Chapter 2

Characterization of Topas 5013 ®

This chapter will characterize a spin coated Topas® film. It tells how to dissolve and spin coat
a film of Topas® . The film will be characterized by measuring the thickness and roughness as
these properties are important for the planar technology platform.

2.1 Introduction
Today polymers or as they are more commonly known - plastics, are used in every aspect of life.
From cars to medical packaging, polymers play an important role. Polymers like carbon fiber
reinforced epoxy are replacing metal and wood in areas like construction and cars, where the
advantages are low cost and low weight. For example in pipes used for water, plastic has replaced
metal, because plastic does not corrode and thus lasts longer.

The development of celluloid began in 1856 by Alexander Parkes. He tried to commercialize

it, but failed. One of his workers; Daniel Spill, later tried to commercialize a similar product,
Xylonite, but also failed. Not until John W. Hyatt invented cellulose in 1869 [20], did the mass
production of polymers begin. This was one of the first thermoplastics to be produced. It re-
placed ivory in billiard balls and made photographic films possible. Cellulose had at that time
exceptional properties like changing the shape by heating at low temperatures, transparency and
a relative toughness, but unfortunately cellulose was also highly flammable.
The next major step was in 1909 when Leo Baekeland presented Bakelit [21]. This polymer is a
thermosetting polymer, which cure to a stronger form (often) by heating.
In the years between the World Wars the need for new and improved materials sparked a rash re-
search into polymers. This created polyvinyl chloride and polyethylene. They were used for cloth-
ing, construction and even in the music industry (gramophone records). Besides these polymers
many others have been developed and commercialized. These includes carbon fiber reinforced
materials, which is composed of carbon fiber embedded in polymers, while others are tailored
to resist certain groups of chemicals. Again others have been customized to withstand heat or
deflect pressure.

One of the newest addition to the class of thermoplastic polymers is Topas® . Topas was
invented in 1990 by Ticona [17] and is fabricated in a process between ethylene and norbornene.

5
6 Chapter 2: Characterization of Topas® 5013

Figure 2.1: The glass transition temperature plotted versus norbornene

content in Topas® .

Topas® is a thermoplastic, which is very flexible regarding the glass transition temperature.
By varying the amount of norbornene it is possible to vary the glass transition temperature tg
between 70 − 170 ◦C, see figure 2.1. Together with a low absorbtion in the visible and infrared
spectrum [19] and a good resistance towards most esters, ketones, alcohols and ethers [18] this
makes Topas® suitable for many applications.
The good optical properties makes Topas® a good choice for optical lenses, and its chemical
resistance [18] together with a low moisture uptake makes it suitable for both packaging and for
medical applications. The sterilization possibilities are also an advantage for medical applications.
These applications use injection molding to give Topas® the final shape.
The wide range of glass transition temperatures also makes Topas® a suitable choice for use
in thermal nanoimprint lithography [13]. Together with good chemical resistance to most acids,
bases and polar organic solvents it is good for cleanroom processing.

Compared to other optical polymers like polymethyl-methacrylate (PMMA), poly carbonat

(PC) and polystyrene (PS) Topas® exhibit a very high Abbe number, low birefringence and a
high transmission. Even if both PMMA and Topas® have a high Abbe number, PMMA does
not exhibit the same good chemical resistance as Topas® . PC has a high elongation at break,
but at the same time suffers from a higher moisture uptake and a low Abbe number. PS has
similar physical properties as Topas® , but exhibits a lower optical transmission, which makes
Topas® the better choice for optical components. Like all the polymers mentioned above Topas®
is relatively cheap.
Previously PMMA has been the preferred polymer for use in nanoimprint lithography due to its
2.1 Introduction 7

availability, ease of use and low cost, but due to the properties mentioned before, Topas® has
gained a greater use as the molding material for thermal nanoimprint lithography at MIC.

For the optical components fabricated in this project, the former polymer has been BCB.
Benzocyclobutene (BCB) is a thermosetting plastic, which exhibits a very planar surface and
a high uniformity across the surface after the spin coating process. BCB is a curable polymer,
which means that by heating BCB will have a stronger form. The curing also makes BCB more
resistant towards most chemicals and gives a high temperature resistance, but unfortunately this
is not usable in thermal nanoimprint lithography.

