First Draft 1

1 Kashka Irani - Fabrication of High-Aspect-Ratio Metallic Micro-structures Using Photolithography
FABRICATION OF HIGH-ASPECT-RATIO METALLIC

MICRO-STRUCTURES USING PHOTOLITHOGRAPHY

KASHKA IRANI

ROYAL MELBOURNE INSTITUTE OF TECHNOLOGY
2013

Acknowledgments

Table of Contents
Acknowledgments
Table of Contents
Summary
List of Tables
List of Figures
1 Introduction
1.1 Background
1.2 Objectives
1.3 Organization
2 Microfabrication techniques
2.1 MEMS
2.2 MEMS fabrication technologies
2.2.1 Conventional fabrication techniques
2.2.2 Non-conventional micro-fabrication methods
3 Mould Fabrication
3.1 Photoresist
3.2 KMPR: Negative Epoxy Resist
3.2.1 Cross-Linking [32]
3.2.2 The polymerization reactions of KMPR
3.2.3 Chemical properties of KMPR
3.2.4 Mechanical properties of KMPR
3.3 Comparison of reverse-side and top-side exposure
3.3.1 Contact between exposure mask and KMPR
3.3.2 No sticking of photomask to photoresist
3.4 Preparing exposure mask [37]
3.5 Copper electroplating
3.6 Photo-Lithographic Process of KMPR
3.6.1 Introduction
3.6.2 Photolithographic steps
3.6.3 Resist Stripping
3.6.4 Micro cracks and delamination
3.6.5 KMPR residual
3.7 Experiment 1- Mould fabrication
3.7.1 Results and discussion
3.7.2 Solution
3.8 Experiment 2- Mould fabrication (soft baking and exposure energy)
3.8.1 Soft-baking time
3.8.2 Exposure time
3.8.4 Solution
3.9 Experiment3 - Mould fabrication (soft baking temperature and relaxation time)
3.9.1 Soft baking temperature
3.9.2 Relaxation time
3.9.4 Solution

3.10 Experiment 4- Mould fabrication (post exposure baking temperature)
3.10.1 Post exposure baking temperature
3.11 Experiment 5- Mould fabrication (two layers)
4 Electroplating
4.1 Electroplating
4.2 Electrochemical deposition
4.3 Electroplating Mechanism
4.4 Electroplating Calculation
4.5 Electroplating parameters:
4.6 Poor coating
4.7 Problems during electroplating
4.8 Equipment
4.9 Electroplating experiments
4.12.1 Experiment 1
Results and discussion
Solution
4.12.2 Experiment 2- Electroplating (low current
Results and discussion)
5 Fabrication of micro-array structure
5.1 Sacrificial photoresist
5.1.1 Sacrificial photoresist
5.1.2 Advantages
5.1.3 Disadvantages
5.2 Sacrificial photoresist and transparent resist
5.3 Using PDMS technique
5.3.1 Advantages of using PDMS technique
5.3.2 PDMS problem
5.3.3 Micro mould fabrication parameters
5.3.4 Experiment
5.4 Chromium and gold seed layer
5.4.1 Chromium and Gold as a seed layer material
5.4.2 Gold evaporation
5.4.3 Removal of seed layers
5.4.5 Results
6 Conclusions and future work
6.1 Conclusions
6.2 Recommendations for future work
References

Summary
High-aspect-ratio micro structures (HARMST) are commonly used in making micro devices.
Additionally, the next generation high performance micro-electromechanical systems (MEMS)
devices require fabrication technologies with high aspect ratio features. Among the methods used
to fabricate HARMST, LIGA is the most well-known and is by far the best option in terms of
fabrication quality. Despite the LIGA fabrication process being well-developed in the laboratory,
it has its draw backs largely due to the expensive X-ray source that is needed for exposing the
resist. An alternative method in producing HARMST is to use a photoresist and expose it by UV-
lithography. KMPR resist is the latest forerunner of all commercial high-viscous photoresist in
high-aspect ratio applications, superior to SU-8. This is due to the vertical absorption of KMPR
near UV range which results in vertical sidewalls and uniform exposure as well as an inherently
easier stripping option in comparison to SU-8. UV sensitive characteristics, high viscosity, and
high functionality are some of the advantages of KMPR. In comparison to LIGA, it has proved to
be a possible replacement to producing HARMST at lower costs.
This project looks into the use of KMPR as an electroplating mould to fabricate high aspect ratio
metallic microstructures. For fabrication of KMPR micro structures without any defects, such as
cracks, key parameters which affect the KMPR photolithography process will be examined.
These include pre bake and post-exposure bake times and temperatures, exposure time and
finally cooling off time after pre-baking and before development. For UV exposure, two methods
exist: top side and reverse side. In this paper the top-side method was chosen as it has several
advantages in comparison to the reverse side method.
After top-side exposure, the fabricated KMPR mould is plated using copper electroplating.
Electroplating parameters have been studies and pulse plating method has been established to
produce efficient electroplating results. The KMPR was then removed from the plated structure
using MicroChems Remover PG (NMP).
Finally, a second and even more cost effective fabrication method has been studied. The aim was
to find a method which did not require lithography. PDMS on PDMS double and PDMS on
PDMS thermal aging casting methods were investigated. The conclusion is that KMPR
lithography is more practical than the other methods. Using this method, high aspect metallic
micro structures has been fabricated in an inexpensive way.

List of Tables

List of Figures

1 Introduction
1.1 Background
Micromachining or micro fabrication is an important development process which refers to the
technique for fabricating 3D structures with dimension in the micrometer to nano meter ranges.
Typical devices include: MEMS devices, microsystems, actuators and sensors, micro dies and
moulds to name a few. These include the use of a set of manufacturing methods that rely on thick
and thin film batch fabrication methods for integrated electronic circuits. In general,
micromaching refers to precise fabrication techniques such as: lithography for fabricating three
dimensional structures, etching to remove ,,.,,, , laser ablation to something something
something
The advantages of the above mentioned is the achievable miniaturization of devices structural
capabilities in comparison to other conventional machining methods. Miniaturization has the
several advantages:
-Decrease weight
-Decrease volume
-Decrease costs
-Increase performance
-Increase reliability
-Increase speed
The demand for High aspect ratio microstructures has been increasing in the MEMS industry.
HARMTS are used in airbag sensors, medical devices, chemical and electrical systems including
micro accelerator, micro-actuator, polymerase chain reaction (PCR) system, micro mixer and
reactor. For example, a high aspect-ratio metallic microstructure with vertical sidewalls can
increase the output force of a micro-actuator [3], and it can hence be used as a mould for replica
multichannel polymer chips [2]. Metallic high aspect ratio microstructures have several
advantages [6]:
-low driving voltage,

-larger displacement in actuator systems,
-increase structural rigidity,
-higher actuation force,
-large magnetic force due to the large volume,
-larger displacement in actuator system,
-higher sensitivity in sensor applications by virtue of large mass
Usual methods for fabricating HARMS include LIGA (Lithographie Galvanoformung
Abformung), LIGA-like, and DRIE (Deep Reactive Ion Etching) processes. LIGA is a well-
known method to make HARMS with a few millimeters in height and aspect ratios of up
100:1. One of the major problems of this method is its need for X-ray source. LIGA-like
is a low-cost HARMS fabrication method using SU-8 as a resist but its aspect- ratio and
resolution are lower than LIGA process. Another option to make HARMS is to use DRIE to
make deep silicon trenches.
Electrodeposition through a mask is one of the main steps for the fabrication of electronic
microstructures and three-dimensional devices in MEMS. Thick microstructures such as
HARMS can be easily fabricated by this additive process, which selectively electroplates
HARMS on conductive layers. By this method, thick microstructures can be easily fabricated in
a simple, low-cost process at ambient temperature.

NEED TO BULD UP.

1.2 Objectives
The aim of this project is to fabricate high aspect metallic micro structures. This aim can be
further categorized into four sections as follow:
1. For the fabrication of high aspect metallic microstructures, the first requirement is to
fabricate a KMPR micro mould. When a thick photoresist is used in the fabrication
process several problems need to be overcome:
- Resist cracking
- Resist delamination
- Resist residue
- Air bubbles
In this project all problems will be solved by optimizing key parameters during the
photolithography process.
2. The second objective is to properly copper electroplate into the micro mould filling the
inside. In this project the parameters of the electroplating test parameters are studied, and
different electroplating parameters are tested to overcome potential problems.
3. The third objective is to find a fabrication method which eases the process of separating
the nickel electroplated structure with the KMPR micro-mould. A good fabrication
method should be fast, readily available and should be able to separate the micro
structure from the mould completely without leaving any excess residue. In this project a
couple of methods have been studied in order to find the best approach.
4. The last objective is to find an alternative and even faster method that the ones mentioned
above. PDMS on PDMS double casting has been studied. A couple a methods have been
tested and the best method has been found.

