental data, but rather matching the imaginary oscillator shape (via
the parameters) to the reference optical constants.
The fit varies the oscillator parameters to match the imaginary curve shape as entered
from the B-Spline. Figure 4-33 displays the fit results; the black dashed curve in the
graph is calculated using the Tauc-Lorentz dispersion formula.
Figure 4-33. Fit to imaginary values of SiC using a single Tauc-Lorentz oscillator.
Figure 4-34. The only parameters required when fitting Real Part Only are the Einf, and
two Poles.
For this example, it is best to adjust the parameter values before starting the fit.
When values are changed in the Layer Parameters box, the optical constant graph
is automatically updated. Enter a UV Pole Amplitude of 100 and the curve raises
upward. The parameters can also be dialed in by positioning the mouse above a
parameter, holding down the Shift key, and rolling the mouse wheel. Figure 4-35
shows the curves after the UV Pole Amplitude has been raised to get closer to the
reference curve. However, the tilt of the curve does not match. This is because
the UV Pole Position may also need to be adjusted. In general, the closer the UV
Pole energy to the measurement range, the more tilting that will occur.
Figure 4-35. Increasing the UV Pole Amplitude also raises the Gen-Osc e1 value, but the
blue curve does not match the tilt of the red reference curve from our B-Spline. This will
require adjustment of the UV Pole Energy.
After adjusting the UV Pole Position and Energy to better match the reference BSpline curve, the two values are fit to give the best match. Afterward, the Einf
value was also added to the fit to get a good match, as shown in Figure 4-36.
Figure 4-36. Fit to the B-Spline reference e1 value was achieved by varying (and fitting) the
UV Pole Amplitude, Energy, and Einf.
Replace Layer
After you are happy the Gen-Osc matches both real and imaginary components of
the reference B-Spline, press Replace Layer to apply the new oscillator model in
place of the B-SPLINE layer, as seen in Figure 4-37. The new oscillator can now be
used to fit data. Press Fit to see how well the Tauc-Lorentz oscillator can describe
the experimental data. If further samples with similar optical constants are
measured, this model could be used directly skipping the B-SPLINE layer step. If
the parameterization is not successful, press Cancel to revert to the existing layer
optical constants.
Note: Fitting data is different than Fitting within the Parameterize Layer dialog box.
In the former, you are allowing the oscillator parameters to adjust to best match the
experimental SE data while the latter is used to get starting oscillator parameter
values by fitting the shape of the reference optical constants.
Figure 4-38. Results using Glass with Absorbing Film model to fit the a-Si on Glass
data file.
Starting Mat
The two models for absorbing films in this example are very flexible and can be used
for many different materials. This is primarily due to the Global Fit of n,k.
However, the data analysis takes longer than average, as the global fit attempts all
possible combinations of n,k starting values. To minimize analysis time, the global
fit range could be reduced. However, this would require knowledge of the
appropriate range for the material. If the material is known, it is also possible to use
a similar material file as the Starting Mat inside the B-SPLINE layer.
For best comparison of analysis time, re-analyze the a-Si on Glass data set using
the Glass with Absorbing Film model. The analysis takes between 20 and 25
seconds on a Pentium D computer. Re-open the Absorbing Film on Glass model
and make the following modifications:
Expand the B-SPLINE layer and left-click Starting Mat = none. Open asi.mat from the Semiconductor Folder.
Press Fit:>Fit.
The data analysis should return the same result as before, but in less than one second.
Thus, this new model would be an alternative for a-Si when the optical constants for
the film are similar to the reference material file.
Dial each oscillator parameter by holding the SHIFT key down while the
mouse-wheel is rolled over top of the oscillator parameter. Holding CTRLSHIFT will make reduce the step-size of each adjustment.
2.
3.
Figure 4-40. Click the number next to an oscillator to view the oscillator and its control
points (in gray). As is often the case, the oscillator is outside the measured energy range, so it
is difficult to locate.
Figure 4-41. Press Expand X-Axis to better view the oscillator and control points (in grey).
The control points can be adjusted with your mouse to change the amplitude, center energy,
broadening and bandgap for the selected oscillator.
Figure 4-42 displays the fit results; the black dashed curves in the graph are
calculated using the specified oscillator. Go ahead and Replace Layer to add this
newly developed oscillator material file to the model. Fit the data and the result will
have an MSE near 1.2.
Next, expand the oscillator layer in the model and add the following fit parameter to
the fit: Ep. This allows added flexibility to the optical dispersion. The MSE
should reduce to 0.8 as shown in Figure 4-43.
Add the additional fit parameters (Einf, Et, and Eu) to the model to check for further
improvement. The question becomes whether slight MSE improvement warrants the
increased number of variables in the modeling process. As a general rule-ofthumb, each additional fit parameter should decrease the MSE by 20% to make it a
clear choice.
Figure 4-42. Optical constants of a-Si from SPLINE fit are matched with a Cody-Lorentz
oscillator.
Figure 4-43. Final results using Cody-Lorentz oscillator to fit a-Si data set.
Saving a Model
After the appropriate Cody-Lorentz fit parameters are determined for the a-Si film,
the model can be saved for future use. This process is quite simple: press
Analysis>Model:>Save. Choose the Folder where you would like to locate this
model and give the model a name. The Common folder is a central location for
easy access by all users. Figure 4-44 demonstrates this procedure.
Figure 4-44. Press Save from the Model: pane to bring up the Save Model dialog box.
4.5.
Open the data file a-Si Multilayer. This sample is similar to those discussed in
Section 4.4. The substrate is silicon with a SiO2 film underneath the a-Si. No model
exists for this type of sample, so it will need to be constructed. It is preferred to start
by modifying a similar model. Lets start with the model saved in Section 4.4 for aSi on glass using a Cody-Lorentz oscillator. Open this model from the Common
location. If you did not save the model, then repeat Section 4.4 before proceeding
and make sure to save the final model as suggested.
Figure 4-45. The Add command opens the dialog box shown. The blue bar can be moved up
or down to designate the location of the added layer.
Left-click the Thickness #1= 0.00nm designation for the SiO2 layer to change its
thickness to a nominal value of 100nm. Also select this thickness as a fit parameter.
Generate data and the graph should appear as shown in Figure 4-46. The generated
data does not match the experimental data. If a fit is attempted from this point, it
will not be successful. Press Fit.
Press CTRL-R to reset the model to the previous values after verifying the poor fit
quality.
The poor fit is due to the unknown thickness values. There are a couple approaches
to determine better starting thickness values for each layer. Both approaches work
better if the Cody-Lorentz oscillator parameters are turned OFF before proceeding:
Expand General Osc layer and right-click on each of the fit parameters that
was previously turned on, toggling them off.
Figure 4-46. Generate data with a nominal thickness of 100nm for SiO2 layer.
Approach #1: Use a Global fit for both Thickness #1 and Thickness #2. Make sure
to turn off the Thickness Pre-fit. This method can take a very long time if the search
includes a wide range of possible thicknesses.
Approach #2: Use the Global fit for the SiO2 layer thickness with the Thickness
Pre-fit turned on. This reduces the global fit dimension and works much more
quickly.
NOTE: Approach #2 will not work if the a-Si layer (Thickness #2) is used within
the Global Fit rather than the SiO2 layer (Thickness #1). The Thickness Prefit
always operates on the thickest layer in the model.
Figure 4-47 shows the successful fit of the sample. The current model does not
include the oscillator parameters, so the a-Si film must have similar optical constants
to the previous sample.
Figure 4-48. The Fit Options section of Model pane allow Derived Parameters to report the
optical constants from any layer in the model. With multilayer films, the correct layer # must
be entered.
The Cody-Lorentz parameters can now be turned on to improve the fit quality by
allowing the a-Si optical constants to vary. Figure 4-47 also points out that the
Derived Parameters are pointing to the wrong layer number. Within the Fit Options
section of the Model pane, change the Layer number from #1 to #2 for both the
Derived n and Derived k as shown in Figure 4-48.
Add the Cody-Lorentz parameters back as fit parameters (right-click on their values).
Turn off the Global Fit and Thickness Prefit (as the thicknesses are already well
known) and press the Fit button. The results should appear as shown in Figure
4-49. Save the new model for future use on similar samples.
Oxide or Roughness?
There is often a question whether the surface is oxidized or rough. In the model
above, surface roughness was implemented with a final value of 2.29 nm. Turn the
roughness off and AddLayer of SiO2_Jaw on top of Layer #2. Fit the data. The
final result should have SiO2 surface layer of 2.70nm with an MSE of 1.25. Thus,
there is very little difference between roughness and oxide for this sample. This is
true in general when the film thickness is below about 10nm. In the case of Silicon,
it is more likely the surface is oxidized than rough, but both models work.
4.6.
Parameter Uniqueness
Graph ScratchPad
Choose the Cr.mat file from the metals directory as the starting mat=
inside the B-Spline layer.
Parameter Uniqueness
Now, you are ready to press Fit. Your initial result should look similar to that
shown in Figure 4-50. Because there is large correlation, pressing fit again-andagain will result in different results with nearly the same low MSE. To visualize the
MSE profile, use the Parameter Uniqueness option under the +OTHER Options
section of the model. Figure 4-51 shows the results for a Parameter Uniqueness test
of the thickness from 10nm to 50nm. Before proceeding, copy this curve to the
Graph ScratchPad for later comparison. Although there is a minimum value for the
MSE around 45nm, the MSE value is below 0.4 for most of the thickness range.
Figure 4-50. Starting fit to only the SE data using a B-Spline layer.
Figure 4-51. MSE profile from the Parameter Uniqueness test of thickness when only fitting
the SE data for a Cr film on glass.
Append Data
To help break this correlation apparent when only fitting the SE data, lets add the
Intensity Transmission to the fit to provide additional information. We want both the
SE data and the Transmission Intensity data present in the fit. To this end, you need
to right-click (R) on the Open data command and select Append Data as in
Figure 4-52. Choose to append the Cr on glass_T data file from the EXAMPLES
directory. This data was collected from the same sample. With both data files
present, you can switch between graphs of SE and T data by choosing the data set
right above the graph, as shown in Figure 4-53.
Figure 4-52. Right-click on the Open command to choose Append Data when adding
Transmission data to the already opened SE data.
Figure 4-53. Choose the data set to graph by selecting entry just above the graph. Here, you
can see that both SE and T data for the Cr on Glass sample are open simultaneously. Thus,
they will both be included in the fit.
Figure 4-54. Adding the Transmission Data Weighting option from the Configure Options
section of the Model.
After turning on the Transmission Data Weighting, you will need to set this value to
100% in the Fit Options section of the Model, as shown in Figure 4-55.
Figure 4-55. Change the Transmission Data % Weight to 100% in the +Fit Options
section of the Model.
Go ahead and press Fit again. If the model thickness starts close to the correct
answer, the fit should find a thickness near 14nm. This is the final result. However,
if the Cr thickness is started too far away from 14nm, the fit can get stuck in a
local MSE minimum. Repeat the Parameter Uniqueness test to check the MSE
profile, which should produce a curve as shown in Figure 4-56. It is obvious that
there is only a single thickness that will provide the best MSE, although we can also
easily see how it is possible to get stuck in the local minimum near 45nm.