Topas® consists of several grades, each specialized for specific applications. Each grade is
named by the following convention: Topas® XXYY. Here XX is the molecular weight given as
the solution viscosity1 . YY is the Heat Deflection Temperature measured after ISO standard
ISO75-2:2004 2 .
The first grade is Topas® 8007, which has a higher elongation at break and a lower elastic modu-
lus compared to the other grades, is used as the standard grade for general purpose applications
like packaging. Topas® 6013 has a higher glass transition temperature making it suitable for
containers in the lab with a good chemical resistance and a high transparency. Topas® 6015
has the highest heat deflection temperature with the same properties as the other two grades.
Other than these grades, two special grades exist suited for different applications. The first one
is Topas® 5013, which is specially tailored for optical applications. The other, Topas® 6017, is a
grade with a very high heat deflection temperature suited for short-term high temperature expo-
sure. The investigations will be focused on Topas® 5013, as it is tailored for optical applications.

It should be noted that it was also possible to obtain wafers made of Topas® . These wafers
were produced during a collaboration between POEM [23], Danish Technological Institute [24]
and Danfoss A/S [25] by using injection molding to shape Topas® into discs with a smaller di-
ameter (∼80 mm) than standard silicon wafers and had a thickness around 1 mm. Unfortunately
they had a very high surface roughness making the Topas® wafers an unusable for the LRSPP
waveguides and the compact terahertz spectrometer..

Investigating the thickness and roughness dependence on the spin speed and solution proper-
ties is crucial in determining the proper fabrication steps. The first step is determining the best
solution to give the required thickness for the optical components, while at the same time having
a relatively low roughness. The next step will be on implementing the spin coating process into
existing standard cleanroom procedures at Danchip [22].
This chapter is divided into 5 sections including this one. The following section will give a brief
introduction to the basic of polymers, followed by how to dissolve and spin coat Topas® , ending
with a characterization of the spin coated Topas® .

1
Solution viscosity: The viscosity of a dilute solution of a polymer, measured under prescribed condi-
tions, is an indication of the molecular weight of the polymer and can be used to calculate the degree of
polymerization.
2
http://www.iso.org
8 Chapter 2: Characterization of Topas® 5013

cyclo norbornene M
pentadiene
x y
ethylene dimer
Figure 2.2: A simplified process for creating Topas® from ethylene and
norbornene using a metallocene catalysation polymerization process.

2.2 Plastic
The word ’polymer’ is a composite word of ’poly’ and ’meros’, where polymer translates into
’consisting of many parts’. A polymer or more commonly known as plastic, which means ’fit for
molding’, is a chain of small identical molecules, called monomers, linked together in a periodic
array.
To create the long molecules which make up the polymer, two polymerization methods are com-
monly employed, chain growth and step growth polymerization. Both methods have the same
overall conversion rate, but the former creates high molecular weight polymers faster than step
growth polymerization.
The first step in the polymerization process is initiation. The initiation makes the monomers
ready to form long molecules. This is usually a process done under a controlled temperature and
pressure.
Chain growth polymerization breaks the bond of two monomers (double or triple bonds) to form
two monomers with dangling bonds. These dangling bonds can bond with other dangling bonds
to create a molecule of two monomers. By continuing this process very long molecules can be
created.
Step growth polymerization occurs when the monomer’s functional group reacts with other
monomers functional groups. By releasing a small molecule the resulting molecule from the
two monomers can be bonded, creating a larger molecule.
Termination of the formation of the long molecules occurs when the long molecules either bond
to themselves or there are no more free dangling bonds available.
Both of these processes have been used for a long time and they have their strengths and weak-
nesses. One of the weaknesses is random termination and little control over the branching of the
molecules. To counter this, new polymerization methods are being developed. One of these is
metallocene catalysation [26–28]. Metallocene catalysation is a very complex process, which in
simple terms consist of two ring structures surrounding a metal atom. The metal atom acts as
a catalyst creating sites for the monomers to "attach". With this process it is possible to have
tight control of the properties of the polymer like molecular weight, tactivity etc.

Topas® is a copolymer created by ethylene and norbornene [29–31] using a metallocene

catalysation polymerization process as shown simplified in figure 2.2. When the chain molecule
2.3 Preparing Topas® 5013 solution 9

consists of two different molecules as is the case in a copolymer like Topas® , the chain molecule is
called a dimer. The process starts with creating norbornene from cyclo pentadiene. Norbornene
and ethylene is then polymerized in a metallocene catalysation process, creating long chains
of dimers as shown. The amount of norbornene denoted ’y’ in figure 2.2 is the weight percent
as shown in figure 2.1. The presence of norbornene also raises the glass transition temperature
and makes the polymer more amorphous. The glass transition temperature increases with the
norbornene content in Topas® as shown in figure 2.1. If the norbornene content is at least 15%,
Topas® is considered an amorphous polymer [26].