NEED TO BULK UP.

1.3 Organisation
Chapter 2 introduces MEMS and discusses the major methods in MEMS fabrication
and their limitations.
Chapter 3 describes KMPR as a negative epoxy resist and discusses its physical,
mechanical and chemical characteristics, UV exposure methods and the selection of
the top side method. It describes the preparation of the exposure mask for KMPR
lithography and the details of the photolithographic process of KMPR.
Chapter 4 describes the electroplating process and the parameters affecting
electroplating and the problems during the electroplating process, fabrication of
KMPR mould without any cracks or delamination and discusses the fabrication
process for separating the electroplating structure from the KMPR mould.
Chapter 5 presents the PDMS on PDMS double casting fabrication process.
Chapter 6 gives the conclusions and recommendations for future works.

2 Microfabrication techniques
2.1 MEMS
Microelectromechanical Systems (MEMS) is technology which integrates functional
microsystems such as micro-electrical, mechanical, optical and other components to enable the
whole system to sense, decide and react [7]. MEMS relates to technologies such as Micro
Systems Technology (MST in Europe), Micromachines (in Japan), Micromechanics,
MicroMachining Technology, Microdynamics, Micromechatronics, MicroEngineering
Technology and MicroInstruments. MEMS technologies are not limited to the micron range
and can reach as high as the centimeter range. MEMS is a vast and broad collection of
microfabrication methods for building microstructures which can be integrated with
electronic circuitry, resulting in new products and new product concepts. In fact, MEMS
applications are more diverse than purely microelectronic integrated circuit (IC) applications.
A map of MEMS applications is shown in Figure 2-1 [8].
The first MEMS Silicon-based pressure sensor application was invented in 1959 [9]. In the
1990s research progresses have seen sensors, actuators and control functions fabricated on a
single silicon IC chip. In 1982, the first 40,000 micromachine pressure sensors were produced
for medical industry [11]. In 1994, this amount reached 20 million. Also, in 1999, MEMS
optical network switches were first developed by Lucent Technologies. In the academic area,
the number of conferences and journals in the MEMS area is increasing and researchers are
combining MEMS sensors and actuators with emerging bioMEMS technology. Hence,
government funding in this area is increasing too. But still there are many challenges in this
technology and commercialization of this area which needs to be solved.


Figure 2-1 A map of MEMS applications [8].
2.2 MEMS fabrication technologies
There are several methods for fabrication of MEMS. This chapter will review
microelectromechanical systems (MEMS) and t hen discusses the major methods in MEMS
fabrication as well as their limitation.
There are three characteristic features of MEMS fabrication technologies: miniaturization,
multiplicity, and microelectronics. Several microfabrication technologies are used profusely and
of those, three are used mainly for MEMS. Bulk micromachining, surface micromachining and
high-aspect-ratio machining utilizing photolithography, chemical and plasma etching, thin-film
deposition and other manufacturing processes were originated from integrated circuits
technology and microelectronics. Manufacturing processes such as surface micromachining and

bulk micromachining refer to conventional techniques. The other methods such as LIGA
developed for MEMS are referred to non-conventional techniques. In the following, the major
MEMS fabrication methods will be discussed.

2.2.1 Conventional fabrication techniques
Bulk micromachining
In the 1960s, this process was developed for precise silicon etching, making it the oldest
micromachining technique. This technique requires the removal of the substrate material in
order to expose miniaturized mechanical components. During this process, a silicon wafer is
covered by a mask and etched in desired orientation by etching solution. The solution
removes unwanted parts and forms the microstructure. Figure 2-2 shows several
constructions formed by this fabrication method such as beams, bridges, nozzle and
membrane [15]. Figure 2-3 shows a construction formed by this fabrication using SEM imaging
[reference].

Figure 2-2 Examples of bulk micromachining structures [16]


Figure 2-3 SEM image of bulk micromachining structures [16]
Surface micromachining
There are many methods of surface micromachining depending on the material and the
etchant used during the fabrication process. In the general method, devices are formed by
methods such as patterning, deposition, and etching of sacrificial and structural thin
films on silicon wafer. By using surface micromachining more complex structures can be
fabricated than the bulk micromaching fabrication technique and this is because the precise
horizontal dimension and vertical direction of a structure can be accomplished. Figure 2-4
shows a typical surface micromachining structure [15]. The main reason that surface
machining is used over various other techniques is because single-sided wafer processing is
required and the process is relatively simple.

Figure 2-4 Standard surface micromachining structure [17]

2.2.2 Non-conventional micro-fabrication methods
In the following, we will discuss about the techniques known as high-aspect-ratio 3D
microstructures.

LIGA
LIGA is a German acronym for Lithographie (lithography), Galvanoformung
(Electroplating) and Abformung (Moulding). This process was developed at the Institute
for Nuclear Process Engineering in Karlsruhe of Germany in 1980s to make nozzles for
uranium enrichment [18]. It can produce small precise structures with the heights of
several hundred micrometres to 1 mm, and with an aspect ratio of more than 100:1.
Figure 2-5 shows the LIGA process.
The limitation of LIGA is its need for X-ray synchrotron radiation source. The problem
of X-ray source is not limited only to its cost, but also it is not adapted to the standards of
cleanroom. So, the LIGA technique is not used widely except by a few research
organizations [14]. This restriction encouraged the researchers to find other alternatives
such as LIGA-like processes.

LIGA-like processes
LIGA-like processes are the techniques similar to the LIGA process but at very low cost
and can still produce the high-aspect-ratio micro structures as LIGA technology. In these
processes, conventional UV is used instead of expensive X-ray synchrotron source.
Hence, the whole process can be done in a cleanroom. This process can be done in two
ways: making moulds using photolithography, or dry etching of silicon wafers.

Photoresist
Many efforts have been done for using photoresist in MEMS fabrication industry.
Polyimide was used by several research groups as electroplating mould for fabrication of
metallic structures [20]. It can be spun in the range of 10-50m thickness just in a single
coating. A thick positive photoresist such as AZ4620 can reach to the thickness of 2-10m
in one single coating. Therefore, it is possible to reach a thickness of 35m by triple
coating and 22m by double coating with good resolution [21]. But all of these photo
resists have two main problems: hard to coat thicker than 50m because of alignment
and their lower resolution.


Figure 2-5 LIGA processing steps [19]

The advent of KMPR made a revolution in the ultra-thick photoresist. The characteristics
of this photoresist will be discussed in the next chapter. By using this photoresist, 1000m
thickness can be achieved in a single coating. Hence it is possible to obtain more than 1
mm by using multiple spin coating. KMPR is a low-cost process and has good mechanical
properties and it can be used as a mould for subsequent process such as injection
moulding and electroplating.

DRIE:
Deep Reactive Ion Etching (DRIE) is a new dry etching method which can obtain high
aspect ratios of more than 50 micron [22]. The advantages of using DRIE are:

-Highly anisotropic
-Enhanced structures
-High resolution

Figure 2-6 shows the DRIE set-up. The fabrication method has two-step process (Figure 2-
7). In the first step, a passivating polymer is coated by the plasma deposition method on
a patterned silicon wafer. In the second step, the desired parts will be etched. In this step,
the fluorine radicals ionized of SF6 remove the protective coating from the area parallel
to the substrate surface. Then, it starts to etch the exposed silicon area anisotropically in
the normal RIE mode until it forms the desired vertical sidewall [23]. DRIE process made a
revolution in bulk micromachining. It can be used to etch shallow and deep structure
into the back side and front side of a wafer. It can also be used to etch through the wafer
completely.
The disadvantage of this method is of course:
-Use of high plasma power
-Specialized hardware is required
-Expensive.
Because of this, DRIE is not a commonly used fabrication method in laboratories.


Figure 2-6 DRIE experimental set up

Figure 2-7 Standard DRIE processing structure
Multi-layer process
The previous method provide a simple 2D pattern extended in the third direction. Only
cylindrical and prismatic shapes can be created and are therefore not true 3D structures
rather 2.5D structures [24]. To solve this problem, many modified LIGA processes have
been developed. The multi-layer process is one of these processes and can provide

stepped structures, conical structures and sloped side wall structures. Initially, this
method was developed for a three-layer structure. Figure 2-8 shows the process involves
photolithography; electroplating and planarization.

