PET biochip fabrication for DNA sample preparation...i PET Biochip Fabrication For DNA Sample Preparation By Micah James Atkin B.Sc.(Hons) B.Elec.Eng.(Hons) A dissertation submitted

i

PET Biochip Fabrication For DNA Sample Preparation

By

Micah James Atkin

B.Sc.(Hons)

B.Elec.Eng.(Hons)

A dissertation submitted in fulfillment of the requirements for the degree of Doctorate of Philosophy in

Microtechnology

in the Industrial Research Institute of Swinburne

at the SWINBURNE UNIVERSITY OF TECHNOLOGY,

HAWTHORN, VICTORIA AUSTRALIA

Supervisors: Professor Erol Harvey

Dr. Karl Poetter Professor Robert Cattral

January, 2010

ii

Abstract


There are many potential applications for the use of microfluidic biochips in molecular

diagnostics. Two areas of research that have received considerable attention for

integration into microfluidic devices are the stages of DNA amplification and detection,

less investigated is that of DNA sample preparation. The focus of the bulk of this

research has been on the traditional methods of micromachining in glass and silicon

materials. More recently polymers have been investigated as a lower cost alternative.

The reports to date of polymeric microfluidic devices have focused on bulk surface

machining, or replication techniques. The alternative approach of machining and

layering polymer films has shown promise for greater 3-dimensional structuring and

continuous production.

It is known that some polymers fluoresce at the wavelengths commonly used for DNA

fluorescence detection, reducing the signal to noise ratio. It is also known that the

surface charge detrimentally impacts many DNA amplification and detection techniques

by the non-specific binding of proteins. There have been few reports of microfluidic

biochips fabricated from Poly(ethylene Terephthalate) (PET) for molecular diagnostics.

This thesis describes an investigation into a novel method for fabricating a DNA sample

preparation biochip by laser machining and the subsequent layering of PET films.

A new method of laser micromachining microfluidic structures in PET film using a

direct-write 355nm frequency tripled Nd:YAG (Neodymium-doped Yttrium Aluminium

Garnet; Nd:Y3Al5O12) laser was investigated, with the results indicating that ablation

was dominated by photothermal processes. A reproducible ablation threshold of 2.0 ±

0.3 J/cm2 and etch rates of up to 25µm per beam pass were achieved. Compression of

melt at the cut edge produced channel dimension close to the beam diameter, having a

channel width down to 30 ± 5 µm for film thicknesses between 12-350 µm.

iii

During process development it was shown that surface oxidation of the film was

important for improving capillary flow, electroosmotic flow, bonding time and quality,

and that it detrimentally impacted biocompatibility and fluorescence. The laser

machined surfaces had a reduction in the ester component compared to the non

machined surfaces. Surface oxidation techniques were developed for improved

microfluidic performance and bonding using chemical saponification and UV photo-

oxidation. These produced a change in the contact angle from 75o for the native PET

film down to 16o and 35o respectively. However, only UV patterning by non-contact

masking enabled localised surface oxidation with features down below 100 µm for

limiting the impact of surface modification on biocompatibility and fluorescence.

To demonstrate the applicability of these techniques polymer microfluidic filtration

chips were investigated for leukocyte filtration and particle retention for solid phase

extraction. Filter membranes were achieved with pore exit dimensions down to 1µm and

porosities up to 50% with non-supported spans of 1x2mm. Cake-layer formation

proved to be an issue during in-line leukocyte filtration causing non-reversible filter

fouling and increased backpressures, with back-flushing having limited success only

when using surface modified membranes. Stable solid-phase matrices were achieved

using irregularly shaped particles (25-75µm) resulting in extraction efficiencies of

approximately 7-14% with the first elution containing approximately 82% of the DNA

recovered.

This thesis concludes that frequency tripled Nd:YAG laser ablation of PET films can be

used to produce components of a biochip for molecular diagnostics, namely

microfluidics and solid phase extraction of DNA. This new laser machining method of

PET film enables the fabrication of microfluidic channels more quickly than excimer

based lasers, and with improved feature resolution in comparison to CO2 lasers.

Importantly, the thin film fabrication process can be used for fabricating microfluidic

filtration devices entirely from polymers. It was shown that the detrimental impact of

the fabrication process on device biocompatibility and fluorescence can be removed by

using masking techniques to pattern the oxidised areas.

iv


Declaration

This thesis contains no material which has been accepted for the award of any other

degree or diploma at any university and to the best of my knowledge and belief contains

no material previously published or written by another person or persons except where

due reference is made.

_________________________________________________________

Micah Atkin

v

Contents

PET Biochip Fabrication For DNA Sample Preparation .................................................. i

Abstract ............................................................................................................................. ii

Contents ............................................................................................................................ v

List of Figures .................................................................................................................. ix

List of Tables .................................................................................................................. xiii

1. Introduction ............................................................................................................... 1

1.1 Molecular Diagnostics ...................................................................................... 1

1.2 Biochip Technology .......................................................................................... 2

1.3 Motivation for This Research............................................................................ 3

1.4 Objectives of This Research ............................................................................. 4

1.4.1 Overview of This Thesis ........................................................................... 5

2. Background ............................................................................................................... 6

2.1 Introduction ....................................................................................................... 6

2.2 Microfluidic Fabrication ................................................................................... 6

2.2.1 Historical Perspective................................................................................ 6

2.2.2 Polymer Fabrication Techniques............................................................... 7

2.2.3 Microfluidic Fabrication Summary ......................................................... 17

2.3 Microfluidic Biochips ..................................................................................... 19

2.3.1 Historical Perspective.............................................................................. 19

2.3.2 Biochip Detection.................................................................................... 19

2.3.3 DNA Amplification ................................................................................. 22

2.3.4 Biochip Sample Preparation .................................................................... 25

2.3.5 Biochip Summary ................................................................................... 32

2.4 Conclusion ...................................................................................................... 33

3. PET Characterisation and Modification ................................................................. 36

3.1 Introduction ..................................................................................................... 36

3.2 Background ..................................................................................................... 36

3.2.1 PET .......................................................................................................... 36

3.2.2 Surface Modification ............................................................................... 37

3.2.3 Oxidation ................................................................................................. 39

3.2.4 Thermal Degradation .............................................................................. 41

3.2.5 Fluorescence ............................................................................................ 41

vi

3.2.6 Biocompatibility ...................................................................................... 44

3.3 Thermal Analysis ............................................................................................ 46

3.3.1 Experimental ........................................................................................... 46

3.3.2 Results ..................................................................................................... 48

3.4 Surface Modification ....................................................................................... 51

3.4.1 Experimental ........................................................................................... 51

3.4.2 Results ..................................................................................................... 54

3.5 Optical characteristics ..................................................................................... 61

3.5.1 Experimental ........................................................................................... 61

3.5.2 Results ..................................................................................................... 61

3.6 Biocompatibility .............................................................................................. 66

3.6.1 Experimental ........................................................................................... 66

3.6.2 Results ..................................................................................................... 69

3.7 Summary ......................................................................................................... 71

4. 3ω Nd:YAG Laser Machining of PET ..................................................................... 72

4.1 Introduction ..................................................................................................... 72

4.2 Background ..................................................................................................... 72

4.2.1 Laser Irradiation of Materials.................................................................. 72

4.3 Experimental ................................................................................................... 81

4.3.1 Laser Machining...................................................................................... 81

4.3.2 Imaging and Profile Measurements ........................................................ 88

4.3.3 Surface Chemistry Analysis .................................................................... 88

4.4 Results ............................................................................................................. 89

4.4.1 3ω Nd:YAG Laser Output Reproducibility ............................................ 90

4.4.2 Ablation Threshold.................................................................................. 91

4.4.3 Etch Rate vs Fluence ............................................................................... 93

4.4.4 Cut Quality .............................................................................................. 98

4.4.5 Raster Scanning ..................................................................................... 109

4.4.6 Surface Chemistry ................................................................................. 110

4.5 Summary ....................................................................................................... 113

5. Biochip Fabrication .............................................................................................. 114

5.1 Background ................................................................................................... 114

5.1.1 Bonding Mechanisms ............................................................................ 114

5.2 Experimental ................................................................................................. 117

vii

5.2.1 Channel Fabrication .............................................................................. 117

5.2.2 Filter Fabrication ................................................................................... 117

5.2.3 Bonding and Sealing ............................................................................. 119

5.3 Results ........................................................................................................... 119

5.3.1 Channel Fabrication .............................................................................. 120

5.3.2 Filter Fabrication ................................................................................... 121

5.3.3 Bonding and Sealing ............................................................................. 130

5.4 Summary ....................................................................................................... 136

6. Performance Evaluation of Sample Preparation Biochip .................................... 138

6.1 Background ................................................................................................... 138

6.1.1 Filtration ................................................................................................ 138

6.1.2 Capillary Flow ....................................................................................... 139

6.1.3 Electroosmotic Flow ............................................................................. 140

6.1.4 Pressure Driven Flow ............................................................................ 141

6.1.5 Sample Preparation ............................................................................... 143

6.2 Experimental ................................................................................................. 144

6.2.1 Capillary and Blood flow ...................................................................... 144




6.3 Results ........................................................................................................... 147

