PLASMA BONDING OF POLY (DIMETHYL) SILOXANE AND GLASS SURFACES AND ITS APPLICATION TO MICROFLUIDICS by SHANTANU BHATTACHARYA, B.E. A THESIS IN MECHANICAL ENGINEERING Submitted to the Graduate Faculty of Texas Tech University in Partial Fulfillment of the Requirements for the Degree of MASTER OF SCIENCE IN MECHANICAL ENGINEERING Approved December, 2003
107
Embed
GLASS SURFACES AND ITS APPLICATION - Semantic · PDF fileGLASS SURFACES AND ITS APPLICATION ... 4.6 Schematic of PDMS-PDMS and Glass PDMS interface 47 ... MEMS equipments
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
PLASMA BONDING OF POLY (DIMETHYL) SILOXANE AND
GLASS SURFACES AND ITS APPLICATION
TO MICROFLUIDICS
by
SHANTANU BHATTACHARYA, B.E.
A THESIS
IN
MECHANICAL ENGINEERING
Submitted to the Graduate Faculty of Texas Tech University in
Partial Fulfillment of the Requirements for
the Degree of
MASTER OF SCIENCE
IN
MECHANICAL ENGINEERING
Approved
December, 2003
ACKNOWLEDGEIVIENTS
I would like to thank my thesis advisor Drs. Shubhra Gangopadhyay and
Jordan Berg for their support, encouragement and valuable guidance throughout
this work. I would not have been able to finish this work without their guidance
and moral support. I am grateful to Dr. Gangopadhyay, La Pierre Chair Professor
at the University of Missouri at Columbia, for giving her valuable time for the
thesis, notwithstanding her extremely busy schedule. I would also like to thank
Dr. Mark Holtz for his valuable comments and suggestions.
I would also like to thank Dr. Arindom Datta of Jack Maddox laboratory
for his help and suggestions throughout this endeavour. Special thanks are due to
all my peers at Jack Maddox Laboratory for being with me and helping me at all
times throughout my stay here.
I have deep gratitude for the help and guidance that I was blessed with
from my family here Finally, I would like to pay my deepest regards to my
parents for bringing me up with values that always help me to succeed in all
endeavors of my life. I dedicate this thesis to my parents.
11
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ii
LIST OF TABLES vi
LIST OF FIGURES vii
CHAPTER
1. INTRODUCTION 1
1. IBackground 1
1.2Processing using soft-lithography 6
1.3Wafer level bonding techniques 6
1.3. IDirect bonding 7
I.40bjectives 7
2. INTRODUCTION TO INDUCTIVELY COUPLED PLASMA AND CONTACT ANGLE 9
2.1 Introduction 9
2.2 Inductively coupled high density plasma system 10
2.2.1 Trion Inductively coupled high density plasma reactor 11
2.3 Contact angle 13
2.3.1 Sessile drop technique 15
2.3.2 Physical reasons for the drop spread 16
3. LITHOGRAPHY BASED POLYMER MOLDING PROCESSES 18
3.1 hitroduction 18
3.2 Materials used 19
111
3.2.1 SU-8 photoresist 19
3.2.2 Poly (dimethyl) siloxane, PDMS 22
3.3 Equipment used 23
3.3.1 Photograph and description of contact angle setup 25
3.4 Glass wafer casting template construction 26
3.4.1 Mask design 26
3.4.2 Wafer cleaning and photoresist coating 27
3.4.3 Exposure of resist coated wafers 31
3.5 Fabrication of blister 36
3.5.1 PDMS molding 36
3.5.2 Imparting hydrophilic properties by exposure to oxygen plasma 37
3.5.3 Fitment of Inlet/Outlet port to the blister 38
4 RESULTS AND DISCUSSION AND CONCLUSION 39
4.1 Introduction 39
4.2 Surface roughness versus bond strength 41
4.2.1 Lapping process 41
4.2.2 Roughness measurement 43
4.3 Measurement of contact angle and bond strength 48
4.3.1 Effect of chamber pressure variation 49
4.3.2 Effect of RIE power variation 52
4.3.3 Effect of time of exposure 55
5 DESIGN CONSIDERATIONS FOR A MICRO REACTOR 64
5.1 Introduction 64
IV
5.2 Micro-mixer mechanisms 68
5.2.1 Passive micro-mixers 69
5.2.1.1 Lamination mixers 69
5.2.1.2 Injection micro-mixer 71
5.2.1.3 Valve micro-mixer 72
5.2.2 Active micro-mixer 73
5.2.2.1 Mixer with pumped fluid inlets 73
5.2.2.2 Ultrasonic mixer 74
5.2.2.3 Magneto hydrodynamic mixers 74
5.3 Experimental (Design, Fabrication and testing) 75
5.3.1 Design 75
5.3.2 Fabrication 77
5.3.3 Testing 79
5.4 Results and discussion 81
6 CONCLUSIONS AND RECOMMENDATION 84
REFERENCES 86
APPENDIX A: Title 93
APPENDIX B: Title 94
LIST OF TABLES
3.1 Chemical composition of SU-8 19
3.2 Spin speeds versus thickness 30
3.3 Pre-exposure or soft bake parameters 30
3.4 Post-exposure bake parameters 31
3.5 Thickness measurements on the molds in Figures 3.5, 3.6, 3.7 and bUster-—35
4.1 Various roughness parameters obtained by Vision 32 44
5.1 Ranges of "Re" values and characterization of the nature of flow 66
A. 1 Important physical properties of SU-8 (Photoresist) 93
VI
LIST OF FIGURES
1.1 Size ranges of micro-fluidic devices 3
1.2 Plot between analyte concentration and sample volume 4
2.1 Schematic of an ICP tube 10
2.2 Schematic of the Trion chamber 12
2.3 Schematic of a contact angle 14
2.4 Infinitesimal expansion of a drop on a surface 14
3.1 Structure of SU-8 molecule 20
3.2 Mechanism of crosslinking of SU-8 21
3.3 Structure of Poly (dimethyl) siloxane 22
3.4 Photograph of the contact angle setup used for experiment 26
3.5 BUster mask 27
3.6 SU-8 mold of thickness 450 microns using multi-layering and single
exposure 33
3.7 SU-8 fabricated on silicon with single layer, single exposure and resist
thickness 150 microns 33
3.8 SU-8 structure of unequal height and its microscopic image fabricated with
multi-layer and multi-exposure 34
3.9 Masks for 1st and 2nd exposure 35
3.I0Aluminum holder for the SU8 master 37
3.11 Schematic of a bUster assembly. 38
4.1 (a~c) PDMS to PDMS failure 40
4.1 (d~f) PDMS to Glass failure 40
4.2 Lapp size used (red bars) versus measured roughness (green bars) 44
4.3 Images of surfaces with different roughness values measured by NTl 100-45
4.4 Bond strength variation of PDMS PDMS with surface roughness 46
4.5 Bond strength variation of glass PDMS with surface roughness 46
4.6 Schematic of PDMS-PDMS and Glass PDMS interface 47
In the first set of experiments, the bond strength is measured by varying
chamber pressure and keeping Inductively coupled power (ICP) at 150 watts,
Reactive ion etching (RIE) power at 20 Watts, Oxygen flow rate at 20 seem and
the time of exposure at 30 seconds. Bond strength is found to increase with an
increase in chamber pressure. In the second set of experiments the RIE power is
varied at a constant chamber pressure (1000 m Torr for Glass PDMS and 700 m
48
Torr for PDMS-PDMS) and all other parameters same as before. In the third
experiment the time of exposure is varied and remaining parameters kept same as
before. The variation in RIE power and time of exposure indicates maximum
bond strength at a certain optimum value of power and time. This observation is
true for both PDMS to PDMS and PDMS to glass bonding. A theoretical
calculation of bond strength has been made for PDMS/PDMS in Appendix B
using Maxwell's equation. The following subsections of this chapter describe
these effects in details.