Figure 2-8 Multi-layer process steps [25]
a) Patterning of substrate b) Deposition of metal c) Stripping of photoresist and plating
with sacrificial layer d) Planarization to reach to the desired thickness e) Repetition of
the process f) Obtaining the 3D microstructure after stripping the sacrificial layer
Microstereolithography
This process is based on stereolithography which is used to produce rapid prototyping
structures. This 3D microfabrication technique was developed in 1992 [26]. Figure 2-9
shows a schematic design of stereolithography machine used in the
microstereolithography. The method of manufacturing is by stacking 2D layers and forms
a 3D structure. UV curable polymer is used as material. In 1996, 2m was reported as
the minimum resolution of this process [27]. Microstructures manufactured by this
method can be used directly or can be used as mould for plating metal structures. Figure 2-
8 shows a fabricated chain by this method.


Figure 2-9 Stereolithography desktop machine [28]
Figure 2-10 Schematic diagram of stereolithography desktop machine [28]

3 Mould Fabrication
3.1 Photoresist
Photoresist materials are organic compounds whose chemical properties change when exposed
to ultraviolet light due to breaking of chemical bonds. Photoresists are classified into two
groups: Negative and Positive resists. When the exposed region becomes more soluble in
the developer, the compound is called positive resist. When the exposed region become
less soluble in the developer, the compound is called negative resist. These resists can be
divided into one-component or two-component systems [29]. One-component refers to
sensitive homogeneous material. PMMA and COP are examples of positive and negative one-
component resists, respectively. A two-component resist consists of a Photo-active component
(PAC) in an inert matrix resin. A classical two-component resist is the novolac positive resist
consisting of a novolac copolymer and a photo-active component called diazonaphthoquinone.
Upon irradiation, the diazonaphthoquinone changes from a base soluble inhibitor to a base
soluble photo-product so, the development of the resist results in direct copy of image
(positive) from the mask on the substrate (Figure 3-1a).
A two-component negative resist is the cyclized polyisoprene synthetic rubber matrix with
bisarylazide photo-active components, such as the Kodak KTFR. When the photoactive
component releases acid due to irradiation, the matrix resin is polymerized to form inverse or
negative patterns on the substrate (Figure 3-1b). Although this resist forms pinhole-free film, it
has two problems. Firstly, oxygen in the resist radicals causes cross-linking. Hence, it is
better that the resist is coated in nitrogen or vacuum environment. Secondly, swelling in the
images of negative resist causes degradation of patterns, limiting the resolution (not greater than
2m) [30]. These problems can be solved by using KMPR resist rather than the negative resist
Kodak KTFR.


Figure 3-1 Photolithography process (a) Positive resist (b) Negative resist
3.2 KMPR Properties
KMPR more or less has the same physical, chemical and mechanical properties as SU-8. It is a
fairly new photoresist that has been invented by Microchem chemicals because of the stripping
problem associated with SU-8. In these next sections, SU-8 chemical, mechanical and . Will
be discussed as it has been well established for many years and has the same properties as
KMPR photoresist.

3.3 SU-8: Negative Epoxy Resist
SU-8 is a negative resist, trademark of Shell Chemicals, and is derived from EPON resin SU-8
[31]. IBM developed a two-component negative photoresist which consists of EPON resin
SU-8 and a photo-initiator called triarylsulfonium salt dissolved in gamma butyrolactone (GBL)
solvent. Chemically, SU-8 is known as glycidyl ether derivative of bisphenol-A novolac, which

is a transparent solid epoxy resin. Photoresists such as SU-8 are based on epoxies. The prefix of
epoxy refers to a bridge which consists of an oxygen atom linked to two other atoms, most often
carbon, already combined in some way. This kind of structure is called 1, 2-epoxide (Figure 3-2).
A molecule which contains one or more 1, 2-epoxy groups is defined as an epoxy resin. These
molecules are able to convert to a thermoset form or three dimensional network structures. This
converting process is called crosslinking or curing [32].

Figure 3-2 1, 2-epoxy ring

3.2.1 Cross-Linking [32]
The process by which one or more types of reactants (i.e. a curing agent and an epoxide) are
changed from a low molecular weight to a highly crosslinked network, is called crosslinking or
curing.
There are three categories for epoxy resin curing agents:

- Active hydrogen compound, which is cured by polyaddition reactions.
- Ionic initiators, which are subdivided into anionic and cationic.
- Crosslinkers, which couple through the hydroxyl functionality higher molecular-weight epoxy
resins.

SU-8 has an average of eight functional epoxy groups (the highest functionality obtained
commercially) to maximize resist sensitivity (Figure 3-3). Due to its low molecular weight
(40001000 amu) dissolution of a wide range of solid SU-8 in solvent GBL is possible.
Therefore, it produces resists of wide range of viscosity; so it can be possible to have film
thickness from a few m to a few hundred m by spin-coating. The melting point of solid SU-8
is 82C but after curing and polymerization, the transition temperature (Tg) will reach 200C.
This high transition temperature causes the resist to have thermal stability and an excellent
resistance to plasma etching. As Figure 3-4 shows, there is negligible absorption of UV radiation

when the wavelength is greater than 360nm. Hence, it makes constant exposure of resist
throughout the film, therefore vertical side walls with aspect ratio of 18:1 is possible. The photo-
active component used in SU-8 resist is an onium salt called tri-aryl-sulfonium salt. This salt
contains three aromatic compounds (aryl) covalently bonded to center sulphur atom, which is

Figure 3-3 Basic SU8 molecule, note the 8 epoxy groups [32]

ionically bonded to a Lewis acid (e.g. BF4-, PF6-, AsF6- and etc.). One specific example of tri
arylsulfonium salt is triphenyl-sulfonium hexaflouroantimonate (Ph3SSbF6).


3.2.2 The polymerization reactions of SU-8
A negative resist can become insoluble through free radical polymerization of bisaryldiazide and
can be polymerized by cationic polymerization. Unlike bis-aryldiazide, cationic polymerization
of the resist films cross link SU-8. When UV irradiates, the triarylsulfonium salt is reduced to
radicals (aryl and diarylsulfonium radicals) and then it reacts and forms strong acids which starts
the cationic polymerization process [30]. There are three reaction steps for the cationic
polymerization (Figure 3-5) of SU-8 [34]. They are photolysis, initiation and preparation steps.
UV radiation breaks down the triarylsulfonium salt to generate Lewis acids. These acids react
with the monometers to form the three-dimensional cross-linking polymers. The polymerization
is done by the ring-opening of the 1, 2-epoxy.

3.2.3 Photolysis of tri-arylsulfonium salts
Photolysis of tri-arylsulfonium salts can be divided into three steps. First step includes the
separation of a carbon to sulfur bond by UV irradiation to form a diarylsulfonium cation radical,
an aryl radical and anion. Secondary radicals will be resulted by further interactions between the
monomer and radical species. Then hydrogen ions will be released to form strong acids (HPF6,

HBF4 and etc) which are the initiators of the polymerization process. In comparison with the
other photo-initiators, using tri-arylsulfonium salts has different advantages. First of all, the
presence of oxygen will not inhibit the photolysis process. Secondly, the photolytic rate and
frequency response of the photo-initiators will be a function of the tri-arylsulfonium cations. By
changing the cations, the rate and the frequency response of the photo-initiators can be changed.
Thirdly, the easy photolysis of the tri-arylsulfonium salts does not relate to thermal
decomposition. In the opposite direction, it can be seen that at higher temperature such as 150C,
after several hours, only small thermal decomposition will occur. [30]

3.2.4 Initiation and propagation of polymerization
As described before, when UV irradiates, strong Lewis acids will form to make the initial
protonation of monomer. One example is in multifunctional novolac-epoxy resin. The epoxy
groups are attacked by Lewis acids and these acids open up the epoxide ring for next
polymerization at room temperature, photo-initiated cationic polymerization can occur but it is
better to have post-baking after exposure for thick resist to increase the rate of cross- linking
process. Oxygen does not inhibit the resist polymerization and long storage time is possible with
mixture of photo-initiators and monometer. [30]

3.2.5 Chemical properties of SU-8
While the swelling of the resist and the existence of oxygen affect free radicals polymerization,
cationic polymerization is not affected but it provides great chemical properties which is good
for LIGA- like processes. These chemical properties are:

Resist sensitivity (minimum dose that gives dimensional equality of clear and opaque feature):
it is optimized because SU-8 has the largest number of epoxy groups per molecule.
Excellent contrast (contrast:1/log(Df/Di), Df: Extrapolated dose for full thickness, Di : idealized
minimum dose): excellent contrast is possible because of the low molecular weight of SU-8
where the unexposed resist in comparison with the polymerized resist dissolves at a faster rate,
so it gives a good edge definition.