6.3.1 Capillary Flow ....................................................................................... 147




6.4 Summary ....................................................................................................... 160

7. Conclusions ........................................................................................................... 161

7.1 Results Summary .......................................................................................... 161

7.1.1 PET Characterisation and Modification ................................................ 161

7.1.2 3ω Nd:YAG Laser Machining of cPET ................................................ 164

7.1.3 Biochip Fabrication ............................................................................... 166

7.1.4 Performance of Sample Preparation Biochips ...................................... 168

7.2 Process Guidelines ........................................................................................ 169

7.2.1 Design ................................................................................................... 170

viii

7.2.2 Fabrication............................................................................................. 171

7.2.3 Operation ............................................................................................... 173

7.3 Conclusion .................................................................................................... 176

8. Publications .......................................................................................................... 178

9. References ............................................................................................................. 179

ix

List of Figures

Figure 1 Schematic of web production system for laminating a seven layer microfluidic circuit (Source:

Micronics) .................................................................................................................................................. 14

Figure 2 Roller embossing with hot a) and cold b) forming rollers ........................................................... 15

Figure 3 Representation of UV embossing process; (1) Resin, (2) substrate, (3) master, (4) UV curing

light. (www.epigem.com)........................................................................................................................... 16

Figure 4 a) Flow through chamber with weir exits designed for even particle packing [137], b) Silicon

post reaction chamber for high flux [132], c) mesh filter [133], d) lateral percolation filter [138]. ........... 28

Figure 5 Microfluidic filter design ............................................................................................................. 35

Figure 6 Norrish Type I reaction [171] ...................................................................................................... 40

Figure 7 Norrish Type II reaction [171] ..................................................................................................... 40

Figure 8 Mechanism of thermal degradation of PET [172] ........................................................................ 41

Figure 9 Energy level diagram for a typical organic molecule; non-radiative transitions and radiative

transitions [173] ......................................................................................................................................... 42

Figure 10 PET a) Fluorescence and b) Luminescence excitation and emission spectra [92] ..................... 43

Figure 11 Standard Thermogravimetric Analysis curves [180] ................................................................ 47

Figure 12 DSC scan of PET after quench cooling [181]. ........................................................................... 48

Figure 13 TGA of 100µm T542 cPET sample ........................................................................................... 49

Figure 14 MDSC of 100µm T542 cPET sample ........................................................................................ 50

Figure 15 Diagram of XPS operation ......................................................................................................... 51

Figure 16 Contact angle of a liquid droplet on an ideal surface ................................................................. 54

Figure 17 Contact angle measurements of Melinex Type 542 and Mylar Type A after NaOH exposure . 55

Figure 18 Proposed ester hydrolysis mechanism for surface modification ................................................ 56

Figure 19 High resolution C 1s and O 1s region spectra for the (a) unmodified and (b) modified Mylar

PET surfaces .............................................................................................................................................. 57

Figure 20 Sessile contact angle of UV irradiated Melinex Type 542 samples ........................................... 58

Figure 21 Wetting of UV patterned Melinex Type 542 with a solution of dye in deionised water (1:100)

.................................................................................................................................................................... 60

Figure 22 Relative fluorescence intensities of cPET samples at 532nm and 635nm as measured on a

GenePix 4000B microarray scanner ........................................................................................................... 63

Figure 23 Surface fluorescence of Type 542 cPET film after 30 minutes UV exposure through a non-

contact mask ............................................................................................................................................... 65

Figure 24 Microarray analysis of Cy5 labbeled oligonucleotides spotted in a cPET microchannel .......... 66

Figure 25 Diagram of custom built computer controlled thermocycler ..................................................... 67

Figure 26 Electrophoresis results of PCR amplifications with PET samples ............................................. 69

Figure 27 PCR assays in cPET chambers with facilitating agents (BSA, Lanes B&G; Skim Milk, Lanes

C&H; PEG, Lanes D&I; controls, Lanes A&F) ......................................................................................... 71

Figure 28 Polymer surface indicating absorption and ablation depths from a laser pulse [197] ................ 76

Figure 29 Etch depth versus fluence for 248nm Excimer laser ablation of PET [198] .............................. 77

x

Figure 30 Laser irradiation through a projection mask causing polymer ablation [199]. ........................... 78

Figure 31 CO2 laser x-y scanning stage ..................................................................................................... 81

Figure 32 Typical CO2 laser output vs duty cycle of RF drive circuit ....................................................... 82

Figure 33 Excimer laser beam delivery system .......................................................................................... 83

Figure 34 Typical Excimer laser fluence vs attenuator calibration curve .................................................. 83

Figure 35 3ω Nd:YAG laser optical configuration .................................................................................... 85

Figure 36 3ω Nd:YAG laser beam profile ................................................................................................. 86

Figure 37 AVIATM 3ω Nd:YAG laser system’s ThermoEQ pulse stabilisation......................................... 86

Figure 38 3ω Nd:YAG laser energy versus current calibration taken one week apart (A-C) and repeated

one month later (D-F) ................................................................................................................................. 87

Figure 39 Typical 3ω Nd:YAG laser frequency versus energy .................................................................. 88

Figure 40 3ω Nd:YAG etch reproducibility at 3.5J/cm2 on a 10kHz scanning vector ............................... 91

Figure 41 Image of heat affected zone on 100µm PET just above ablation threshold ............................... 92

Figure 42 Etch rate versus fluence for the 248nm Excimer laser ablation of cPET using 128 pulses. ...... 94

Figure 43 Etch rate versus fluence for the CO2 laser ablation of cPET. ..................................................... 95

Figure 44 Etch rate versus fluence for the 3ω Nd:YAG laser ablation of cPET. ...................................... 96

Figure 45 Energy Level versus etch depth per pulse for the 3ω Nd:YAG laser ablation of cPET (Energy

levels 1= 2.5, 2=10, 3=20, 4=28 J/cm-2) ..................................................................................................... 98

Figure 46 Example cross sections illustrating the effect of the wall angle and thermal damage produced

by a-b) CO2, c-d) 3ω Nd:YAG, and e-f) Excimer laser machining of cPET. ............................................. 99

Figure 47 A comparison of the typical cut quality between a) 248nm Excimer laser ablation, b) 355nm

frequency tripled Nd:YAG laser ablation, and c) CO2 laser ablated channels to 100µm depth in 250µm

thick PET films. ....................................................................................................................................... 101

Figure 48 Thermal effect of frequency variation of 10 shots at 27J/cm2 at a) 1Hz b) 1kHz and c) 10kHz

using the 3ω Nd:YAG laser ablation of cPET .......................................................................................... 103

Figure 49 Heat affected zones after 3ω Nd:YAG laser ablation of cPET at 3.5J/cm2 10kHz for a) 10, b)

50, and, c) 500 shots (cut entirely through layer) ..................................................................................... 103

Figure 50 Typical images of drilling through a 100 µm PET substrate with reducing energy distributions

using masked 3ω Nd:YAG laser beams a) 28J/cm2 40 shots, b) 60 shots at 28 J/cm2 with 8mm mask), c)

500 shots at 28 J/cm2 with 4mm mask. .................................................................................................... 104

Figure 51 Cut quality on cPET after single 3ω Nd:YAG laser beam pass at 10kHz with fluences of a) 4.4

J/cm2, b) 8.7 J/cm2, c) 18 J/cm2, d) 26 J/cm2, e) 33 J/cm2 ........................................................................ 105

Figure 52 Cut quality on cPET after multiple (10) 3ω Nd:YAG laser beam passes at 10kHz with 30sec

b/w passes and at the fluences of a) 4.4 J/cm2, b) 8.7 J/cm2, c) 18 J/cm2, d) 26 J/cm2, d) 33 J/cm2 ........ 106

Figure 53 Confocal image of lip profile for 10-350µm thickness cPET films machined by the 3ω

Nd:YAG laser at a fluence of 18 J/cm2 .................................................................................................... 108

Figure 54 Scanning electron microscope image of 350µm PET machined using the 3ω Nd:YAG laser at a

fluence of 18 J/cm2 ................................................................................................................................... 109

Figure 55 Structures induced in 100µm cPET film by raster scanning of the 3ω Nd:YAG laser beam at a)

10 J/cm2 and b) 30 J/cm2 ......................................................................................................................... 110

xi

Figure 56: Chemical structure of Poly(Ethylene Terepthalate) ............................................................... 111

Figure 57 Sphere embedded in tapering hole at position of maximum interference ................................ 118

Figure 58 Microfluidic chip layers ........................................................................................................... 120

Figure 59 a) Three and b) Two Sided channel formation ......................................................................... 120

Figure 60 Stress distribution for a 0.35 mm pore stretched to 0.5 mm .................................................... 122

Figure 61 Excimer laser machined cPET membrane with 15 x 30 µm rectangular pores ........................ 124




Figure 65 Excimer laser machined filter membrane with a square array of pores having a) 8µm entrance

holes and b) 1µm exit holes ..................................................................................................................... 127