4.3.1 Effect of chamber pressure variation
Figures 4.8 and 4.9 illustrate plots for contact angle and bond strength
versus chamber pressure for a fixed RIE power and exposure time for glass-
PDMS and PDMS-PDMS bonding respectively. Bond strength is measured as the
value of pressure at which interfacial separation of the pressurized blister starts
occurring. The maximum bond strength obtained for glass to PDMS bond is 72-
psi. This corresponds to a contact angle of less than 5 degrees.
There is a decrease in bond strength below 100 mTorr pressures. Normally,
at a chamber pressure of 100 mTorr or less the plasma etching becomes highly
directional and anisotropic [19]. The high level of anisotropy in etching leads to a
damage in the siloxane backbone instead of etching the surface methyl groups.
The change in contact angle measured on the glass surface does not show a
sporadic decrease like that of PDMS for reason explained later. For PDMS-PDMS
bonding, the maximum bond strength is found to be 58 psi. The corresponding
49
80
70-
60-
50-
40-
30
20
10
• Bond strength (psi) • Contact angle (deg) ^
200 400 600 800
Chamber Pressure (mTori)
1000
Figure 4.8 Plot of contact angle and bond strength with chamber pressure (range of variation =20mTorr) for glass-PDMS bonding
60-
50-
40'
g> soil) to •g 20-o m
10-
Bond Strength (psi) Contact angle (deg)
— I — 100 200
— I — 300
— I — 400
— I — 500
— I — 600
— I — 700
Chamber Pressure{mTorr)
60
50
O) 40 0)
O) 30 5
Co
20 g O
10
Contact angle Below 5 deg
800
Figure 4.9 Plot of contact angle and bond strength with chamber pressure (Range of variation =20mTorr) for PDMS-PDMS bonding
50
contact angle is found to be less than 5 degrees. The bond strength curve in the
low pressure region(<100 mTorr) is similar to the curve for glass to PDMS
bonding. In the high-pressure region (>100mTorr) there is a gradual increase in
the bond strength and decrease in contact angle with an increase in pressure. The
behavior of the data in the high-pressure region can be explained in the following
way. As the chamber pressure increases the mean free path of the gas molecules
reduce, and the plasma becomes more and more isotropic. This homogeneity
drives the plasma to act uniformly over the substrate. Also with an increase in
pressure the sheath of charged particles formed near the electrode move closer to
the substrate [19]. So, newly formed ions near this sheath have smaller distances
to travel before striking on to the substrate resulting in less momentum transfer.
Less energetic oxygen ions generated in this way remove methyl groups from the
surface without damaging the material.
One more important behavior of the trend is reflected at pressures below
100 mTorr where the contact angle of PDMS rises faster than that of glass. This
can be explained by considering, that glass is more rigid structurally. So at a low
pressure and greater mean free paths when the ionic momentum transfer increases
it is sufficient to damage the fiexible siloxane backbone in PDMS but not
sufficiently strong to affect the sturdy surface structure of glass. Thus the contact
angle in the case of glass does not increase so much at lower pressures as in case
of PDMS. The behavior of the bond strength is by and large reverse to that of
contact angle, which fits our theory very well. However, the point corresponding
to the pressure value of 50 mTorr [Figure 4.9] shows a sudden reduction in the
51
bond strength value which may indicate an extraordinary damaging of the PDMS
structure and thus a substantial loss of surface silanol bond density. Thus, in this
case although the alteration of glass surface is relatively less but the damage to
PDMS surface causes a huge decrease in bond strength. All measurements have
been taken for a constant value of PDMS substrate thickness (2.5 mm) although
the effect of changing thickness can be estimated. If thickness of the substrate is
reduced, then at lower pressures, the ionic sheath near the electrodes is nearer to
the substrate surface causing the ions to hit the surface with less impact than in
the case of thinner substrate. Thus, the surface damage at lower pressure is lesser
for thicker samples and the bond strength value should rise up slightly at lower
pressure. However, at higher pressure, there may be a possibility of the surface
rising above the dark space and thus facing negligible plasma activity. So thick
ness cannot be abruptly increased
4.3.2 Effect of RIE power variation
Figures 4.10 and 4.11 show plots of contact angle and bond strength for
variation of RIE power. The data shows an interesting trend wherein the bond
strength peaks at 20-watt RIE power for glass to PDMS and PDMS to PDMS
bonds. The peak value of bond strength here in both cases are similar to those
obtained before, with 68 psi for PDMS-Glass and 54 psi for PDMS-PDMS. The
contact angle trend follows an inverse behavior to bond strength. The contact
angle curve dips down to below 5 degrees at 20 watt RIE power level and then
52
70-
60-
^ 50-'</) a.
£ 40-OJ c 2 30-w
• D
§ 20-m
10-
0-
0
g
1* •
' 1
20
• Contact Angle (deg) D Bond strength (psi)
^°~~~~~----.c
• ^ ^ - ^ m^-^^
• ^
40 60 80 100 120
RIE Power (Watt)
~~~~~---5
• ,
140
-
_
.
-
160
70
60
50 S •D " • " ^
O 40 'i
CS
30 iS c o o
20
10
Contact angle below 5 deg
Figure 4.10 Plot of contact angle and bond strength with RIE power( range of variation= 2W) for glass-PDMS bonding
60-
50-
in n .c
c (i>
r« T3 c o m
40
30
2U
10-
0-
- l — ' — I — I — I — ' - I — I — I — ' — I — • — I — ' -
- Bond strength (psi) Contact angle (deg)
60
50
1-40 §>
<i>
1-30 o)
20 I c o
h i o "
Contact angle Below 5 deg
0 20 40 60 80 100 120 140 160
RIE power (W)
Figure 4.11 Plot of contact angle and bond strength with RIE power (Range of variation=2W) for PDMS-PDMS bonding
53
goes up on either side of this value. Simultaneously, the bond strength goes down
as the contact angle goes up.
This behavior can be explained by considering the plasma behavior for
various bias levels dictated by RIE power. At low power levels, the kinetic energy
of ions incident on the substrate reduces. This coupled with the ambient high
chamber pressure leads to a large reduction in the number of reactive ions on the
substrate. This is so because a lower power level reduces the electron acceleration
within the plasma environment thus leading to a reduction in the radical density.