Therefore, the resist is stable at high temperature; therefore it prevents degradation of resist
contrast which is good for electroplating of high-aspect-ratio structures in hot bath.
Obtaining different thickness from 1m to 650m by single resist spin. Low molecular weight
makes dissolution of resist in solvent easier, so it is possible to obtain different thickness.
Due to Figure 3-4, during exposure, UV absorption is negligible, so it results into the
relationship between thickness and exposure time.
Excellent thermal resistance and stability during process and etch resistance to reactive ion
etching (RIE) because of high glass transition for cross-linked resist.
High adhesive strength, so it allows higher film stress without peeling due to wafer dishing or
bowing [30].

3.2.6 Mechanical properties of SU-8
Compared to metals and ceramics, polymers are mechanically weaker in toughness and
mechanical strength. But if densities are considered, they have equivalent specific strength and
toughness [35]. Mechanical properties of SU-8 can be easily modified by energetic beam to
shape cross-linked network which results into an increase in molecular weight but also film
embitterment. Automotive industry has several efforts to develop ways to minimize such effect
to increase the polymers life period, while, semiconductor industry has used this effect to make
polymeric masks for next dry or wet etching of the under layer.
SU-8 can easily dissolve in solvent (GBL), so it allows the resist thickness to be from few
microns to hundreds of microns by single spin-coating. When UV exposes, polymerization
occurs to make highly cross-linked regions with high hardness but low toughness. Furthermore,
the adhesive strength of epoxy bond allows highly stress SU-8 deposits on the substrate without
any delamination. Table 3-1 shows an overview of several mechanical and chemical properties of
SU-8 photoepoxies:


Table 3-1 Mechanical and chemical properties of SU-8 [32]
3.3 Preparing exposure mask
The first step for fabricating the KMPR mould is preparing the exposure mask for SU-8
photolithography. For preparing of exposure mask, there are several photoresist options such as
KMPR 1005, KMPR 1010, KMPR 1025, KMPR 1035 and KMPR 1050. All these
photoresists can be used the difference remains in the structural dimensions you want to achieve.
In this project, KMPR 1035 has been chosen.

KMPR 1035:
The photoresist is a high contrast i-Line epoxy photoresist and can be stripped easily from a
wafer. KMPR 1035 is designed to coat 30 80 m in a single step using the four standard
viscosities. KMPR 1035 has excellent adhesion, and is ideal for many MEMS, Electroplating
and Deep Resistance Ion Etching applications (DRIE). 500mJ/cm
2
is the minimum exposure
requirement. For enhancing thermal stability, a post exposure bake cycle of up to 100C
is used. To use this photoresist its suppliers recommendation is given as Table 3-2:


Table 3-2 Process parameters of KMPR1035 for 5m structure
Depending on conditions of the processes mentioned above as well as the substrate being used,
the parameters for testing need to be changed in order to maintain sufficiently better results.
Following values from table. were not good as can be seen from figure ..

Figure 3-6 KMPR experimental result
Figure . shows that there is some amount of photoresist left in the structure hence we can
conclude that the exposure time is not enough. Calculating this we found that an exposure time
of .. was sufficient.


Figure 3-7 KMPR experimental result
As it can be seen in Figure 3-, there is no photoresist residue left inside the structure and the
sample is ready for the next step. If in any case we observe a little photoresist residue in the
structure, we can simply to do post exposure bake at 115C for one minute. Due to the fact that
the photo-lithography process of KMPR1035 in most of the parts is the same as the other
photoresists, the details of each step will be discussed in section 4 of this chapter. Here the
process of KMPR1035 will be discussed very briefly.

Steps of this process are as follows:

Surface preparation

Silicon wafer substrates are rinsed in an acetone bath for 15 minutes, and then they are sprayed
with isopropanol alcohol (IPA) and de-ionised water (DI). After cleaning the samples,
dehydration by hotplate at 100C for 10 minute is recommended.

Spin-coating

Two step spin-coating is used:
First step: Spin at 500 rpm with acceleration of 100 rpm/second for 5-10 seconds
Second step: Spin at 3000 rpm with acceleration of 300 100 rpm/second for 30 seconds


Relaxation time

One minute is enough time for photoresist to stabilize itself

Soft baking

Soft baking is done on hotplate at 100C for 15 minutes.

Exposure

The exposure energy which is needed is 500-645mJ/cm
2
.

Post exposure baking

Post exposure baking is done on hot plate at 100C for three minutes.

Development
The samples are developed in TMAH aqueous developer and DI water (1: 1) for 6 minutes then
rinsing with DI water for another 6 minutes and finally dry with N2 gas.

3.4 Copper electroplating
After patterning of KMPR1035 on a silicon wafer, copper electroplating must be done to make
the exposure mask. We will discuss about the copper electroplating and its parameters in detail in
the next chapter. By copper electroplating, copper is deposited in the left open spots.
Electroplating is carried out in an acid copper sulfate bath at 40-50C with a current density of 1-
1.5 A dm
-2
for 60 minutes. Figure 3-10 shows the results of electroplated sample (exposure
mask). As can be seen in this figure, the two sides of silicon wafer are opened in the KMPR
photolithography process until they can be electroplated. These two sides are used for having
connection in both electroplating, first and last one.

3.5 Photolithographic Process of KMPR
3.5.1 Introduction
The physical and chemical properties of KMPR were discussed in the previous chapter. In this
chapter, lithographic process of KMPR will be discussed. We can divide the whole lithographic
process to eight steps as follows (Figure 3-11):

Figure 3-8 Photolithographic processing steps for KMPR:
1-Apply KMPR1000 resist, 2-Spin Coating, 3- Pre-baking, 4- UV-Exposure,
5-Post-baking, 6- Development, 7-Hard baking, 8-Final product

1- Surface preparation
2- Spin Coating
3- Soft baking
4- Exposure
5- Post Exposure Baking

6- Development
7- Hard baking
8- Stripping

Each of the above steps will affect resolution, measure of vertical side wall, density defects,
control of thickness, and adhesion to substrate and etch resistance.
Hence, we must try to find the optimum value for the parameters in each step.

3.6.2 Photo-lithographic steps
1-Surface Preparation
Surface cleaning of the substrate is one of the important parts of the photolithography process
because all the structures will be built on the surface. To achieve maximum process reliability,
the substrates should be cleaned and dried before KMPR coating. If the substrates are clean
baking at 90C- 100C for ten minutes is sufficient for the desorption of water. This baking
process prevents any occasional resist lift-off (poor adhesion) due to moisture absorption at
substrate and SU-8 interface. If a substrate is contaminated with impurities, it should be cleaned
with Piranha wet etch (H2SO4 and H2O2) followed by rinsing in DI water. There are other
methods for substrate cleaning such as using reactive ion etching (RIE), or submerging substrates
in an ultrasonic acetone bath to remove organic impurities followed by rinsing in isopropyl
alcohol to remove the acetone on the substrate.

2-Spin Coating
The coating step should be performed exactly after surface preparation to avoid reabsorption of
water or any particle. There are two methods to coat resist in semiconductor industries: static and
dynamic dispensing method. Dynamic dispensing method is not suitable for KMPR thick
photoresist because it traps bubbles in the resist. So, the only choice will be static dispensing
method. In this method wafer is placed on a suction chuck and the resist is poured manually on
the substrate. Typically, it is enough to dispense 1 ml of resist for each inch of substrate. The
wafer is initially rotated at a speed of 500 rpm for 5-10 seconds with acceleration of 100
rpm/second. After that, according to the type of KMPR to use and thickness to achieve, we

choose our rotation speed by the data provided by the suppliers. Figure 3-14 shows the spin
speed versus thickness for KMPR 1005, 1010,1025,1035,1050.

Figure 3-9 Spin speed vs. Thickness
for KMPR 1000 resist (21C US & EU)

When the spin coater rotates in high speed, the resist is pushed to the wafer edge and the threads
of resist will be spun off the wafer and deposited around the coater. This two-step process makes
an initial thick resist coating to be spread to the edge before it ramps up the speed to obtain a
uniform film. This process hence decreases the edge bead of resist.