Again, copy this curve to the Graph ScratchPad. Now, we can compare the MSE
profiles from the previous two Parameter Uniqueness tests by viewing the Graph
ScratchPad, as shown in Figure 4-57.
Figure 4-56. MSE profile when fitting both SE and T data simultaneously for a thin Cr layer
on Glass.
Figure 4-57. Graph ScratchPad comparing the MSE profile when fitting both SE and T data
or only the SE data from a thin Cr layer on glass.
This chapter includes further examples to demonstrate both advanced and specialized
topics. Advanced topics include multi-sample analysis, creation of composition and
temperature dependent material files and working with anisotropic or non-ideal
models. A few of the examples use data from mapped samples, which help
demonstrate how to work with data collected at multiple points from a sample.
However, not all ellipsometers have this capability, so modeling of mapping data is
only necessary for specific systems.
The section examples from this chapter are listed below, along with the primary
CompleteEASE features that are discussed.
Section 5.1 Map Data and Model Non-Idealities
FEATURES IN THIS EXAMPLE
Show Map Data
Sellmeier Layer
Model Calculation
(Non-idealities)
Thickness Non-Uniformity
Bandwidth (nm)
Set Ranges
Zoom on Map
Multi-sample Analysis
Dynamic Fit
OC Library Mode
Build Library
(OC Compare)
Draw Graph
Difference Mode
Biaxial Layer
5.1.
Sellmeier Layer
Model Calculation
(Non-idealities)
Thickness Non-Uniformity
Bandwidth (nm)
Set Ranges
For this example, we will open a data file that contains multiple points scanned
across a silicon wafer with thick SiO2 coating. Our goal is to effectively determine
the SiO2 thickness and refractive index, along with the uniformity of the thin film
properties across the wafer.
Figure 5-1. Map Data for SiO2_Map data file. With symbols shown, we can visualize each
point on the wafer that was measured.
Figure 5-2. Spectroscopic Data from a single-point of the map. This view was gained by leftclicking the mouse on a map point.
Wavelength Units eV
To better view short wavelengths, go ahead and convert the graph Wavelength
Units to photon energy (eV) instead of nm. This will expand the UV region to give
near-equal spacing between the interference oscillations. Wavelength and photon
energy are inversely related. Thus, large photon energies are actually the shortest
wavelengths. To convert to photon energy, either use the short-cut CTRL-ALT-W
or go to the Options tab and change the Wavelength Units to eV within the Display
Units pane. Figure 5-4 shows the data and model versus photon energy. For
simplicity, I have turned off the Double-Y axis for the graph. It is now very clear
that the short wavelengths (high eV) are not matched by the model. This is a
problem with the final Cauchy parameters that were found.
Figure 5-3. Fit results for a single-point of the SiO2 map using the Silicon with Transparent
Film model.
Figure 5-4. Data fit for Psi shown versus photon energy (in eV) to better visualize the
discrepancies at short wavelengths (large eV).
Set Ranges
Next, extend the data range to 5.0eV. To add this extra range, lets try using the
Data: > Set Ranges command. This will allow you to type in the new MAX
photon energy. The Set Ranges dialog box is shown in Figure 5-6. After selecting
up to 5eV, press Fit. Keep adding extra photon energy in 0.5eV increments and
fitting the data until you have all data graphed. The final fit should appear as in
Figure 5-7. The MSE is significantly improved near 40 compared to the previous
MSE of over 200.
Figure 5-6. Set Ranges command allows you to set the wavelength range and choose which
data point you have selected.
Sellmeier
Although we are able to get the Cauchy to better match the data with a little work, it
is not designed to handle such a wide spectral range. A better model for transparent
films that cover a wide spectral range is the Sellmeier relationship. It has the added
benefit of remaining Kramers-Kronig consistent thus ensuring physical shape for
the resulting index of refraction.
Replace the Cauchy layer in your model by pressing the Cauchy name with leftmouse click and go to the Dielectrics directory. Choose SiO2 (Sellmeier) material
file in place of the Cauchy. Press Generate and the starting point should be close to
the experimental data, as in Figure 5-8. Go ahead and fit the SiO2 thickness and all
four Sellmeier parameters. Your MSE should be near 30 (as compared to values as
high as 200 and as low as 40 with the Cauchy). Thus, this model better matches the
true shape of SiO2 index dispersion over a wide spectral range.
Figure 5-8. Generated data with the Sellmeier model for SiO2 (before fitting).
Both options are found in the +FIT Options section of the Model. After both are
turned off, press the Fit Scan Data from the Fit: panel.
If you leave the thickness prefit and global fit ON during analysis of a full scan, it
will take excessively long. However, if there are large variations across the sample,
these may need to be turned back on using caution and narrowing the search range
of a Global fit.
After the data from all points are analyzed, their results will be shown in the Fit:
panel, as in Figure 5-9. Now, if you press your left-mouse button on any of the
blue parameters, the graph will update with uniformity map of that parameter. As
an example, the Thickness #1 parameter is graphed, as shown in Figure 5-10.
Figure 5-9. Scan Data results are shown in the Fit: panel. Press any of the blue
parameters in this list to view the map of that parameter.
Model Calculation
Now, lets return our attention to the fit at a single-point. Try pressing a point on the
Thickness map and you will find that the results maps do not allow you to jump
directly to the data. Therefore, we first need to switch the graph back to Psi or Delta.
To do this, simply check the Show Map Data box at the top of the Graph panel.
After the graph is switched, go ahead and left-click your mouse on a point toward the
center of the graph. Again, the spectroscopic data from that point is graphed and
automatically fit with our model.
Psi and Delta are matched reasonably well. However, there may be more
information about this sample to consider. Take a look at the depolarization curve,
which was NOT currently included in the MSE calculation. Press CTRL-Z to see a
depolarization graph as shown in Figure 5-11. The increasing depolarization toward
short wavelengths (large eV) is very common for thick layers. The cause of this
depolarization is due to a non-ideal consideration. The most common non-idealities
for thick layers are thickness non-uniformity or instrument bandwidth. Both can be
included in our model.
Figure 5-13. Depolarization data with 2.4% Thickness Non-uniformity in model calculation.
Figure 5-15. Press Configure Options to select hidden menu Include Depolarization Data.
Figure 5-16. Final Thickness Map using Sellmeier model and including the Bandwidth.
Figure 5-17. Final MSE Map using Sellmeier model and including the Bandwidth.
Figure 5-18. Final Index at 633nm (1.960eV) using Sellmeier model and including
Bandwidth.
5.2.
Multi-Sample Analysis
FEATURES IN THIS EXAMPLE
Selecting Multiple Points from Map
Zoom on Map
Multi-sample Analysis
Once again, we will be working with mapped data from multiple locations across a
sample. This sample is more complex than the previous example. In addition to an
SiO2 coating across the wafer, there are 5 patterned locations on the substrate with an
extra layer of Cr over the SiO2 film. The Cr layers were deposited to nominally 10,
15, 20, 25, and 40nm thick. These layers are thin enough to allow light to penetrate
through to the underlying SiO2 film. They were part of a study of different
ellipsometric methods to measure thin absorbing layers.3 Figure 5-19 shows the
location of each of the Cr regions, along with their nominal thicknesses in
nanometers.
25
20
40
10
15
Figure 5-19. Overview of sample. The blue region represents a 100nm SiO2 layer. At the
five orange points, there is an additional Cr layer on the surface of the SiO2 film, with the
nominal thicknesses shown for each (in nm).
J.N. Hilfiker et al. Survey of methods to characterize thin absorbing films with Spectroscopic Ellipsometry, Thin
Solid Films, 516 (2008) pp. 7979-7989.
150 4BData Analysis 3 Advanced
SHORT CUT
FUNCTION
CTRL-Click on Point
Select/De-Select Point
CTRL-ALT-Click on Point
Delete Point
CTRL-SPACE Bar
Figure 5-22. From right-click menu on graph, select the Set Symbol Size option.
Zooming on Map
While this graph is not too complex, it is possible to collect hundreds or even
thousands of points on the same sample. In these cases, it can be helpful to zoom
into a certain location to better isolate a specific point. For demonstration purposes,
drag your mouse across a region of the sample, as shown in Figure 5-24. The
zoomed area is shown in Figure 5-25. To zoom back out to all points, simply rightclick in the graph area and choose Zoom All.
Figure 5-24. Drag your mouse over a graph area to zoom in to that region. A rectangle will
show you the location of your zoom region.
Figure 5-27. Graph with multiple points selected as shown by the larger grey circle
surrounding those points (4 pts in upper-left section).
Figure 5-28. When multiple data points are selected, you can choose which point to view, or
select the check-box to Graph All Data Sets simultaneously.
Figure 5-29. Choose Set Ranges and select only the 65 degree data.
Figure 5-30. Graph of all four Psi curves from multiple data point selection.
Figure 5-31. Right-click over the Data Set text above graph to get extra choices for
selecting, unselecting, deleting, and clearing data sets.
Figure 5-32. Drop-down list of points shows which are (Not selected).
Multi-sample Analysis
Thus far, we have fit each of the four selected points individually and as a combined
group. However, when we fit them all together, we had a single set of fit parameters
for the entire group. This may be okay for the Cauchy parameters as the film is the
same everywhere but the thickness is NOT the same at each point. There is an
advanced feature in CompleteEASE that allows multiple points to be fit
simultaneously with some of the fit parameters common to all points and other fit
parameters allowed to vary between points. Although the four points that are coated
with SiO2 do not benefit from this powerful technique, the demonstration for these
points will help understand the basic concepts before applying to the Cr films
which can benefit from multi-sample analysis.
To use this method, first go to the bottom of the Model and choose Configure
Options. Under the Model Options section, check Multi-Sample Analysis as an
Available option. This is shown in Figure 5-33.
Figure 5-33. Check the Multi-Sample Analysis option from Model Options section.
Next, go to the +Model Options section and expand the newly available +MultiSample Analysis section. Press the Add Fit Parameter text and choose Thickness
#1. The thickness for each selected data point will show up in the list below (as a
fit parameter). Now, right-click on the Data Set text and choose to Select ALL
Data Sets. This will allow us to vary and fit the thicknesses for all 4 locations in the
Multi-sample analysis section. The Cauchy parameters will still be fit but the same
values will be used to fit all data sets simultaneously. Go ahead and press Fit.
Figure 5-34 shows the final results after the multi-sample analysis. The graph shows
the model matching each of the 4 curves as four different thicknesses are fit along
with a common set of Cauchy parameters.
Figure 5-34. Fit results for multi-sample analysis of four locations of SiO2 on Si. A common
set of Cauchy values is used, while the thickness of each SiO2 layer is allowed to vary in the
+Multi-Sample Analysis section of the Model.
Figure 5-35. Right-click in the Fit panel and turn off the Auto Fit check-box.
Figure 5-36. Select the center point of graph (shown surrounded by grey).
I show this data in Figure 5-37 with double-Y axis turned back on to show both Psi
and Delta. I have also generated data to show how different the center-point data are
compared to a model for the single-layer SiO2 film.
