Optical Analysis of Titania: Band Gaps of Brookite, Rutile and

Anatase

Ryan Lance
Advisor: Dr. Janet Tate

A thesis presented in the partial fulfillment

of the requirements for the degree of

Bachelors of Physics

Department of Physics
Oregon State University
May 5, 2018
Contents
1 Introduction 2

2 Optical Phenomena of Thin Films 3

2.1 The Index of Refraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Absorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.3 The Band Gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

3 Methods 6
3.1 The Grating Spectrometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.2 SCOUT for Optical Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . 9

4 Results and Discussion 12

4.1 Band Gap Dependence on Thickness . . . . . . . . . . . . . . . . . . . . . . 14

5 Conclusion 15

6 Appendix 16
6.1 Grating spectrometer settings . . . . . . . . . . . . . . . . . . . . . . . . . . 16
6.2 Filtering 2nd Order Light . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
6.3 Band gap of the substrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

7 Using SCOUT 17
7.1 User configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
7.2 The Layer Stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
7.3 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

8 Acknowledgments 18

1
List of Figures
1 Indirect and direct band gaps . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 The grating spectrometer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3 TiO2 Raw Film Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4 Transmission, reflection, and corrected transmission spectra. . . . . . . . . . 8
5 High-energy region of raw spectra . . . . . . . . . . . . . . . . . . . . . . . . 9
6 Screenshot of the SCOUT interface . . . . . . . . . . . . . . . . . . . . . . . 10
7 Refractive index model constructed in SCOUT. . . . . . . . . . . . . . . . . 11
8 Density of states in the OJL band gap model. . . . . . . . . . . . . . . . . . 11
9 High-fraction brookite film on SiO2 . . . . . . . . . . . . . . . . . . . . . . . 12
10 High-fraction anatase film on SiO2 . . . . . . . . . . . . . . . . . . . . . . . 13
11 High-fraction rutile film on SiO2 . . . . . . . . . . . . . . . . . . . . . . . . 13
12 Gap energy vs. Thickness for many polyphase TiO2 films. The phase plots
(Rutile, Brookite, Anatase) show how much of each phase is present in each
film. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
13 Measurements taken on film 056-1. There is an inverse relationship between
the gap energy and film thickness–an artifact of the analysis. . . . . . . . . . 15
14 SnCaSe measurement. T , R, and T /(1 − R) with and without a color filter. . 17

2
List of Tables
1 Literature on the band gaps of rutile, anatase, and brookite. . . . . . . . . . 2
2 Band gaps and thicknesses of high-fraction films of anatase, rutile and brookite. 12
3 Grating Spectrometer Settings . . . . . . . . . . . . . . . . . . . . . . . . . . 16

3
Abstract
The optical transition in high-fraction polymorphs of titania (TiO2) were investigated to
determine the band gap behavior of the most common polymorphs—brookite, rutile, and
anatase—the values of which are varied in the literature. The direct optical band gaps of
brookite, rutile, and anatase, were determined to be 3.37(7)eV, 3.41(11)eV and 3.59(2)eV
respectively.
TiO2 was grown in-house via pulsed-laser deposition onto fused silica. The reflection
and transmission spectra of high-phase fraction brookite, rutile and anatase, were measured
with a grating spectrometer. Optical analysis software, SCOUT, was implemented with
dielectric function models to perform a parameter fitting on transmission and reflection
spectra, produced by the optical models, to the experimental data. From the parameter
fitting, the optical absorption and index of refraction were extracted. The index of refraction
was approximately 2.5 at 500 nm. A relationship in the high-phase brookite film showed a
relationship between the thickness and band gap. Initially this appears to be a quantum well
effect. However, the relationship is in face determined to be a consequence of our calculations
favoring band gap fitting at low thicknesses–leading to an artificial trend.

1
1 Introduction
TiO2 is used abundantly in industry for paint pigmentation, as a photocatalyst, in dye-
synthesized solar cells, as a UV-protectant and more [1][2][3][4]. TiO2 is a wide band gap
semiconductor, with three common polymorphs; rutile, anatase and brookite. The ground
state, rutile, is easy to make and the most common found in industry, while brookite is the
least. As a result, rutile has been studied extensively, and its band gap energy is the most
well-known band gap of 3.0 eV. The band gaps of brookite and anatase are not known to the
same degree of precision as rutile—anatase has been reported to have a band gap of about
3.2 eV, and brookite has been reported to have a band gap anywhere between 3 eV to 3.6 eV.
There is a lack of consistency in the literature values of the band gaps of TiO2’s polymorphs
as shown in Table 1. TiO2 is usually assumed to be a direct band gap semiconductor, so
many labs measure for the direct band gap (labeled D). Often the indirect gap is measured
as well (labeled I).