Thus less number of active sites formulate on the substrate surface after
etching in such a plasma environment, which leads to a reduction in surface
bondage. The ions tend to thus graze on the surface of the substrate without
producing much chemical or physical change of the surface. The reverse
behavior at higher power levels suggests an increase in the ion bombardment.
Thus the Si-O-Si, whose dissociation energy (445KJ/mol) is much higher
than the Si-C bond dissociation energy (306KJ/mol.), is affected resulting in
damage of the overall uniquely fiexible Siloxane backbone [19].
Contrary to the chamber pressure variation case, one important observation
in this trend is a general homogeneity in variation of the contact angle and bond
strength in both glass-PDMS and PDMS-PDMS bonds. This can be attributed to
the constancy in the chamber pressure due to which directionality never arises in
the etching. This helps in preventing the differential nature of trends in both cases
by eliminating the anisotropicity levels as had happened in the low chamber
pressure case.
54
4.3.3 Effect of time of exposure
The time of exposure has a similar trend as RIE power [Figure 4.8(a) & (b)].
The bond strength peaks in this case for an exposure time of 20 sees. The values
of strengths are similar to that obtained in the earlier cases with a rise in contact
angle and subsequent fall in bond strength at a longer or shorter exposure time.
The least contact angle value at highest bond strength is again less than 5 degrees.
One possible explanation can be obtained from Owen and Smith's [56]
investigation on the PDMS surface after high RF power and longer treatment.
The progressive oxidation of the surface leads to the formation of an
extremely brittle silica layer on the surface. Owen and Smith have clearly seen
cracking on the surface under scanning electron microscope (SEM). They
mentioned that less harsh, lower RF power, and shorter treatment times produced
un-cracked surfaces with a layer of Silica [SiOx], which retards the migration of
low molar mass molecules from the bulk of the structure [Fig. 4. 12 and 4.13]. As
this layer is exposed longer in a plasma environment this layer cracks and
promotes transport of low molar mass molecules to the surface, which covers the
oxidized layer. Another possible explanation of this trend can be obtained from
the work of Hillborg and Gedde [57] who suggest that prolonged UV exposure in
a plasma environment makes the surface undergo fast hydrophobic recovery.
They have mentioned about the various transformations, which take place on
prolonged UV exposure [Figure 4.12 and 4.13]. The longer exposures to oxygen
plasma and UV of a treated PDMS surface results in the formation of excessive
surface silanol concentration [58].
55
70-
60
50 Q.
40
i 30-tn
T3 g 20
CD
10-
0-— I —
10
a Bond Strength (psi). Contact ngle (deg)
— I — 40 20 30 40 50
Time of exposure (Sec)
60
70
60
50 o>
40 ra
30
- 20
10
Contact angle Below 5 deg
Figure 4.12 Plot of contact angle and bond strength with time of exposure for Glass-PDMS bonding
60-
50-
•tf)
a. ,c r a> (0
T3 c o m
40
30
?0
10-
• Contact angle (deg) D Bond strength (psi)
- 60
—1 1 1 1 1 1 I ' I
10 20 30 40 50
Time of exposure (Sec)
60
- | Contact Angle Below 5 deg
Figure 4.13 Plot of contact angle and bond strength with time of exposure for PDMS-PDMS bonding
56
As the silanol bond density increases it results in chemical transformation on the
surface known as polymer surface chain scission reactions. In such a situation a
marked reduction in the number of surface silanol bonds occur by back biting
scission reactions [Figure 4.17, la] and a physical surface cracking and a gradual
migration of the mobile, low molar mass PDMS oligomers to the surface [Figure
4.14 and 4.15] [56]. If chemical transformation takes place by scheme 2 and 3 as
shown in Figure 4.16 the scission reactions occur in a variety of ways like [Figure
4.17,1b, n, and III].
In all cases due to a rearrangement of the surface molecules in short and
long cyclical chains, the surfaces get cracked and damaged at several places
which promotes surfacing out of the low molecular mass oligomers and thus a
reduction in the hydrophilicity [58].
A general trend was plotted using all values of contact angle and bond
strength for all different plasma parameters [Figure 4.18 and 4.19]. The plot gives
a universal increase in bond strength with decrease in contact angle for both glass-
PDMS and PDMS-PDMS cases for variation of RIE power and pressure. The
time of exposure if plotted doesn't follow this universal trend on the longer
exposure side. Although the bond strength reduces with increase in contact angle
for longer exposures but still the slopes of the graph on both sides of the optimum
time are dissimilar. This is true for both glass-PDMS and PDMS-PDMS bonding.
Thus it is clear that the trend shown by the universal curve works till and until the
surface is not damaged severely causing the cracking of the surface. There is a
relatively large spread in the data points in case of glass/ PDMS bonding. This
57
Thin Silica Layer
PDMS Bulk
Figure 4.14 The hydrophilic silica like surface retarding the bulk molecules
Cracks on the Silica Layer
Short chain oligomers coming p from the bulk
PDMS
Figure 4.15 Cracking of the silica like surface promoting the low molar mass molecules to rise
58
Plasma
UV ^fl' CHjOOH
v.si_o • ~vsi_o-^ •^s i—o + OH I o, I V I
CH, ' ' CH, CH,
CH,
Scheme (1)
CHj I UV
" ^ S i O • -^ + CH
OOH -OH
Q
CH3 ^ CH3
o o
^ • ^ S i — o S i — o
CH3
Scheme 2
CEL
I / / 2 ^ S i O • HjC-Si-CH 2CH 2-Si- H3C
I o o CH3
Scheme 3
Figure 4.16 Various schemes of surface modification in Silicone rubbers on corona and UV radiations
59
O — Si
OH -• O H +
lA
IIA
II
Si
0 +
Si
0 III
Figure 4.17 Mechanisms for chain scission reactions (la) 'Back biting' yielding short cyclics, (lb) random formation of short cyclics along chain, (II) random
formation of high molar mass cyclics along the chain (III) Intermolecular chain scission.
60
Bond strength vs Contact angle for various plasma parameter variation (Glass-PDMS)
30 1
^25
S20-
!15
l i o c o
Below
I
• Chanter pressure"
A RE power
•
A •
5 deg i i i i i i
25 35 45 55 65 75 85
Bond strength (psi)
Figure 4.18 Universal trend for variation of chamber pressure and RIE power for Glass-PDMS bonding
Bond strength vs Contact angle for various plasma parameter
variation (PDMS-PDMS)
330 <u "5)25 1=
:!2o o •0 , _ - l b o o 10
5 Below 5 degree
• Chamber pressure
A RIE power
•
A A
• • • A •
40
Bond strength (Psi)
80
Figure 4.19 Universal trend for variation of chamber pressure and RIE power for PDMS-PDMS bonding
61
may be attributed to the fact that the plotted contact angle on glass may not be the
right parameter influencing the bonding strength.
One more explanation of this effect can be the reconstruction of the surface
dangling bonds on the surface of treated PDMS with progression of time in the
plasma environment, which is full of oxygen free radicals.