2.3 Preparing Topas 5013 solution ®

The first step is determining how to dissolve Topas® 5013 into a usable solution - liquid enough
to be used for spin coating, but still viscous enough to give the right thickness.
It has previously been found that Topas® is resistant towards most solvents [18] so new solvents
should be used. Because Topas® is a cycloolefin copolymer [30], chemicals with similar chemical
structures should be capable of dissolving Topas® 5013. Among these are toluene (methylben-
zene, CAS nr.: 108-88-3)) and sec-butylbenzen (CAS nr.: 135-98-8), which both are aromatic
hydrocarbons.
In this project the first choice was toluene due to its availability, while a switch to sec-butylbenzene
was made late in the project due to better results obtained by Micro Resist Technology GmbH
[32]. The technique for dissolving Topas® 5013 using toluene can be applied to using sec-
butylbenzene as solvent.
It should be noted that the lid used to close the bottle needs to be coated with an inside teflon
layer to prevent the solvent from dissolving the lid.

It is very simple to make a Topas® 5013 solution using toluene. The process consists of five
simple steps.

1. Clean the bottle with ethanol and let it dry for an hour.

2. Calculate the required amount of Topas® 5013 and toluene.

3. Pour the required amount of Topas® 5013 into the bottle.

4. Pour the required volume of toluene into the bottle.

5. Put the bottle into an ultrasonic bath (US) for some days, depending on the weight percent.

The weight percent was varied from 15 - 40. A solution with a weight percent of 50 was also
made, but it was not possible to use the solution due to a high viscosity after a very long time
in the ultrasonic bath.
If a thick solution (wt > 20%) has not been used for a couple of days it is necessary to put the
solution into an ultrasonic bath for a day before use. By doing this, the solution will be fluidic
enough to be used for spin coating.
10 Chapter 2: Characterization of Topas® 5013

Clean the bottle

To avoid contamination of the solution from dust or particles it is necessary to clean the bottle
before use. This is done with ethanol as it is capable of removing fat, particles and dust, while
at the same time evaporates very fast - leaving no trace. Using water is not an option as drying
the bottle will take too long with a high risk of contaminating.
Pour a small amount of ethanol into the bottle and let it roll around, so the ethanol will touch the
whole bottle. Afterward pour the ethanol into the specified wastecontainer and let the remaining
ethanol in the bottle evaporate in a fumehood for at least an hour.

Calculate the required amount of Topas® 5013 and toluene

To distinguish different Topas® 5013 solutions it is necessary with a labeling system, which
has been chosen to be the weight percent of Topas® 5013 dissolved in the solution. The weight
percent, wt%, is calculated as shown in equation 2.1, where mT opas is the mass of Topas® , ρtoluene
is the density of toluene and Vtoluene is the volume of toluene:
mT opas 5013
wt% = (2.1)
ρtoluene Vtoluene + mT opas 5013

This can be rewritten to find the necessary amount of Topas® 5013 given a volume of toluene
and the weight percent:

wt% ρtoluene Vtoluene

mT opas 5013 = (2.2)
1 − wt%

Pour Topas® 5013 into the bottle

Topas® 5013 is delivered from Ticona GmbH as small pellets in sizes up to 5 mm. The calculated
amount of Topas® 5013 is measured and poured into the bottle.

Pour toluene into the bottle

The given volume of toluene is poured into the bottle.

Put the solution in an ultrasonic bath

Put the solution into an ultrasonic bath and let it stand for some time. The required amount of
time is determined by the volume of the solution and weight percent e.g. a small solution with
a weight percent of 20 requires less time than a bigger solution with a weight percent of 30.
To determine if the solution is ready for use - the bottle is tipped to see how much of the solution
is viscous. If the solution is fluidic enough it is deemed ready for use.
2.4 Spin coating Topas® 5013 11

(a) 5 scratches with a spacing of ap- (b) Measurements of the roughness. 24

proximately 1 cm used for measure-
ments of the thickness. 20 measurements
(red, dashed lines) were performed from
scratch to scratch. The arrow indicates
the scanning direction.
measurement (red lines) spaced with a
distance of approximately 1 cm, each 500
µm long.

Figure 2.3: Sketch of the measurements of height and roughness on the wafer.