Ref/Year Rutile (eV) Anatase (eV) Brookite (eV) TiO2 (B) (eV)
[3],2008 3.01(I)-3.37(D) 3.20(I)-3.53(D) 3.13(ID)-3.56(D)
[4],2018 3.45 (D)
[5],2014 3.2 3.0-3.2
[6],2012 3.0 (I) 3.2 (I) 3.03,3.05,3.13 (I)
[7],2014 3.0 (D) 3.2 (D) 3.29 (D)
[8],2013 3.03 3.20
Table 1: Literature on the band gaps of rutile, anatase, and brookite.

The goal of this thesis is to provide measurements for the band gap of high-fraction
brookite, anatase, and rutile films. TiO2 films of high-phase brookite, rutile, and anatase
were grown and characterized. The films are deposited in a an amorphous phase using
pulse-laser deposition, then annealed at 400◦ C. Most of our films are annealed into some
combination of the three polymorphs. It is not understood yet how to make TiO2 consistently
choose the brookite phase, although ways to make pure-phase TiO2 nanoparticles from amor-
phous TiO2 and other techniques for growing high-fraction brookite films are in development
[3][4]. We have produced several films of high-phase brookite, anatase and rutile in thin film
form, and transmission and reflection spectra are measured using a grating spectrometer.
The direct band gap of the high-phase films were then calculated using an optical analysis
software.
The band gaps of brookite, anatase, and rutile films are characterized using optical
spectrometry. Transmission and reflection data are taken between λ = 200 → 2000 nm via
grating spectrometer. Rudimentary calculations are performed using the Sellmeier equation
to model the refractive index and standardized thin film analysis. A method developed for
calculating the band gap developed by linearization of the absorption curve, colloquially
called Tauc analysis, is used to measure the band gap [10]. Tauc analysis provides us with
a way to measure the spectra for a direct or indirect band gap by linearizing the absorption
curve two different ways.

2
 Another way we measure the band gap is by using an optical analysis software called
SCOUT. SCOUT provides a model constructed by [12] for incorporating direct band gap
transitions in the dielectric function models of our film. The direct gap model in SCOUT is
used to efficiently measure the spectra of many TiO2 films. SCOUT calculates the theoretical
transmission and reflection spectra from modeling a film with a dielectric function model
on a substrate of our choice, then fits the theoretical spectra to the experimental data
by adjusting parameters in the dielectric function model. This is a powerful method for
film characterization, and is capable of distinguishing the band gaps of rutile, brookite and
anatase.

2 Optical Phenomena of Thin Films

Our lab grows TiO2 phases in thin film form. To determine the optical properties, the
transmission and reflection spectra are measured using a grating spectrometer. The spectra
are then analyzed for absorption, refractive index and thickness. The films contain two
major features that help in the determination of the thickness and band gap: interference
fringes and strong absorption at the band gap. The interference fringes are a product of the
interference of internal reflections happening inside the film. The periodicity and size of the
fringes in the transmission and reflection spectra let us determine the thickness d of the film.
Once the thickness is known, the absorption can be determined by removing the fringes from
the transmission spectra, then applying Beer’s law and solving for the absorption α(λ).

2.1 The Index of Refraction

The index of refraction ñ(λ) of a material contains all the optical quantities we might want
to study. In the spectral region accessible by our grating spectrometer, the real part of the
index of refraction is approximated by the Sellmeier equation (equation 1).
N
2
X Ai λ
n (λ) = 1 + (1)
i=1
− λ20i
λ2
The Sellmeier equation is expanded to the first order for approximating the index of refrac-
tion of an unknown material. To compute the index of refraction of a thin film on a substrate,
the transmission and reflection coefficients are computed from the Sellmeier equation and a
preliminary guess for film thickness. The absorption is calculated from the corrected trans-
mission, then related to the extinction coefficient κ by α = 4πκ/λ. Theoretical transmission
and reflection coefficients are computed from the index of refraction of substrate and film
by the definitions of transmittance and reflectance. In practice, the angle of incidence is
θi = 5.5◦ . This is a sufficiently small angle to assume cos θi ≈ cos θt ≈ 1. The transmission
and reflection coefficients
Ir  ñf ilm − ñsubstrate 2
R= = (2)
I0 ñf ilm + ñsubstrate


It cos θt  ñf ilm 2

T = = (3)
 
I0 cos θi ñf ilm + ñsubstrate


3
Where the index of refraction is the complex function:

ñ(λ) = nr + iκ (4)

In practice, an optimization algorithm changes parameters Ai and λ0i in the Sellmeier equa-
tion as well as thickness d to minimize the difference between theoretical transmittance and
reflectance to the experimental data. The the model fits the data in the low absorption
region, where only fringes are present. Completion of the fit yields the real part of the index
of refraction nr and the thickness d. Once the absorption is determined, the wave vector κ
is simulated from the absorption by the relationship:
αλ
κ= (5)
4π
Which completes the dielectric function ñ(λ) for our material.