3-Pre-Baking
Pre-baking step is another important part of the photo lithography process. This step influences:
structure definition, moulds aspect ratio, moulds side wall, and development time and exposure
energy [40]. The spin coated film from the previous step has homogeneous dissolution KMPR
monomers and triarylsulfonium photo-active components in a huge reservoir of solvent. The
presence of solvent in resist is for facilitating the spin-coating process but it creates problems for
UV absorption and causes the resist to stick to the photo-mask during UV exposure. Therefore,
removing the solvent from the resist by pre-baking is necessary until it can be used for UV
exposure. For pre-baking, a level hot plate with good thermal control is used. It is not good to
use convection oven for this step because it forms a layer on the resist which inhibits the
evolution of solvent. By using a hot plate, the heat flows to the substrate from the back side and
this avoids formation of the crust at the resist surface. Therefore, ramped temperature is

recommended. When substrate is put on the hotplate directly, it is observed that the thick resist
flows to the centre of the substrate but this phenomenon is reduced by using ramped temperature.
Pre-baking time is very important. If a proper time is not chosen, three problems will happen:

1- Problem in UV absorption
2- Sticking the resist to the photo-mask

To find the optimum time for pre-baking step at baking temperature (T=100C), the structural
dimensions that is required should be known. After this simply refer to the KMPR1035
Datasheet. From the data sheet, for a structural thickness of 31-55 microns, a time of 5 minutes is
a sufficient period for pre-baking.
There are other methods for optimizing the baking time conditions. The first alternative method
could be to measure the change of KMPR1035 on 1 inch wafer by using a micro-balance after
pre-baking in different periods of time to measure amount of solvent vaporized. The second
alternative method could be to remove the wafer from the hotplate after prescribed intervals and
cool down in the clean rooms temperature. Then putting it on hotplate again, repeat this step
until the wrinkles in the film leaves the film completely and are no longer seen in the film [39].
After this step, it is better to give a time to photoresist to rest and stabilize itself.

Table 3-3 Soft Bake Times
4-Exposure
The next step after pre-baking of KMPR resist is UV exposure by using Karl Suss Mask Aligner
(MA8). This machine uses a source of mercury light which has a beam source of between 300-
450 nm which is near UV peak at 365nm. With reference to Figure 3-4, the optical absorption of
KMPR at 365 nm wavelength is negligible. So, we can conclude that exposure dose for KMPR is
directly proportional to the thickness of film. Upon irradiation, photolysis of tri-arylsulfonium

salt occurs; it results in a strong acid which starts the cationic polymerization process. Cross
linking (Polymerization) happens by nucleophilic interaction of epoxide oxygen with the
carbocation produced by ring opening and protonation of an epoxide of another polymer chain.
Table 3-18 Thickness vs. exposure dose for different film resist thickness. As it can be seen from
this figure, 235mJ/cm2 is the minimum dose which needs for 5 m film thickness. Therefore for
30 m film thickness which is used in this project, a minimum dose of 665mJ/cm2 is needed. But
in practice 665mJ/cm2 dose is not enough, and higher dose must be used until it compensate the
absorption and sufficient exposure to achieve sharp edge profile. Hence, for those substrates
which are not reflective, a higher dosage is needed to compensate the absorption in the resist and
substrate interface. The resists which are more viscous such as KMPR1050, they have less tri-
arylsulfonium salts for improving the edge profile, therefore they require longer exposure time,
too. Due to the fact that we use reverse side exposure method in our project, so, light
transmittance from the substrate (glass) also plays an important role in the calculation of
exposure energy. In the next section we will discuss about it in details.

Table 3-4 Exposure Energy

Table 3-5 Exposure Doses for Substrates

5- Post Exposure baking
Post exposure baking must be done after UV exposure. Cross-linking of KMPR happens during
this step. Post exposure baking is very necessary because in the solid state when molecular
motion is little, little reaction can occur. For having the highest efficiency while baking, the
temperature must be greater than 55C (in this temperature the resist is solid). During the cross-
linking formation, some shrinkage will happen due to densification and out-gassing of the
solvent. Baking time is dependent on the baking temperature. Usually baking is done on hotplate
at 95C for 15 min but this depends of the thickness of the structure. For stress reduction, it is
better to heat the substrate slowly to the baking temperature and cooling it down slowly to 65C.

6-Development
In order to remove the unexposed region from the structure, the development step must be done.
A good resist has a high resist selectivity and minimum swelling. Resist polymerization begins in
exposure time and accelerates during the post exposure baking until it minimizes the diffusion of
photo-active elements from the diffusion to the unexposed area therefore it keeps the
dimensional accuracy of side walls. In the development step, the regions which are highly
polymerized remain safe and the un-polymerized regions dissolves. MicroChems TMAH
(0.26N) aqueous alkaline developer I used but SU-8 developer can also be used. Other solvents
such as diacetone alcohol and ethyl lactate also can be used as an alternative. When high-aspect-
ratio or thick film structures are developing, high agitation must be done. Therefore it is better to
use ultrasonic or magnetic bath. The development time is dependent on the thickness of
structures and can be found in the suppliers data sheet such as MicroChems 1035 [39].

Table 3-6 Development Times for 2.38% TMAH

7-Hard baking
The mechanical properties of KMPR are very good. However if the imaged resist is to be used as
the final device, hard baking can be beneficial. Therefore this step is useful for those parts which
are to be subject to thermal processing during their operation or for ensuring that the properties
of KMPR do not change during the application [39]. In this step the developed structure changes
to a very hard material and cross-linking process will be completed. After this step, the stress due
to the polymerization of resist is negligible but the stress due to the difference between thermal
coefficients of the resist and the substrate during the baking process is significant. So this stress
may lead to film delaminating or wafer bowing. The baking temperature is about 150C to
250C and the time of baking is between 5 to 30 minutes.

8 Resist Stripping
KMPR has been designed to strip easily from any substrate. The same cannot be the same for
SU-8 which has been designed for permanent and highly cross-linked epoxy material. Several
SU-8 techniques can be used to remove KMPR. These methods can be classified into two
groups; physical and chemical techniques. Physical techniques are the techniques used to
demolish or destroy the cross-linked KMPR directly such as water jet and laser ablation. These
methods are fairly selective to the KMPR polymer over metal microstructure, but they are not
able to remove all small areas of KMPR resist. In chemical methods, a chemical solution is used
to remove the SU-8. Some of the methods for removing SU-8 are: NANO Remover PG, nitric
acid (100%), piranha (H2O2:H2SO4), hot remover PG followed by wet etchant (H2SO4: H2O2:
H2O=3:1:1)[43], hot NMP(methyl : pyrrolidone =1:2), oxygen/luorine plasma etching [44], KrF
excimer laser [45] or water jet as a mechanical way but it is not good for fragile structures. A
simpler and more efficient method is to use MicroChems Remover PG (NMP). Simply heat the
bath to 80C and immerse the substrate for 10-20 minutes. The advantage of using remover PG
is that it does not demolish the micro structure leaving intact and without damage or excess
residue.

3.6 Experiment 1- MAKING MASK
A vector based graphics editor Adobe Illustrator was used to create a 10x10 pin array to
achieve higher resolution, higher pixel and better contrast as compared to using computer based
graphics such as a bitmap image.
A 10x10 pin array was created with a circle diameter of 1.50 mm and gap between each circle of
4.50 mm. Because KMPR solution will be poured on top of the silicon wafer in this experiment a
negative photoresist (Figure 3-1) is required to produce the pins, so the patterns created in Adobe
Illustrator match the silicon wafer dimensions. When the pattern is exposed to ultra violet light
the black patterned regions will be blocked while the white patterned regions will be formed onto
the silicon wafer.

Figure 3-10 (a) (b)
Conclusion
The mask was printed onto UV sensitive paper and will thus be used for subsequent experiments
which involve photolithography.

3.7 Experiment 2- Evaporation
The hypothesis for experiment 2 was to successfully create a conductive substrate using
evaporation techniques so that KMPR photoresist could be dispensed during the
photolithography process. The idea is that the conductive layer on the substrate could help when
electroplating.

Figure 3-1 shows the evaporation process. The whole process was started with coating the 3
silicon wafer substrate with a thin metal layer also called the seed layer (80 nm Chromium and
250 nm Gold) to form a conductive substrate which could then be used for electroplating (Figure
3-2). Originally, the seed layer consisted for 20 nm Chromium and 50 nm Nickel. However when
testing the conductivity it was found that the substrate remained unconductive and this was due
to the strong nickel oxide present on the surface. A way to remove the nickel oxide present on
the surface is to use a nickel oxide etchant described in the figure below:

Figure 3-11 KMPR mould fabrication steps



photoresist by acetone 6-Coating a thick SU-8 photoresist by spin coater

3.8 Experiment 3- Mould fabrication using photolithography
The hypothesis for experiment 3 was to successfully create a KMPR mould using
photolithography so that the mould could be copper electroplated into. Once the plating was
successfully, the KMPR resist would be able to easily be removed using MicroChems Remover
PG (NMP) so that a standalone copper structure remained.