2.4 Spin coating Topas 5013 ®

To investigate the thickness and roughness of spin coated Topas® 5013 a standard procedure was
used for measuring both quantities. To measure the height of a spin coated wafer, 5 scratches
(black lines) with a spacing of approximately 1 cm were made as shown in figure 2.3(a). In
making these scratches it was assured that the scratches went all through the polymer down to
the wafer. Afterwards the thickness was measured with 20 measurements from scratch to scratch
(red, dashed lines) with a vertical spacing of 1 cm using a Dektak 8 Advanced Development Pro-
filer from Vecco Instruments Ltd. [33]. Parameters for the height measurement program using
the Dektak 8 Advanced Development Profiler can be found in Appendix A.

As initial investigations showed that spin coated Topas® 5013 is very rough, it was necessary
to make the surface flat. Two methods were proposed to flatten the spin coated Topas® 5013
surface. The first one is called ’Flat stamp’, while the other is called ’LOR-de-bonding’. The
former method uses a smooth stamp with a roughness below 10 nm to press the irregularities on
the Topas® surface away. The latter method involves using LOR [34] as sacrificial layer, bonding
and wet etching. Both methods includes spin coating Topas® 5013 onto a wafer. The spin coating
machine used is a Specialty Coating System P6204 [35] from Cookson Electronics.

The roughness measurements are carried out using a Dektak 8 Advanced Development Pro-
filer by making 24 measurements each 500 µm long spaced with approximately 1 cm as shown
in figure 2.3(b). The scan length of 500 µm have been chosen to calculate the average roughness
over a distance comparable to at least 10 times the feature size of the optical components. To
have a measure for the roughness, the parameter Ra known as the Arithmetic Average Devia-
12 Chapter 2: Characterization of Topas® 5013

tion from the mean line have been chosen, which is explained in Appendix A. Parameters for
the roughness measurement program using the Dektak 8 Advanced Development Profiler can be
found in Appendix A.

It should be noted that the word substrate wafer can either be a silicone wafer or a silicone-
dioxide wafer (Pyrex). Each type of substrate wafer needs different ways of cleaning as described
below.

Silicon wafer

1. The wafer is dipped in buffered HF solution3 (BHF) to remove the natural oxide and give
a better adhesion, followed by a rinse in de-ionized water and a spin drying. (BHF: 35 sec.,
H2 0: 5 min.)

Pyrex

1. Clean and polish with soap Triton-X. 3 x 20 seconds on each side.

2. To remove organic materials and improve adhesion pyrex wafers are put into a piranha
bath. This is followed by a rinse in de-ionized water and a spin drying. (time: 10 min.,
temp.: 80 ◦C, H2 0: 5 min.)

2.4.1 Thickness of raw Topas® 5013

First it was necessary to know which weight percent gave the required thickness. The weight
percent of the solutions was varied from 15 wt% to 40 wt%. A standard silicon wafer was cleaned
with BHF and prebaked at 200 ◦C for 10 minutes to evaporate moisture on the wafer. The wafer
was spin coated with the given Topas® 5013 solution using a Specialty Coating System P6204
(spin time: 60 sec., acceleration: moderate, lid: on) with varying spin speeds from 1000 to 3000
revolutions per minute (rpm) in steps of 500 rpm. Afterwards the wafer was postbaked at 200 ◦C
for 10 minutes to evaporate the solvent. A Dektak 8 Advanced Development Profiler was used to
measure the thickness as described before. The thickness measurements are shown in figure 2.4.
The results have been fitted to equation 2.3 [36], where T is the thickness, ω is the spin speed,
K is a calibration constant, C is the polymer concentration in grams per 100 ml. and η is the in-
trinsic viscosity. α, β and γ are experimental constants. As most of these variables are unknown,
the equation has been reduced to a more simple expression used for fitting the results, where G
and α are fitting constants.

KC β η γ G
T = α
= α (2.3)
ω ω
The correlation coefficient of the fit to Topas® 5013 15 wt% is 0.88, while Topas® 5013 30
wt% is only 0.60. The best correlation is achieved by Topas® 5013 40 wt% with a value of 0.99.