Before fitting, we need to select ALL angles again.
To fit this data, we need to add another layer to our model. Press Add and choose
location above SiO2 film to add B-Spline layer from the Basic folder. Expand the BSpline layer and press the Starting Mat = none. Choose Cr.mat from the Metal
directory. This will calculate the B-Spline points to match Chromium optical
constants as published in Paliks Handbook of Optical Constants. Hold the SHIFT
key and roll the mouse wheel over the thickness until it matches the data.
Unfortunately, this film does not have the same optical constants as book values.
This is common for metals, which will differ optically with different deposition
methods and processing conditions. Roll the thickness to 40nm, which was the
nominal thickness for the Cr layer at this location. The model with 40nm is
compared to experimental data in Figure 5-38.
Figure 5-37. Data for center-point compared to model with single-layer SiO2.
Figure 5-38. Model with 40nm of Cr (starting mat in B-Spline is Paliks Cr) compared to
experimental data at center point.
Because the results are not very close, we need to either global fit the B-Spline
parameters or fix the thickness to start. We will choose the latter, as we were given
the nominal Cr thickness. With thickness FIXED at 40nm, go ahead and press Fit.
The data fit reasonably well (see Figure 5-39), with an MSE around 4. However,
there is a small bump in the model at 3.5eV that dont match. This is an
interference enhancement from underlying SiO2 film. We can only match this
region with the correct Cr thickness. Now, turn on the Cr thickness and fit it as well.
The final results appear as shown in Figure 5-40 with a good match across the
spectrum and MSE near 2.
Figure 5-39. Fit to center point with 40nm fixed B-Spline thickness. The miss-match near
3.5eV is due to interference enhancement from the underlying SiO2 film. The correct Cr
thickness will match this region as well.
Figure 5-41. Map view with all 5 locations selected that have Cr coating over SiO2/Si.
Figure 5-42. Spectroscopic Graph comparing the Psi curves from each of the 5 locations
with only 65 degree data selected.
Now, lets prepare the Multi-Sample Analysis. In the +MODEL Options section,
you will see that the Thickness #1 parameter is still selected (Figure 5-43(a)). Press
the Delete All Parms option to remove Thickness #1, as we will start with a common
SiO2 thickness and only allow the Cr thickness to vary. Next, press the Add Fit
Parameter selection and choose Thickness #2 (Figure 5-43(b)).
(a)
(b)
Figure 5-43. (a) Press Delete All Parms to remove the Thickness #1 parameter from the
multi-sample analysis. (b) Next, press Add Fit Parameter and choose Thickness #2.
Before we begin fitting, we need to tune the different thickness values closer to
their final result. For this purpose, un-check the Graph All Data Sets. Now, go
down the list of points and SHIFT-Mouse Roll the different thicknesses shown under
the Multi-Sample Analysis section until the Generated curve better matches the
corresponding Experimental data curves. For more fine adjustment of the
thickness, hold both the SHIFT-CTRL buttons and roll the mouse button. This
decreases the thickness increment. An example is shown in Figure 5-44, where only
Data Point #5 is selected and the 5th thickness is being rolled to better match the
curves.
Figure 5-44. Match to Data Point 5 from rolling in the thickness under Multi-Sample Analysis
section of the model.
After all data points have been matched individually, press the Graph All Data
Sets menu again. Your graph should look similar to that in Figure 5-45. You
hopefully noticed that the curves for thinner layers were not perfectly matched when
we rolled the thickness. This is okay, as the thickest Cr layer does not match the
optical constants for the thinner layers. In fact, we will later remove this point from
the multi-sample analysis. For now, go ahead and press Fit. All five data points
will be fit simultaneously with a single B-Spline layer to describe the common Cr
optical constants and individual thicknesses for the Cr layers. The fit result should
appear as shown in Figure 5-46. The MSE should be just under 20.
Figure 5-45. Preliminary data match by rolling individual thicknesses to pre-tune our guess
to all 5 data sets before fitting multiple sample data.
Figure 5-46. Fit result for all 5 locations with multi-sample analysis.
Now, lets remove the thickest layer from the multi-sample analysis. To do this,
choose the Point location for the center point from the drop-down list above the
graph. Next, right-click the mouse over Data Set: and choose to Un-Select Data
Set. The curve for the center point should disappear leaving only four curves and
their corresponding models. Press Fit and the MSE should drop from above 19 to
less than 14. The center (thickest Cr) point does not share the same optical constants
as the thinner layers and thus is best left out of a multi-sample analysis. However,
the problem is not confined to just the thickest layer. It turns out that the Cr layer
optical constants appear to vary with thickness even for the thinner layers.
NOTE: The multi-sample analysis stayed intact by un-selecting the suspect center
point. If instead we had deleted this point, the thicknesses listed in our multi-sample
analysis would now be pointing to the wrong sample locations and we would need to
roll these starting values over again.
Some of this variation is probably due to microstructure, possible roughness at the
surface and oxidation. To test this theory, turn on surface roughness in the model
and add it as another fit parameter in the Multi-Sample Analysis list. This will allow
each point to have different amounts of roughness. Press Fit and the MSE drops to
near 9. The thickness of the rough surface increases with the thickness of the film
(as shown in Figure 5-47). With this model, we could now add the thickest Cr point
back into the model to see if this trend (increasing roughness) will better match the
center point and be consistent (we expect larger roughness). Sure enough, the MSE
for all five points drops below 12 with the results shown in Figure 5-48. At this
point, research would continue by testing the samples with other techniques such
as Atomic Force Microscopy to see if the roughness result is true for this sample.
Figure 5-47. Fit to four thin Cr films with roughness added to the multi-sample analysis.
Notice the resulting roughness increases as the Cr film thickness increases.
Figure 5-48. Fit results for all five points, showing increasing roughness with increasing Cr
layer thickness.
As a final exercise press Data: > Set Ranges and select all the angles back. Notice
when multiple data sets are selected (with more than 1 angle), each data set is
graphed with the same color. For example, all curves for Data #1 are in magenta and
all curves for Data #2 are in blue. This helps to group them and reduce the total
number of different colors graphed. Before fitting again, turn on the SiO2 thickness
as a fit parameter and add Thickness #1 to the list of Multi-Sample parameters. With
all the angles selected and three different thicknesses included in the Multi-Sample
Analysis, press Fit and the results should appear as shown below. You will notice
that the center SiO2 thickness is larger than the remaining 4. However, this was
verified on a separate wafer that was not coated with Cr. Thus, the larger SiO2
thickness at center is to be expected. The final result is shown in Figure 5-49.
Figure 5-49. Final fit to all angles for 5 locations with Cr, SiO2, and Roughness thicknesses
included in the multi-sample fit.
5.3.
Dynamic Fit
In Situ data provides ellipsometry measurements that were taken versus time. This
often involves data as the thin films are being processed. The model may require a
description of changes in film thickness, optical constants, composition, or even
temperature versus time. The navigation, manipulation, and analysis of in situ data
have many parallels with the corresponding processes for uniformity maps and
multi-sample analysis. Thus, it can be helpful to review these examples before
proceeding with this example.
This example is split into two sections. First, we will discuss navigating and
selecting data sets from in situ time-dependent data set. These topics can be
considered by any user-level. Second, we will work to analyze the in situ data via a
multi-sample analysis approach and finally a time-dependent data analysis
procedure. As this data comes from an amorphous silicon film and will require
analysis methods including the B-Spline and Genosc layers, those examples should
be reviewed in detail before proceeding with this example.
SHORT CUT
FUNCTION
CTRL-SPACE Bar
Figure 5-51. Left-click mouse on graph near 3 minutes selects this time and shows the
selection as a vertical gray bar.
Figure 5-52. To view the spectroscopic data at the selected time-slice, press CTRL-SPACE
BAR or un-check the Show Dynamic Data button just above the graph panel.
Figure 5-53. Right-click in the Fit panel region and turn off the Auto-Fit check-box.
Building a Model
We will start with the assumption that we measure a coating that has known optical
constants. This will allow us a model and find the thickness at any point in time.
For our model, start with BK7 Glass substrate from the Dielectrics subdirectory
and then add a-si parameterized from the Semiconductor directory. Start by fitting
the a-si parameterized layer thickness only. You will find the fit is not adequate, as
shown in Figure 5-54.
Figure 5-54. Fit result for data near 3 minutes when only fitting the a-Si layer thickness (no
surface layer).
To improve the fit, we must consider that the surface of any silicon layer will be
oxidized and possibly rough. To fit the surface, we can either add an SiO2 layer or
simply turn on the surface roughness thickness as a fit parameter. The improved fit
result is shown in Figure 5-55.
Figure 5-55. Fit result for time near 3 minutes when including the surface roughness.
Fit Dynamic
Now, lets fit these two parameters versus time to see how they change as the data
vary across the measured time span. Simply click the Fit Dynamic button in the Fit
panel and the software will progress from the earliest time slice and fit the data,
continuing toward larger times, as shown in Figure 5-56.
You may notice problems starting a dynamic fit with model that is tuned for a thicker
region of the data. If the fit fails, it may appears as shown in Figure 5-57. Here, the
fit finally matches the data at around 3 minutes. To solve this problem, simply
tune the fit for an earlier time before pressing Dynamic Fit. Go ahead and select a
time slice near 0.3 minutes. Roll the thickness values for the roughness and a-Si
layer until they are closer to correct for this new time. Press Fit for the single-point
and you should get a result like shown in Figure 5-58. Now, press Dynamic Fit
again and this time it should be successful. Press Show Dynamic Data to ensure
the model curves match the Experimental curves over all times, as shown in Figure
5-59.
NOTE: It can be beneficial for many applications to start the fit from end-point
rather than starting time, as the film thickness is often larger at the end of a data run.
To fit the data in reverse-time order, right-click the Fit Dynamic button and select
Fit Backwards in Time.
Figure 5-57. Fit fails until about 3 minutes because the starting thickness was not close for
early time region.
Figure 5-58. Fit for a-Si and rough layer thickness at time slice near 0.3 minutes.
You will also notice in Figure 5-59 that the Fit Panel shows the final results (from
last time). If you select the MSE, Roughness, or Thickness #1 with your mouse, the
time-dependence of these values will be graphed, as demonstrated in Figure 5-60 and
Figure 5-61.
Figure 5-62. Holding CTRL button down while left-clicking mouse on graph allows multiple
time-slices to be selected.
Figure 5-64. Right-click on the words Data Set above a graph to access additional options
related to the multiple time-slice data sets.
Figure 5-65. Choose Multi Sample Analysis from the Available Model Options within the
Configure Options selection.
Figure 5-66. Model Options section is expanded to view the newly added Multi Sample
Analysis section. Press Add Fit Parameter to designate a fit parameter to vary with each
time-slice.
Figure 5-67. Starting points for all time-slices have been adjusted by rolling the mouse over
each individual thickness from within the multi-sample analysis section of the Model.
You can now press Fit and all thicknesses (for each time slice) along with a single
roughness value that is used to fit all data sets is varied. This result is shown in
Figure 5-68. If the roughness was also changing with time, its thickness could be
added as a time-dependent variable within the multi-sample analysis section.