Figure 66 Excimer laser machined filter membrane with an offset pore array of pores having a) 8µm

entrance holes and b) 2µm exit holes ...................................................................................................... 128

Figure 67 Excimer laser machined filters fabricated a) in focus b) 20µm out of focus and c) 30µm out of

focus ......................................................................................................................................................... 129

Figure 68 Scanning electron microscope image of 3ω Nd:YAG laser cut 350µm thick PET film before a),

and after b) bonding ................................................................................................................................. 130

Figure 69 Channel cross sections of 3ω Nd:YAG laser machined channels in cPET film illustrating a)

poorly and b) well bonded devices ........................................................................................................... 132

Figure 70 Sealed cPET fluidic channels showing excimer cut a) two sided and b) three sided, and 3ω

Nd:YAG cut c) two sided and d) three sided channels with the insert showing the 3 sided channel prior to

bonding..................................................................................................................................................... 133

Figure 71 Channel cross section of 3ω Nd:YAG laser machined channels in cPET film bonded by the a)

single step procedure, and b) two step procedure ..................................................................................... 134

Figure 72 Cross section of an of 3ω Nd:YAG laser machined channel in cPET bonded using the UV

thermal bonding method........................................................................................................................... 135

Figure 74 The Fahraeus effect [231] ........................................................................................................ 142

Figure 75 Axial accumulation of red blood cells [231] ............................................................................ 143

Figure 76 Custom computer controlled pumping apparatus..................................................................... 145

Figure 77 Picogreen dsDNA Calibration Curve ....................................................................................... 146

Figure 78 Capillary flow experimental wall angle (a) between channels x and y. ................................... 148

Figure 79 Graph of pressure differential across a filter membrane (100x100 array of 3.5µm pores) during

initial wetting showing the point of maximum pressure (bubble point) required to force liquid entirely

through the membrane. ............................................................................................................................. 149

Figure 80 Backpressure profile of filter membrane with 100x100 array of 3.5µm pores ........................ 151

Figure 81 Axial accumulation of blood cells in 250µm wide 100µm deep microchannels fabricated by 3ω

Nd:YAG laser machining ......................................................................................................................... 152

Figure 82 Particle retention of 5µm silica microspheres in a a) tightly packed sealed channel, and b) an

image of packed microspheres in a channel with the capping layer removed .......................................... 154

xii

Figure 83 Membrane filter with 8µm entrance pores a) before filtration, and b) after filtration of

methylene blue stained white blood cells ................................................................................................. 155

Figure 84 Solid phase DNA extractions from six different microchips packed with 25-75µm irregularly

shaped silica particles ............................................................................................................................... 157

Figure 85 Amount of DNA present in multiple elutions from six solid phase extraction microchips ..... 158

Figure 86 Gel electrophoresis result after PCR amplification of eluted DNA from a cPET extraction

microchips with a silica particle solid phase ............................................................................................ 160

xiii

List of Tables

Table 1 Surface treatment technologies [159]............................................................................................ 37

Table 2 Relative % atomic concentrations of the elements present on the surface of clean T542 cPET and

NaOH treated T542 cPET .......................................................................................................................... 56

Table 3 XPS analysis of UV exposed Melinex Type 542 .......................................................................... 58

Table 4 XPS analysis peak assignment of UV exposed Melinex Type 542 ............................................... 59

Table 5 Ablation hole size for increasing frequency at 8.7J/cm-2 .............................................................. 97

Table 6 Cut dimensions for 100µm thick cPET film machined by 3ω Nd:YAG laser with increasing

fluence ...................................................................................................................................................... 107

Table 7 Dimensions of laser machined cuts using different thickness substrates .................................... 108

Table 8 Distribution of carbon species on uncut and 3ω Nd:YAG laser machined cPET as determined by

XPS .......................................................................................................................................................... 111

Table 9 Commonly used mask patterns for 10x10mm pore arrays .......................................................... 118

Table 10 FEA results of pressure versus wall angle to force a 5 µm sphere through a pore having a 3.5 µm

exit hole diameter. .................................................................................................................................... 123

Table 11 Bonding parameters .................................................................................................................. 132

Table 12 Summary of capillary flow experimental results ....................................................................... 147

Table 13 Normalised DNA elution data ................................................................................................... 158

1

1. Introduction

Currently most molecular diagnostics are carried out in centralised laboratories in

hospitals or specialised research medical centres. The equipment and protocols these

laboratories use require expensive and time-consuming steps to be performed, adding to

the overall experimental cost and thereby prohibiting genetic screening to be performed

as a regular diagnostic tool. Biochip technology, microfluidic microchips designed for

biological experiments, has the potential to reduce the cost of genetic testing by

automating and integrating the previously laboratory based processes onto a single

disposable platform.

1.1 Molecular Diagnostics

Genomics is the field of molecular biology dealing with an organism's genome, i.e. its

complete set of deoxyribonucleic acid (DNA). Although the cells within an organism

are considered to contain identical genomes, the genomes from different individuals

within a given species are estimated to be about 99.9% identical. This 0.1% difference

among individuals is significant as it relates to approximately three million bases. The

importance of this knowledge brought about the formation of the U.S. Human Genome

Project in 1990 by the U.S. Department of Energy and the National Institutes of Health.

On February the 16th 2001 the HGP announced the completion of the first rough draft of

the human genome sequence [1], and at about the same time Venter et al. of Celera

Genomics announced the completion of their initial sequencing of the human genome

[2].

In the area of diagnostics the availability of this knowledge brings about a simpler

means to examine individual humans for genetic mutations rather than examining their

entire genome [3-5]. This molecular diagnostic technique can be applied to parental

screening, newborn screening, carrier screening, forensic screening and susceptibility

screening. Parental screening discerns whether a foetus is at risk for various identifiable

genetic diseases or traits. Newborn screening involves the analysis of blood or tissue

samples taken in early infancy in order to detect genetic diseases for which early

intervention can avert serious health problems or death. Carrier screening identifies

individuals with a gene or a chromosome abnormality that may cause problems either

for offspring or the person screened. Forensic screening seeks to discover a genetic

2

linkage between suspects and evidence discovered in criminal investigations.

Susceptibility screening involves the screening of selected populations for genetic

susceptibility to environmental hazards.

Other examples of the benefits that this kind of research promises are in the areas of

evolution for comparing mutations and historical events, agriculture to produce disease,

insect and drought resistant crops, and in identifying disease causing mutations and the

genetic characteristics of infectious agents . These advances promise to revolutionise the

healthcare industry by providing the ability to diagnose disease and tailor

pharmaceuticals at the molecular level. Rapid, sensitive and inexpensive methods of

detection for these mutations in the human genome and infectious agents will be highly

beneficial for control of inherited and infectious diseases.

With these potential benefits there has been much emphasis placed on developing tools

and techniques to improve this sequencing and implementing the knowledge that can be

gained with further analysis. One promising new technique to reduce the cost of

molecular diagnostics and move it out of centralised laboratories is that of microfluidic

biochip technology.

1.2 Biochip Technology

In this context biochips are microfluidic devices used to perform biochemical analysis

in a small planar package, typically the size of a common microscope slide. The term

substrate is used to describe the bulk material in which the microfluidic device is

fabricated and is typically made of glass, silicon or plastic.

Microfluidics is the ability to move, mix, pump and otherwise control fluids on the

microscopic scale. The main advantages for working at the micron scale for chemical

and biochemical analysis are: the increased response time, due to the large surface to

volume ratios and small diffusion distances; the smaller footprint, leading to higher

levels of integration; and the reduction in sample and reagent volumes. Liquid flow in

this micro regime is dominated by different physical properties compared to its

macroscale counterparts [6]. The substrate choice and manner of processing is critical in

microsystems where the liquid, solid and gas phase boundaries play an important role in

3

the fluid dynamics of the system, unlike their macro counter parts where gravity and

inertia have considerable effect.

Early work in this field was pioneered over 30 years ago by Tuab et al. [7] at IBM with

the development of ink jet printer nozzles and at Stanford University with the

fabrication of a miniaturised gas chromatograph by Terry et al. [8]. Since then there has

been substantial research into this area including the development of lab-on-a-chip

devices, or micro Total Analysis Systems (µTas) [7-10] with the aim of reducing

laboratory costs by lowering reagent volumes, shortening analysis times, reducing

capital equipment expenditure and lowering operator intervention. Much of this early

work was performed in silicon, borrowing from the semiconductor industry, with a

more recent shift towards polymers for low cost production [11-14]. Initial work

focused on system components such as pumps, valves, and miniaturising well

understood detection technologies that could benefit from the scaling effects such as

capillary electrophoresis (CE) [15] and microarrays [5] for DNA analysis.

A problem in the operation of many of these microfluidic systems is their interface to

the outside world and their requirements for sample preparation. Of crucial importance

to many of these devices is the incorporation of sieves and filters for the prevention of

blockages or to retain particles for further processing, such as with solid phase

extraction [16]. Much of the previous work on integrating filtration systems in

microdevices has focused on silicon and glass technologies [17-21] . Of more difficulty,

due to the different manufacturing methodology and lower mechanical strength, is the

integration of filtration systems in polymer microfluidic chips.