2.2 Absorption
Calculation of the band gap starts by measuring the absorption of a film. We first assume
the normalized intensity of reflection R, transmission T and absorption A add up to 1.

T +R+A=1 (6)
This is an assumption that works well for determining the absorption from bulk materials,
but the transmission and reflection data in films contain interference “fringes” that don’t
permit us to treat the absorption this way. We define the ‘corrected transmission’, Tc , to
be the transmitted intensity of light when there are no interference fringes. It is shown by
McIntyre [13] that by dividing the transmission by 1 minus the reflection, removes the fringes
from the spectra (for low absorption regions).
T
Tc = (7)
1−R
The corrected transmission will correctly remove the fringes from the film. The absorption
is defined by the change in intensity per unit length. For an absorption α(λ) and film
thickness d, the transmitted intensity for a film with no interference effects would be It (λ)
in accordance with Beer’s Law.

It = Io e−(αd) (8)
We can treat the corrected transmission as the transmission of light through our film with
the interference fringes removed (as if it were a bulk sample). The corrected transmission is
then the ratio between the throughput intensity and the incident intensity.
It T
= = e−αd (9)
I0 1−R
If the sample thickness is known (from analysis of the fringes), then absorption α can be
solved for by

4
 1 − R
α = ln d−1 (10)
T
The absorption α(λ) contains all the relevant information for analyzing the band gap. Com-
puting the band gap can be done by linearizing the absorption and checking for the onset of
strong absorption (Tauc method) or by fitting a parameter model to the relevant spectra.

2.3 The Band Gap

The band gap is indicated by the region of strong absorption. One way to calculate the
band gap is using a method of analysis proposed by Tauc, Grigorovici, Vancu [10] which
involves linearizing the absorption–this is called the Tauc method. In the Tauc method, the
density of states is assumed to be parabolic, so that it can be linearized by a square-root
function. When the conduction band is directly above the valence band, the transition is
called a direct gap. If the conduction band is offset in momentum space with respect to the
valence band, the transition is called an indirect gap. Both gap types are shown in Figure
1. The absorption is linearized by the function (αE)n , where n = 2 for direct gap materials
and n = 1/2 for indirect gap materials. The absorption α is multiplied by the energy E to
make the quantity dimensionless.

Figure 1: Indirect and direct band gaps

Each polymorph of TiO2 that we find in our films has a different band gap, and each band
structure has indirect and direct transitions. Since all three phases show up in most of our
films-optical data is taken on regions where all three phases are present and multiple kinds
of transitions are taking place in each spectrum. We attempt to calculate the band gap by
assuming that the film has only a direct transition and only an indirect transition–which
means that both Tauc linearizations of absorption are used in band gap calculations. The
idealized representations of the density of states used in the Tauc method loosely match
real absorption in experiment. Due to the complexity of features such as tail states, we
use optical modeling software for computations. The optical models in SCOUT can fit
the experimental data by changing the parameters of a direct gap model and the index of
refraction. Performing measurements this way let the model adjust the dispersion relation
and the band gap simultaneously to obtain the best possible fit.

5
3 Methods
The thin films are measured optically by analyzing the transmission and reflection spectra
of the film the in ultra-violet (λ = 250 nm) visible and near infrared regions (λ = 2000 nm).
The transmission and reflection spectra contain interference fringes, which are analyzed to
measure the thickness of the film and dispersion relation of the material. The absorption and
band gap of the film are analyzed by computing the corrected transmission spectra, which
contains information about the absorption of the film. The spectra are measured using a
grating spectrometer, then analyzed with optical modeling software.