Figure 5-68. Fit results when a-Si thicknesses were allowed to vary at each of the five timeslices, along with a single roughness for all time slices.
Figure 5-69. Starting with a B-Spline layer, choose a-Si for the Starting Mat to get the
node values in the correct general shape for this type of film.
You will notice that the starting point is not the correct final optical constants for the
time-dependent silicon simply by Generating data, as shown in Figure 5-70.
However, we have five time slices to provide adequate information about this film at
different thicknesses. Go ahead and press Fit. The node values will adjust in the
B-Spline to best fit the amorphous silicon data from ALL time-slices simultaneously.
This fit result is shown in Figure 5-71, with the B-Spline optical constants shown in
Figure 5-72.
Figure 5-70. Generate Data and you see that the book-value a-Si optical constants used as a
starting point in our B-Spline do not match the experimental data. However, they are close
enough to give us a good starting point for fitting.
Figure 5-71. Fit results with B-Spline Optical constants allowed to vary.
5.4.
OC Library Mode
Build Library
(OC Compare)
Draw Graph
In this example, we will demonstrate how to build a composition- or temperaturedependent optical constant library. This procedure can be used to create special
material files for any series of samples that show a systematic shift in optical
properties related to composition, temperature (or other material property or process
condition). During the process, B-splines are used to smooth the constituent optical
constant spectra for each known composition or temperature. The library is then
built using an interpolation algorithm based on the critical point shifting algorithm of
Snyder et al.4 In the critical point shifting algorithm, the dielectric function at an
arbitrary composition or temperature is evaluated from a weighted sum of the nearest
reference dielectric function spectra, after the reference spectra are wavelengthshifted to line up the critical point features. CompleteEASE simplifies this process
by automatically determining the wavelength shifting polynomials (that describe the
path of critical points versus composition/temperature) via a non-linear regression
analysis.
The full procedure for creating a composition or temperature dependent material file
is to obtain optical constants from a series of samples with varying, known
composition or temperature. This example will not demonstrate this initial step, as
the procedures for each individual film will be similar to those taught in earlier
examples. After successfully modeling the experimental data to determine accurate
optical constants for a specific composition/temperature, save the tabulated optical
constants for each material file. In this example, we work with a series of Si1-xGex
material files that were obtained by fitting a series of layers with different
germanium compositions (x). The actual composition was measured by other means
in this case XRD. Thus, we now have the optical constants (n,k) for a series of
known Si1-xGex compositions.
Before starting this example, change the Wavelength Units to eV, and the Optical
Constant Units to e1 & e2 from the Options Tab.
P.G. Snyder, J.A. Woollam, S.A. Alterovitz, and B. Johs, Modeling AlxGa1-xAs Optical Constants as Functions of
Composition, J. Appl. Phys. 68 11 (1990) 5925.
Table 5-3. Compositions for each Si1-xGex material file used in this example.
GERMANIUM
COMPOSITION % (X)
Si_jaw.mat
0% (x=0.000)
SiGe-1.mat
5.2% (x=0.052)
SiGe-2.mat
9.8% (x=0.088)
SiGe-3.mat
15.5% (x=0.155)
SiGe-4.mat
20.4% (x=0.204)
SiGe-5.mat
26.3% (x=0.263)
SiGe-6.mat
31.9% (x=0.319)
SiGe-7.mat
37.1% (x=0.371)
SiGe-8.mat
42.3% (x=0.423)
SiGe-9.mat
49.3% (x=0.493)
SiGe-10.mat
53.9% (x=0.539)
Ge.mat
100% (x=1.000)
For this example, populate the seven layers with Silicon, SiGe-2, SiGe-4, SiGe-6,
SiGe-8, SiGe-10, and Germanium (all shown in green in table). Before you will be
able to set the compositions for each layer, you must turn ON the OC Library
Mode. Type the corresponding compositions for each layer as shown in Figure
5-74. The temperatures can remain at 25 Celsius for all files. If we were making a
temperature dependent file, we would adjust the corresponding temperatures to
match each material file.
Figure 5-74. OC Compare model setup to build a Library for SiGe films of different
composition.
Before proceeding to build the OC Library from these reference files, set the
wavelength range to match the material file. For the SiGe material files, the optical
constants were measured from 248nm to 1240nm (1eV to 5eV). Also, the # of
Wvls may need to change to allow the B-Spline to more accurately match the
reference files over this wide spectral range. The original material file contains
about 540 individual wavelengths. Thus, I have set the # of Wvls to 500
although this will significantly increase the amount of time needed for the
calculation. When ready, press Build Library, and a graph will appear as shown in
Figure 5-75. In this graph, the colored curves are the reference optical constants at
the specified compositions, and the black dashed curves are interpolated spectra at
reference and midpoint compositions. Thus, each colored graph should be matched
with a corresponding dashed black curve. The additional dashed black curves should
demonstrate whether the shifting algorithm has worked.
Draw Graph
This graph can be very busy, but you can change the view by selecting a different
setting at the bottom of the model under Draw Graph. Pressing the e1 & e2
selection will open the dialog box shown in Figure 5-76. Figure 5-77 and Figure
5-78 show the e1 only and e2 only spectra for this example, respectively. In
addition, the graph can show the Shifted Spectra and Wvl Shift Function to help
assess whether the composition library is successful. For advanced users, the
Library Resolution and Spectra Shifting Parameters can be modified, as shown in
Figure 5-79.
Figure 5-77. Graph of e1 only after building the SiGe composition library.
Figure 5-78. Graph of e2 only after building the SiGe composition library.
Figure 5-79. Advanced features to customize the OC Library algorithm, including spectral
shifting parameters and resolution in eV.
To further assess whether the library algorithm was successful, you can zoom in on
each of the graphs. Figure 5-80 shows a zoomed view of the E1 peak for the SiGe
films. This shows that the peaks are shifted to higher energy with increasing silicon
composition.
Figure 5-80. Zoomed view of the E1 peak for the SiGe compositions, showing a nice shift in
critical point toward higher energy for increasing silicon composition.
Save OC Library
After successful calculation of a composition or temperature dependent optical
constant file, press Save OC Library at bottom of the Model panel. This material
file can now be used to fit future data files. To demonstrate what this file looks like
and how it operates, click the Reference Material = none, which is the bottom
layer in our OC Compare model. Add your newly created material file in this entry.
Change the Graph Type in the Model to e2 only and roll your mouse over the
composition in this Reference Material. You will be able to visualize the optical
constants shifting compared to your refence constants as the composition is varied.
An example graph is shown in Figure 5-81, where the composition has been moved
to x=0.75 to see how well the shifting works between the two most widely spaced
reference values.
Figure 5-81. The newly created Composition dependent file has been added as the Reference
material to compare to the original reference material files as the composition is varied.
5.5.
Anisotropic Films
FEATURES IN THIS EXAMPLE
Convert to Anisotropic
Difference Mode
Biaxial Layer
In this example, we will allow a layer to have anisotropic optical constants, where
the index normal to the sample surface is different than the index in the sample
surface plane. This is common with many polymer films that orient along a specific
direction due to processing.
To start, open the Aniso-Organic on Si data file from the Examples folder. The
data should appear as shown in Figure 5-82. For simplicity, turn off the Double-Y
Axis and view the Psi curves. The oscillations show that this film is transparent over
the measured spectral range. Open the Si with Transparent Film model and press
Fit. The results of this fit should appear as seen in Figure 5-83.
Figure 5-83. Fit to the anisotropic film using standard Transparent Film model.
60
in degrees
50
40
nx>nz
30
20
nz>nx
10
0
300
600
900
1200
Wavelength (nm)
nz>nx
nx>nz
1500
1800
Figure 5-84. Simulation for transparent films showing the effects of anisotropy with either Nz
> Nxy or Nxy > Nz as compared to isotropic data.
Convert to Anisotropic
Right-click over Cauchy Film to access the menu of options to modify the Cauchy
layer. From this list, choose to Convert to Anisotropic. The Cauchy layer will
now become embedded as part of a new layer named Biaxial. Biaxial stands for
anisotropy where Nx, Ny, and Nz are all different. However, the layer name is used
here to describe any anistropic film. Within the Biaxial layer, you can choose either
Uniaxial or Biaxial anisotropy. For our case, we will use Uniaxial anisotropy, as it
would require more advanced measurements to study the potential Biaxial anisotropy
of a sample. This will be described in more detail in a later Example.
Biaxial Layer
The biaxial layer can be used in two different ways and both will work equally well
for this example. The difference is whether to have Difference Mode = OFF, as
shown in Figure 5-86 or Difference Mode = ON, as shown in Figure 5-87.
Figure 5-86. Biaxial layer with Uniaxial anisotropy and Difference Mode OFF.
Figure 5-87. Biaxial layer with Uniaxial anisotropy and Difference Mode ON.
With the Difference Mode = OFF, each direction is described by an individual layer.
In our case, each direction is seeded with the Cauchy layer, as that is the layer we
started with. The layers could just as easily be B-Spline, Gen-osc, or a mixture of
these.
With the Difference Mode = ON, the x-direction is defined by a material layer, while
all other layers are determined by summing the x-direction optical constants with the
difference value calculated from the extended Cauchy type dispersion described as
the dZ (z-difference). This method can be extremely useful when the fit is primarily
sensitive to the index difference and not sensitive to the absolute value of the index.
For this example, turn the Difference Mode to ON and then roll the mouse wheel
over the dZ_A parameter. If you roll this parameter to positive numbers, the
simulated curve will look worse than the isotropic fit (tilt in wrong direction).
However, if you roll the dZ_A parameter to negative values, it will start to match
better. Go ahead and fit the dZ_A and dZ_B parameters. Make sure you are still
fitting the Cauchy parameters and the thickness, but it is wise to now turn off the
Global Fit and Thickness Prefit, as we are close to the final answer. The final result
should appear as shown in Figure.
5.6.
6. Reference - Graph
6.1.
6.2.
Graph Pane
The Graph pane is an important part of the CompleteEASE user interface.
Spectroscopic data, model fits, and optical constants are all plotted in the Graph
pane. The features described in this section can be used regardless of the type of
information displayed in the graph window.
If the mouse is stopped in the plot area for a few seconds, a tool tip box will appear
which reports the X and Y values corresponding to the mouse position. The mouse
can be moved to other positions and the X-Y values will be updated.
The legend can be repositioned by clicking (L) in the legend area and dragging the
legend box to the desired location.
Show Data
If optical constants (or something else besides data) are being graphed, you can press
this button to go back to the data graph.
Range-Select
To zoom in on a specific spectral range, hold down (L) and drag over the
desired spectral range (Figure 6-2). The graph will then redraw, plotting only the
selected spectral range.
Figure 6-2. To select a specific spectral range (hold down the left mouse button and drag the
mouse over the desired range).
Zoom All
To plot the entire spectral range, simply press the Zoom All button at top-right of
the Graph pane (Figure 6-3). This same option can be found at the top of the RightClick menu from within the graph.