1.3 Motivation for This Research

The successful commercialisation of biochip technology is heavily dependent on the

development of low-cost fabrication techniques for the manufacture of these disposable

biological sensors. Polymers provide a suitable substrate with a wide range of

processing capabilities allowing for cost effective manufacturing. Conventional

approaches of fabrication for complex microfluidic devices have involved the assembly

of several injection-moulded or embossed components to create the more complex 3-

dimensional (3-D) devices. A promising fabrication alternative is to manufacture 3-D

4

microfluidic devices by stacking, aligning and bonding several microstructured films.

The use of thin polymer films for device fabrication enables continuous manufacturing

strategies to further lower device cost, however, process compatible methods for

microstructuring these films still need to be developed.

The applications of high-throughput detection methods like capillary electrophoresis

and microarray analysis are still limited by the time and labour intensive sample

preparation and amplification stages. Laboratories require cost-effective and time-

efficient systems that provide the appropriate quality, quantity and purity of DNA. With

microfluidic technology it is now possible to integrate the stages of sample preparation,

amplification and detection, as has been demonstrated on silicon and glass based

substrates [17]. However, polymer systems are less developed in this area due to the

impact the fabrication and assembly methods have on device performance, their

intrinsic fluorescence making detection difficult [22], and mechanical strength limiting

filter designs for sample preparation [23].

1.4 Objectives of This Research

The motivation behind this research is to investigate a polymer film biochip fabrication

technique. This fabrication technique is based on frequency tripled (355nm)

Neodymium Yttrium Aluminium Garnet (3ω Nd:YAG) laser micromachining of

polyethylene terephtalate (PET) and developed for the fabrication of a sample

preparation biochip compatible with on-chip polymerase chain reaction (PCR)

amplification and microarray analysis. The research question can be stated as, “Is 3ω

Nd:YAG laser machining of polyethylene terephtalate (PET) suitable for fabricating a

biochip for DNA sample preparation?” and can be broken down into the following

main objectives.

Objectives:

- To characterise and evaluate structured PET film as a suitable material for this

application, investigating bulk thermal characteristics, fluorescence, surface

chemistry and methods of modification, and biocompatibility;

- To characterise the application of direct-write 3ω Nd:YAG laser microstructuring of

PET for biochip fabrication;

5

- To develop a fabrication method involving a polymer biochip for DNA sample

preparation using laser processing; and

- To evaluate the biochips’ performance in terms of their fluid flow, backpressure,

filtration, and finally DNA extraction from whole blood.

This study will provide a method for, and an insight into, some of the issues associated

with laser processing of polymer films for biochip fabrication. The method of 355nm

Nd:YAG laser micromachining of PET is investigated in terms of its cut characteristics

and effect on biochip performance, with biochip sample preparation focused on as a key

stage towards the goal of a fully integrated microfluidic device for point-of-care DNA

testing.

1.4.1 Overview of This Thesis

Chapter 2 reviews the historical development and current state of the art for polymer

microfluidic fabrication and biochip development, focusing on the film microstructuring

techniques and the three stages of on-chip DNA analysis - sample preparation,

amplification, and detection. Chapters 3-6 address each of the four main objectives with

their relevant background, experimental, results, discussion and conclusions. The PET

film is characterised and methods for overcoming its fluorescence and improving its

biocompatibility are then discussed in Chapter 3. The technique of 3ω Nd:YAG laser

micromachining of PET is investigated in Chapter 4. In Chapter 5, the laser

micromachining method is combined with other microfabrication techniques to develop

a biochip platform for sample preparation by filtration. In chapter 6 these biochips are

then tested for fluidic operation, filtration and DNA extraction. Finally, Chapter 7

concludes the thesis by summarising the results and suggests possible directions for

future work.

6

2. Background

2.1 Introduction

This review begins with a brief history of the development of microfluidic fabrication,

followed by a discussion on methods for their low cost manufacturing in terms of

materials and processes. Specifically state-of-the-art methods for polymer device

fabrication and thin film fabrication methods are explored. Finally the review focuses

on biochip devices in the areas of sample preparation, amplification and detection.

2.2 Microfluidic Fabrication

Microfluidic fabrication deals with the processes for creating structures to control

liquids in the micron domain. The fundamental features common to most microfluidic

devices are the microchannels used to confine fluids. The processes used to fabricate

these systems can be broadly categorised into three methods:

a) Techniques that form 3-sided channels into a substrate and are then sealed with

another process;

b) The layering of gaskets, where the channels are cut directly through the substrate,

producing 2-sided channels that are then sealed on opposite sides; and

c) Deposition techniques that build up the wall in a layered fashion, as with print-head

deposition.

2.2.1 Historical Perspective

The rapidly growing field of micrototal analytical systems (µTAS), or lab-on-a-chip

devices, began with the integration of a gas chromatograph onto a silicon wafer by

Terry et al. in 1975 [24]. However, it wasn’t until the late 1980’s that further work was

published describing the fabrication of micropump and valve components [25]. The

following explosion in the number of microfluidic publications was preceded by a paper

by Manz et al. in 1990 on the concept of the “micrototal analysis system”; devices that

are miniaturised and integrated onto a single silicon wafer to perform sample

pretreatment, separation and detection [26].

7

Much of this early work was performed on silicon or glass substrates using techniques

developed in the 1970’s and 1980’s for the semiconductor industries, described in the

reviews [25],[27],[28],[29],[30]. Investigations into polymer fabrication techniques for

mass producing these devices began around the mid 1990’s, with early developments

described in the reviews [12],[13],[31],[11],[32] .

2.2.2 Polymer Fabrication Techniques

The manufacture of polymer substrates in microfluidics offers several advantages over

their silicon and glass based counterparts. Generally these include:

- a reduction in material cost;

- a large selection of materials with a variety of physical and chemical properties;

- simplified and lower cost manufacturing processes;

- and in some cases, such as injection moulding, more flexibility in geometric

configuration in comparison to the etching processes used for silicon and glass.

These advantages have resulted in the large growth of literature reporting polymer

microfluidic devices over the past decade [14],[33],[34], [35],[36],[37],[239]. A key

component of this body of work has been the development of microfabrication

techniques in these polymer devices. These polymer microfabrication techniques can be

classified into two categories:

a) Direct machining methods where the pattern is created directly on the substrate.

These methods include laser-based lithography and beam scanning, plasma etching,

wet chemical UV lithography using photoresists, x-ray lithography and print-head

deposition. Of these techniques, the laser based machining and print-head deposition

methods are of particular interest due to their ability for rapid design change and

part manufacture using direct-write computer driven systems;

b) Replication processes that use a master template to form the pattern. The replication

processes that have been developed for microstructuring include embossing,

injection moulding, thermoforming, and casting.

2.2.2.1 Print-head Deposition

The term print-head deposition is used here to describe methods for directly depositing

material using a scanning print-head, such as an inkjet cartridge or dispensing syringe.

Current commercially available rapid-prototyping techniques, such as Stereolithography

8

(SLA) and Selective Laser Sintering (SLS), typically have tolerances that are too large

(>250µm) and use materials that are unsuitable for most microfluidic devices.

Techniques that have been developed for microfluidic fabrication include polymer

micro-stereo lithography [38] and ink-jet delivery [39]. More recently Lago et al.

described the use of a laser printer to deposit toner as the walls, a technique allowing for

the simple implementation of microfluidic circuits, however its effectiveness as a

microfluidic system was limited since the walls were porous and only 6µm high [40].

Print-head deposition techniques can provide a cost effective method for rapid

prototyping and an ability to fabricate complex 3-D structures. However, these systems

are also typically limited in terms of the available materials suitable for deposition, and

the suitability of these materials for microfluidic applications.

2.2.2.2 Laser Micromachining

2.2.2.2.1 Excimer lasers

UV laser ablation of polymers was first reported in 1982 by Kawamura et al. [41] and

Srinivasan et al.[42]. The strong optical absorption of some polymers at these UV

wavelengths, combined with a pulse duration less than the time associated with the

diffusion of heat into the bulk material, produces a characteristically photochemical

ablation [43]. Since then extensive work has been done to characterise the laser cutting

process and understand the ablation mechanisms [44;45]. This will be discussed in more

detail in Chapter 3.

Reports on laser machining of microchannels in polymers began appearing in the late

1990’s. Roberts et al. [46] were the first to demonstrate electroosmotic flow, a non-

mechanical method for microchip pumping and a key factor affecting capillary

electrophoresis resolution, in excimer laser ablated channels in PET capped with a

PET/PE laminated layer. Pethig et al. [47] used patterned gold and polyimide layers to

realise channels and electrode arrays for a travelling wave dielectrophoresis concept . A

gasket approach was developed by Martin et al. [48] using laser machined polyimide

layers to realise a microdialysis chip incorporating a molecular weight cut off filter and

microdialysis membrane .

9

A novel laser microstructuring approach was reported by Prins et al. [49] for the

fabrication of multiple parallel channels from micrometer thin PET films. In this

process areas of PET are exposed by laser ablation of aluminium coated PET. The

machined material is then folded upon itself to form a multi-layer laminate. The

laminate is then diffusion bonded, allowing only the exposed regions of PET to form a

bond. The laminate is then pulled slightly apart to form a honeycomb-like structure.