This reconstruction comes from the dangling bond on Si combining with
Oxygen and another dangling bond forming Si-O-Si cross-linkage over the whole
surface [Fig 4.13]. This cross-linking causes the surface to develop a high contact
angle against water [58].
CH,
CH^-—SI O-
CHn
CH3—Si-I
CH.
CH,
CH,
CH3 —
CH,
^ CH,
CH, CH,
v.,0 Si O^v-Si——O Si-
CH, CH-!
V
Oxygen Plasma
I CH, CH, CH,
-CH,
Dangling onds
^ i ^ ^^JO—SI O v ^ i O Si CH
\7
Long Time of Exposure Contact angle more
due to Si-OsSi
Figure 4.20 Bond reconstruction with increased time of exposure
62
Depending on the plasma properties, the surface that influences the bond strength
could be glass, PDMS or both.
Another important observation is the loss in the hydrophilicity when the
surface is kept outside after exposure to oxygen plasma. This can be attributed to
the reconstruction of the dangling bonds by molecular oxygen [58] and various
organic contaminants present in the atmosphere after placing the exposed surfaces
in open air.
63
CHAPTER 5
DESIGN CONSIDERATIONS FOR A MICRO REACTOR
5.1 Introduction
As mentioned in Chapter 1 the field of micro-fluidics has diverse
applications in chemistry and life sciences. Some examples of applications
include: measurement of the reaction gradient, screening of chemicals for drug
discovery or DNA synthesis, enzyme and substrate reactions and high
temperature and light induced reactions [1]. The advantages of moving to the
micro-scale are seen in terms of cleaner reactions due to frequent flushing out of
products, a very homogeneous temperature distribution for endothermic reactions,
higher safety levels particularly in case of hazardous reactions by reduction in the
volumes, and increased sensitivity [60]. In most of the chemical and biomedical
analysis, a sample solution is to be tested with a reagent [61]. Two or more than
two solutions are to be mixed, to make such reactions physically possible. Most of
the macro-scale mixers utilize the features of turbulent flow (fluctuating and
agitated) causing the formation of eddies and vortices across a large range of
length scale [62]. Mixing on the micro-scale relies mostiy on inter diffusion of
various species participating in the mixing process. This is due to the laminar
nature (smooth and steady) of flows at the micro-scale.
Although the flow behavior of fluids has been widely studied both
mathematically and physically, still no general analysis of fluid motions exist
[63]. The reason for this is that there is a profound and vexing change in fluid
behavior at moderate Reynolds number, which, is the ratio between the viscous
64
forces to the inertial forces. Mathematically, the Reynolds number is represented
by
Re = pvd
(5.1)
where p is the mass density of the fluid, v is the velocity, p is the fluid
viscosity, d is the hydraulic diameter given by
A (5.2)
Here, P = Perimeter wetted by the fluid and A= Area wetted by the fluid.
At certain moderate values of Reynolds number the flow ceases to be
laminar and becomes turbulent. This process is called transition to turbulence.
Schematically this is represented in Figure 5.1. (a,b and c)[5].
Small natural disturbances damp quickly
_JWL ^ -
Intermittent bursts of turbulence
Continuous turbulence
(a) (b) (c)
Figure 5.1 Three regions of viscous flow (a) laminar flow (Re low) (b) Transition flow (moderate Re) (c) turbulent flow (high Re)
Table 5.1 sorts out the flows into various categories with a brief
description of its nature including Re dependence [63].
65
Table 5.1 Ranges of 'Re' values and characterization of the nature of flow
S.No.
1
2
3
4
5
6
Range of 'Re'
0<Re<l
l<Re<100
100<Re<10^
1000<Re<10^
10VRe<10*'
10VRe<oo
Description
Highly Viscous, laminar "creeping motion"
Laminar, Strong Reynolds Number dependence
Laminar, Boundary layer theory is useful
Transition to turbulence
Turbulent, Moderate Reynolds Number dependence
Turbulent, Slight Reynolds Number dependence
Micro-scale mixing rates strongly depend on the flux of diffusion O
mathematically expressed by
ax (5.3)[63]
where D is the diffusion coefficient in m^ / sec and c is the species
concentration in kg/ m l The diffusion coefficient is inversely proportional to the
fluid viscosity at a constant temperature.
Figure 5.2 [4] represents a range of coefficient for various states of
matter.
The average diffusion time x is given by:
d' T =
2D
where d= Length of the mixing path
D= Diffusion coefficient.
(5.4)
66
IQ 10 10' 10' 10- 10" 10°cm^/sec
i L
Solid
i L
Polymers glasses
t t Liauid Gases
»
Figure 5.2 Diffusion coefficient range
Because of their small sizes, the micro-mixers decrease the diffusion time
significantiy. In general, fast mixing can be achieved with smaller mixing path
and larger contact surface area. The smaller mixing path is automatically taken
care off due to the micro-scale. However, area of contact between the interacting
fluid streams can be manipulated to promote or disable the mixing.
In cases of small channel dimension, the fluid interaction level with the
channel walls finds prominence over that with other molecules. In such cases the
diffusion occurring is the Knudsen diffusion [66] characterized by a
dimensionless quantity signifying the ratio between mean free path of the
molecules and channel size [Eq. 5.5]
X Kn =
Dh (5.5)
where Dh is the hydraulic diameter and A, is the mean free path, Kn is the
Knudsen number.
67
As X in case of liquids is very low in the Angstrom range [4], Knudsen
diffusion is not important for liquids. However, this becomes vital in the study of
gas flows.
5.2 Micro-mixer mechanisms
Mixers are categorized into two broad categories based on their
architecture. The first category of mixers with non-moving parts is known as
passive or static mixer category. The second kind known, as active or dynamic
mixers comprise of moving parts, which are used to manipulate, or control
pressure gradients in mixing area [67]. Turbulence and mechanical agitation that
are the main causes of mixing in a variety of length scales has a very less say in
determining mixing at micro scale. This is primarily due to the very low values of
Reynolds number, typically between 10 and 0 [4] which, develops a totally
laminar flow regime. In such cases, all mixing is diffusional. The three general
requirements of any micro mixer are:
• Smaller device size,
• Compatability with complex systems,
• Minimum mixing time.
The mixing time can normally be minimized by reducing the path length
[Eq. 5.4] and increasing the mixing surface. However, there is an extent up to
which lengths can be reduced after which, some physical factors like particulate
matter in fluid, high throughput and high pressures cannot tolerate any further
reduction in size. After such lengths have been reached, mixing can be enhanced
68
by splitting the flow streams into n different sub-stream and rejoining them again
in a single stream. Such a mechanism for mixing can be normally found in
passive mixers.
5.2.1 Passive micro-mixers
The passive micro-mixers do not have any moving parts. They are
categorized further into lamination mixer, injection mixer and valve mixers.
5.2.1.1 Lamination mixers
They split and laminate fluid layers or segments thus accelerating the
mixing process. The lamination mixers can be further subdivided into parallel and
sequential lamination. Parallel lamination is a simple concept and is most suitable
for planar mixing systems. Figure 5.3 [4] shows the schematic of a parallel
lamination with two splits.