NOTE: Zoom All does not appear if you change the graph to Use Clipboard
Aspect Ratio.
Zoom All
Press to graph all data.
Statistics
The graphed data undergo a variety of statistical calculations. For example, a
straight-through measurement of air can be used to verify the measurement accuracy.
The statistics of this measurement can then be calculated, as in Figure 6-5.
Figure 6-6. The Graph ScratchPad can be used to compare all types of graph curves, and
customize the view for easier comparison.
Show Legend
Turn On/Off the graph legend.
Show Symbols
Turn On/Off graph symbols for the Experimental Data. The model curves are still
represented as solid lines. Figure 6-7 shows an example data graph with symbols on.
Auto Y-Scale
Y-Axis is automatically selected to show all data.
Manual Y-Scale
User can select lower and upper Y-axis limits. All data are still included when
fitting, but only points within this range are graphed.
Auto X-Scale
X-Axis is automatically selected to show all data.
Manual X-Scale
User can select the lower and upper X-axis limits. All data are still included when
fitting, but only points within this range are graphed.
Cancel Menu
Close the Graph menu.
The Measurement tab is shown in Figure 7-1. This tab consists of three panels:
System Status, Measurement Controls, and Fit Results. A brief description of all
commands within this tab is provided.
7.1.
7.2.
MODE:
Fast
Ellipsometry, ~ 3 seconds
Standard
Ellipsometry, ~ 10 seconds
Long
Ellipsometry, ~ 30 seconds
Transmission
M.M.
Mueller-Matrix, ~ 10 seconds
M.M. Long
Mueller-Matrix, ~ 30 seconds
For the Standard and Long modes, the motor home sensors are checked before
and after measurement. Also, they both implement Zoneaveraging, which collects
data with input polarizer set to +45 and -45 and then reports the average. This
improves data accuracy.
The M.M. data modes allow measurement of 12 Mueller matrix elements. The
absolute intensity of m11 is measured, while all other elements are normalized to
m11.
To improve accuracy of Transmission Intensity and Mueller matrix measurements,
perform an S-T Baseline or Off-Sample Baseline prior to the sample
measurement (from the Hardware tab).
Sample Alignment:
Specifies how the Z-stage should be used to adjust the sample height before the
measurement. Table 7-2 provides details for each choice.
Table 7-2. Alignment Options
OPTION
DESCRIPTION
None
Stage remains at current location, which is below beam for straightthrough (90) measurements.
Standard
Robust
A thorough alignment of the light beam that searches over the full
translation range. Use for glass substrates with multiple reflected beams.
Fixed Height
Prompt Height
Allows user to enter the sample stage height for measurement. Perform a
z-stage scan from Hardware tab to help determine appropriate height.
Model
Specifies which model will analyze the data. If it is set to None, data will be
acquired, but no data analysis will be performed.
Measure
Begins a measurement using the settings specified in the Mode: and Sample
Alignment: fields. After data is acquired, the Model: is used to analyze the data.
Results are placed in the box in the upper right corner of the screen.
7.3.
7.4.
MSE = 2.837
Thickness # 1 = 96.71 0.018 nm
n of Cauchy Film @ 632.8 nm = 1.865
Copy to Clipboard - Table
Copy results to the clipboard in a format ready to paste into a table or spreadsheet, as
shown below.
PARAMETER
MSE
Thickness # 1 (nm)
n of Cauchy Film @ 632.8 nm
VALUE
2.837
96.71
1.865
ERROR
BAR
0.018
7.5.
Select a single data file from the Files panel and you can Open File. This will open
the data file, but not the model that was associated with the previous fit.
Use the mouse to select multiple results in the Files panel and you will be allowed to
Calc. Multi-Sample Stats. An example of this is shown in Figure 7-4.
If the selected data file is a map of multiple points across the sample, the Graph
Data button appears. When selected, a new window appears that allows the user to
select the fit result they wish to view, as shown in Figure 7-5.
The Analysis tab is shown in Figure 8-1. This tab consists of three panels: Data,
Model, and Fit.
8.1.
Data: Panel
Open
(L): Open a data file.
(R): Append Data file to existing data. The data files must share the same
wavelengths.
Save
(L): Save a data file.
(R): Choose to Save Data Subset or Export to Text File.
Info
Shows information about the current data set opened in CompleteEASE. An
example of the information box is shown in Figure 8-2.
Set Ranges
Choose the Minimum and Maximum wavelengths for use with current data set
during fit.
8.2.
Fit: Panel
Generate
Generate data based on the current model and data set. Shortcut keystroke to
Generate data is Alt-G.
Fit
Fit data based on the current model and data set. Shortcut keystroke to Fit data is
Alt-F.
Fit Dynamic
(L): Fit dynamic data that has been taken versus time.
(R): Choose to Append Fit Results, Fit Backwards in Time, or Fit
Backwards and Append.
Reset
(L): Restore the previous model parameters, from before the fit.
(R): Choose to Show Reset List.
Figure 8-3. Reset List allows user to reset to any recent positions, which are automatically
saved before and after each fit.
8.3.
Model: Panel
The box in the right half of the tab displays all the options and specifications for the
current data analysis model. An example model is shown in Figure 8-4. Details of
each option within the model are given below.
Open
(L): Open a new model.
(R): Open Snapshot or Merge models.
Open Snapshot
Right-click on Open button to open saved snapshots. A snapshot is a single file
that compresses all information from the CompleteEASE screen, including the data
measurement, model, and fit results. This is a convenient way to save and store
information.
Merge Models
Right-click on the Open button to merge models together.
Save
(L): Save current model and all model settings.
(R): Save Snapshot.
Save Snapshot
Right-click on Save button to save snapshot file. A snapshot is a single file that
compresses all information from the CompleteEASE screen, including the data
measurement, model, and fit results. This is a convenient way to save and store
information. The file will be given .ss file extension.
Clear
Remove the current model.
Add
Press this button to add a layer. The Add Layer to Model dialog box appears. Select
the position of the new layer by moving the mouse to adjust the blue-bar and press
the mouse button (Figure 8-5).
Delete
Press this button to delete a layer from the model. Choose the layer with the mouse
(making it blue) and click to delete (Figure 8-6).
Save
Press this button to save optical constants for a layer. Before clicking on the layer
you wish to save (Figure 8-7), choose the Save Type (either Dispersion Parameters
or Tabulated).
Model Pictorial
The model is shown as a layer-by-layer picture with details regarding thickness and
other parameters to describe the individual layers. Any layer with a + can be
expanded to show more details about that layer. This is demonstrated in Figure 8-8.
(a)
(b)
Figure 8-8. (a) Model with layers collapsed and (b) with Cauchy Film layer expanded.
Editing Layers
Each layer has a name, show in blue. Place your mouse over the layer name to do
the following:
(L): Open a new material file in place of the current layer.
(R): Choose from a menu to Graph Layer Optical Constants, Append Text to
Layers Fit Parameters, Save Layer Optical Constants, Parameterize Layer,
View Layer Comment, Convert To EMA, Convert to Anisotropic, or Grade
Layer. This menu is shown in Figure 8-9
Parameterize Layer
Right-click a layer to parameterize the optical constants of that layer using a
dispersion model, such as the General Oscillator layer. This will open the layer
optical constants as reference values in the Parameterize Layer dialog box, as shown
in Figure 8-10.
Convert to EMA
Right-click to replace the layer with an EMA, in which the original layer is now
embedded as the host material (Material 1). The second material is default to Void
with EMA% default to 0, but selected as a fit parameter. An example is shown in
Figure 8-11.
Figure 8-11. Example of the EMA layer where the Cauchy Film was automatically converted
to the host (Material 1) with void as the second material.
Convert to Anisotropic
Right-click to replace the layer with an anisotropic Biaxial Layer, in which the
original layer is now embedded as the starting material for both the Ex and Ez
directions. The default configuration is to have no orientation (determined by Euler
angles) and anisotropy set to Uniaxial (Nxy Nz). An example is shown in Figure
8-12.
Convert to Isotropic
After converting a layer to Anisotropic, you can right-click on the Biaxial layer to
unwrap the host film back to an isotropic layer.
Grade Layer
Right-click to replace the layer with Graded Layer, where the original film is
embedded as the average refractive index. The % inhomogeneity is default to 0%,
but can be varied to allow the index to increase or decrease from the bottom to top of
the layer. An example of this is shown in Figure.
Remove Grading
After grading a layer, you can right-click on the Graded layer to unwrap the host film
back to a layer without grading.
Angle Offset
Apply an offset to the angle when calculating the model. This parameter can be fit.
8.4.
8.5.
Layer types
The CompleteEASE software provides many different ways to represent the optical
constants of a material versus wavelength. These representations range from simple
tabulated lists for the optical constants to complex dispersion models. All of these
different representations are classified as layers in the CompleteEASE program.
The layers currently available are listed in table 5.1. The rest of this chapter
describes the operation and typical uses of the different layers in detail. This chapter
makes extensive use of the OC Compare model to demonstrate the layers so
reviewing that section of the Working with Models chapter would be advantageous.
Table 8-1. Layer types in CompleteEASE.
Layer Name
Standard (Tabulated N,K)
Description
Contains tabulated values for the optical constants of a
material.
Void
EMA
EMA-Coupled
Cauchy
Cauchy_Wvl
Cauchy_Extended
Sellmeier
Coupled
WvlByWvl
Biaxial
Temperature/Alloy
General Oscillator
B-Spline Layer
Tabulated n,k
Most of the layers in the CompleteEASE material library are standard tabulated
optical constant lists. The optical constants for this type are represented as a table of
values versus wavelength. The table can contain the real and imaginary parts of the
materials' dielectric function (1 and 2) or the materials' complex index of refraction
(n and k). If the user finds optical constant tables in the literature and wishes to use
them in CompleteEASE, simply type the table into a file in the correct format and
they can be opened.
Figure 8-17 shows how typical standard layers are represented in the Model. Unlike
the rest of the layer types described in this chapter, these have no user adjustable
parameters to define the optical properties making it the simplest layer type to use.
Left clicking on the layer name ('SiO2_JAW' in Figure 8-17) will bring up the 'Open
Material' dialog allowing a different material to be chose for the layer. Right
clicking on the layer name 'SiO2_JAW' will open extended choices that include:
Parameterize Layer
Convert to EMA
Convert to Anisotropic
Grade Layer
Figure 8-17. A model constructed of standard layers each set of optical constants is
described by a table of values versus wavelength.
Void
The Void layer is a special case of Tabulated N,K layer where the optical constants
are fixed for all wavelengths at the nominal values for air (vacuum): n=1, k=0.
EMA
The EMA layer uses the Effective Medium Approximation to calculate the optical
constants of a mixed material. Calculation is based on mixing optical constants of
two or three constituent materials, the percentage of each material, a depolarization
screening factor, and the calculation type. An EMA layer is shown in Figure 8-18.