This approach allows very high density parallel channels; it does not however, have the

flexibility to allow complex channel designs.

There has been a significant amount of research investigating the fabrication of

microfluidic devices by excimer laser ablation, as outlined in the review by Malek [35].

The work has included more fundamental studies on the effect that excimer laser

machined surfaces have on the microfluidic systems, with reports showing changes in

hydrophobicity, non-specific protein binding, and electrokinetic flows. Most of these

studies have used polyimide (PI), polycarbonate (PC), or polymethylmethacrylate

(PMMA) substrates for the UV machining of polymer microfluidic devices. Prins et al.

and Roberts et al. are among the few groups that report the use of excimer laser

structuring of PET films for microfluidics applications.

2.2.2.2.2 CO2 Lasers

CO2 lasers, producing radiation with a wavelength of approximately 10µm, have been

widely adopted by industry and are considerably cheaper than UV lasers. For this reason

they have generated much interest in the microfluidics community. These lasers produce

light with longer wavelengths and therefore lower photon energies than UV systems,

resulting in photon-substrate interactions that are typically photothermal [50].

Researchers investigating microfluidic fabrication with CO2 lasers have focused mostly

on using PMMA as the substrate [51],[52],[53],[54]. This is due to the materials optical

clarity and high absorption at this wavelength. Klank et al. demonstrated a microchannel

fabrication method using multiple passes of the laser beam across the material surface,

forming 3-sided channels with Gaussian profiles having widths of typically 250µm and

depths up to 350µm. This work was extended by Bowden et al. who overlapped

successive beam passes in a raster pattern to form channels and cavities with widths in

the millimetre range. Klank et al. noted that the channel surface roughness from

10

redeposited droplets of melted material caused by the ablation process was as large as

15µm. Cheng et al. demonstrated channel widths as small as 100µm with aspect ratios

of up to 7, and improved surface roughness with an annealing stage that allowed the

softer debris to reflow into the substrate. The formation of microfluidic structures with

two machined sides by CO2 machining of gaskets has been demonstrated with stainless

steel and polyimide [55]. These microfluidic systems required post processing of the

machined steel before bonding to remove the lip formed from debris and reflown

material at the cut edge.

CO2 laser ablation of PET has been shown to have a predominantly photothermal

ablation mechanism, with resultant ablated material having a morphology dependent on

the irradiated wavelength. For CO2 laser irradiation it was shown that PET has a

significantly stronger adsorption at the 9.25µm wavelength [50] resulting in a smaller

heat effected zone around the ablation site.

2.2.2.2.3 Nd:YAG Lasers

A recent feature of industrial lasers has involved the use of optical harmonic generation

as a method of changing laser wavelength. For example the 1.06µm fundamental

wavelength of the Nd:YAG has been frequency multiplied to produce 532nm, 355nm

and 266nm radiation. These higher harmonic Nd:YAG lasers have been developed for

printed circuit board drilling to remove polyimide, resins, copper, glass, ceramics and

other metals. These frequency multiplied systems have been used for micromachining

silicon using frequency doubled (2ω=532nm), tripled (3ω=355nm), and quadrupled

(4ω=266nm) modes [56]. Of particular interest to polymer microfluidic fabrication are

the higher harmonic modes of operation, the frequency tripled (3ω) and quadrupled

(4ω) modes. The higher photon energies from these lasers provide the possibility for

photochemical ablation and thereby the reduction of thermal damage during

micromachining.

The photochemical ablation of polymers for microfluidic fabrication using frequency

multiplied Nd:YAG lasers has been demonstrated using polyimide [57]. The bond

energy of the CN functional group in polyimide suggests that the material can undergo

direct valence electron absorption at the wavelengths produced from these frequency

11

tripled (3ω) and quadrupled (4ω) modes. This dominantly photochemical ablation was

reported by Paul et al [57] who demonstrated cleanly ablated structures showing little

thermal damage . McGinty et al [58] demonstrated the difficulties in achieving channel

formation on polystyrene (PS), polycarbonate (PC), and polymethylmethacrylate

(PMMA) using a frequency tripled Nd:YAG laser. The PMMA failed to produce any

channels, whilst the optimised machining of PC and PS resulted in a volumetric

removal of approximately 20 µm3/µJ with lip formation of up to 33% with respect to the

channel height.

There is little literature reporting the ablation of PET using frequency tripled Nd:YAG

lasers. Although the individual photon energies of frequency tripled Nd:YAG lasers are

not high enough for direct electronic excitation of the molecular bonds in PET, it

remains to be seen how these higher frequency lasers perform in micromachining

microfluidic channels in PET.

2.2.2.2.4 Ultra Short Pulse Lasers

Ultra short pulse lasers (<<1ns) have been investigated as a method to achieve greater

precision and higher aspect ratio structures. The ultra-short pulse widths allow very

high peak laser intensities with relatively low pulse energies. For example, a laser

focused to a 20µm diameter spot having an energy approximately a third of a millijoule

and a pulse width of 100fs has a peak intensity of 1015 W/cm2. This high energy density

enables machining of materials that are normally transparent at the laser wavelength.

Early work by Kuper et al. [59] showed clean etching of PMMA using 300 femtosecond

duration Excimer laser pulses at 248nm. Since then other polymers have been

investigated for ultra short pulse laser micromachining including PTFE, PDMS,

PMMA, PI, FEP, PC, and PET [60],[36]. Generally the results have shown advantages

over nanosecond pulse lasers for micomachining in terms of; lower ablation thresholds,

less thermal damage, no plasma shielding effects, and the ability to form higher aspect

ratio structures. However, the machining rates are often slower and the laser systems are

more expensive than some of their longer pulse counterparts. Typical applications for

ultra-short pulse lasers have involved high-aspect ratio drilling [61], however, examples

of microfluidic structures have been demonstrated by beam scanning to produce 3-D

structures in doped glass [62] and PMMA [63].

12

2.2.2.3 Bulk Replication Technologies

The replication techniques used for structuring polymers typically offer the advantage

of reducing fabrication costs for high volume manufacture by using a master template.

The casting, hot embossing and injection moulding techniques described below have all

been commonly employed to fabricate microfluidic devices by replicating the

template’s structure into the surface of a polymer, where the bulk of the polymer

substrate is typically much greater in depth than the replicated structures. These

techniques are typically batch-based and as such offer advantages in terms of process

flexibility when combining with other stages of device production.

2.2.2.3.1 Casting or Soft Lithography

Casting or soft lithography is the process whereby a liquid polymer is cast onto a

mould, cured, and then removed from the mould. A degassing stage is often required to

ensure the liquid polymer does not contain any air bubbles. For microfluidic

applications the elastomer polydimethylsiloxane (PDMS) is the most commonly cast

polymer and the term soft lithography is often used to describe this method. Due to the

simplicity of this technique there are many examples of such cast microfluidic

prototypes in the literature [64],[65],[66]. Other elastomer and thermoset plastics have

been reported [37]. However, compared with the other replication techniques casting

requires much longer cycle times, typically in the order of minutes to hours, limiting the

suitability of this technique for mass production.

2.2.2.3.2 Hot Embossing

Developments in embossing for microfluidic replication started in the late 1990’s using

imprinting or stamping processes [67] and progressed into the hot embossing process,

now the most common tool used by researchers for channel replication in polymers

[68], [69], [70], [33], [37]. The technique commonly employed is simple. The polymer

substrate and master template are placed together and heated to just above the polymer’s

glass transition temperature. Pressure is then applied against the two items replicating

the mould structure into the polymer and the system is cooled before the sample is

released from the template. The whole process can be performed under vacuum to help

remove any air or vapour bubbles. Cycle times are relatively large in comparison to

injection moulding techniques (typically a few minutes compared with seconds). The

13

most common materials that have been used in this replication process include

polystyrene, polymethylmethacrylate, polyethylene terephthalate glycol, and

polycarbonate substrates.

A similar process to hot embossing performed at room temperature has also been

employed, however, the quality of the replicated structures are more dependent on the

processing parameters, requiring greater pressures, longer residence times and careful

material selection [71].

2.2.2.3.3 Injection Moulding

The injection moulding process involves the melting of thermoplastic pellets which are

then injected into the heated mould cavity. The cavity is then cooled and the resolidified

plastic parts are ejected. The injection moulding of micro-structures is similar to

standard injection moulding used in the plastics industry, except the mould is evacuated

prior to injection to avoid bubble formation and high-aspect ratio micro-structures

require the mould to be heated [12]. A further variation on this technique is the

combination of hot embossing and injection moulding, where the polymer melt is

compressed inside the mould cavity in a similar manner to CD/DVD production [72],

[37].

Injection moulding of micron sized structures has been under development for 30 years

and micro-moulding machines are commercially available today. For a review of this

process and the wide variety of thermoplastic materials that have been used for micro-

moulding see reference [33].

2.2.2.4 Thin Film Replication Technologies

Recently, in an effort to develop more cost effective manufacturing processes, there has

been a shift away from batch based methods towards continuous manufacturing [73].