Fig. 5.3 Mechanism of a parallel lamination mixer
69
The streams split into n sub-channels and then recombine back into a
single stream. This way the mixing time decreases by a factor of n ( Eq. 5.6) [9].
The disadvantage of such mixers is the occupancy of a large chip area.
(5.6)
where, Tnew is the diffusion time after splitting and x is the ordinary diffusion time.
1"'stage
(a)
2" * stage
(b)
3'^ stage
(c)
nzD
2"* stage
1''stage
(d)
Fig. 5.4 1'*, 2"'^, and 3"* stage of mixing and the 3 dimensional representation of a sequential mixer with 2 different fluid inputs
70
In sequential mixers the joined stream is split into 2 channels [4] and recombined
to form 2" layers in a way as shown in the schematic [Figure 5.4]. The time of
mixing is 4""' times faster [Eq. 5.7]
mew X ,n-\ (5.7) [68]
This form of mixing essentially occurs in 3 dimensional fluidic
structures. There are several T and Y mixers in the lamination micro mixer
category. As the T mixers have typically two laminae, it requires long channel
length because of shorter mixing times. [69] Sometimes in order to break the
mixing stream, fluid pulsing is provided resulting in intermittent supply of both
fluids.
5.2.1.2 Injection micro-mixer
Injection micro-mixer works by splitting one of the streams into several
sub-streams and injecting them into a second one. The nozzles meant for injection
create many micro- aggregates of the split fluid into the continuous one. This way
the contact surface increases, thus promoting mixing. Figure 5.5 shows a
schematic of an injection mixer. The two fluids A and B enter the various layers
of the device. The interface has a set of 400 micro-nozzles etched in silicon [70].
The liquid B gets split up into micro-aggregates and gravity causes these
aggregates to intermix in the stream of A.
71
• ^ :r
Fig. 5.5 Injection mixer
5.2.1.3 Valve micro-mixer
In this type of mixer, a passive valve is used for releasing one of the two
fluids into the stream of the other. The valve is actuated by capillary forces of the
second stream or external pneumatic forces. This way mixing occurs due to inter-
diffusion of both streams. A schematic of valve micro mixer is shown in Figure
5.6 [4]. It shows two fluids A and B mixing with the action of passive valve C
ry 1 ] Passive valve C Mixing Chamber
B A+B B A-hB
Figure 5.6 Valve micro-mixer
72
5.2.2 Active micro-mixer
Active mixers use dynamic parts to agitate the fluids and thus promote
mixing. The moving actuators can be externally operated pumps or valves or
some kind of energy source like acoustic or magnetic sources. Primarily, the
active micro-mixers can be categorized into:
• Mixers with pumped fluid inlets,
• Ultrasonic mixer,
• Magneto-hydrodynamic mixers.
5.2.2.1 Mixer with pumped fluid inlets
This design uses pumps to actively improve the mixing of fluids [72]. As
low Reynolds number flow is characteristically symmetrical and reversible,
mixing is usually not possible by external stirring. However, by using a chaotic
flow field generated by external pumps two fluids can be mixed A chaotic flow
field is achieved by a set of two pumps connected via source and sink to the
mixing chamber. The mixing chamber is filled with fluid A at the top and B at the
bottom [Figure 5.7]. After several cycles of pumping through the chamber
complete mixing is attained.
73
(a) (b)
Fig. 5.7 Chaotic mixer (a) Two liquids are introduced into the chamber (b) Complete mixing after several pump cycles.
5.2.2.2 Ultrasonic mixer
Such a mixer uses acoustic streaming to stir the fluid. Fluids are mixed
rapidly in a piezoelectric pump chamber by ultrasonic vibrations that are induced
from an external piezoelectric actuator in contact with the chamber wall [Figure
5.8] [73]. Mixing Chamber
^%1 €1
Fig. 5.8 Ultrasonic mixer
5.2.2.3 Magneto hydrodynamic mixers
In this type of mixers external magnetic or electric fields can be used to
mix fluids. The principle of this type of mixer is based on Lorenzt force acting on
a conducting solution.
74
F = IxBw (5.8)
where I is the electric current, B is the magnetic flux density and w is the distance
between the electrodes.
This body force stirs the fluid. Using complex electrodes patterns and
switching schemes, complex mixing patterns can be generated. [74]
5.3 Experimental (Design, fabrication and testing of a micro-mixer)
5.3.1 Design
The project of micro-mixer design was carried out in Jack Maddox
laboratory at Texas Tech as a curriculum requirement. Several student pilot
groups were formulated, and they designed, fabricated and tested some micro
mixer designs under a set of design constraints provided ab-initio. The remaining
part of this treatise will focus on the various designs considered by the student
pilot groups and mixing results found after physical testing of the designs 5.3.1
followed by calculation of the Reynolds number and a qualitative study of the
flow behavior.
The project intends to produce a "T" mixer with two time varying inlets
entering through the arms of a "T". The flow sources are elevated at about 25 cm
above the device, resulting in approximately 2.45 Kpa of static pressure head. The
pressures driving the two flows are nominally equal, but may be modified through
the two independently actuated pneumatic valves, one for each source line.
75
(a) Tube bank (b) Triangular type
(c) Figure eight (d) Serpentine channel
Figure 5.9 Different mask designs [Acknowledgements, MEMS 1 Class, 2002]
These valves are controlled by a computer operated Lab-view code. The two
fluids meet at the stem of the "T", and flow from there into the mixing chamber or
the structure to be designed, which also serves as a mixer. Adobe illustrator is
used for defining the various designs in black and white full-scale layout. This is
then developed using a 3200 dpi printer. The white areas used for exposing the
spun on photo resist comprise of the flow channels and the dark areas used to
mask the light define the areas where the photo resist should come out. The
following designs are planned:
1. Tube bank type design [Figure 5.9 (a)],
2. Triangular type design [Figure 5.9 (b)],
3. Figure eight type design [Figure 5.9 (c)],
4. Serpentine channel design [Figure 5.9(d)].
76
5.3.2 Fabrication
The fabrication starts with the pattern construction using SU8 2075
. The device is planned in two layers of PDMS and one layer of glass. The two
PDMS layers contain the fluidic channels and the pneumatic valves. This is
shown in Figure 5.8 for a simple T mixer.
The ultra-thick negative photo-resist SU8- 2075 is spun on a piranha cleaned
2.5 inches diameter glass wafer to a depth of between 100 and 225 microns using
the procedure in section 3.4.2. The thickness can be varied by controlling the spin
speed. The SU8 is patterned lithographically, by selective exposure to a UV light
source using the masks.