Figure 3-31 shows the whole KMPR mould fabrication steps. The whole process was started
with coating a thin layer of KMPR1035 on 3 silicon wafer and patterning it with the help of
routine UV lithography. Then KMPR1035 was developed, to lift up the exposed area. By the
help of nickel electroplating, nickel was plated in the open areas. Then KMPR1035 resist was
removed using MicroChems Remover PG (NMP) to form the nickel mask. For mould
fabrication, the challenging part is to find the photolithography parameters of KMPR and to
UV LIGHT
(1)
(2)
(3)
(4)
(5)
KMPR Photoresist
Silicon Wafer
Copper
1- Coating a thin layer KMPR negative photoresist (30m) on Silicon
Wafer 2-Patterning of KMPR
photoresist 3-Development the photoresist 4-Electroplating copper 5-
Removing the KMPR photoresist
photoresist by acetone 6-Coating a thick SU-8 photoresist by spin coater

control them to have a good, accurate mould. Table 3-3 shows the photolithography parameters of
KMPR for the first experiment.

Parameters Setting
Cleaning method Acetone bath (15 min) + IPA + DI water +

Dehydration on hotplate 120C (10 min)
Relaxation time 5 min
Spin-coating Spin at 500 rpm (10s), accelation +

Spin at 3000 rpm (30s), acceleration
Soft baking 100C for 15 minutes
Exposure energy 665 mJ/cm2
Post exposure bake 100C for 3 minutes
Development AZ 326 developer (6 min) + Acetone (50s) + IPA +

DI water + Pressurized nitrogen

After the first experiment two phenomena were observed:
Some holes were able to plated into (Figure 3-31)
Not all holes were able to be plated into (Figure 3-32)
Not all holes were plated into. Therefore, it is apparent that delamination has happened. In
several cases, complete delaminating happens as it can be seen in Figure 3-32.


Figure 3-15

Figure 3-16

Solution

The separation of KMPR from substrate and cracks on the pattern depends on the internal stress.
The internal stress in the resist depends on many parameters, such as exposure dose and soft
baking time, so for overcoming these problems, we decreased these two parameters in the
following experiment.

3.9 Experiment 4- Mould fabrication with adjusted
parameters (soft baking, exposure energy and post exposure
baking).

The hypothesis for experiment 4 was to overcoming crack, delamination and electroplating
problems by adjusting the soft-baking, exposure and post exposure baking times, which were
subsequently increase increased. This would create a KMPR mould where all holes could be
plated into.

Soft-baking time
To find the sufficient time of KMPR thick resist, for pre-baking (T=95C) referring back to
Table 3 - ., the minimum required pre-baking time is minutes. By adjusting the pre-baking
time to . the majority of surface solvent had vaporized and the resist was sufficiently rigid for
contact exposure without damaging the photo mask.

Exposure time

Exposure time was increased to mJ/cm2. This amount is about 1.5 times more than the amount
of energy used for silicon substrate.

Post Exposure Baking time

As has been reported by other researchers, it is possible to decrease post exposure baking
temperature to 55C [63].
Table 3-4 shows the photolithography parameters of KMPR for the second experiment.


Parameters Setting
Cleaning method Acetone bath (15 min) + IPA + DI water +

Dehydration on hotplate 120C (10 min)
Spin-coating Spin at 500 rpm (10s), accelation +

Spin at 3000 rpm (30s), acceleration
Soft baking 100C for 15 minutes
Exposure energy 700 mJ/cm2
Post exposure bake 100C for 3 minutes
Development AZ 326 developer (10 min) + Acetone (50s) + IPA +

DI water + Pressurized nitrogen


There was a noticeable change in results from experiment 3. By changing the exposure energy
we were able to plate into the KMPR structure.

As can be seen in Figure 3-37, with the help of all modification, there was no crack on the resist
surface and no delamination happened. So, with the help of Table 3-6 we can make the moulds
for the next step (nickel electroplating) without any problem.


4 Electroplating

4.1 Electroplating

Electroplating is the process of transferring metallic cat ions (metallic pieces) though electrolyte
(solution) and deposits onto the substrate by an electric current. Many applications for
electroplating exist in the modern age and there is good potential to produce high aspect ratio
structures such as those made in the LIGA process. Additional advantages of this process are
listed below:

Cheap capital equipment
Low operating cost
Reliability
Precise control
Quick process
Very low temperature process (50C), so it is IC-compatible processing
Magnetic structural material available
High conductivity structural material available
No clean room needed
Few process variable
High-aspect-ratio devices can be achieved

The electroplating process involves three key concepts:

1- Preparation of substrate
2- Preparation of electrolyte/solution/bath
3- Electroplating process (current supply needed)

Key characteristics of the resultant electroplated structure, such as uniformity, thickness and
mechanical properties can be controlled by plating parameters, such as electrolyte composition,
plating temperature, current and pH of electrolyte, discussed below.
Electroplating can be applied to a wide range of materials, including conductive polymers,
metals, metal alloys and some semiconductors. Almost 17 metals can be used in this process,
namely as: Ni, Co, Fe, Cr, Pb, Ir, Sn, Pd, Zn, Rh, Ru, Cu, Ag, Cd, In, Pt and Au. But only a few
of them can be used for electro-deposition of microstructures. Among all the metals which can
be electroplated, nickel is commonly used in MEMS industry. It has good mechanical properties,
such as hardness, yield strength and Youngs modulus and good resistance against corrosion; so
it has been chosen as the structural material in this project.

4.2 Electrochemical deposition
Electrochemical deposition or electroplating can be divided into two groups:

- electrodeposition
- electroless deposition

In this thesis we will discuss only electrodeposition.

Electrodeposition refers to the decrease of metal ions from electrolyte solution to the solid metal
by an external current. LIGA and LIGA-like process are examples of this process by which metal
and metal alloys are plated onto the mould structure to make microstructures.

4.3 Electroplating Mechanism

Figure 4-1 shows a simple electroplating set up used in this project. Both the sample and copper
are immersed in the electroplating solution. The sample and copper are connected to the negative
and positive terminals respectively. So, the sample is the cathode and copper is the anode. When
there is enough bias between cathode and anode, electrons will move from the anode to the
cathode. At the cathode, a reaction occurs and results in the deposition of nickel on the sample.
This reaction is as follows:


Ni2++2e- Ni (4-1)

At the other side, which is the anode, another reaction occurs whereby the nickel dissolves into
the electrolyte.

Ni 2e-+Ni2+ (4-2)

This reaction helps the electrons to flow to the power supply and nickel ions flow inside the
solution.

Figure 4-1 Typical setup for electroplating [65]

4.4 Electroplating Calculation

For calculating the required current density needed for obtaining a certain thickness in a definite
time, Faradays law is used. Faradays law saying that when there is 96485 coulombs of
electrical charge through the solution of electrolytic, one gram equivalent weight of anode
substance is electroplated on the cathode. One gram equivalent weight refers to equivalent
weight in grams. Copper ion (Cu2+) which is in the plating bath needs 2 electrons to change to
Cu atom until it can be deposited. The gram equivalent weight gives the value of the material
being plated out according to the number of required electrons to neutralize the atomic weight in

grams from the ions of the material [66]. This value can be calculated by dividing the molecular
or atomic mass of the substance to the moles number of electrons achieved or lost.

Therefore times can by calculated by using Faradays Law:

()
() (

) (

) (

)
(
) (
) (
)

where,

The desired time for a 3 silicon wafer with structural dimensions thickness of 35m was 1h at a
current of 150 mA.

4.5 Electroplating parameters:

There are several parameters which affect electroplating that we will discuss about them in the
following:

1-Current density
Current density can be calculated by dividing the plating current to the plating area (4-3).
According to Faradays Law increasing in the current density results in an increase in the rate of
plating. Furthermore, studies have shown that the deposited films with high current density are
smoother than the films plated with lower current density [67]. However, from train and error
increasing the current density drastically corrode the copper finish resulting in a dull plated
surface. Results were based on trial and error.

2- pH
pH depends on the chemical composition of the electrolyte but it must be kept in a range of 1.5
to 5 to have a good electroplating [68]. During electroplating, pH must always be controlled. If

pH is greater than desired value, drops of dilute hydrochloric acid are used to decrease the pH. If
pH is less than desired value, potassium hydroxide is used to increase the pH. [69]

3- Temperature
Temperature plays an important role in electroplating. It has been observed that at high bath
temperature, the surface finish is poor. The recommended temperature for copper electroplating
is 40C to 60C.

4-Agitation
When electroplating, the concentration of ions near the cathode will decrease. In order to
minimize the regions with low concentration, agitation is needed. Thus, during the process
agitation will remove hydrogen bubbles from the cathodes surface. It was reported that with
increase in agitation, surface roughness will decrease [71]. Therefore, agitation must be in a
proper value which prevents pitting and poor surface finishing.