This manufacturing approach typically involves the use of thin film substrates that are

presented to each stage of the manufacturing process as parts of a continuous polymer

web. The method of processing continuous webs of polymer is commonly employed in

the printing and packaging industries and is also referred to as reel-to-reel processing.

An example of such a web based system used for microfluidic fabrication is shown in

14

Figure 1, where seven-layer devices are fabricated by laser machining, cutting, and

laminating [73].

Figure 1 Schematic of web production system for laminating a seven layer

microfluidic circuit (Source: Micronics)

Furthermore the batch based methods described for fabricating microstructures into bulk

substrates, enable limited device complexity due to the inability of these methods to

machine structural overhangs and internal three dimensional geometries. In comparison

the use of thin polymer films enables the stacking and aligning of multiple machined

thin films, which can enable the use of standard 2D micro-fabrication methods to form

complex 3-D micro-structures [74].

Stacked 3-D microfluidic manifolds are not novel, and early examples have been

fabricated with photolithographic techniques using silicon substrates [75]. For structure

replication in thin polymer films the processes of embossing, UV embossing, and

thermoforming have been demonstrated and are described below.

2.2.2.4.1 Thin Film Embossing

The hot embossing technique described earlier has been successfully demonstrated for

microfluidic replication in bulk amorphous substrates, however, there are no reported

Copyright restricted image

15

cases of microfluidic feature replication in PET thin films. These semi-crystalline thin

films have greater mechanical strength and thermal stability than their amorphous

counterparts. Embossing of sub-micron sized features on thin films has been used for

many years for surface finishing and the replication of microoptic structures such as

holograms, corner cube reflectors, and anti-reflective structuring. The replicated

structures are typically embossed on a continuous web based system using either a

heated roller or heated film topology as depicted in Figure 2. With a similar method

Dreuth et al. demonstrated submicron structuring of very thin (<10µm) PET film for

electrostatic actuators [76].

Figure 2 Roller embossing with hot a) and cold b) forming rollers

2.2.2.4.2 UV Embossing

UV Embossing is a technique employed in the micro optics industry for high accuracy

replication to improve the planarity and dimensional accuracy compared to its hot

embossing counterpart [77]. It has recently been applied to the commercial production

of microfluidic devices. The process, depicted in Figure 3, works by coating the

substrate (2) with a thin UV-curable resin (1) then using a master template (3) to

emboss the pattern and cure (4) the resin during contact with the template. By

embossing into a fluid layer and then curing around the structure, much of the stress

inherent in thermal bonding can be avoided.

Microfluidic applications often require the selection of materials that have suitable

chemical properties. This UV embossing process currently requires UV cured resins,

which may not have the surface chemistry compatible with many microfluidic

applications.

a) b)

16

Figure 3 Representation of UV embossing process; (1) Resin, (2) substrate, (3)

master, (4) UV curing light. (www.epigem.com)

2.2.2.4.3 Thermoforming

Thermoforming processes developed in the packaging industry, including vacuum

forming and blow moulding, are replication methods for thin film structuring used to

shape the entire thickness of a film. A prototype genetic diagnostic instrument was

reported by Findlay et al. in 1993 incorporating thermoformed fluidic chips with

channel dimensions of millimetres [78]. The system utilises thermoformed raised

channels as a pumping mechanism by guiding a roller along the surface.

These moulding processes have been adapted by Truckenmuller et al. for microfluidic

fabrication in polystyrene [79]. Replicated channels were 150µm wide, 50µm deep, and

had corner radii down to 100µm. Truckenmuller further demonstrated the simultaneous

thermoforming and bonding of a thick polystyrene substrate containing electrodes to the

formed film.

Despite the simplicity of the thermoforming technique there have been few reports of its

use for microfluidic fabrication. This may be due to technical difficulties in achieving

micron sized dimensions, and the relatively large corner radii introducing the possibility

of dead volumes around the bonded edges of the channel walls.


17

2.2.3 Microfluidic Fabrication Summary

Polymers represent a low cost alternative to the traditionally preferred silicon and glass

substrates. The literature shows that significant advances have been made in the

development of microfluidic fabrication methods into the surfaces of bulk polymers. Of

particular interest for the mass manufacture of disposable microfluidic devices are the

recent trends in:

- reel-to-reel fabrication. Indicating a shift away from batch based processing towards

continuous low-cost manufacturing.;

- the structuring, aligning and bonding of thin polymer films. A technique for forming

complex three-dimensional devices with traditional two-dimensional machining

approaches. An approach compatible with reel-to-reel processing.

PET film is readily processed on these web based systems but suitable microstructuring

techniques need to be developed. Of the direct machining techniques that may be

compatible with reel-to-reel processing:

• Print-head deposition techniques have been reported but show structural chemical

compatibility limitations.

• Excimer laser systems are widely used and capable of producing sub-micron sized

features, however, they are relatively expensive and have slow machining times.

• Infrared lasers offer higher speed processing but introduce large heat affected zones

with channel dimensions down to 100µm.

• Newer solid state diode lasers have not been reported for microfluidic device

fabrication in PET, but offer possibilities of high speed processing with finer

resolutions than the infrared lasers.

• Femtosecond lasers offer high aspect ratio machining but are less widely used due to

equipment cost and relatively slow machining times.

Of the web-based process compatible replication technologies

• Hot embossing is commonly used for amorphous substrates but is not suitable for

high aspect ratio structure replication in orientated films.

• UV embossing has been reported commercially for microfluidic replication but is

limited in the surface chemistry.

18

• Thermoforming has only been demonstrated for surface structure formation with

relatively large dimensional constraints and dead volumes.

19

2.3 Microfluidic Biochips

2.3.1 Historical Perspective

The landmark paper by Watson and Crick in 1953 [80] proposed the double helix

structure of DNA and marked the beginning of a molecular biology revolution. Since

then important developments have been made in nucleic acid analysis such as DNA

hybridisation, the polymerase chain reaction (PCR), electrophoretic techniques, and

microarray analysis. These tools have led to greater sensitivity and throughput for

molecular analysis allowing whole genomes to be sequenced effectively, as has been

demonstrated with the completion of the sequencing of the human genome in 2001

ahead of schedule [81]. For molecular diagnostics to become a viable tool in the health

care industry, cost effective tools need to be developed for the screening of DNA

directly from biological samples.

Microsystems technology with the advantages of reduced reagent requirements, speed,

and integration have the potential to drastically reduce the current costs of clinical

diagnosis and move it out of the centralised laboratories to the point-of-care. To

increase the throughput of conventional equipment, Microdevices have been developed

in the areas of detection, amplification, and more recently involving sample preparation.

2.3.2 Biochip Detection

There are many gel-based electrophoretic techniques for determining single nucleotide

polymorphisms (SNPs), including denaturing gradient gel electrophoresis (DGGE),

temperature gradient gel electrophoresis (TGGE), and polyacrylamide gel

electrophoresis analysis (PAGE) for the analysis of small PCR-amplified DNA

fragments (<400 nucleotides). Many of these techniques can be inexpensively set up in

a modern molecular biology laboratory, however, they are labour intensive and time

consuming. For high-throughput analysis the newer technologies of multiple-lane

capillary electrophoresis (CE) and microarrays are capable of simultaneously screening

multiple polymorphisms in a single gene or in several genes.

Capillary electrophoresis, developed in the early 1980’s [82], enabled greater separation

resolutions compared to traditional gel electrophoresis by allowing the use of higher

20

separation voltages. The capillary format can dissipate more of the energy formed from

Joule heating. The operating principle between the two electrophoretic systems is the

same; the discrimination between mutant and normal genes is based on amplifying the

mobility differences that are due to the differences in their structures. Changing

parameters such as pH, temperature, and the surface chemistry on the capillary walls

can amplify these mobility differences leading to enhanced resolution. The sample is

then visualised by techniques such as autoradiography, absorbance, fluorescence, and

electronic detection .

Microarrays are an increasingly popular technique in molecular research that allow

researchers to simultaneously analyse thousands of DNA fragments. Microarray

technology exploits both the hybridisation features of DNA and its ability to bind to a

solid support and yet still allow hybridisation to occur. DNA microarrays consist of

sequence-defined nucleotide probes arranged on a solid substrate. For mutation

detection purposes these probes can represent all types of single nucleotide changes,

such as substitutions, deletions, and insertions. Fluorescently tagged PCR-amplified

products hybridise to target DNA and then fluoresce when energised. Comparison

between the signal obtained and the normal profile can then determine if mutations are

present [5].

2.3.2.1 Microarrays

The development of microarray technology began with Gillespie et al. in 1965 when

they observed that DNA attached strongly to a nitrocellulose membrane [83]. The solid

support prevented the single DNA strands from forming their duplex structures together

but still allowed complementary RNA to hybridize. Another major development was the

blotting method by Edwin Southern in 1975 who combined filter hybridization with gel

electrophoresis of DNA fragments and showed that labelled nucleic acid molecules

could be used to interrogate nucleic acid molecules attached to the solid support [84].