Glass Wafer
PDIVIS layer cont£iining device
PDIVIS layer containing pneumatic lines
Fig. 5.10 Device planning in three layers
77
Upon such exposure, and subsequent heating, the SU8 forms extensive
cross-linking and becomes extremely resistant chemically. The areas that are
screened by the black portions of the mask do not crosslink, and are easily
dissolved by a chemical developer, which does not affect the exposed areas. Thus,
a negative of the desired device is obtained, with SU8 features defining the
channels in the portions the mask was transparent and plain glass in a portion that
was dark. This negative is used to cast a device out of the silicone elastomer poly
(dimethyl) siloxane (PDMS) (GE silicones RTV 615). The procedure followed for
this is mentioned in 3.5.1. The edge effects [Figure 5.11] are carefully removed
without damaging the engraved structures.
Molding tool
PDMS layer
Edges due to f .surface tension
Glass wafers
Bonding Improper
(b)
Figure 5.11 Formation of Edge effects due to unbalanced forces of cohesion and adhesion between PDMS and the aluminum plate and its effect on bond quality
If edge effects are not removed, the bond quality is affected by the
induced non-planarity [Figure 5.11 (b)]. Both the channel layer and the blister
78
layers are made in this way. Finally, three 0.75mm holes are drilled into the glass
wafers, one inlet for each fluid and one combined outlet. This glass wafer serves
as a support structure for the device and also helps in attachment of the fluidic
ports. The PDMS fluidic layer and the glass support plate are oxidized in an
Oxygen plasma (Trion Inductively coupled plasma). Immediately upon removal
from the chamber the two pieces are pressed. After bonding process is complete,
the same process is repeated with glass-PDMS. Stainless steel tubes with an
outside diameter of 0.3 mm (McMaster Carr) are inserted into the pneumatic layer
to provide attachments to an off-chip compressed nitrogen gas tank. Barbed
polycarbonate tubing connectors are epoxied to the outside surface of the glass
wafer over the drilled holes to provide access for the mixing stream and an outlet
for the product. The inlet connectors are connected with silicone tubing
(McMaster-Carr) to two bottles containing water, one dyed fluorescent green and
the other undyed. Silicone tubing carries the outlet liquid to a waste beaker. The
pneumatic lines from the two valves are connected, through a Lab-view controlled
solenoid valve, to a bottie of nitrogen gas regulated to 15-20 psi gas pressure.
Figure 5.12 shows the various micro mixer devices developed. Primarily four
types of designs are made as clear form the masks [Figure 5.9 (a) to (d)].
5.3.3 Testing
The Reynolds number of flow in each of these devices has been
calculated with a qualitative visualization of degree of mixing. Surprising and
counter intuitive observations come up on testing of flow behavior through these
devices
79
(a) Tube bank (b) Triangular type
(c) Figure eight (d) Serpentine channel
Figure 5.12 Various micro-mixers
[Acknowledgements, MEMS 1 Class, 2002]
For testing of the micro-mixer, a set of valve opening / closing driver
control programs are written in Lab-view. Here, the basic intent is to have a
simultaneous and altemately running flow. Opening and closing the valves
simultaneously and altemately respectively achieve this. The mixer is connected
to two fluid reservoirs containing a highlighter dye and DI water located at
approximately 25-cm. height from the level of the mixer. The flow velocity is
calculated by measuring the rate of fluid output. Since the channel dimensions are
known from the fabrication stage, thus by using the channel dimensions and the
flow velocity, the Reynolds number is calculated. The level of mixing in the flow
80
is also observed by eyes and microscope, photographed and filmed respectively.
The highlighter dye shines on being illuminated by a UV source. Thus, as mixing
takes place, the shine is dimmed as the concentration of dye in the mixture
decreases.
5.4 Results and discussion
The first two designs are the "tube bank type design" and the "Triangle"
design. Both these promote little or no mixing. The input streams after entering
the chamber run out of the exit side parallel without any substantial change in the
coloration. There is no inter-diffusion of species at the molecular level as they
move in parallel laminae. Although the design promotes eddies and vortices in the
flow and a laminar boundary layer, developed about the rough inside of the mixer
trips, but it happens so far away from the interface of two streams that they cause
insignificant mixing. Slight mixing is seen as the flow is made time varying using
lab-view controlled pneumatic valve set. This is primarily because, as both layers
move parallel to each other with a different velocity, there is a high shear rate
between the layers promoting mixing. The flow velocity in the stem of the "T" is
monitored and found to be 2.84 mm/sec and 3.97mm/sec respectively. Also the
wetted perimeter to area ratio in this case is .00074. Thus Reynolds number is
calculated using Eq. 5.1 and 5.2 and found to be 1.86 E-03 and 2.59 E-03. The
Reynolds numbers for the remaining designs are calculated to be within the same
range. This explains the highly laminar nature of the flow and also the reasons for
Its repulsive behavior towards the mixing process. Figure 5.13 (a) and (b) show
the parallel flow and the time varying flow in case of "triangle" design.
(a) (b)
Figure 5.13 Parallel flow of DI water and highlighter dye (a) and time varying flow by valving (b)
[Acknowledgements, MEMS 1 Class, 2002]
The "figure eight" type design was intended to split the flow and bring it
together multiple times. This acts similar to a parallel lamination mixer. The
diffusion time in parallel lamination in reduced by a factor of (2)", where n is the
number of times the fluid is split apart. Mixing by diffusion, which is the only
mechanism of mixing at the micro scale, will only occur when the flow residence
time in a mixer of the twin fluid is more than the diffusion time of the mixer. This
is also proportional to the square of path length and inversely proportional to the
diffusion coefficient. Now the diffusion coefficient to liquids is of the order oflO'^
sees as shown in the range diagram in Figure 5.2. In the "Figure eight" design,
the mixer splits the liquid apart twice. Thus the diffusion time is reduced four
folds. However, as the fluid velocity at inlet is nearing 7.52 mnVsec, the residence
time of the fluid in the mixer is very less (near about 3-4 sees) and instead of
reduction in the diffusion time to one fourth, the diffusion time is far in excess of
82
the residence time and the species go out of the mixer without any substantial
mixing. Figure 5.14 shows the flow pattern of the highlighter dye and water seen
under an UV source and photographed with a digital camera.
Fig. 5.14 Flow pattern in a "Figure Eight" micro-mixer
Acknowledgements, MEMS 1 Class, 2002
The fourth design, which had some mixing, is a planar serpentine
channel. The reason for mixing in this case is an increase in the interface area
between the fluids due to expanse in the channel length. Although the diffusion
time proportional to the square of the channel length should increase many folds,
a simultaneous increase in the interface area between the fluids results in typically
higher diffusion rates and more mixing [4].
83
CHAPTER 6
CONCLUSIONS AND RECOMMENDATION
The originally hydrophobic surface of PDMS becomes hydrophilic upon
oxygen plasma treatment under certain process conditions. We have found that a
uniform oxygen plasma exposure of lower RF power with shorter duration makes
a thin layer of undamaged oxide on the surface of PDMS with active silanol
groups that largely facilitate in obtaining an irreversible sealing. The results
follow a general trend in terms of bond strength and contact angle measured on
plasma treated surfaces. It is observed that one gets stronger bonding as contact
angle decreases subsequently when PDMS or glass is treated by O2 plasma. An
excellent correlation between different plasma parameters and surface wettability
of PDMS or glass surface measured in terms of contact angle is found. All the
results indicate that a contact angle below 5 degree is a general requirement for
getting very good bond strength and thereby one can obtain the correct plasma
parameters for surface treatment by investigating and monitoring contact angle. A
nice correlation has been obtained between surface roughness and bond strength
of PDMS-PDMS and PDMS-glass bond. The fall in bond strength with increasing
surface roughness is greater in case of glass-PDMS for reasons, already discussed.