The EMA layer in the EASE software provides three different EMA mixing
methods for the user to choose. These include a simple Linear combination,
Bruggemann EMA, and Maxwell-Garnet EMA. The Linear mixing simply
interpolates between the constituents dielectric functions to get the optical constants
of the composite material. The Maxwell-Garnett and Bruggemann EMAs are more
common. The Maxwell-Garnett assumes that spherical inclusions of materials #2
and #3 exist in a host matrix of material #1. The Bruggemann EMA makes the selfconsistent choice of the host material. The Maxwell-Garnett and Bruggemann
EMAs are very useful for modeling surface and interfacial roughness as well as
modeling poly-crystalline materials by mixing together amorphous and crystalline
optical constants of the material.
EMA-Coupled
This is a special case of EMA layer, just as described above, except with the first
material using the Coupled description. This allows the Material #1 to obtain optical
constants from a separate layer in the model, designated by the layer number. An
example of the EMA-Coupled layer is shown in Figure.
Figure 8-19. Example of the EMA-Coupled layer, where Material #1 receives its optical
constants from a separate layer in the model, referenced by the layer number.
Cauchy
The Cauchy layer is a commonly used layer for determining the optical constants of
a transparent or partially transparent film (Dielectrics and Semiconductors below the
fundamental bandgap). Over part of the spectral range, the optical constants of these
materials can be represented by an index that varies slowly as a function of
wavelength and an exponential absorption tail. The index of refraction of the
Cauchy layer is represented by an inverse power series containing only even terms
and the extinction coefficient is represented by a simple exponential tail. These
representations are as follows:
The Cauchy is a dispersion layer that describes the index of refraction using a threeterm equation, as in Equation 8-1. In addition to the three terms for index of
refraction (n), there are three additional terms to describe an Urbach absorption tail,
as in Equation 8-2. The Cauchy layer and graph of the optical constants is shown in
Figure 8-20.
n ( ) = A +
(8-1)
(8-2)
Figure 8-20 shows the Cauchy layer as displayed in the CompleteEASE model. The
A, B, C parameters are variable fit parameters that determine the index dispersion.
The k Amplitude and Exponent are fit parameters for determining the shape of the
extinction coefficient dispersion. The Band Edge parameter can be set manually but
is not a fit parameter since it is directly correlated to the k Amplitude parameter (the
extinction coefficient is equal to k Amplitude at the Band Edge.
Cauchy_WVL
This special representation of the Cauchy layer is found in the ADVANCED
materials directory. It provides the same shape as the Cauchy, but is described with
the terms shown in Figure 8-21. This allows the user to specify a known index of
refraction at a specific wavelength, without concern for the dispersion terms.
Figure 8-21. Cauchy_Wvl layer is same dispersion shape as Cauchy, but allows entry of the
index at a design wavelength in-place of the A parameter of the standard Cauchy.
Cauchy_Extended
This special representation of the Cauchy layer is found in the ADVANCED
materials directory. It provides a similar shape to the Cauchy, but with ability to
decrease index quickly toward NIR spectral region by adding an IR term. An
example layer is shown in Figure 8-22 with the resultant optical constants shown in
Figure 8-23.
A 2
2
n = ( ) + 2
B2
(8-3)
where () is an index offset, A is the amplitude, B is the center energy, and E is the
position of a pole in the infrared. These are represented in the layer as Einf,
Amp., Center En., and IR Pole Amp. respectively. Figure 8-24 shows the layer
as it appears in a model.
Coupled
Special layer used to match the optical constants of a separate layer within the
model. The Coupled layer expands to allow designation of the layer number where
the optical constants should come from.
WvlByWvl
Layer which allows the optical constants to vary independently at each wavelength.
This layer has the utmost flexibility, but also the largest number of fit parameters.
As it does not maintain Kramers-Kronig consistency or even smooth continuous
curves for the optical constant dispersion, it should be used with extreme caution.
Biaxial
The biaxial layer allows the user to model anisotropic materials. This layer is used
for both uniaxial and biaxial anisotropy and has 3 Euler-angle terms to rotate the
orientation of optical axes relative to the ellipsometer measurement orientation.
There are two modes which can be used within the Biaxial layer. With Difference
Mode = OFF, the optical constants in all directions (2 for uniaxial, 3 for biaxial) are
independently described by material files. This is shown in Figure. With Difference
Mode = ON, the x-directed optical constants are described by a material file while an
extended Cauchy difference is established for the index difference between the xdirection and the other orientations. This is shown in Figure.
Graded Layer
All other layer types can be graded to model the optical variation with depth through
the film. This is achieved via the GRADE button within the Model: panel. An
example Graded Layer is shown in Figure 8-27. There are three types of grading
that can be achieved in this layer: Simple, Linear, and Non-Linear. Grading is
accomplished by dividing the thin film into a series of slices that will have varying
optical constants to approximate the index gradient profile. The number of slices can
be designated. The primary fit parameter is the % Inhomogeneity which will be
positive numbers when the index is larger toward the surface than the substrate and
negative numbers if the index is smaller toward the surface. If you right-click on the
title Graded Layer, the optical constants will be graphed, showing n and k at both
top and bottom, as shown in Figure 8-28.
Gen-Osc
The Gen-Osc layer is a General Oscillator material that allows the summation of
different oscillator line shapes. Figure 8-29 shows a Gen-Osc layer with two
oscillators of different types, along with associated graph of the resulting optical
constants.
Within the Gen-Osc layer you can add a new oscillator by pressing Add Oscillator.
To change the oscillator type, click on the Type = selection and you will get a
dialog box with your choices, as in Figure 8-30. The various options are described in
detail in an Appendix.
Figure 8-29. Gen-Osc dispersion layer with both a Cody-Lorentz and a Gaussian oscillator.
Figure 8-30. Choices for Oscillator Types within the Gen-Osc layer.
The General Oscillator layer currently supports many different types of oscillators:
Lorentz, Harmonic, Gaussian, Tauc-Lorentz, and Cody-Lorentz, Tanguy, and Drude
type dispersion.
Press Show Dialog to bring up the Editor to visualize and manipulate the oscillator
parameters. This dialog editor is shown in Figure 8-31.
Figure 8-31. Dialog Editor to view and manipulate the Gen-Osc parameters.
The oscillators can be manipulated by pressing the number to their left. Use care
when pressing the number, as each mouse-button does a different function:
(R) Delete the oscillator from list.
(L) Selects the oscillator and shows gray drag-boxes to allow manipulation of
the oscillator parameters on graph.
The Dialog Editor allows the oscillator parameters to be matched to reference optical
constants. This procedure is demonstrated in the Analysis Example on use of the
Gen-Osc layer.
Further details regarding the Gen-Osc Layer are given in the next section.
B-SPLINE
The B-Spline layer specifies the optical constants versus wavelength using a series of
control points which are equally spaced in photon energy (eV). The approximate
spacing of the control points is controlled by the Resolution (eV) = field and the
resulting number of control points within the current spectral range is listed after this
value. Basis spline curves (which are by definition smooth and continuous in the
0th, 1st, and 2nd derivatives) are used to interpolate between the control points.
Clicking on the Draw Node Graph displays the location of the spline control
points. The B-Spline layer always works in terms of the complex dielectric function
= 1 + i2 = (n + ik)2, which is often written as e1 & e2. One set of control points
(and associated spline curve) is used to specify the dispersion in e1 and a second
set of control points is used to describe e2. Using a smaller Resolution value
adds more control points, resulting in more flexibility to accommodate sharper
features in the dielectric constant spectra. However, setting the Resolution too low
can result in unwanted noise or unphysical artifacts in the dielectric function spectra.
The B-Spline will start with all nodes set to the Initial Values, as described within
the layer. However, a Starting Mat can be entered and the spline points will
automatically match the optical constants from this reference file. Figure 8-32 shows
the B-Spline layer and spline points matched to Germanium starting material.
The B-Spline layer uses b-spline (also known as basis spline) curves to represent the
dielectric function (1 and 2) vs. eV. B-spline curves are defined by a series nodes
(or knots), which have a position and amplitude. Basis polynomial functions at
each node are defined by the recursive B-spline formula (Cheney and Kincaid,
Numerical Mathematics and Computing, Third Edition, Brooks/Cole Publishing
Company, 1994) in terms of the positions of the adjacent nodes. Summing the basis
functions (weighted by the amplitude at each node) produces the resulting curve. Bspline curves have many desirable properties for modeling dielectric functions:
b-spline curves and their derivatives (up to the spline order minus one) are
continuous (the B-Spline layer uses 3rd order b-splines)
the b-spline node amplitudes affect only the local shape of the curve, e.g.,
changing a node amplitude in the UV does not affect the curve in the VIS
b-spline curves exhibit a convex hull property, e.g., if all the node
amplitudes are non-negative, then the b-spline curve is also non-negative,
thus avoiding non-physical negative 2 values
since the basis functions depend only on the node positions, the node
amplitudes which define the resulting curve are linearly independent, which
greatly increases computation efficiency
Figure 8-33 shows the default settings for the CompleteEASE B-Spline layer. The
node amplitudes can be initialized to constant n and k values (vs. eV) by clicking on
the n= or k= fields, or they can be initialized by specifying a Starting Mat.
The Resolution (eV) field defines the nominal spacing of the nodes, which are
equally distributed in eV over the selected spectral range of the data set. Clicking
the Draw Node Graph field plots the positions and amplitudes of the b-spline
nodes, along with the calculated dielectric function of the layer (Figure 8-34). The
Fit Opt. Const. setting turns ON and OFF all the optical constant fitting parameters
for the layer (e.g., node amplitudes and positions, pole amplitudes, etc.). If Use KK
Mode = OFF, separate b-spline curves are calculated and fit for 1 and 2. When
Use KK Mode = ON, a b-spline curve is used to define 2, and 1 is calculated
from via the K-K transform of 2 (plus contributions from absorption outside the
measured spectral range). Figure 8-35 shows the Node positions when KK Mode is
turned on. Notice, there are nodes to describe 2, along with a few nodes outside the
measured spectral range to help with the Kramers-Kronig integration.
All the available options for the B-Spline layer are shown in Figure 8-36. A brief
description of each option follows:
E Inf: This stand for 1 and infinity, and when in KK mode it accounts for the
absorption far outside the measured spectral range by adding a constant offset to the
1 curve.
IR Amp: With KK mode = ON, this value is the amplitude of a zero energy
oscillator which accounts for absorption at low energies, typically due to free
electron absorption (i.e., a Drude term).
IR Br: The broadening of the IR oscillator; often this can be fixed at zero, but
allowing to vary can improve the fit to some metallic dielectric functions.
Show Parms: Enables display of the all node amplitude fit parameters in the Fit
Results; by default they are hidden to reduce the number of displayed parameters.
Intermix
If you add a layer and choose Intermix layer type before selecting where to add the
layer (Figure 8-37), a special Intermix layer will be added that mixes the optical
constants from the layer below and above in a 50-50% ratio. An example of this
layer is shown in Figure 8-38. The intermix layer steals thickness away from the
layer above and below to avoid moving the interference oscillations.
Figure 8-37. Choose Intermix before adding a layer between two layers to create a special
layer that mixes the optical constants of the layers above and below wherever this layer is
inserted.
Composition Library
Layers that are marked as Comp Library or Comp & Temp Library are created
to show variation in optical constants as the composition or temperature are varied.