5- Bath Composition
Electrolyte bath for copper electroplating is made from sulphate salts of copper and chloride.
Sulphate bath decreases the cathode efficiency but gives good surface finish on the other hand,
chloride bath increase cathode efficiency but gives poor surface. So, for having a good result,
sulphate salts and chloride are mixed together to create copper sulphate bath. Therefore, it must
be noticed that some additives are usually added to the copper sulfate bath such as polyethylene
glycol diacrylate and janus green B. This additives tend to change the properties of the copper
finish. Polyethylene glycol diacrylate will tend to give a reddish tint finish to a structure while
janus green B will give a turquoise or black colour. Table 4-1 shows the chemical concentration
for electroplating solution. Between all these chemicals, copper sulphonate (CuSO4) plays the
role of copper carrier during plating.


4.6 Equipment

Here is the list of equipment used in the electroplating process:
Current Source (Keithley 220 Current Source): Used to provide DC Current and has constant
current (CC). Its resolution is 100mA when it is set in constant current mode.
Hotplate magnetic stirrer: Used to raise the temperature of the solution to the desired value
which is needed for electroplating. A stirrer is used during the electroplating.
Multi-meter: Used to measure current output
SEM: Used to determine the thickness of micro-mould to calculate electroplating time.

[INSERT PICTURE HERE]

Figure 4-2 Set up for experiments

4.7 Experiment 1

In this experiment, the fabricated mould in chapter 3 was electroplated. In all experiments,
before electroplating, all the moulds were evaporated until they became conductive. The
condition of electroplating is listed in Table 4-3.

Table 4-3 Electroplating parameters

pH 4.0
Temperature 40C
Current 170mA
Duration 2 (hr)
Remark Not all holes were filled


As can be seen in Figure 4-5, after a period of electroplating, the mould surface will be covered
and it is not possible for plating to continue and fill the mould. The reason for this phenomenon
is that when the current is high, the plating rate is high. So the surfaces which are nearer to the
anode get plated sooner than the other places. As a result, it does not allow time for other places
to be plated, especially the gaps. This issue is one of the main issues of high-aspect-ratio plating.

Solution

The simplest way to overcome this problem is using low current to decrease the rate of
electroplating.


4.8 Experiment 2- Electroplating (low current)

In this experiment, for overcoming the problem of incomplete filling of the mould, low current
was used. The condition of electroplating is in the Table ... As Table shows, the current and the
duration which were both calculated according to equation from section

Table 4-4 Electroplating parameters

pH 4.5
Temperature 45C
Current 10mA
Duration 24 (hrs)
Remark Excellent Plating


As can be seen in Figure 4-6, the electroplating was improved, but still needed to be improved
further to decrease the pores inside the micro-structure.

5.1 Using PDMS
This is another indirect method for removing the plated metallic structure. In this method instead
of using KMPR as a mould for microstructures, PDMS (Polydimethysiloxane) is used. The
whole fabrication process is as follows:

PDMS elastomer is used most widely in the fabrication of nano-lithography, rapid prototyping
and various microfluidic devices. PDMS has several advantages including:
[94, 95]:

optically transparent
chemically inert
non-toxic
longer life and resistance to chemicals and decomposition
light

Aluminium
master mould by
milling
Aluminium master
mould with
PDMS, cured at
70C, 30 min
Inverted PDMS
master
Evaporation of metal to
make PDMS structure
cond uctive
Replica PDMS master
mould and
aluminium master mould

non flammable
durable and compatible to microelectronics
can be used in a wide range of temperature from -100C-100C,
high flexibility(the shear modulus is between 100kPa-3MPa)
very low loss tangent (tan <<0.001)
high dielectric strength (~14V/m)
low glass transition temperature (Tg ~ -125C)
can be made fast

Due to PDMSs properties, it has been used as inverse mould for high-aspect-ratio
microstructures [96, 97].

5.2 Advantages of using PDMS technique

Surface
This technique has a good smooth surface result. Metallic microstructure can be released from
PDMS simply, without using any chemical material which means no damage happens for the
surface of microstructure. This is a good advantage for this method in comparison with the other
methods of removing SU-8.
Cost
This method is a cheap and very simple method and the price of PDMS is also cheap. By a SU-8
master mould, easily we can fabricate a lot of PDMS replica in a short time.
Rapid
This is a rapid method for making high-aspect-ratio structures because this method omits most
part of the fabrication process.
Simple removing
Removing PDMS from microstructure is simpler and easier than removing SU-8 from
microstructure.
Non-clean-room process
Using PDMS does not need any cleanroom environment which decreases the need for special
equipment.


5.3 PDMS problem
Casting PDMS over PDMS is challenging, due to the strong adherence of both layers. De-
moulding is only possible with careful cutting and peeling [1]. A less troublesome, but rather
elaborate, method is the derivatisation of the first PDMS copy by silinisation in order to reduce
adhesive interactions [1]. Thermal aging is widely used in fabrication techniques as it modifies
the hydrophobicity of the PDMS surface [2] and it has been reported that low molecular weight
(LMW) chains are modified so that the surface is able to retain its hydrophilicity [3,4,5].
Two approaches to PDMS double casting have been proposed in this paper. The first method we
investigated was double casting PDMS onto PDMS, and the second method included an extra
step, where the surface of the inverse structure was thermal aged. All techniques are thoroughly
tested by repetitive moulding steps and the quality of the different copies are thoroughly
inspected and applied directly for electroforming [6, 7].
The aim of this technique is to achieve the same results as KMPR photolithography without
using the lithography process while still being cost effective and producing high aspect ratio
microstructures. For the two methods explored, a metal master mould imprinted with the features
was used for the initial mould. This master mould was fabricated by milling an array of holes, 5
mm in depth, into a rectangle aluminium block, 7 mm length, 1 mm in width (Fig. 1). Four extra
pieces of thin aluminium were used as frames for the master mould, where 2 pieces were 70 x 10
mm and 2 were 10 x 10 mm. The frames were attached onto the aluminium block and cleaned
with isopropyl alcohol and deionised water prior to use. Standard PDMS preparation technique
was followed for both methods, where the PDMS (Sylgard 182 Silicone Elastomer Kit, Dow
Corning, USA) was prepared in a 10:1 ratio of base to curing agent. The mixture was then
degassed in a vacuum oven for 15 mins, to remove air-bubbles, and then poured into the
aluminium mould and degassed for a further 10 min. The PDMS filled mould was then cured in
an oven at 70C for 30 min. The solid PDMS structure was then removed from the master mould
producing a structure with inverted features (Fig. 2).


5.4 Experiment 1 PDMS on PDMS casting

Hypothesis:
By applying the PDMS double casting method we are able to create an inverted PDMS replica of
the PDMS master. The idea would be, to be able to easily remove the PDMS replica from the
master. The inverted structure will be used for electroforming.
Method:
The inverted structure was then fitted with the aluminium frames and more PDMS was poured
on top, degassed and cured for 70C for 30 mins.

Aluminium
master mould
Aluminium master
mould with
PDMS, cured at
70C, 30 min
Inverted PDMS
master
Inverted PDMS master
with PDMS, cured at
70C, 30 min
Thermal aged
inverted PDMS
master
with PDMS,
cured at 70C, 30 min
Inverted PDMS master,
thermal aged at 100C,
24-48 hrs
Replica PDMS master
mould and
aluminium master mould
(a) (b)


Figure 5-3 Unsuccessful result after method 1 fabrication technique was used. PDMS
on PDMS casting failed due to strong bonding (inset: zoomed-in image of pins
encapsulated within the PDMS block)
Result:
In the first method, double casting was conducted onto normal PDMS. This method, although
time efficient, was not a successful attempt as the two structures could not be detached from one
another. The result of this method produced a solid block of PDMS, where the features (pins) of
the inverted mould could be seen inside the block (Fig. 3). The visibility of these features
indicates that the two structures did not merge or melt together, despite the difficulty in
disassembling the two structures. This attachment of the two PDMS structures can be attributed
to strong bonding forces at the interface of the two structures [1].

Discussion and Conclusion:
This bonding affinity may be due to the hydrophobic nature of PDMS surfaces, which has been
linked to the migration of low molecular weight (LMW) chains to the surface [3]. This migration
has been suggested to occur within cured PDMS to recover the hydrophobic surfaces by
covering up thermodynamically unstable hydrophilic surfaces [1]. Thus, if a surface of PDMS
were to remain hydrophilic for extended periods of time, the presence of LMW species needs to
be reduced.

5.5 Experiment 2 PDMS on PDMS casting Thermal Aging

Hypothesis:
To be able to create an inverted structure for electroforming by double casting PDMS on PDMS
with an added step of thermal aging. The result would be an inverted PDMS replica of the PDMS
master. The idea would be to be able to easily remove the PDMS replica from the master by
utilising thermal aging as the non-chemical release agent which evidently did not appear using
method 1.

