1. The document summarizes an experiment using an ellipsometer to measure the thickness and optical properties of dielectric and metallic films. Key findings include measuring indium tin oxide thickness at 6445.28 angstroms and observing relationships between conductivity, wavelength, and optical constants n and k for various metals. 2. Methods included using an ellipsometer and software to obtain thickness and optical constant data for standard samples and test films, then calculating properties like conductivity from the constants and relationships. 3. Results showed thickness variation in standard samples, optical constant and conductivity trends differed between metals like platinum and gold, and conductivity was found to depend on wavelength and the n and k relationship.

1. 1
Ellipsometer – Dielectrics and Metallic Films
Tyler Goerlitz
California State University East Bay
Abstract
Purpose: During this experiment we wanted to measure film thickness of different
dielectrics. We also aimed to calculate various properties of different metallic films
using measured film thickness and optical constants.
Methods: We used a J.A. Woollam Co. Inc. -SE spectroscopic ellipsometer along
with CompleteEASE software. We were able to obtain our data by performing
measurements, then fitting to model data using the CompleteEASE softare. Optical
constants were obtained by using various formulas.
Results: For indium tin oxide (ITO) we measured the transmission to be .903, and
the reflectivity to be .0385. We were able to find a relationship between
conductivity, optical constants n and k, and the wavelength of light. We used this
relationship to observe the conductivity of different metallic films.
Conclutions: After learning how to properly use the ellipsometer, we were able to
take data and fit it to different models provided by the CompleteEASE software.
Using the measured data we observed thickness of the films and were able to find
relationships between optical constants, wavelength, and conductivity.
1. Introduction
We initially aimed to learn how to
properly use the J.A. Woollam Co. Inc. -SE
spectroscopic ellipsometer. We used two
different standard samples to understand the
fundamentals of the machine and the
CompleteEASE software. After we were
comfortable using the ellipsometer and the
software we moved to measure and calculate the
properties of different dielectrics and metallic
films. We measured the thickness of indium tin
oxide (ITO) as well as the optical constants n and
k. N being the refractive index and k being the
extinction coefficient. From these constants we
were interested in finding a relationship between
the optical constants, wavelength, and
conductivity. We were also interested in
calculating the transmission and reflectivity of
the different materials. In the next section I will
go over the relationship we found for the
conductivity.
2. Methods
When initially learning to use the
ellipsometer we measured the thickness of two
standard films. The first was a 250-angstrom
silicon wafer. We made sure the source and
detector arms were at 70 degrees. We measured
the wafer and used a model to fit the raw data.
The model used, Si with Termal Oxide, was given
in the CompleteEASE software. To see if there
were any variances within the film thickness we
moved the wafer to multiple positions and re-
measured the thickness. We repeated this process
for a 600-angstrom wafer.
After the standard films we moved to
measure ITO. We first had to determine which
side of the ITO was the film side. This initially
proved to be troubling. Since the ITO is a
dielectric there should be some resistivity on the
film side. Given this we used a DMM to see which
side showed resistivity. After we determined the
film side we measured the film thickness. We ran
into some problems here, the data and error that

2. 2
we were getting seemed to be extremely wrong.
We were using the ITO on glass model. After
trying different models and observing the errors
associated with each model, we decided to use
the ITO on glass with backside reflection model.
This model gave us the most reasonable thickness
and mean squared error (MSE) values. These
values will be discussed in the next section. Once
we had reasonable data for the ITO we observed
the optical constants. From the optical constants
we were able to calculate conductivity, resistivity,
and transmission. To calculate the conductivity
we used the following formula. where
sigma is conductivity, n and k are the optical
constants, c is the speed of light, and lambda is
the wavelength. Data findings will be discussed
in the next section.
Using these same methods we measured
and calculated the conductivity, reflectivity, and
transmission for gold, copper, and platinum.
Given that all of these data points are dependant
on either n and k or n, lambda we graphed and
observed different trends for the conductivity
given different wavelength. It is important to note
that when taking measurements for the three
different metallic films we were not able to fit our
raw data to a model.
3. Experimental Data
When we measured the 250-angstrom
wafer, we saw that the wafer was not uniform in
film thickness. We did find an average film
thickness of 243.00-angstroms. This proves to be
close to the marketed 250-angstroms. The first
measurement that was made was 239.55 .079-
angstroms, with an MSE of 1.89. The MSE
associated with this measurement and the
successive measurements was very good. We
never observed a MSE higher than 2.0 when
measuring the 250-angstrom wafer.
When we repeated the same process for
the 600-angstrom wafer. We found that the
results were not as consistent as the 250-
angstrom wafer. We found that there was a
decent amount of variance. The highest value
measured was 604.27-angstroms, the lowest was
559.49-angstroms. Although the thickness varied
from the marketed 600-angstroms, we observed
very good MSE values with the highest being .986.
After proper determination of ITO film
side. We measured the thickness of the film to be
6445.28 79.67 with an MSE of 64.50. Although
the MSE here is not ideal, we ran the
measurement multiple times and observed MSE
values ranging from 64 to 66.33. After multiple
measurements we decided that the MSE values
we were getting were reasonable. In figure 1 we
can see the trend of the optical constants n and k
vs. wavelength.
Figure 1:
It is clear to see that at about 380nm we have a n
value of 1.52 and a k value of .0033. From this
graph we can also see that since the n and k
values are not constant at all wavelength that the
conductivity, reflectivity, and transmission all
with change with wavelength.
For the next three metals, copper,
platinum, and gold, we did not have a model to fit
the data to. As stated from above the conductivity
will change with wavelength. Figures 2, 3, will
show the relationship between conductivity and
wavelength for platinum and gold, respectively.

3. 3
Figure 2:
Figure 3:
Notice here the difference between the two
graphs. The conductivity of platinum is almost
linear while the conductivity of gold decreases to
about 600nm and levels off. Figures 4 and 5 will
show the optical constants vs. wavelength for
platinum and gold, respectively.
Figure 4:
Figure 5:
Notice how in figure 4 we have a nice linear
relationship between the optical constants and
the wavelength. Both n and k are increasing at
almost the same rate. Where as in figure 5 we
have n decreasing with wavelength and k
increasing with wavelength, this is why in the
above graph we see that the conductivity is
decreasing with the wavelength.
4. Conclusion
From the above data we were able to
observe some important relationships between
optical constants, wavelength, and conductivity.
The most important relationship to observe is the
conductivity. Perhaps the most interesting metals
to look at are platinum and gold. Seeing the
conductivity curve of the two metals is so
different. It is very important to note that
conductivity does in fact change with wavelength.
We saw the relationship from the equation
. Not only will the conductivity change
with wavelength, but the conductivity also
heavily depends on the relationship between n, k,
and wavelength. We were able to directly see this
from figures 2 and 4. Each graph showed a nice
linear relationship. Where as with figures 3 and 5
we saw decreasing graphs.
0
20
40
60
80
0 500 1000
Conductivity (Ohm*m)^-1vs.
Wavelength (nm)
0
10
20
0 500 1000
Conductivity (Ohm*m)^-1vs
Wavelength (nm)

4. 4
Acknowledgements:
1. Professor Erik Helgren, private
communication
2. Louis Rotella, private communication
3. Michael Pheng, private communication