To obtain a thick polymer layer a low spin speed should be used. However a low spin speed
also causes a decrease in the homogeneity of the spin coated layer. The best choice for the spin
3
The buffered HF-solution is composed of 12.5% (50% HF) and 87.5% (40% ammonium fluoride)
yielding a solution with 6.25% HF, 35% NH4 F and 58.75% H2 O
2.4 Spin coating Topas® 5013 13

Figure 2.4: Spin curves of the different Topas® 5013 solutions. Weight
percent varies from 15 to 40. Spin parameters: spin time = 60 sec., ac-
celeration / deceleration = moderate, lid = on.

speed is around 2000 - 3000 rpm, which is low enough to make a thick layer, while at the same
time ensures a usable homogeneity. From figure 2.4 it is seen that Topas® 5013 solution with a
weight percent of 40 gives a thickness around 25 µm at 2000 rpm, while at 3000 rpm the thickness
has decreased to around 20 µm. Decreasing the weight percent to 30, reduces the thickness by
more than 50 %. A weight percent of 15 gives a thin layer of Topas® 5013. Film thickness from
a few to several micrometers can be achieved by using the appropriate weight percent with the
correct spin speed as shown.

It should be noted that spin coating a solution with a weight percent of 40 gives a different
roughness profile than using a solution with a lower weight percent. The surface of a wafer spin
coated with a solution of weight percent 15 gives a bumpy surface, while using a solution with a
weight percent of 40 gives a flat surface only distorted by the occasionally small particles visible
by the naked eye.
Of course this also affects the roughness. As there is no significant difference in the surface of
wafers spun with solutions of a weight percent lower than 40 only one measurement series has
been done for Topas® 5013 15wt% and one for Topas® 40 wt%. Results are shown in table 2.1.

It can be seen from table 2.1 that a solution of Topas® 5013 40 wt% has a roughness much
smaller than a solution of Topas® 5013 15 wt%. But the lower roughness for Topas® 5013 40
wt% is not good enough, so the spin coated layer still needs to be flattened.
14 Chapter 2: Characterization of Topas® 5013

stamp

Topas Topas Topas

substrate substrate substrate

Figure 2.5: The concept in the Flat stamp method. Topas® is spin coated
onto the substrate wafer. After the spin coating process Topas® is im-
printed with a flat coated stamp. After imprint the result is a smooth
Topas® layer.

From the results shown above it can be concluded that a solution of Topas® 5013 40 wt%
gives the thickest layer with the lowest natural roughness at an average spin speed (2000-3000
rpm). Even with a low natural roughness it is still necessary to flatten the wafer to ensure the
lowest possible roughness. The next step is determining how to flatten the spin coated Topas®
5013 layer.

2.4.2 Flat stamp

’Flat stamp’ is a simple method and uses a silicon wafer coated with an appropriate coating, a
flat stamp, to ensure an easy release between Topas® and the stamp. The concept is shown in
figure 2.5, where a substrate wafer is spin coated with Topas® 5013 and followed with an imprint
with the flat stamp in Thyra at 150 ◦C for 10 minutes. After imprint the Topas® 5013 layer is
smooth with a low roughness. Unfortunately this process is serial and the required process time
scales linearly with the number of substrates.
A vital part of this method is the coating. The coating must adhere well to the silicon substrate,
but at the same time have no adhesion to Topas® 5013. Fulfilling this criteria it should be possible
to use the same stamp to flatten several wafers. The most successful of these coatings was F17 -
triclorosilane, which was deposited on a few wafers at the courtesy of Applied Microstructures
[37]. The coating was deposited using molecular vapor deposition and it was proven that more
than 30 substrates were flattened using F17 -triclorosilane as the coating.
Unfortunately the cost to deposit F17 -triclorosilane proved to be too expensive, so another coating
was tested as well. A coating using fluorocarbon [38] was also tried yielding adequate results at
a low cost. This coating was chosen as the standard anti-sticking layer, which was deposited
using an Advanced Silicon Etcher (ASE) [39] and C4 F8 as feed gas. The recipe was coil power:
300 W, platen power: 20 W, flow rate: 120 sccm and the deposition time was 120. This gives an
estimated thickness of 200 nm for the deposited layer [40].

The criteria for a successful flattened substrate is no sticking between Topas® and the anti-

Topas® 5013 15 wt% Topas® 5013 40 wt%

Roughness 883.7 nm (429.4) 19.6 nm (18.3)

Table 2.1: Roughness of Topas® 5013 for solutions with weight percent
of 15 and 40.
2.4 Spin coating Topas® 5013 15

Flat stamp method

# Step Parameter Comment
1 Clean the substrate wafer. Use either a silicon
wafer or a pyrex wafer
and clean as described
in section 2.4
2 Prebake Temperature: 200 ◦C
Time: 10 min.
3 Spin coat Topas® 5013 onto Spin speed: 3000 rpm Specialty Coating Sys-
the substrate wafer Acceleration: moderate tem P6204
Spin time: 60 sec.
Volumen: ∼4 ml.
Lid: on
4 Postbake Temperature: 200 ◦C
Time: 10 min.
5 Flatten Temperature: 150 ◦C Imprint in Thyra us-
Time: 10 min. ing a flat stamp