Thus contact angle becomes a new scale of reference for measurement of bond
strength. The oxygen plasma parameters developed have been successfully used
in stacking together and irreversibly bonding the multiple layers of glass and
replica molded PDMS. Such a technique has been used to build several designs of
84
parallel lamination micro-mixers. Flow behavior has been qualitatively studied in
such designs. The results, which come are very different from real life macro
scale mixers, which primarily use turbulence for better and quicker mixing. The
mixers show that the success of mixing at the micro-scale is primarily attributed
to inter-diffusion of molecular species. The results show an altogether different
regime of mixing, which can be utilized in future designing of refined versions of
these micro-mixers.
A software package can be envisioned for future with a correct statistical
foundation wherein using design of experiments, and specifying contact angle the
exposure dose can be gauged. Thus for any new plasma etcher the correct set of
parameters can be obtained by specifying an angle below 5° and the other
parameters which are kept constant during the process such as the inductively
coupled plasma power, the oxygen flow rate, the chamber size, etc.
85
REFERENCES
1. Stephen D. Centuria, Microsystem Design. Kluwer Academic Publishers, Boston / Dordrecht / London, 2001.
2. W.S. Trimmer, Editor, Micromechanics and MEMS: Classical and Seminal Papers to 1990. IEEE Press, Piscataway, NJ, 1997.
3. "Miniaturized total chemical analysis system: A novel Concept for chemical sensing," Sensors and Actuators B, Manz, A., Graber, N., and Widmer, H.M., Vol.1, 1990,pp.244-248.
4. N.-T. Nguyen and S. T. Werely, Fundamentals and Applications of Microfluidics, Artech House, Boston, 2002.
5. The mechanisms of hydrophobic recovery of polydimethylsiloxane elastomers exposed to partial electrical discharges, Kim J, Chaudhury MK, Owen MJ, Orbeck T., Journal of Colloid and hiterface Science, 244 (1): 200-207, DEC 1,2001.
7. Ma X.,Cierhart, Collins S.D., Smith R.L., Low temperature bonding for wafer scale packaging and assembly of micromachined sensors paper, Final Report 1998-99 for MICRO Project 98-144 Industrial Sponsor(s): Kumetrix, Inc.
8. Younan Xia and George Whitesides, Soft lithography, Angew. Chem. Int. Ed. 1998, 37.
9. General reviews on photoUthography: a) S. Okazaki, J. Vac. Sci.Technol. B 1991, 9, 2829 -2833; b) H. J. Jeong, D. A. Markle, G.Owen, F. Pease, A. Grenvill'e, R. von B.nau, Solid State Technol.1994, 37, 39 47; c) M. D. Levenson, ibid. 1995 , 38, 57 66; d) L.Geppert, IEEE Spectrum 1996, 33(4), 33 ± 38.Actuators A 1994, 41/42, 593 - 596.
10 Cast molding on the micrometer scale: a) H. C. Haverkom van Rijsewijk, P. ' E J Legierse, G. E. Thomas, Philips Tech. Rev. 1982, 40,287 - 297; b) J. G. Kloosterboer,' G. J. M. Lippits, H. C. Meinders, ibid. 1982, 40, 198 309.
11 B. D. Terris, H. J. Mamin, M. E. Best, J. A. Logan, D. Rugar, Cast molding " on the nanoiiieter scale, Appl. Phys. Lett. 1996, 69, 4262 - 4264.
86
12. Allen MG, Senturia SD, Analysis of critical debonding pressures of stressed thin films in the blister test. Journal of Adhesion 25(4): 303-305, 1988.
13. Dr. Arindam Dutta, Personal communication, Nanotech Center, Texas Tech university, 2002.
14. Mechanisms of Anodic bonding of silicon to Pyrex glass, Kevin B. Albaugh, IBM general technology division, Essex Junction, VT 05452, Don H ' Rasmussen, Department of chemical engineering, Clarkson University, Potsdam. NY 13676.Langmuir I. and Mott-Smith H.M., Physics Review 28 727(1926).
15. Yasuda H.K., Plasma polymerization and plasma treatment of polymers. Journal of Applied polymer science. Applied Polymer Symposia 42, April, 1987, Interscience Publication.
16. Boley Forrest I., Plasmas -laboratory and cosmic, D. Van Nostrand Company, Inc., Princeton, New Jersey, 1996.
17. Auerbach R.A., Tuma D.T. and Blenner D.R., Lord Corporation, March, 1978, Carnegie Mellon University, Pittsburgh.
18. Campbell, Stephen A., Science of Microelectronic Fabrication, Oxford University Press, New York, 1996
19. Chapman B., Glow Discharge Processes, Wiley, New York (1980).
20. Spitzer Lyman, Jr., Physics of fully ionized gases, Interscience Publishers, New York, 1962.
21.Sugawara M., Plasma Etching-Fundamentals and Applications, Oxford University Press, Tokyo, 1988.
22. Manabu Tokeshi, Kichi Sato and Tahehiko Kitamori, Integration chemistry for bio chip: Integration of immunoassay and biochemical lab on chip, Kanagawa academy of science and technology. Department of applied chemistry. Graduate school of Engineering, University of Tokyo, 1995
26. Ephrath, L.M. and Petrillo, E.J., Parameter and reactor dependence of selective Oxide RIE in CF4 and O2 plasma. Journal of Electrochemical Society, 129:2282(1982).
27. Boumans, P.W.J.M., Inductively coupled plasma emission spectroscopy, Interscience publication, New York, 1993.
28. Service and warranty manual ,Trion Technologies, 1998. www.triontech.com
29. Adamson, Arthur W., Physical chemistry of surfaces, Interscience publication. New York, 1983.
30. Hydrophobicity changes in Silicone rubbers, Hillborg H., Gedde U.W., IEEE transactions on dielectrics and electrical insulation, Vol. 6 No. 5, October 1999.
31. Ristic, L., ed.. Sensor technology and devices, Artech House Boston, MA., (1994).
32. Chan, J.H., A.T. Timperman, D. Qin and R. Aebersold. Microfabricated Polymer Devices for Automated Sample Delivery of Peptides for Analysis by Electrospray Ionization Tandem Mass Spectrometry, Analytical Chemistry, vol. 71, no. 20, pp.4437-4444 (1999).
34. JSR materials for C4 processes, JSR Corporation. Specialty Materials Laboratory Fine Electric Research Laboratories, ApiA semicon west seminar, 1996.
35. Alexei L. Bogdanov, Use of SU-8 Negative Photoresist for Optical Mask Manufacturing, MAX-Lab, University of Lund, SE-221 00, Lund, Sweden, 1996.