Several of the material files provided with the CompleteEASE program contain
information on the optical constants of a material as a function of temperature and/or
alloy fraction. One example of this type of material is AlxGa1-xAs. The file
AlGaAs_T.MAT supplied with the CompleteEASE software contains information
on the optical constants of AlxGa1-xAs over x values from 0 to 1 and temperature
values from 577C to 650C. Using this type of layer is very similar to using the
standard layer type for a material, the only difference being that there are
temperature and/or alloy parameters that can be edited and fit for. Figure 8-39 shows
how such a layer is represented in the model.
Figure 8-39. A Temperature/Alloy layer as displayed in the model. The composition and
temperature values are editable fit parameters.
Coupled
Layer that has same optical constants as another layer in the model. An example is
shown in Figure 8-40
Figure 8-40. The Coupled layer gets optical constants from another layer in the model. In
this case, it is using the optical constants from Layer #1, the Cauchy.
Figure 8-41. Example of the Uniaxial layer. Enter any material file for both the ordinary (inplane) and extraordinary (out-of-plane) optical constants.
Figure 8-42. Example of the Uniaxial-Diff layer. Enter any material file for the ordinary (inplane) optical constants and then use the 4-parameter dispersion (Cauchy-like) to describe the
index difference..
8.6.
Pole
Poles describe zero-broadening oscillators. As such, they only affect the real part of
the dielectric function (1). They can describe dispersion created by absorptions that
occur outside the measured spectral range. They are defined in the equation:
n _ pole =
An
2
En E 2
Lorentz
This is the classic Lorentz oscillator model, described as:
(A j + iZ j )B j C j
+
2
j =Osc # C j iB j
Harmonic
This is the classic Harmonic oscillator model. When the broadening becomes much
less than the center energy (Brn,<<En), a Harmonic oscillator behaves as a Lorentz
oscillator. The equation is described as:
( A j + iZ
j = Osc #
1
2
1
1
+
+
B
C i (1 B ) C + i (1 B )
j
j
j
j
2
2
Gaussian
This Oscillator type produces a Gaussian line shape in 2 with a Kramers-Kronig
consistent line shape for 1.
1 = +
A C
i k B
i = Osc .#
G
C
kG B
2 = Ai exp
i
+C
+
kG B
+ exp
+ Z i exp
+C
kG B
kG B
+ Z C
i
kG B
+ exp
+C
kG B
+C
+
kG B
where A is the real part of the amplitude, Z is the imaginary part of the amplitude, B
is the broadening, and C is the center energy. The constant kG is 0.600561294, and
is the photon energy in eV. The function is a convergence series.
The oscillators in this layer are defined so that the amplitude is the height of the
oscillators 2 peak at its center energy and the broadening is the full width of the
oscillator at half maximum value (FWHM). This way, when an oscillators type is
changed, the new oscillator has the same height and FWHM as the old oscillator.
Drude
The classic Drude model describes free carrier effects on the dielectric response. Its
form is a Lorentz oscillator with zero resonance energy. The Drude dispersion type
is described in the equation below. It is defined in terms of the dc resistivity () and
mean scattering time (). The units of are cm and the units of are
femptoseconds.
~ ( ) =
i 4h 2
(h + i2 )
G.E. Jellison, Jr. and F.A. Modine, Parameterization of the optical functions of amorphous materials in the interband
region, Appl. Phys. Lett. 69, 371 (1996), Erratum, Appl. Phys. Lett. 69, 2137 (1996).
6
A.S. Ferlauto, G.M Ferreira, J.M. Pearce, C.R. Wronski, R.W. Collins, Xunming Deng, and Gautam Ganguly,
Analytical model for the optical functions of amorphous semiconductors from the near-infrared to ultraviolet:
Applications in thin film photovoltaics, J. Appl. Phys. 92, 2424 (2002).
8.7.
+MODEL Options
The full list of +MODEL Options are shown in Figure 8-44. To view many of
these, you first need to select Configure Options.
Model Calculation
Choose different model calculations (Figure 8-45) to handle non-ideal sample
situations. For most samples, the default setting of Ideal is preferred. However,
there are choices to Include Bandwidth (eV), Include Thickness Nonuniformity,
and Include Bandwidth(nm), when these non-ideal situations are suspected. These
options are best used in conjunction with Depolarization data fitting.
Delta Offset
Used to correct for windows.
Multi-Sample Analysis
When multiple data sets are appended, but from different samples, they can be fit
simultaneously and this section allows different fit parameters to have multiple
values. An example of multi-sample analysis is shown in Figure 8-46, where three
oxides on silicon are fit with a common Cauchy layer to describe same index, but the
thicknesses are allowed to differ for each data set.
Figure 8-46. Example of Multi-Sample Analysis, where all three data sets are fit
simultaneously with common Cauchy optical constants and different film thickness.
8.8.
+FIT Options
The full list of +FIT Options are shown in Figure 8-47. To view many of these, you
first need to select Configure Options.
Selected Data
Choose which type of data to include in the Fit. Choices include Standard
Ellipsometry (SE), generalized-SE, Mueller-matrix, and Intensity data only, as
shown in Figure 8-48.
Figure 8-48. Choose the selected data type to use during fit.
Fit Weight
Select the fit weighting, with choices as shown in Figure 8-49.
Clicking the mouse on the Type parameter allows you to change the derived
parameter calculation. The available types are shown in Figure 8-51. Some of the
types allow the specification of a wavelength and layer number for the parameter.
After fitting data using a model with derived parameters, the fit results section shows
the model fit parameters and the derived parameters. The derived parameters are
shown in gray instead of black.
Figure 8-52. Fit Results window showing the Derived Parameter for index of the SiO2 layer
at 632nm.
A new derived parameter type called User Equation has been added to
CompleteEASE (new in version 3.38) so simple calculations can be performed after
a fit has finished. If there is an error in the equation then the derived parameter will
not be displayed in the Fit section after a fit is completed. This derived parameter
has 4 items that can be edited a name, equation, and low and high specifications.
Figure 8-53 shows how the derived parameter displays in a model.
Figure 8-53. Display of the Add Derived Parameter within the model.
To edit the equation press the Edit Equation item. This brings up the Edit Equation
dialog as shown in Figure 8-54. In addition to specifying the equation this dialog
allows the user to enter the number of decimal places that should be displayed.
The equation can contain numbers, operators, variables, and parentheses for
grouping. The available operators are: + (add), - (subtract) , * (multiply), /
(divide), and ^ (power). In addition there are a number of constants and functions
that can be used. These are shown below.
Functions
sin
cos
tan
log
abs
sqrt
Constants
PI
8.9.
+OTHER Options
Wvl. Range Expansion Fit
Press this command to execute a Wavelength-Range Expansion Fit. The model is
restricted to longer wavelengths and fit. Then, the data range is increased toward
shorter wavelengths in increments. As each new range of data is added, the data are
fit again. This method works well in combination with the B-Spline layer and is
demonstrated in Example of Section 4.3
Figure 8-55. Example of the Try Alternate Models report that is generated.
Parameter Uniqueness
This command allows user to test the uniqueness of a fit parameter. A single fit
parameter is selected to vary over a range of values. At each value, the fit parameter
is fixed while all other fit parameters are varied to find the best fit (lowest MSE).
The MSE is recorded to visualize the shape of MSE profile versus the varied fit
parameter. An example MSE profile is shown in Figure 8-56.
Simulate Data
This option is used to simulate data based on the existing model. When the
command is pressed, you will see the simulation settings as listed in Figure 8-57.
Configure Options
Choose this command to show hidden features in the +MODEL Options, +FIT
Options, and +OTHER Options sections of the model. The choices are shown in
Figure.
Figure 8-58. Configure which hidden options to show within the Model.
The controls available in the Hardware tab are different for the different
ellipsometer configurations. As a result, this chapter is divided into two major
sections: one section describing the AlphaSEs Hardware tab and another section
describing the M-2000 and RC2 Hardware tab.
9.1.
Hardware
Re-initialize
This button re-initializes the ellipsometer hardware, which initiates communication
over the USB link, scans the spectrometer, moves the motors to home position, and
tests the motors. Details of any errors that occur during hardware initialization can
be found in the Hardware and/or Error logs.
Park Z-Stage
Moves the Z-stage to the park position such that the Z-stage shipping lock can be
engaged.
Signal
Display
Displays signal intensity of the light collected by the spectrometer. Average,
maximum, UV, and IR intensity values are also reported in the status box, along with
the current spectrometer Dark Count and electronics Temperature. An example
is shown in Figure 9-2.
Z-Stage Scan
Scans the Z-stage to generate intensity vs. sample height profile. A sample should
be mounted and the ellipsometer unit set to the Off-Sample configuration before
clicking this button. The resulting signal intensity profile should appear
symmetrical, similar to the graph shown in Figure 9-3. The black dashed curve is a
polynomial fit to the measured profile, and the reported MSE= value quantifies the
symmetry of the profile. If the MSE value is greater than 0.02, contact your J.A.
Woollam Co., Inc. representative.
Calibration
S-T Baseline
Acquires baseline spectra for the Straight-Through configuration such that
accurate transmission intensity and Mueller matrix data can be acquired. For best
data accuracy, always perform a S-T Baseline after moving the ellipsometer optics,
or if it has been a long time (> 1 hour) since the previous baseline.
Off-Sample Baseline
Acquires an Off-Sample baseline such that qualitative reflection intensity data can
be acquired. This option also calibrates and stores the angle of incidence which is
used for subsequent measurements. For best angle of incidence accuracy, perform
an Off-Sample Baseline after moving the source or receiver units, or after mounting
the sample chuck.
Show Logs
Hardware and Error Logs
These buttons display the Hardware and Error logs, which contain information useful
for diagnosing and debugging problems with the instrument. If the instrument is not
working properly, your J.A. Woollam Co., Inc. representative may require this
information to diagnose the problem: click the Save Debug File button that
appears in the log file, and email C:\CompleteEASE\CompleteEASE_Debug.zip file.
An example Hardware Log is shown in Figure 9-4 and Error Log is shown in Figure
9-5.
9.2.
Hardware Status
The Hardware Status panel displays the current state of the hardware. This is the
same information displayed in the Hardware Status panel of the Measurement tab.
System Information
The System Information panel displays details about the ellipsometer configuration
and provides access to the hardware configuration dialog and hardware log. The
information provided includes the type of ellipsometer, wavelength range, and
current motor positions. Please have this information available when contacting the
J.A. Woollam Company for assistance.
Edit Hardware Config.
Pressing this button will bring up the Hardware Configuration dialog. This dialog
contains many settings for configuring the behaviour of the hardware. The dialog
displays a description of each parameters function so the items will not be described
here.
View Hardware Log
This button displays the Hardware log which contains information on calibrations
and initializations.
Controls
The Controls panel provides access to hardware functions other than those needed
for acquiring data. The hardware functions are separated into groups displayed as
tabs. The tabs are General, Calibration, System, and Misc..