Table 2.2: Flat stamp: Use of Flat stamp method to flatten a Topas® 5013 surface.

sticking coating, and that there are no visible cracks or marks present on the surface after imprint.
The machine used for flattening the substrates was first Thyra, a "home-made" thermal bonder,
while in the last part of the project a switch was made to EVG520HE [41], a hot embossing
system. Unfortunately it was shown that Thyra was not properly calibrated and as such the
parameters from this machine are not reproducible at the present time due to a recalibration.
The sandwich structure is shown in figure 2.6 and parameters for the different Topas® grades
are shown in table 2.3.

A sandwich structure as shown in figure 2.6is used in both Thyra and EVG520HE and consist
of the substrate and the stamp. The stamp is coated with anti-sticking layer, while the substrate
is spin coated with Topas® . On top of the stamp is placed a 1 mm vitton mattress to ensure an
homogenous imprint. To protect the hotplates from an eventual polymer flow, sheets of tin foil
have been inserted between the sandwich structure and the heat plates.
Experimentally an estimate for the processing time was found to be around 100 minutes for a
single substrate.

Roughness of Topas® 5013

An investigation of using Flat stamp with different temperatures and times was carried out.
The flat stamp was a wafer coated with F17 -triclorosilane by molecular vapor deposition. The
machine used for the investigation was Thyra. The temperatures on Thyra were 130 ◦C, 140 ◦C
and 150 ◦C and time was varied between 5, 10 and 20 minutes. Both the height and the roughness
16 Chapter 2: Characterization of Topas® 5013

hotplates

foil
Figure 2.6: The "sandwich" structure used for flattening a Topas®
layer using either Thyra or EVG520HE. The structure is inserted vitton
between two hotplates. The next layer is tin foil to protect the heat
flat stamp
plates from an eventual flow of the viscous polymer. A vitton mattress
is used to ensure a homogenous imprint. The stamp and the substrate substrate
is put into the middle of the "sandwich".
foil
hotplates

were measured using Dektak 8 Advanced Development Profiler and the programs in Appendix A.
The results of the thickness measurements done on flattened Topas® 5013 40 wt% spin coated at
2000 rpm are shown in figure 2.7(a). The measured thicknesses all lie within the natural deviation
(±5µm) of a standard silicon wafer. From this it is seen that there is no significant change in
thickness using a higher temperature nor a longer time. Comparing figure 2.4 and figure 2.7(a)
it is observed that flattening the stamp reduces the thickness with approximately 2 µm.
At the same time the roughness was also measured. From figure 2.7(b) it is seen that neither
temperature nor time has a significant impact on the roughness as all measurements lies between
1 and 3 nm.

As such standard parameters have been chosen to be 150 ◦C and 10 minutes. A temperature
of 150 ◦C gives a slightly improved roughness compared to the other temperatures at 10 minutes.

Roughness investigation using a NanoMan Atomic Force Microscope from Vecco Instruments
Ltd. [33] was carried out. The roughness was measured over an area of 10x10 µm2 and the results
are shown in table 2.4

Because the measured area is small, the corresponding values for the roughness will also be
smaller than usual. It is seen that at microscale the flattened wafer still has a lower roughness
than a non-flattened wafer. The difference is greater for a solution with 20 wt% than for a solution
with 40 wt%, where the latter gives a more smooth spin coated surface. This is expected since the
amount of toluene is greater in a 20 wt% solution. The higher amount of toluene will evaporate
faster leaving a more rough surface during the postbake. BCB was also measured and it is seen
that at microscale a flattened Topas® surface has a lower roughness.

Flattened
# Time Temperature Pressure Machine
Topas® 8007 10 min. 150 ◦C 2 kN Thyra
Topas® 8007 10 min. 180 ◦C 10 kN EVG520HE
Topas® 9506 10 min. 130 ◦C 2 kN Thyra
Topas® 5013 10 min. 170 ◦C 2 kN Thyra
Topas® 6013 10 min. 175 ◦C 2 kN Thyra

Table 2.3: Imprint parameters using EVG520HE or Thyra. The parameters

are valid for thick films with t > 10µm.
2.4 Spin coating Topas® 5013 17

(a) Thickness of flattened Topas® 5013 at temper- (b) Roughness of flattened Topas® 5013 at temper-
atures from 130 ◦C to 150 ◦C and times from 5 to atures from 130 ◦C to 150 ◦C and times from 5 to
20 minutes. 20 minutes.