Further work in this area brought about the ‘dot-blotting’ procedure [85] which led to

the formation of arrayed nucleic acid probes on porous surfaces.

DNA microarrays differ from the dot-blots mainly in their size and thereby the density

of their arrays [5]. This is largely attributed to the impermeable solid substrates that are

used for the microarrays rather than the porous membranes and gel pads used in the dot

21

blotting techniques. Solid surfaces enhance the rate of hybridisation as the target nucleic

acids have direct access to the probes rather than having to diffuse into the pores first.

Likewise the washing step after hybridisation is unimpeded.

2.3.2.2 Polymer Biochips

Microarrays are typically manufactured on glass microscope slides with fluorescently

labelled probes due to their compatibility with existing laboratory equipment, well

understood surface modification methods, flatness, low intrinsic fluorescence and low

cost [5]. Transferring this technology into polymer microfluidic devices can lower the

cost of microarray analysis by allowing the integration of fluid handling components.

This integration of reagent handling can reduce capitol equipment requirements, shorten

analysis times, and simplify user protocols. The major issues in transferring this

technology into polymer substrates are the surface topology and chemistry, which are

critical factors that dictate the background fluorescence, and deposited spot size and

reproducibility. Small spots allow dense arrays with concentrated dots of fluorescence,

and the reproducibility is critical for avoiding false readings during signal acquisition.

Due to its importance in microfluidic operation there has been an extensive amount of

work done on surface modification of polymer substrates for microfluidic devices [10].

Techniques for modifying the surface chemistry to alter capillary driven and

electroosmotic flow characteristics, biocompatibility, and microfluidic channel sealing

have been developed for a wide variety of polymers. For microarrays the surface

chemistry is of particular importance for the reproducible formation of small (typically

200µm diameter), high-density oligonucleotide spots, reduced non-specific binding, and

availability of selected surface functional groups for covalent attachment [5].

There have been reports of new microarray detection technologies, including labelled

and unlabelled methods utilising electrochemical and electronic strategies, however,

none of these techniques have yet replaced the more widely understood and readily

available method of fluorescently labelled probes [86]. The intrinsic fluorescence of

organic polymers has hindered the uptake of polymer substrates for this application due

to their tendency to have a high background fluorescence, and therefore drastically

reduce the signal-to-noise ratio during detection [87-90], .

22

Liu et al. [88] compared the relatively high background fluorescence of PC with

PMMA, PP and PS with glass on a GenePix array scanner with the results showing that

PC had a lower fluorescence at 635nm, compared to 532nm, which although five times

higher than glass was sufficient to yield a satisfactory result with 20µM Cy5.

In a paper by Hawkins et al, the background fluorescence of some thin films were

compared with 20mM fluorescein and a nonlinear decay of the background intensity

was observed for most of the materials including PET over a period of up to three hours

[91]. They suggested that a burn-in time to overcome the initial intensities and to allow

for a more consistently reduced background level.

It has been shown that with careful selection of the optical excitation and detection

wavelengths, fluorescently active polymers can be used for microarray detection [88]. A

look at the intrinsic fluorescence of PET, performed by Allen et al., showed that the

monomer unit dimethyl terephthalate does not fluoresce above approximately 470nm,

and although its phosphorescence emission spectra does extend up to approximately

560nm, it requires excitation below 340nm [92]. This suggests that microarray detection

using labels such as Cy5 would be feasible with a pure PET substrate.

2.3.3 DNA Amplification

The relatively low sensitivity of traditional DNA detection techniques has required that

DNA amplification be performed to increase the availability of the target sequence to be

analysed. There are many techniques available for the amplification of specific DNA

sequences, by far the most commonly used and widely understood method is the

polymerase chain reaction (PCR), first developed in 1986 by Kary Mullis and co-

workers [93].

2.3.3.1 PCR Amplification

The polymerase chain reaction is recognized as a breakthrough in molecular

diagnostics, simplifying and increasing the reliability and specificity of detecting DNA

sequences. There are many varieties of the PCR processes, all of which fundamentally

operate by the enzymatic synthesis of specific DNA sequences. The process is cyclic

23

with three distinct stages; denaturation, primer annealing and extension. Initially the

template is denatured and two primers are annealed to two separate sites on opposite

strands, then the annealed primers are extended by the polymerase enzyme to give

overlapping copies of the original template. This cycle is repeated, typically 20-30

times, to give an exponential increase of the specific fragment whose ends are

terminated by the two primers [94].

The exponential nature of the amplification causes the process to be susceptible to

contamination, mis-priming and the formation of false artifacts. Furthermore the

efficiency of the amplification is very sensitive to variations of the experimental

conditions, and different experimental conditions often dictate an adjustment to the

delicate balance of reagents used [94]. Many groups are investigating microfluidics as a

method to reduce the volumes used, typically 20-50µl, and speed up the process.

2.3.3.2 MicroPCR

Of the many amplification techniques available the PCR lends itself well to

miniaturisation. This is due to the lower thermal cycling times required and need for

uniform heat distribution which are improved by the reduction in bulk materials.

Increased levels of integration enabled by the miniaturisation process leads to simplified

experimental protocols, reducing operator handling errors and sources of contamination.

For these reasons there has been a great deal of activity in miniaturising the PCR

protocol onto a microchip platform [95]. Of particular importance in these microfluidic

devices is the increase in surface-to-volume ratios that amplify any surface interaction

problems.

The earliest microfabricated PCR devices were made from glass and silicon due to the

pre-existing fabrication knowledge and good heat conduction properties of the material.

Woolley et al [96] demonstrated successful amplification in simple wells etched in

silicon substrates with glass lids and external heating. A further study by Shoffner et al.

showed that untreated silicon microchips inhibited the PCR reaction [97]. They

investigated several surface passivating strategies to overcome this problem with the

best results achieved from the use of an oxide layer [98]. Other passivation techniques

such as silanisation followed by molecular grafting have been demonstrated [99]. This

24

grafting technique has been used for pacifying silicon walls in capillary electrophoresis,

which has met with varying degrees of success due to the quality of the deposited layers

[98].

These simple single well devices have evolved from thermally isolated multi-well chips

for individually controlled parallel reactions, to more complex biochips having

microfluidic channels and valves for reagent delivery, and integrated heaters for faster

response [95]. Alternative approaches to these static PCR methods, which hold the

reagents stationary and thermocycle the chambers, use continuous flow PCR systems.

These continuous flow devices pass the PCR mixture between fixed temperature zones.

This configuration shortens the cycle times by reducing the thermal delays from the

temperature cycling of the substrate.

Nakano demonstrated a continuous flow system by pumping the reagents back and forth

over three fixed temperature zones in a capillary [100]. Kopp et al. further extended this

concept onto a planar substrate with the micromachining of a glass chip [101].

Continuous flow systems with integrated functionality have been reported. These

include the silicon and glass system developed by Schneegass and coworkers with

integrated heaters and temperature probes [102], and the novel circular approach from

Liu et al. with pneumatically driven pumps that provides the flexibility to choose the

amount of temperature cycling during the experiment [103].

Various detection strategies integrating PCR have been reported. These devices range

from the standard T-shaped electrophoretic chips where the PCR is performed in the

inlet well before injection [104], to the more complicated chips with integrated valves

[105-106]. The research in these glass and silicon based devices has progressed to the

point where single molecule amplification and subsequent detection is now possible in a

single integrated microfluidic chip [107].

More recently polymer based microfluidic PCR devices have been reported, and have

highlighted the key issues of surface effects inhibiting the PCR process [108],[109]. Zou

et al. demonstrated PCR amplification in a multiwell disposable microchip using a PET

substrate formed with a silicon mould and sealed with a polypropylene tape [110].

Complex polymer devices have been demonstrated by Aclara with a poly(cyclic olefin)

25

microchip for PCR and capillary electrophoresis with integrated electrodes and gel

valving [111], and Motorola Labs produced an integrated amplification and microarray

detection device in polycarbonate with Pluronic heat actuated valves [112]. A

comprehensive review of PCR microfluidic devices for DNA amplification was

presented by Zhang et al. [95].

2.3.4 Biochip Sample Preparation

In the clinical environment DNA samples may be collected from a variety of sources

including whole blood, body tissues, sputum, urine, and faeces. Whole blood was

chosen as the sample of interest in this thesis because it is used as a regular source in

forensics and blood banking, and is used extensively in current PCR-based diagnostics

of microbial infections and genetic diseases [113].

Whole blood is a complex mixture of cells, peptides, proteins, lipids, carbohydrates and

other low molecular weight compounds. Many of these can interfere with the detection

processes, as has been shown in recent studies for immunoglobulin G in plasma [114],

polypeptides and haemoglobin in erythrocytes and lactoferrin in leukocytes [113].

These inhibitors need to be minimised before PCR can be carried out. The addition of

amplification facilitators such as Bovine Serum Albumin (BSA) has been used to

counteract some of these inhibitory effects as investigated by Wl-Soud et al. for

standard PCR mixtures [115], and on microfluidic platforms [71],[116].