3.1 The Grating Spectrometer

The grating spectrometer (Figure 2) is used to measure the transmission and reflection
spectra of our samples. With all the available settings (listed in the appendix), the grating
spectrometer has a range of λ = 175 nm to λ = 1998 nm. This is made possible by two lamps,
four gratings, six color filters, and two detectors for measuring various ranges of wavelengths.
Table 2 in the appendix contains the working ranges of each component. It is important
that the temperature of the lamp (Xe or W) does not change throughout the measurement,
so lamps are given sufficient time to reach a steady state temperature and intensity. The
calculated reflection and transmission data contain artificial noise if a steady intensity has
not been achieved. An ozone eater is secured to the lamp to remove the ozone, protecting
the experimentalist from exposure. The lamp light enters a double monochromator fitted
with the correct diffraction gratings to isolate a particular wavelength. The gratings are
optimized to strongly reflect particular wavelengths. The 0.25 micron grating is optimized
to select 250 nm, the 0.50 micron to select 500 nm light etc. The ‘single wavelength’ light
leaves the monochromator, is collimated then focused into the detector. When measuring a
typical spectrum, the power is measured at a single wavelength, then the monochromator
scrolls to the next wavelength. This is done in increments of 1 nm until the whole spectrum
is measured.
The raw lamp spectrum is measured first. The raw lamp spectrum is the transmitted in-
tensity directly from the monochromators, with no film in place. The background spectrum
is measured by the detector when the beam is blocked. Next the sample is placed in the
beam’s path to measure the reflection and transmission intensities. The sample is placed
at an angle so that when the reflection intensity is measured, the detector is not placed in
the optical path. The angle of incidence is kept very small θi < 6◦ so that we can use the
small angle approximation sin θ ≈ θ for computing the transmission and reflection. The raw
transmission is collected by passing the beam through the sample, and the raw reflection
is collected from the reflected light. Figure 3 shows the set of raw data collected from a
TiO2 sample with a Xe lamp source. Notice that the largest spectra is the lamp spectra
(blue), which contains the total intensity at each wavelength. The raw transmission and
reflection spectra (grey and yellow respectively) are lower. On this scale, the dark spectrum
(orange) appears to be zero.
The spectra gathered from measuring the reflected and transmitted beam intensities are raw
spectra. The actual reflection and transmission spectra are computed by subtracting the
background (dark) spectrum then dividing by the lamp spectrum. The resulting spectra are

6
Figure 2: The grating spectrometer.

Figure 3: TiO2 Raw Film Spectra

7
heretofore referred to as transmission and reflection, and they will be used in the analysis.
Once the raw data are collected, the transmission (T ) and reflection (R) spectra are calcu-
lated by first subtracting the dark spectrum, then dividing by the lamp reference spectrum.
This will yield the actual percent intensity (between 0 and 1) of reflection and transmission
for all wavelengths measured. The reflection and transmission are calculated by:
Ir − Idark
R= (11)
Ilamp − Idark
It − Idark
T = (12)
Ilamp − Idark
The spectra shown in figure 4 are the reflection and transmission spectra calculated
from the raw data. The gray curve is the corrected transmission T /(1 − R). The corrected
transmission is the transmission of the sample with the fringes removed. This is useful for
calculating the band gap using Tauc analysis, for which you need the corrected transmission
(equation 9).

Figure 4: Transmission, reflection, and corrected transmission spectra.

The data for TiO2 film 072-4 was taken in the 250 to 999 nm range. The transmission
and reflection curves (blue and orange) contain fringes in the region of low absorption (above
λ = 350 nm). We can see in the transmission and corrected transmission the data drops to
zero around λ = 350 nm. The band gap is seen where the absorption begins to set in around
360 nm.

8
 Figure 5: High-energy region of raw spectra

Figure 5 shows the intensities of the raw data in the high-energy region. The background
intensity (Raw Dark) is about 10−12 W through the whole range. Raw data collected with an
intensity less than 10−11 W are unlikely to yield physical information because the intensity
is comparable to the background level. Spectra taken in the ultraviolet region of our lamp
in Figure 5 show the absolute minimum of resolvable intensity starts around 175 nm or 7
eV. At this energy, the effect of the lamp warming causes the reflection spectra to be larger
than the intensity of the lamp—which is a nonphysical result. This is cured by letting the
lamp warm up for a longer time. The optimal time to take data is at least an hour after the
lamp has been turned on.