Method:
Following the PDMS procedure aforementioned, the inverted structure was cured further
(thermal aged) at 100C for 24-48 hrs. The hardened structure was then fitted with the
aluminium frames and more PDMS was poured into the frames. The structure was then degassed
for 10 min and then cured at 70C for 30 min. The thermal aging process was repeated two
further times at 90 and 110C.

Result:
Thermal aging is a common technique that is used to modify the surfaces of PDMS structures.
This technique exposes the PDMS to high temperatures for prolonged periods of time. Thermal
aging causes gradual crosslinking of LMW chains and affects the rate of migration of the LMW
species, extending hydrophobic surface recovery from minutes to days. Eddington et al. has
shown that by varying the aging time (from 48 hours to 14 days), it is possible to eliminate the
hydrophobic surface recovery completely. While K. Zikowska et al. has proven that the
optimal temperature was between 90C and 120C. Therefore, the rate of hydrophobic recovery
is dependent on aging times, temperatures and volume. This extra step of thermal aging proved
valuable in our second method as detaching the two PDMS structures was successful. The two
PDMS structures were separated without difficulties and replicated the master mould seamlessly,
i.e. the pins from the inverted/thermal aged structure did not break (fig. 4). The thermal aged
structure was also abled to be used multiple times to create replica moulds with high
reproducibility and without the need for further thermal aging. Different exposure times were
applied in the range of 120C, 100C and 90C.


Thermally aged masters, baked for 24-48 h, were used for replication. It was observed that 24 h
thermal aging was sufficient and provided over 95% replication accuracy; while 48 h did not
significantly improve the results. Therefore 24 h thermal aging was established as the optimal
time for PDMS double casting technique.

Figure 5-4 Successful result after method 2 fabrication technique was used. The
PDMS master was successfully inverted after the thermal aging step was added.


5.6 Experiment 3 Electroplating PDMS with seed layer 20nm
Chromium and 50 nm Nickel

Hypothesis:
By conducting copper electroplating on the PDMS double casting thermal aged sample method,
which has a seed layer (20nm Chromium and 50nm Nickel) we are able to create an inverted
metallic replica of the PDMS master. The idea would be, to be able to easily remove the PDMS
replica from the master.

Method:
The PDMS master became the anode and the copper substrate became the cathode.
The following electroplating parameters were used:

Result:
Substrate not conductive [Figure ..].


Conclusion and Discussion

The result was a nonconductive substrate and this was due to the strong nickel oxide present on
the surface. A way to remove the oxide layer is to use an etchant, however, the result did not
produce a conductive substrate.

5.7 Experiment 4- Electroplating PDMS with seed layer 20nm
Chromium and 250 nm Gold

Hypothesis:
By conducting copper electroplating on the PDMS double casting thermal aged sample method,
which has a seed layer (20nm Chromium and 250nm Nickel) we are able to create an inverted
metallic replica of the PDMS master. The idea would be, to be able to easily remove the PDMS
replica from the master.

Method:
The PDMS master became the anode and the copper substrate became the cathode. The
following electroplating parameters were used:


Result:
Figure . shows a layer of copper on top of the PDMS substrate which indicates that the
parameters used from the method were correct and that the substrate was conductive [Figure
.]. This was because gold unlike nickel does not contain an oxide layer on its surface which
meant that using chemicals to remove the oxide layer would not be necessary and would not
affect the results.

A seed layer of 20 nm chromium and 250 nm gold was evaporated onto the PDMS substrate.
This substrate was then used for copper electroplating. It was observed that the substrate was

conductive and it was found that it could be plated into and a result could be obtained. Therefore,
the seed layer of 20 nm chromium and 250 nm gold on a thermally aged PDMS substrate is
optimal to be plated into.

5.8 Experiment 5 Removing PDMS from metallic structure

Hypothesis
To easily peel off the hardened epoxy resin from the PDMS surface and produce an epoxy mould
with the metal over it.

Method
3M Scotch-Weld Epoxy Adhesive 2216 B/A Gray was used in the experiment because of its two
part room temperature curing with high peel strength and shear strength. The key characteristics
provided by the manufacturer are shown below:


The manufactures data sheet instructions were followed point to point. It specified that:

The metallic surface needed to be clean. This was done by using acetone and
Isopropanol.
Mix both the base and accelerator with a weight proportion ratio of 5 to 7.
Pour the epoxy adhesive over the metallic surface and cure at room temperature for 24
hours.
Peel off the epoxy adhesive mould.

Note: All tests procedures were followed in the clean room.

Result
The epoxy resin was poured on top of the PDMS structure and left to cure at room temperature
for 20 hours as specified by the data sheet. The result after curing show that the epoxy resin
adhered to the PDMS mould, firmly (Figure 2222).
Table 1. 3M Scotch-Weld Epoxy Adhesive 2216 B/A Gray key characteristics.


Tape was used around the PDMS structure to hold the epoxy resin. This is shown below:

After peeling off the PDMS, the result is shown below:

Figure 1. Result after curing for 20 hours.
Figure 2. Result after curing for 20 hours.
Figure 2. Result after peeling off PDMS.

Discussion and Conclusion
The desired outcome was to easily peel off the hardened epoxy resin from the PDMS surface
and produce an epoxy mould with the metal over it however this was far from the case. It was
incredibly hard to peel off the PDMS and it is not possible to remove the excess PDMS from the
holes. However, the epoxy adhesive did successfully remove the metal from the PDMS surface
and this was shown in figure 3.
A solution to fix the peeling problem may be to heat the PDMS/Epoxy moulds before removal.
This may thermally expand the PDMS and we may see a better peeling result. A solution to
producing better results may also be to cure the epoxy resin at higher temperatures as specified
by the manufactures data sheet:

However this remains to be tested.

6 Conclusions and future work

6.1 Conclusions

In this project, a suitably developed KMPR mould by the top side exposure method, followed by
an electroplating process, and a sacrificial method, high-aspect-ratio metallic microstructures
have been fabricated. The parameters of mould fabrication are studied by which a set of
optimized parameters are identified to produce a KMPR mould without any cracks, delamination
and KMPR residue. We need to increase soft-baking time and temperature, post exposure baking
time and temperature and to have relaxation time after soft-baking and before development.
The next step after mould fabrication is to suitably electroplate it. It is found that the only way to
electroplate KMPR mould completely without any pores inside the structure is to use low current
by controlling electroplating parameters such as pH, temperature, agitation and bath
composition. KMPR resist was easily removed from the electroplated metallic structure by using
MicroChems Remover PG (NMP).
Lastly, an alternative in direct method of forming a structure, electroplated plating into the
structure and separating the metal from the structure easily, was studied. PDMS double casting
technique. This process has several advantages and disadvantages as follows:

PDMS on PDMS double casting technique:
This technique has been worked on by researchers, but needs to be studied further. It has several
advantages such as: Fabrication process is simple, rapid prototyping, no lithography technique
required, cost effective, and thickness can be controlled. A few problems with this technique
remain: Cannot simply cast PDMS on PDMS due to cross-linking and sufficient time and effort
is needed in order to produce an inverted structure with high aspect ratio dimensions.

PDMS on PDMS double casting with thermal aging technique:
This technique has also been worked on by researchers. It has same advantages as PDMS on
PDMS double casting. A problem with this technique remains: Thermal aging of at least 24 h is
needed in order to produce an inverted PDMS structure. Another inherit problem is that when the
mould is plated into, it is not able to be easily separated from the metal.

6.2 Recommendations for future work

There are several aspects in this project that can be investigated in the future:
In mould fabrication, we just spin-coat one layer of KMPR and able to achieve a maximum of
150m thickness KMPR mould. But there is potential to have thicker mould, the suggestion for a
method to have a thicker mould by spin coating KMPR after each electroplating process can be
achieved.

In mould fabrication, we created a structural dimension of 30m. There is potential to increase
and decrease this amount to produce a high aspect ratio greater than 1.

The evaporation of gold onto the silicon substrate produces a conductive substrate but does not
provide a reliable option when trying to remove KMPR photoresist from the silicon substrate
with the gold evaporated on it. An alternative seed layer such as titanium could be investigated to
provide the same conductive layer as gold.

In the part of fabricating structure by PDMS technique and electroplating into the mould, it
was found that removing the metal from the PDMS was not easy. Parameters to improve this
technique could be studied further to that the process is smooth.


References

First Draft 1

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

First Draft 1

Hochgeladen von

Copyright:

Verfügbare Formate

1 Kashka Irani - Fabrication of High-Aspect-Ratio Metallic Micro-structures Using Photolithography

FABRICATION OF HIGH-ASPECT-RATIO METALLIC

Das könnte Ihnen auch gefallen