Figure 2.7: Measurements of flattened Topas® 5013.

AFM measurements
Polymer Roughness
Topas® 5013 20 wt% 0.416 nm
Topas® 5013 20 wt% (flattened) 0.391 nm
Topas® 5013 40 wt% 0.425 nm
Topas® 5013 40 wt% (flattened) 0.274 nm
BCB 0.398 nm

Table 2.4: AFM measurements of the roughness over

an area of 10x10 µm2 .

2.4.3 LOR-de-bonding
The second method, LOR-de-bonding, is more complicated. This involves the use of a chemical,
LOR [34]. LOR is a chemical based on polydimethylglutarimide (PMGI) used for bilayer lift-off
process. The roughness of spin coated LOR is below 10 nm, and by using this layer as a sacrificial
layer, it is possible to obtain a smooth layer of Topas® 5013. The concept is shown in figure
2.8 and described in details in table 2.5, where 1) LOR is spin coated onto the carrier wafer 2)
followed by a spin coating of Topas® 5013. 3) The substrate wafer (most often a pyrex wafer) is
bonded using Thyra under specific temperature and time. The bonded structure is submerged
into SR-1165 (Shipleys Remover 1165) [42] for some time, until the Topas® has separated from
the carrier wafer i.e. the SR-1165 has removed LOR.
The carrier and substrate wafers were bonded in Thyra, where the temperature was set to 150 ◦C
for 10 minutes, because this temperature is just above the tg of Topas® 5013, making the polymer
slightly fluid. Slightly fluidic makes Topas® 5013 "melt" onto the substrate.
18 Chapter 2: Characterization of Topas® 5013

LOR-de-bonding
# Step Parameter Comment
1 Passivate carrier wafer Chemical: buffered HF
Time: 35 sec.
2 Prebake carrier wafer Temperature: 200 ◦C
Time: 10 min.
3 Spin LOR on carrier wafer Spin speed: 7000 rpm Specialty Coating Sys-
Acceleration: high tem P6204
Spin time: 45 sec.
Volumen: ∼4 ml.
Lid: on
4 Interbake carrier wafer Temperature: 200 ◦C
Time: 10 min.
5 Spin Topas® 5013 on carrier Spin speed: 3000 rpm Specialty Coating Sys-
wafer Acceleration: moderate tem P6204
Spin time: 60 sec.
Volume: ∼4 ml.
Lid: on
6 Postbake carrier wafer Temperature: 200 ◦C
Time: 10 min.
7 Clean substrate wafer Use either a silicon
wafer or a pyrex wafer
and clean as described
in section 2.4
8 Bond the substrate and car- Temperature: 150 ◦C Thyra
rier wafer in Thyra Time: 10 min.
9 The bonded wafers are im- Temperature: 50 ◦C
mersed in an ultrasonic Time: several days
bath with SR-1165

Table 2.5: LOR-de-bonding: Using the LOR-de-bonding method to prepare a substrate.

Roughness of Topas® 5013

The success rate of this method was very poor, as Topas® did not stick to the substrate wafer. A
success was determined by visual inspection of the amount of Topas® residing on the substrate.
The substrates had only small patches of Topas® and as such it was not usable for further
processing. 10 point measurements of the roughness on the small patches of Topas® 5013 using
the same parameters as described in Appendix A for the roughness program showed that the
roughness was around 1-3 nm. This indicated the method might work, if it was possible to
2.4 Spin coating Topas® 5013 19

carrier
Topas LOR
LOR LOR Topas Topas
carrier carrier substrate substrate
1. 2. 3. 4.
Figure 2.8: The LOR-liftoff concept. By using the LOR layer as a sacri-
ficial layer it is possible to obtain a smooth surface of Topas® .