3.2 SCOUT for Optical Modeling

An optical modeling software called SCOUT is used to calculate the index of refraction,
thickness and the band gap. SCOUT computes the transmission and reflection spectra from
a user-defined layer stack (Figure 6). The layer stack for measurements has a thin film
on a substrate. Each layer is given properties based on the material that constitutes it.
The material which is being studied is given a model for the index of refraction—in this
case the Sellmeier model. The model usually contains information about dispersion and
absorption, features which are captured in the Sellmeier equation OJL interband transition
model (discussed later).
Figure 6 shows a screenshot of the SCOUT interface just after a fit has been performed
on a TiO2 film on a 1 mm glass SiO2 substrate. Theoretical (blue) and experimental (red)

9
transmission and reflection spectra are on the right–a preliminary glance tells us this is a good
fit. The center displays the current layer stack: TiO2 on SiO2 , and the left side displays the
sample name, measured thickness, fit deviation, fit rating, and values of all the fit parameters
used in the measurement. The fit rating says this measurement is “Acceptable”, but the
user must judge for themselves whether or not the fit parameters are physically reliable. For
TiO2 , we expect our band gap to be between 3.1 eV and 3.7 eV (25,000 cm−1 and 30,000
cm−1 ).

Figure 6: Screenshot of the SCOUT interface

To measure a thin film on glass, a substrate and thin film are defined in the layer stack,
with air in the half spaces on either side. The user defines the substrate (usually amorphous
glass), and the layer is given the properties of that material which are defined in the SCOUT
database or added by the user. The layer has some unknown thickness, index of refraction,
and absorption. These properties are given models with unknown parameters. The Sellmeier
equation is used to model the index of refraction, and the absorption is modeled by the OJL
band gap model. From the model for index of refraction, transmission and reflection spectra
are produced and compared to experimental data. The experimentalist gives SCOUT the
power to manipulate the model parameters (Sellmeier coefficients, band gap strength etc.).
In a optimization routine, SCOUT will change the model parameters to produce reflection
and transmission spectra that match up to experiment. For our measurements, SCOUT is
given the Sellmeier coefficients and parameters to the OJL band gap model. A typical model
of index of refraction is shown in Figure 7. The black and red lines are the real and complex
parts of the index of refraction respectively.
SCOUT can handle any number of layers, as well as the coherence of each layer. The
plane wave will retain its phase in layers that are coherent, and lose its phase in layers that
are incoherent. If the film is thin with respect to the coherence length of the light, then the
layer should be made coherent so that the light retains its phase through that layer. This is
a useful feature in SCOUT because only thin layers are coherent. The thin film layers are

10
 Figure 7: Refractive index model constructed in SCOUT.

generally about 50-300 nm thick, while the substrates are generally 1 mm thick. As a result,
the plane wave will retain its phase through the film and not the substrate.

Figure 8: Density of states in the OJL band gap model.

The O’Leary, Johnson, Lim (OJL) interband transition model (Figure 8), simulates direct
band gap absorption with tail states. The model has four parameters: gap energy, strength,
gamma, and decay. The gap energy expresses the width of the band gap, the gap strength
determines how sharp the gap onset is, the gamma parameter expresses the sub-gap ab-
sorption which could come from tail states or general disorder, and the decay expresses how
much the absorption decays after the gap in the high-frequency range.
The OJL model is similar to the Tauc model in that it assumes a parabolic density
of states–but it also contains a decaying exponential in the gap region. This is useful for
modeling tail states and line broadening in experimental data. OJL is used in tandem
with the Sellmeier equation to produce the complete index of refraction model used in the
measurement. The parameters of the OJL model and coefficients in the Sellmeier equation
become the “fit parameters” SCOUT uses to perform measurements.

11
4 Results and Discussion
Each TiO2 polymorph of interest (Brookite, Anatase and Rutile) has been created in a high-
phase fraction thin film form. Several regions of interest on each film were measured and
calculated for thickness, absorption and band gap behavior. The amorphous precursors were
created under specific conditions using pulse-laser deposition, then annealed at 400◦ C to
crystallization. The films were then measured using the grating spectrometer in the UV-Vis
region and calculations were performed in SCOUT.

Film No. Phase Composition No. Thickness (nm) Egap (eV, D)

079 Anatase (100%) 2 75.1 3.58
5 70.2 3.61
080 Rutile (100%) 2 52.0 3.33
5 31.8 3.49
056 Brookite (88%) (multiple) 55(6) 3.37(7)

Table 2: Band gaps and thicknesses of high-fraction films of anatase, rutile and brookite.

Table 2 lists the band gaps of anatase, rutile and brookite. The band gaps of anatase and
brookite are close to those found in literature: anatase has a band gap of about 3.6 eV,
and brookite a band gap of 3.4 eV, which is confirmed by Alotaibi. Rutile was measured to
have a larger band gap (3.3 − 3.5 eV) than is usually found in the literature. One spectrum
(reflection, transmission and corrected transmission) are plotted against wavelength for one
measurement on each film shown in the table in Figures 9, 10, and 11.