General
The General tab shown in Figure 9-7 is nearly the same for all of the different
ellipsometer configurations. The available controls are as follows:
Align Sample
Align the sample without acquiring data. For systems with automated alignment
capabilities a dialog is displayed allowing the user to choose between manual
alignment and automated alignment (Figure 9-8). Choosing No in the dialog
displays the manual alignment dialog.
Figure 9-8. The Sample Alignment dialog gives the user a choice of automated alignment or
manual alignment.
Display Signal
This function displays the raw signal intensity. This screen is useful for debugging
light intensity problems.
System Check
A System Check is the most common procedure for calibrating the ellipsometer. See
the hardware manual for details on the System Check procedure.
Move To Load Pos.
This function configures the system to make loading a sample easier. This applies
only to systems with automated z-stages and translators. Generally, pressing this
button will result in the z-stage moving to its maximum value and the translator
moving towards the user.
Calibration
The Calibration tab appears different depending on the type of ellipsometer. Figure
9-9 shows the Calibration tab for an M-2000 and Figure 9-10 shows the tab for an
RC2. It is not common for a user to need the controls in this tab as a System Check
will perform many of these functions.
Initialize Hardware
This button re-initializes the ellipsometer hardware, which initiates communication
over the USB/TCPIP link, scans the spectrometer and moves the motors to home
position. Details of any errors that occur during hardware initialization can be found
in the Error log.
Coarse Calibrate and Fine Calibrate (M-2000 Only)
System
The controls in the System tab are dependent on the type of base the ellipsometer is
attached to. The main types of bases are the ESM (horizontal sample mount,
automated z-stage and angle), ES130 (horizontal sample mount, manual z-stage and
angle), Fixed Angle Focused (horizontal sample mount, automated z-Stage), Vertical
(vertical sample mount, automated angle), and Accumap. Some of the bases have
multiple attachments which can also change the controls available on the System tab.
Figure 9-11 through Figure 9-13 show the System tab for some of these bases. Brief
descriptions of the different buttons on this panel can be found below. See the
hardware manual for a detailed description of the use of these functions.
Figure 9-11. System tab for ESM, ES130, and Fixed Angle bases.
automatically detect which stage is being installed. The other bases will display a
dialog with the different stage options and the user will select the one being installed.
Misc.
The controls in the Misc. tab cover functions that did not fit into the General,
Calibration and System tabs and are dependent on the type of base the ellipsometer
is attached to and which attachments are installed. Some systems will not have a
Misc. tab.
10.
The Options tab is shown in Figure 10-1. This tab consists of four panels: Display
Units, Configuration Controls, Miscellaneous, and About CompleteEASE. A
brief description of all commands within this tab is described.
10.1.
Display Units
Sets the Wavelength Units:, Thickness Units, Optical Constants, and Graph
Ellipsometric Data: settings to their default values.
Wavelength Units
Choose from , nm, m, eV, and 1/cm for the wavelength units.
Thickness Units
Choose from , nm, m, and mm for the thickness units.
Optical Constants
Display and report optical constants as either n & k or e1 & e2.
These settings are saved when the program is exited, and restored when the program
is restarted.
Use Defaults
Return all settings to their default values, which include wavelength in nm, thickness
in nm, and optical constants as n & k.
10.2.
Miscellaneous
Reset Win. Size
Return the CompleteEASE window size to the default size.
Manage Users
Press Manage Users to setup the user list, along with access level and password, as
shown in Figure 10-2. The user privileges can be defined for Engineer level, by
further pressing Define User Privileges, which opens the window shown in Figure
10-3. The User customization is further defined in the CompleteEASE
Configuration (next section).
Figure 10-2. Manage Users window, where multiple users, passwords and privileges can be
established.
10.3.
Configuration Controls
Edit Configuration
Allows customization of the CompleteEASE software configuration. Press this
button to open the window shown in Figure 10-4. Each item to the left can be
expanded by pressing the circle to the left of the name. Also, there are many hidden,
Advanced options, which will not appear until you check the box at bottom of the
screen: Show Advanced Config. Options. The list with Advanced options grows
to that shown in Figure 10-5.
Figure 10-7. Memory Management and Display Units section of the Configuration
Parameters window.
Figure 10-8. Remote Communications, Window Setup, and Graph Clipboard Parms sections
of the Configuration Parameters window.
Figure 10-9. Hardware Simulation Mode, In Situ Tab Parameters, and Shutter Controls
sections of the Configuration Parameters window.
10.4.
About CompleteEASE
Displays the About box shown in Figure 10-10, which contains the software
version number, memory availability, and copyright information.
11.
11.1.
Appendices
Troubleshooting Guide
Please contact your J.A. Woollam Co., Inc. representative with any questions
relating to CompleteEASE or your ellipsometer system. To reduce the support time,
please create and email the debug file described in the following section before
contacting your representative.
Your closest Woollam Representative can be found in Section 11.3.
2.
3.
4.
Email this file to your J.A. Woollam Co., Inc. representative to aid in debugging
problems with your ellipsometer system.
1.
Choose View Error Log from the Help menu. This opens the log view.
2.
Press the Save Debug File button at the bottom of the dialog. This creates the
file CompleteEASE_Debug.zip in the \CompleteEASE directory.
3.
Email the file along with a description of your problem and any supporting data
to [email protected].
10BAppendices 259
11.2.
Figure 11-1. Choose the CompleteEASE Data Transfer menu from FILE menu of the
Experimental Data Window.
2.
260 10BAppendices
3.
Browse to select the directory where you wish the new files to be saved.
Imported files will be saved in WVASE32 format in the directory chosen.
4.
After the files are imported, a message will appear, as in Figure 11-4.
Previous .SE and .iSE files are now saved in WVASE32 .dat format.
10BAppendices 261
From the Experimental Data window in WVASE32, select the file menu
and choose CompleteEASE Data Transfer. Then, select the Export Data
to CompleteEASE option as show in Figure 11-5.
2.
262 10BAppendices
3.
Figure 11-7. Browse for location to place the newly created CompleteEASE files.
4.
After the files are exported, the following message will appear. Previous
.dat and .bdt files are now saved in CompleteEASE .SE and .iSE
formats.
Figure 11-8. Message when all data files are exported successfully.
10BAppendices 263
11.3.
Japan
J.A. Woollam Japan
Fuji 2F 5-22-9 Ogikubo
Suginami-ku
Tokyo 167-0051 Japan
PH: 81-3-3220-5871
FX: 81-3-3220-5876
[email protected]
www.jawjapan.com
China
Lamda Pacific
Room 906, Block C
No. 70 CaoBao Rd.
Shanghai, 200235, China
PH: 86-21-64325169
FX: 86-21-64326125
[email protected]
www.lamdapacific.com
United Kingdom
L.O.T.-Oriel UK & Ireland
1 Mole Business Park
Leatherhead, Surrey
United Kingdom KT227 BA
PH: 44-1372-378822
FX: 44-1372-375353
[email protected]
www.lotoriel.co.uk
Korea
Wizoptics
#501, Taebo B/D
240-5 Gocheon-Dong
Uiwang-si, Gyeonggi-do
437-801 Korea
PH: 82-31-477-3785
FX: 82-31-477-3786
[email protected]
www.wizoptics.com
Taiwan
Titan Electro-Optics Co.
14 Fl., No. 19-11
San-Chung Road
Taipei 115, Taiwan
PH: 886-2-2655-2200
FX: 886-2-2655-2233
[email protected]
www.teo.com.tw
France
L.O.T. Oriel France
4, Allee des Garays
Palaiseau, France 91120
PH: 33-1-69194949
FX: 33-1-69194930
[email protected]
www.lot-oriel.fr
Singapore
Crest Technology
4 Loyang Street
Loyang Industrial Estate
508839 Singapore
PH: 65-6546-4811
FX: 65-6546-4822
[email protected]
www.crest-technology.com
Australia
Scientific Solutions
P.O. Box 4033
Caloundra BC
Queensland, Australia 4551
PH: 61-7-5438-2222
FX: 61-7-5438-2233
[email protected]
www.scisol.com.au
Italy
L.O.T. Oriel Italy
Via Francesco Sapori 27
Rome, Italy 00143
PH: 39-06-5004204
FX: 39-06-5010389
[email protected]
www.lot-oriel.it
India
Sinsil International
New Bramhand Annex
Phase 8, G7/G2
Thane (W) 400607
Mumbai, India
PH: 91-22-2489-5077
FX: 91-90-2722-5513
Israel
VST Service Ltd.
P.O. Box 4137
19 Imber Street
Petach-Tikva, 49130 Israel
PH: 972-3-92477-10
FX: 972-3-92477-11
[email protected]
www.vacuumltd.com
264 10BAppendices
File Structure 13
Fit 204
Fit Dynamic 205
Fit Log 87
FIT Options 234
Fit Results 55
Fit Scan Data 144
Fit Weight 236
Fitting the Data 50
B
Backside reflections 75
Backside Reflections 68
Bandwidth (nm) 147
B-SPLINE 107, 223
C
Calibration 244
Cauchy 63, 142, 216
Cauchy_WVL 217, 218
Changing the Graph 155
Clear Multi-Data Set Mode 158
Cody-Lorentz 124, 231
Compare Optical Constants 89
Comparing Results 88
Composition Library 228
Configuration Controls 255
Configure Options 241
Create Debug File 258
G
Gaussian 230
Generalized SE 192
Generate 204
Generate Report 90
Gen-Osc 114, 221, 230
Glass with Absorbing Film 122
Glass with Transparent Film 79
Global Fits 66, 111, 235
Graded Layer 49, 81, 220
Graph 193
Graph All Data Sets 155
Graph ScratchPad 70, 135, 196
H
Hardware Tab 242
Harmonic 230
HTML Clipboard 87
E
Editing Layers 208
Effective Medium Approximation 48, 215
Ellipsometry 22
EMA See Effective Medium Approximation
Error Logs 244
Exporting WVASE32 Data 262
L
Limit Wvl Range for Fit 111
Lorentz 230
M
Manage Users 254
10BAppendices 265
N, C, & S 50
T
O
Off-Sample Baseline 244
Options Tab 253
OTHER Options 239
P
Parameter Error Bars 55
Parameter Uniqueness 101, 132, 240
Parameterize 113
Park Z-Stage 242
Perform Thickness Pre-Fit 235
Pole 230
Pseudo Optical Constants 47
U
Uniaxial 229
Uniaxial-Diff 229
Use Clipboard Aspect Ratio 197
R
Range-Select 194
Range-Selecting Wavelengths 109
Re-analyzing Multiple Data Sets 90
Re-initialize 242
Return Path Ellipsometer 233
Roughness 131
S
Sample Alignment: 199
Save 98, 208
Saving a Model 127
Screen Layout 193
SE + T 94, 132
Selecting a Point 140
Selecting Multiple Points on a Map 154
Self-Assembled Monolayers 100
Sellmeier 143, 218
Set Ranges 142
266 10BAppendices
V
View Previous Fit Results 201
Visualizing Model Changes 62
W
Wavelength Range Expansion 109
Wavelength Units eV 140
Wvl Shift (nm) 233
Wvl. Range Expansion Fit 239
Z
Zoom All 194
Zooming on Map 153
Z-Stage Scan 243