36. Michael Koch, Alan Evans, Arthur Brunnschweiler, Microfluidic Technology and applications. Research Studies Press Ltd., Baldock, Hertfordshire, England, 1998.
39. Armani, D., C. Liu and N. Alum (1998) "Re-configurable fluid circuits by PDMS Elastomer Microm achining, "12"^ MEMS, MEMS-1999, pp. 222-227, Orlan, EL PDMS Elastomer Microm achining, "12"^ International Conference on
40. Wilbur, J. L., A. Kumar, E. Kim and G.M. Whitesides (1994) Advanced Materials, 6, pp. 600-604.
41. Sean O'Brien, On MEMS talk , Sci. Chem.(1997), http://mail.mems-exchange.org/pipermail/mems-talk/.
42. Standard operating procedure Microsystem Technology Laboratories MIT, Bldg-39Room, 32160, VassarStreet, Cambridge, MA 02139,2001.
43. Home page of Solid state equipment corporation, Horsham, PA San Jose. www.ssecusa.com
44. Michial Duff, Actively enhanced etch fabrication and application of optical fiber capillary devices in the 300-500 Mhz Range by Howell. http://acoustics.stanford.edu/IRP/duffthesis
45. Homepage of Analytical Instmments,LLC1200 Mendelssohn Avenue, Suite 50, Golden Valley, MN 55427-4366. www.aibltd.com
46. Homepage Headway Research, Inc. 3713, Forest Lane Garland, TX 75042-6928 USA. www.pitt.edu/AFShome/g/i/gilbertw/ public/html/jason/links
48. Dannenberg. H, Phenyl-bridging in the 2-phenylethyl radical. A molecular orbital study, Joumal of Organic Chemistry, 66 (18): 5996-5999 (2001).
49. Hinkley, J.A., Oxidative aging of cured polymer from phyenylethynyl terminated imide oligomers, Joumal of Adhesion, 16, 115-125 (1983).
50. Allen, M.G., and Senturia, S.D., Proceedings of the American Chemical Society Division of Polymeric Materials: Science and Engineering 56, 735-739 (1987).
51. Gent A.N., Zhang L.Q., Strain-induced crystallization and strength of elastomers I. cis-l,4-polybutadiene, Joumal of polymer science phys 39 (7), pp. 811-817,2001.
52. Home page of smg diamond company (www.smg-diamond.com/lapping )
53. R. R. Kunz, H. R. Clark, P. M. Nitishin, and M. Rothschild, Lincoln Laboratory, Massachusetts Institute of Technology, Lexington, Massachusetts 02173-9108, B. S. Ahem, Rome Air Development Center, Hanscom Air Force Base, Massachusetts 07131-5000, High resolution studies of crystalline damage induced by lapping and single point diamond machining, 1999.
54. NT 1100,Users manual, Veeco Systems for Research, 100-1039-17th Ave, S.W. Suite 405 Calgary, Alberta T2T 0B2,1996.
55. Plasma treatment of Poly (dimethyl) silaxane, Owen M.J., Smith P.J., Joumal of adhesion science and technology, Vol.8, No. 10, pp. 1063-1075, 1994.
56. Hydrophobicity changes in Silicone mbbers, Hillborg H., Gedde U.W., IEEE transactions on dielectrics and electrical insulation. Vol. 6 No. 5, October 1999.
57. Siloxane polymers by Stephen J. Clarson and J. Anthony Semlyen, Ellis Horwood- PTR Prentice Hall, 1996, www.prenhall.com.
58. Henry, S., et. al., Micromachined needles for transdermal delivery of dmgs, proceedings for MEMS' 98, 11* IEEE intemational workshop microelectromechanical systems, Heidelberg, Germany, Jan 25-28, 1998, pp. 494-498.
59. Madou, M., Fundamentals of microfabrication, CRC Press, Boca Raton., 1997.
60. P.G. Drazin, Introduction to hydrodynamic stability, Cambridge University Press, Cambridge, UK,2002.
61. Frank, M., White, Fluid mechanics, McGraw-Hill Companies, New York, 1992.
62. R.W. Fox and A.T. McDonald, "Introduction to fluid mechanics", 5* Edition, John Wiley and Sons, New York, 1998.
63. N.-T. Nguyen and S. T. Werely, Fundamentals and Applications of Microfluidics, Artech House, Boston, 2002.
64. Cussler, E.L., Diffusion Mass Transfer in the Fluid Systems, Cambridge University press. New York, 1984.
65 Erbacher, C , et al.. Towards Integrated Continuous Flow chemical reactors, ' Mikrochi'mic'a Acta', Vol. 131, 1999, pp. 19-24.
66. Deshmukh, A.A., Liepmann, D., and Pisano, A.P., Characterization of a micro mixing, pumping, and valving systems, Proceedings of transducers '01, 11* International Conference on solid state sensors and actuators, Germany, June 6-7, 2001, pp. 779-782.
67. Deshmukh, A.A., Liepmann, D., and Pisano, A.P, Continuous Micromixer with pulsatile micropump. Technical digest of the IEEE solid state sensors and actuators workshop, Hilton head island, SC, June 4-8, 2000, pp. 73-76.
68. Larsen, U.D., Rong, W.,Telleman P., Design of rapid micro mixers using CFD," Proceedings of transducers 99, 10* intemational conference on solid state sensors and actuators, Sendai, Japan, June 7-10, 1999, pp. 200-203.
69. Seidel. R.U., et al., "Capillary force mixing device as sampling module for chemical analysis," Proceedings of transducers '99, 10* intemational conference on solid state sensors and actuators, Sendai, Japan, June 7-10, 1999,pp.438-441.
70. Evans, J., Liepmann, D., and Pisano, A.P, "Planar laminar mixer. Proceedings of MEMS 97, 10* IEEE Intemational workshop micro electro mechanical system, Nagoya, Japan, Jan 26-30, 1997, pp. 96-101.
71.Woias, P., Hauser, K., and Yacoub- George, E., "An active silicon micrimixer for microTAS applications," Micrototal Analysis System 2000, A. van den Berg et. al.(eds.), Boston, MA: Kluwer Academic publishers, 2000, pp.227-282
72. Bau, H.H., Zhong, J., Yi, M., A minute magneto hydrodynamic mixer, ' Sensors and actuators B, Vol.79, No.2-3, 2001, pp. 207-215.
APPENDIX A
Physical properties of SU8 (Photoresist)
92
Table A. 1: Important Physical properties of SU-8 (Photoresist)
Characteristics Value Conditions
Glass temperature: Tg
-50C unexposed film (not polymerized)
>200C
-55C
fully crosslinked film (exposed and post (hard?) baked)
MCC blend before PEB
Degradation temperature: Td ~380C fully crosslinked film (exposed and post (hard?) baked)
Coefficient of thermal expansion : CTE
52.0 +1-5.1 ppm/C
postbaked at 95C, thermal cycling test on Si wafer
Thermal conductivity
Polymer shrinkage
Kinematic viscosity
Viscosity
0.2 W/mK general value for thermoplastic not for SU-8