Figure 9: High-fraction brookite film on SiO2

All three high-fraction films exhibit features common of TiO2 spectra. In the visual and
near infrared regions we see fringe behavior onset in the reflection and transmission spectra.

12
 Figure 10: High-fraction anatase film on SiO2

Figure 11: High-fraction rutile film on SiO2

These are removed in the corrected transmission–as it is approximately equal to unity in the
visual to near infrared regions for all films. In the ultraviolet region, photons have enough
energy to excite the electrons across the band gap–which causes strong absorption, at which
point the transmission goes to zero. The strong absorption also causes the reflectivity increase
to about 0.35 in each film.
Figure 12 contains four plots with data from TiO2 polyphase films. The the top left
plot shows the distribution of all the measurements taken. In the “All Films” figure, an ‘x’
indicates one measurement. All four plots contain the same films, but the size of the markers

13
Figure 12: Gap energy vs. Thickness for many polyphase TiO2 films. The phase plots
(Rutile, Brookite, Anatase) show how much of each phase is present in each film.

in the Rutile, Anatase and Brookite plots indicate the respective amount of Rutile, Anatase
and Brookite in each film. There is no obvious trend in the anatase phase, but the brookite
tends to favor thicknesses below 80nm, and rutile favors higher thicknesses—-although trace
amounts are present in many films. This is consistent with an observation in the literature
which points out that the brookite favors this same thickness region [1].

4.1 Band Gap Dependence on Thickness

The data in Figure 12 shows a subtle relationship between the film thickness and the size
of the band gap. In the high-fraction brookite films and anatase films, it appears as though
the band gap is getting larger as the film thickness decreases. Figure 13 contains the 23
measurements taken on film 056. If we look at the measurements performed on this film,
which is primarily brookite, there is a linear relationship between the band gap and film
thickness. I will argue that this feature is a artifact caused by how our films are measured.
The feature that most significantly affects the measurement in the UV region is the band
gap–where the transmission undergoes strong absorption. The visible region of the spectra is
mostly affected by the thickness d–which contributes to the number and frequency of fringes.
For any given film, SCOUT will try to fit the fringe region as well as the band gap region.
SCOUT will often have to sacrifice precision in the UV region (where the gap is) for precision
in the visible region (which is dominated by fringes). This is standard because our models

14
Figure 13: Measurements taken on film 056-1. There is an inverse relationship between the
gap energy and film thickness–an artifact of the analysis.

do not match the behavior of TiO2 perfectly.

When the film becomes very thin, the number of fringes in the region of measurement
decreases to nearly zero. When this is the case, SCOUT will sacrifice additional precision
in the thickness of the film, in order to make the UV region (and therefore band gap) fit
better. Changing the thickness of a very thin film does not significantly affect the fit in the
visible range but changing the band gap significantly affects the fit in the ultraviolet range.
Consequently, for very thin films, the band gap is measured with increasing flexibility–which
causes an artificial relationship between the band gap and thickness to emerge. In figure 12,
there is no relationship between band gap and thickness in the rutile or anatase phases, and
the relationship across multiple brookite films is less certain.

5 Conclusion
Our lab produced high phase fraction rutile, brookite and anatase from amorphous precur-
sors. The transmission and reflection spectra were measured using a grating spectrometer.
The data were then characterized in SCOUT, which used models for the index of refraction
and band gap to produce theoretical data then perform a minimization routine to change the
model parameters to fit the theoretical to the experimental spectra. The spectra of our high
phase fraction films were measured and the optical parameters were calculated in SCOUT,
yielding the values in table 2. A trend emerged in our analysis of the high phase fraction
brookite film, but it was argued that the relationship is an artifact of the analysis. In future
work, we will determine a more accurate model for TiO2 , and compare the calculations of
the band gap for various optical models that were not used in our analysis.