Figure 2.9: A sketch of how pieces of

Topas® were cut off before bonding with
the substrate.

properly separate the carrier and substrate wafers. As the method did not work, a series of
experiments were carried out to investigate the adequacy of SR-1165 to etch LOR.
• Can SR-1165 remove LOR:
Yes, within a few minutes and no heating required.
• Can SR-1165 separate a Topas® layer from a wafer using LOR as sacrifical layer:
Yes, but only after a few days (3 days) in a constant ultrasonic bath at 50 ◦C.
• Can SR-1165 separate bonded wafers, where the intermediate layer is LOR:
Partially, after several days. The progress of etching the LOR was tested each day by
inserting a scalpel into the structure to determine if the wafers had separated.
As the possible cause for the lack of success, it was suggested that Topas® during the spin
coating process might block for access of SR-1165 to LOR, thus making etching of LOR a very
slow process. It was investigated if cutting off pieces of Topas® 5013 before bonding as shown
in figure 2.9 had any effect, or breaking pieces of the substrate wafer and afterwards cutting
pieces of the spin coated Topas® 5013 off would help. The goal was to remove Topas® 5013 from
the edges to give access for SR-1165 to etch LOR. Unfortunately neither of these approaches
improved the success rate.

From these investigations it was estimated that a batch of bonded substrates would require
several days to separate using this method and thus it was determined that the method required
more time and resources than was available for this project, when at the same time the Flat stamp
method yielded satisfactory results. It was then decided to discontinue further investigations into
this method.
20 Chapter 2: Characterization of Topas® 5013

2.4.4 Comparison
From the estimated times for each process, it was seen that the Flat stamp method is faster than
the LOR-debonding method, as long as the number of processed substrates are less than 31. A
summary of the two processes is listed in table 2.6, where the (dis)advantages of using either
method are listed.
While both processes yielded the required roughness, only the Flat stamp method yielded usable
results, while the the LOR-de-bonding method was only able to yield small patches of Topas®
on the substrate.

Flat stamp LOR-de-bonding

Table 2.6: Advantages and disadvantages Fast Slow
of using either Flat stamp or LOR-de- Serial process Parallel process
bonding to prepare a substrate. Simple Complex
Usable results Unusable results

2.5 Switching from toluene to sec-butylbenzene

Results presented at the NNT 2004 conference [43] by a collaboration between Micro Resist Tech-
nology GmbH [32] and MIC under the European Union Research Sixth Framework Programme,
project NaPa [44], indicated that using sec-butylbenzene as solvent for Topas® yielded a more
uniform surface and lower roughness compared to using toluene. Based on these results a switch
was made to sec-butylbenzene.
At the same time a switch was also made from using Topas® 5013 to using Topas® 8007 as
explained in chapter 3.

The experience obtained using toluene as solvent also applies for using sec-butylbenzene,
although there are a few minor differences. The highest weight percent of a usable solution using
sec-butylbenzene was determined experimentally to be 30 compared to 40 for toluene. Using
sec-butylbenzene instead of toluene yields a very smooth surface with a roughness around 2-4
nm. With a natural roughness below 10 nm, it is not necessary to flatten the substrate, thus
simplifying the process. The lower roughness is attributed to the lower evaporation rate of sec-
butylbenzene compared to toluene. With a lower evaporation rate, Topas® remains dissolved for
a longer time during spin coating and will be fluidic for a longer time, enabling a more uniform
spreading.
A solution of Topas® 8007 using sec-butylbenzene as solvent with a weight percent of 30 was spin
coated with Specialty Coating System P6204 at 2000 rpm to give a thickness of approximately
20 µm required for the LRSPP waveguides. A lower weight percent of 20 gave a thickness of
around 5 µm at the same spin speed necessary for the compact terahertz spectrometer.
2.6 Conclusion 21

2.6 Conclusion
By spin coating Topas® 5013 dissolved in toluene a thickness from a few to several micrometers
was achieved. Because the roughness of spin coated Topas® 5013 is very rough, two methods were
proposed to flatten the surface. The first, Flat stamp, used a flat stamp to flatten the surface,
while the second, LOR-de-bonding, uses a smooth sacrificial layer together with bonding and wet
etching. Only the former yielded usable results, where a roughness of 1-3 nm over a distance of
500 µm was achieved.

Experimental results by Micro Resist Technology GmbH showed a lower roughness after spin
coating by using sec-butylbenzene compared to using toluene. Due to problems in using Topas®
5013 as explained in the next chapter, a switch was made to Topas® 8007. Using Topas® 8007
and sec-butylbenzene eliminates the need for flattening the substrate, hence simplifying the pro-
cess. The acquired thickness for Topas® 8007 dissolved in sec-butylbenzene was shown to vary
from 5 to 20 µm, which are the required thickness of the dielectric layers for the devices fabri-
cated in this project. The measured roughness after spin coating was shown to be between 2-4 nm.

The next step is to investigate how Topas® can be used with standard UV-lithography.