Oxidation Effects on the Optical
Constants of Heavy Metals in the
Extreme Ultraviolet
Amy Grigg
R. Steven Turley
Brigham Young University
1
Why Extreme Ultraviolet?
Thin Film or Multilayer Mirrors
Soft X-Ray Microscope
EUV Lithography
Earth’s Magnetosphere in the EUV
2
Oxidation Problem



High absorption
requires thin films.
The surfaces of many
materials oxidize.
At optical wavelengths,
this oxidation is often
negligible. It is a major
issue for our thin films,
however.
3
X-Ray Photoelectron Spectroscopy
4
How XPS works
K max  hv  
5
Electron Binding Energy
4500
Th
4000
4f5/2
3500
4f7/2
Counts
3000
2500
2000
Th
4d3/2 4d5/2
O
1s
1500
C
1000
1s
500
0
1000
800
600
400
Th
5d3/2 5d5/2
200
0
Binding Energy (eV)
6
Peak Shifts

3.6K
3.4K
3.2K
3K
2.8K
2.6K
2.4K
2.2K
2K
1.8K
1.6K
1.4K
1.2K
1K
800
600
400
200
354.9
352.9
350.9
348.9
346.9
344.9
342.9
340.9
338.9
336.9
334.9
332.9
330.9
328.9
Thorium peaks
on surface
326.9

3.2K
3K
2.8K
2.6K
2.4K
2.2K
2K
1.8K
Thorium peaks
after oxygen is
gone
1.6K
1.4K
1.2K
1K
800
600
400
200
354.9
352.9
350.9
348.9
346.9
344.9
342.9
340.9
338.9
336.9
334.9
332.9
330.9
328.9
326.9
7
Depth Profiling: Method 1
 Sputtering:
Argon ions knock
off individual
atoms and drill a
hole through the
sample
8
Depth Profiling: Method 2
 Variable
angle scans:
Depth is obtained as the analyzer is
moved towards incidence
θ
eee-
Sample
9
Variable Angle Results
Only penetrates about 80 Angstroms
into the sample
 This allows us to see surface
contamination, but not composition
with depth
 Results are averaged: cannot obtain
resolved composition with depth

10
Sputtering Results
11
Results


This isn’t the true interface because the
probability of electron escape is
exponential with depth
Ideal interface:
12
Results
Si interface is not ideal
 This is likely due to the sputtering
process

Implantation
 Sputtering roughness
 Shape of sputtered area

13
Too Much Oxidation
These samples were only a few hours
old.
 We need more uniformity for singlelayer reflectors.
 Solution: Make ThO2 mirrors.
Reflection is also high in EUV and it
should be more uniform.

14
ThO2 Results
15
ThO2 Conclusions
 Fully
oxidized thorium is much
more uniform.
 ThO2 shows definite promise as a
single-layer reflector in the EUV.
16
Ultimate Goal

Determine the optical constants of
our materials
Ν  n  ik  1    i
17
Fitting Reflection/Transmission


Others have taken the oxide layer into
account, but have assumed an ideal interface
Measured oxidation (XPS) shows that our
approximation is inaccurate
18
Put the Data to Use


Incorporate what we know about
oxidation into the way we determine our
optical constants.
Write a data-fitting program
Inputs: measured oxidation,
reflection/transmission data
Outputs: optical constants of measured
element (δ and β)
19
Data-Fitting Program


Task: Find the most likely optical
constants (δ and β) to have produced
the measured data that is input
Least squares routine: Minimize
N
   ( f ( xi ;  ,  )  yi )
2
2
i 1
20
Function f ( xi ;  ,  )

Function f uses Matrix Method with
Fresnel coefficients to determine
theoretical reflection and transmission
 D1 
 D2 
   B 
 U1 
U 2 
21
Fit and Data
22
23
Continued Research



Make a sample, measure it with XPS,
model the oxide layer with the program
based on the XPS results, and test method
Make user interface for program if method
works well
Determine cause of non-ideal Si interface
24
Acknowledgements
A special thanks to
R. Steven Turley
David Allred
Matt Linford
BYU Thin Films Group
Physics & Astronomy Department Funding
ORCA Mentoring Grant
NASA Space Grant
Images from www.schott.com/magazine/english/info99/ and www.lbl.gov/Science-Articles/Archive/xray-inside-cells.html. 25
Other Results of Interest
There was an increase in oxygen
when the sample sat for more than 4
or 5 minutes in between
sputtering/scans.
 This was observed for 5 out of 5
samples that sat still between scans.

26
* *
*
*
* indicates where the sample stood for more than
4 or 5 minutes in between scans
27
What Could This Be?
 Hypothesis:
This is likely due to
preferential sputtering.
 The argon ions will knock off
oxygen atoms more readily than
thorium.
 While sputtering, scans would
show less O than actually exists.
28
Future Research
 Test
preferential sputtering
hypothesis.
 Investigate other peak anomalies:
N, Ar
29