15
6 Appendix
6.1 Grating spectrometer settings
Table 3 lists the grating spectrometer settings available and the wavelengths they can be
safely be used at to obtain correct data. In a typical measurement of a TiO2 film, the grating
spectrometer is set to measure from 250 nm to 1000 nm in 1 nm increments with the 0.25
micron grating, Xe lamp, silicon detector, and no filter. This allows the user to take data of
a large enough region for fitting in SCOUT, while preventing the need to switching filters or
gratings. The 0.25 grating is strongly reflective in the UV region, making the band

Setting Value Range λ (nm)

Lamp Xenon (Xe) 175→1750
Tungsten (W) 250→1998
Color Filter None 200→400
345 nm 400→650
651 nm 650→1220
1200 nm 1220→2000
Grating (suggested) 0.25 micron 200→1000
0.50 micron 200→1500
1.0 micron 500→2000
2.0 micron 1000→2000
Detector Silicon (Si) 200→1120
Indium-Gallium-Arsenic (InGaAs) 200→2000

Table 3: Grating Spectrometer Settings

6.2 Filtering 2nd Order Light

When measuring a sample, the correct color filter must be used to ensure the second-order
light is blocked. This only applies for regions of the film 2nd order light is not absorbed by the
band gap. For example, TiO2’s largest band gap is about 3.2 eV, so the user needs to start
using the color filter at 860 nm. At that point, not all the 380 nm light will be absorbed by
the band gap, and the intensity from that light will be measured. For measuring a TiO2 film
up to 1000 nm, the second-order light is dim enough that there is not a big impact. But if we
look at a SnCaSe sample which has a lower band gap, we can see clearly how the intensity
from second-order light affects the near infrared measurement.
Figure 10 shows the transmission and reflection spectra of a SnCaSe film which has been
measured with and without a color filter. The spectra labeled “filter” were taken correctly
by inserting a color filter to stop second order light. The spectra R, T, and T/(1-R) has
been taken without a filter. The band gap of the SnCaSe film stops absorbing at about 800
nm. We can see that the second order light (which has different interference fringes) skews
the spectra above 1600 nm (see the red dashed line). The effect is strong above 1700 nm,
when more second-order light is showing through.

16
 Figure 14: SnCaSe measurement. T , R, and T /(1 − R) with and without a color filter.

6.3 Band gap of the substrate

Our films were grown on was either fused silica, soda-lime silicate (SLS), Eagle Corning XG
(EXG) glass, or a silicon wafer. Glass has a band gap of about 9 eV, well above TiO2 ,
but some of our substrates have a band gap similar to TiO2’s band gap of about 3.5-3.8 eV
depending on the polymorph. EXG has a band gap of about 5.8 eV (EXG optical properties)
and SLS has a band gap of about 5 eV. If the band gap of the substrate is too close or smaller
than the film, the calculated absorption might end up being that of the substrate. The correct
substrate needs to be chosen when performing the analysis in SCOUT. SCOUT contains the
optical properties of many common materials, and new material properties can be added to
the SCOUT database.

7 Using SCOUT
7.1 User configurations
All the films were measured using an optical fitting software SCOUT produced and dis-
tributed by Wolfgang Thiess. In order to measure the optical properties of a film in SCOUT,
SCOUT needs to be configured to fit the transmission and reflection spectra. A configura-
tion can be created and saved for a later time. Configurations can be loaded by clicking
F ile → Open. A window opens up for the user to navigate to a configuration to load. Con-
figurations can be saved by clicking F ile → SaveAs then selecting the folder and naming
the new configuration.

17
7.2 The Layer Stack
To customize the configuration for measuring the properties of thin films, a new layer stack
should be created with at least two layers: a thin film and thick layer, where the thin film
is arranged above the thick layer. A half-space of air or vacuum is added on either side to
simulate the atmosphere. Something to notice is that the thin film is labeled as “coherent”
and the thick layer is “incoherent”, which is the only real difference between these two
objects. A simple layer gives the user the option to have coherence or not. The material of
the substrate can be chosen by drag-and-dropping a material from the materials database
listed on the right of the layer stack. If the film has a known dielectric function, it can be
chosen from the materials database, otherwise they will create a new material with some
susceptibilities.

7.3 Materials
A new material can be created and added to the “materials” menu. A new material will have
properties that are called “susceptibilities” in SCOUT. The built-in susceptibilities include
features to model simple index of refraction, oscillators, interband transitions, and more. For
a semiconductor, we can add two susceptibilities to model it accurately: a Sellmeier index
of refraction, and a band gap. The real part of the index of refraction is added by creating
a new “index of refraction” susceptibility, then entering the Sellmeier equation into the real
part of the susceptibility. The band gap model we generally choose is the OJL2 band gap
model. This was proposed in a paper by O’Leary, Johnson and Kim. It models a direct
band gap with some effects from tail states. This does a good job of simulating the band gap
of a semiconductor, while modeling flaws in materials by an exponentially decaying density
of states. The susceptibilities contain parameters that can be adjusted to change features
in the material. In the OJL2 model, the Gap Energy parameter changes the energy of the
band gap.

8 Acknowledgments
I would like to thank my research advisor Dr. Janet Tate who taught me about the physics
of thin films and gave me tutoring in technical writing and professional development. I would
like to thank James Haggerty, James May, Okan Agirseven, and Bethany Matthews for giving
me insight into materials physics, trusting me with their films, and supporting my endeavors.
I would like to thank Dr. McIntyre for letting me use his lab for film characterization. I
acknowledge support from the Dr. Russ and Dolores Gorman Faculty Scholar fund.

18
References
[1] Haggerty, J. E., Schelhas, L. T., Kitchaev, D. A., Mangum, J. S., Garten, L. M., Sun, W.,
Stone, K.H, Perkins, J.D., Toney, M.F., Ceder, G., Ginley, D.S., Gorman, B.P., Tate, J.
(2017). High-fraction brookite films from amorphous precursors. Scientific Reports, 7(1).

[2] Landmann, M. Rauls, E. Schmidt, W. (2012). The electronic structure and optical re-
sponse of rutile, anatase, and brookite TiO2 . J. Phys.: Condens. Matter. 24, 3.

[3] Reyes-Coronado, D., Rodrı́guez-Gattorno, G., Espinosa-Pesqueira, M. E., Cab, C., Coss,
R. D., Oskam, G. (2008). Phase-pure TiO2 nanoparticles: anatase, brookite and rutile.
Nanotechnology,19(14), 145605.

[4] Alotaibi, A. M., Sathasivam, S., Williamson, B. A., Kafizas, A., Sotelo-Vazquez, C.,
Taylor, A., . . . Parkin, I. P. (2018). Chemical Vapor Deposition of Photocatalytically
Active Pure Brookite TiO2 Thin films. Chemistry of Materials.

[5] Chimupala, Y., Hyett, G., Simpson, R., Mitchell, R., Douthwaite, R., Milne, S. J.,
Brydson, R. D. (2014). Synthesis and characterization of mixed phase anatase TiO2 and
sodium-doped TiO2 (B) thin films by low pressure chemical vapour deposition (LPCVD).
RSC Adv.,4(89), 48507-48515.

[6] Lin, H., Li, L., Zhao, M., Huang, X., Chen, X., Li, G., Yu, R. (2012). Synthesis of
High-Quality Brookite TiO2 Single-Crystalline Nanosheets with Specific Facets Exposed:
Tuning Catalysts from Inert to Highly Reactive. Journal of the American Chemical So-
ciety,134(20), 8328-8331.

[7] Kim, D., Kim, W., Kim, S., Hong, S. (2014). Brookite TiO2 Thin Film Epitaxially Grown
on (110) YSZ Substrate by Atomic Layer Deposition. ACS Applied Materials & Inter-
faces,6(15), 11817-11822.

[8] Scanlon, D. O., Dunnill, C. W., Buckeridge, J., Shevlin, S. A., Logsdail, A. J., Wood-
ley, S. M.,. . . Sokol, A. A. (2013). Band alignment of rutile and anatase TiO2 . Nature
Materials,12(9), 798-801.

[9] Swanepoel, R. (1983). Determination of the thickness and optical constants of amorphous
silicon. Journal of Physics E: Scientific Instruments, 16, 1214-1222.

[10] Tauc, J., Grigorovici, R., Vancu, A. (1966). Optical properties and electronic structure
of amorphous germanium. Phys. Stat. Sol. 15, 627.

[11] Hishikawa, Y., Nakamura, N., Nankano, S., Kishi, Y., Kuwano, Y. (1991). Interference-
free determination of the optical absorption coefficient and the optical gap of amorphous
silicon thin films. Japanese Journal of Applied Physics, 30(5), 1008-1014.

[12] S.K.O’Leary, S.R.Johnson, P.K.Lim (1997). The relationship between the distribution
of electronic states and the optical absorption spectrum of an amorphous semiconductor:
An empirical analysis. J. Appl. Phys., 82(7), 3334.

19
[13] D. H. McIntyre, “Thin Film Interference” (2010).
http://physics.oregonstate.edu/∼tatej/TateLabWiki/doku.php?id= mustread:start

[14] Ventura, S. D., Birgin, E. G., Martı́nez, J. M., Chambouleyron, I. (2005). Optimization
techniques for the estimation of the thickness and the optical parameters of thin films
using reflectance data. Journal of Applied Physics,97(4), 043512.

[15] Chiang, T. (2008) Quantum Physics of Thin Metal Films. AAPPS Bulletin. 15, 2.

20