The major steps in DNA isolation from biological samples are cell isolation, cell lysis,

purification, and recovery of DNA. There are many conventional techniques and

protocols available that combine some or all of these steps. Typically cell isolation is

performed by centrifugation or vacuum filtration then the desired cells are lysed with

the resulting DNA being purified and isolated by the binding of genomic DNA to a

solid phase [117]. The purified nucleic acid can then be reduced in size by enzymatic or

chemical methods, which usually occurs within the PCR amplification process.

Most current microfluidic DNA sample preparation systems can be characterised in two

ways - cell isolation then purification before lysis, or solid phase extraction. In both

cases it is necessary to trap or filter either the leukocytes or the solid phase extractant in

26

a microchannel. Filtration is of interest not only to filter particles, but to other

disciplines, such as adsorption and ion exchange chromatography, reactor engineering,

and separation chemistry. For this reason microfluidic devices with integrated filtration

that have not been used for DNA purification are discussed next, followed by a review

of the microdevices that have been specifically made for DNA extraction. A review of

solid supports fabricated in microdevices is described by Peterson [118] and a review of

artificial molecular sieves is described by Fu et al [119].

2.3.4.1 Filtration in microchips

In microfluidic devices many real-world samples require filtration before analysis to

avoid large particles from fouling or blocking the channels. Methods for particle

trapping are also required to immobilise microspheres or other solid phases for

subsequent chemical reactions. These microfiltration devices can be categorised into

two types; continuous flow systems that require streaming fluids, and systems that trap

particles in one location.

2.3.4.1.1 Continuous flow filters

It remains to be seen if continuous flow systems can effectively extract DNA from

whole blood. Chen et al demonstrated a continuous flow mixer for cell lysis prior to

solid phase extraction [120],[121],[122]. Although barrier free devices have been

reported for plasma separation from blood based on diffusion [123] and centrifugal

force [124], greater separation resolutions are required to distinguish between cell types

or extract DNA from the complex mixture of lysed whole blood without the use of a

solid binding phase.

2.3.4.1.2 Particle trapping

The typical method for filtration used in microsystems is the use of inchannel

restrictions, however, the implementation of these systems in both fabrication and

design vary between researchers. Particle trapping without channel restrictions has been

performed using surface functional groups for chemically binding particles [125], with

magnetic fields for use with paramagnetic particles [126],[112], or localised potential

wells for trapping charged particles [127].

27

In-channel restriction can include the incorporation of external filter membranes

between laminated layers as shown by Karp et al [128]. The fabrication of filters via in-

situ polymerisation was demonstrated by Morothy et al [129], where an emulsion was

photo-polymerised to form a porous network of polyHEMA spheres. The filters were

fabricated reproducibly in glass chips however, the researchers noted the importance of

the channel surface chemistry and the difficulty in achieving polymerisation in polymer

microchannels.

Ion track etching is a technique commonly used in fabricating polymer based laboratory

filters and has been adapted for microfluidic manufacturing by Metz et al. with

polyimide films [130]. With this method they were able to fabricate filters within 3-D

microfluidic structures with pore dimensions between 50nm and 2µm and successfully

show the retention of 500µm and passage of 300nm particles. Limitations of the system

include the random pore distribution, which causes larger pores to be formed from pore

overlap and a weakening of the membrane for higher pore densities, the instability of

many of the etchant solutions, the surface modification from the etchant process, and

the implementation of such a fabrication method in continuous manufacturing.

Other filters fabricated with traditional silicon fabrication techniques include silicon

weirs [67],[19], posts [67],[21],[131],[132],[133],[134] and membranes [23],[135].

Examples of devices with smaller effective pore sizes used for molecular sieves and

filters are reviewed by Fu et al [119].

Wilding separated leukocytes from whole blood for DNA extraction and his work is

described in detail in the next section. Crowley used a weir filter <1um high etched in

silicon to separate blood plasma from whole blood [136]. Initially high capillary

pressures of 20psi in the system caused haemolysis, however, this was eliminated by

altering the design for reduced pressure and residence time. For the trapping of

microsphere particles for chemical and biochemical applications Paul et al demonstrated

a flow through chamber with weir side exits to give regular particle packing (Figure 4a)

[137].

28

Another silicon etched device was produced by Anderson et al. [132] with high aspect

ratio pillars to form a high flux reaction chamber inside a larger waste chamber (Figure

4b). Similar in-channel posts have been fabricated by Sato et al. [133] and Yoon et al.

[134] by angled exposure photolithography to produce thick mesh screens with pore

sizes down to 10µm (Figure 4c). He et al. [138] demonstrated a novel approach to

sample inlet filtering by using lateral percolation filters etched in quartz (Figure 4d).

Figure 4 a) Flow through chamber with weir exits designed for even particle

packing [137], b) Silicon post reaction chamber for high flux [132], c) mesh filter

[133], d) lateral percolation filter [138].

Kuiper has performed extensive work using intereference lithography to form silicon

membranes for lager beer filtration [23]. Leoni and Desai microfabricated membranes

by silicon etching with pore sizes down to 7nm with tailored surface chemistries for cell

immuno-isolation and viral filtration [135].

There are few reports of filters integrated into polymer microfluidic devices. With the

increased focus on disposable microdevices there is a need to develop reliable,

reproducible and cost effective fabrication methods to integrate filtration into polymer

a) b)

c) d)

Copyright restricted images

29

microdevices. Photolithographic laser machining of thin films has been shown to be a

suitable method to fabricate complex microstructures with sub-micron accuracy [139]

and represents an opportunity for filter fabrication.

Kuiper fabricated membranes using both photolithographic and hot embossing

techniques but these were not incorporated into any microdevices [23]. Multiprocess

steps were required to produce the sub-micron thickness membranes and difficulties in

separating the mould without damaging the sample were encountered for the embossed

filter. Weir based filters provide a simpler design for replication in polymers due to

their low aspect ratio, but require a rigid sealing layer to maintain the gap tolerance.

2.3.4.2 DNA extraction in microchips

Two common methods of DNA extraction used for isolating genomic DNA from whole

blood are leukocyte isolation followed by lysis, and solid phase extraction.

2.3.4.2.1 Leukocyte isolation

The first integrated whole blood sample preparation microchips were published in 1998

by two groups: Wilding and Kricka’s group at the University of Pennsylvania who

used microfilters and PCR for DNA amplification from whole blood with no on-chip

detection [67]; and Cheng et al. who used dielectrophoretic separation of E.coli and

cultured cervical carcinoma cells from whole blood [140].

Continuing on from their previous work on performing polymerase chain reaction in

microchips directly from isolated leukocytes [141], Wilding and his team fabricated an

integrated cell isolation and PCR device using silicon microfilter chambers [67]. Filter

designs were investigated to optimise the white blood cell (WBC) isolation and red

blood cell (RBC) removal on the basis that a 10% efficiency of WBC isolation and a

99.9% removal of RBC is sufficient for PCR of genomic targets. Due to the

deformability of the cells the filter gaps needed to be substantially smaller than the

actual size of the cells to be isolated. Weir type filters were preferred due to their ease of

manufacture in comparison to comb filters. This was due to the silicon wet etching

fabrication process used that has difficulties in etching fine gaps (<5um). With a weir

filter containing a 3.5µm gap and being 250µm in length in a channel 2-4mm wide they

30

were able to achieve a WBC isolation efficiency of between four and 15% sufficient for

electrophoretic analysis in fewer than eight minutes.

The approach by Cheng et al. was to manipulate charged particles with an electric field

[140]. Electronic microelectrodes were used to separate cells from the complex solution,

then lyse the isolated cells, before attracting the DNA for hybridisation. After

hybridisation the polarities of the electrodes were reversed to allow the nonspecific

sample DNA and un-hybridised probes to be removed selectively from the specific

probe location. They demonstrated these techniques for the separation and isolation of

cultured cervical carcinoma cells from normal blood cells [142], and the separated

E.coli cells from whole blood [140].

Other authors have presented articles on integrated systems that perform cell lyses,

amplification and detection on trapped cells, however they are from less complex

mixtures than whole blood. For example Waters et al. developed a microchip for the cell

lysis, multiplex PCR and CE separation of E.coli cells in aqueous solution [143].

2.3.4.2.2 Solid Phase Extraction

Solid phase extraction is a very common laboratory procedure used for the isolation and

purification of DNA from complex mixtures including whole blood. There has been

much activity into miniaturising this process due to the effectiveness and ease by which

silica can be incorporated into microdevices through either surface coatings, silica

deposition on silicon, or silica microparticle trapping. A review of solid phase extraction

methods in microfluidic devices is found in reference [144].

The solid phase extraction procedure relies on the properties of DNA binding to silica

under chaotropic conditions, which was first discovered by binding DNA fragments to

powdered flint glass in the presence of sodium iodide [145]. The procedure was then

developed for other nucleic acids using different chaotrops [146], with the binding

process dependent on the ionic strength and the pH of the lysis and binding solution.

Elution of the DNA from the silica is performed washing with a low salt buffer or

water.

31

The surface of silica is covered by OH groups bound to a SiO2 skeleton. It is at these

points that water molecules, and many other organic molecules with polar groups are

adsorbed through hydrogen bonding. Amorphous Silica is arranged in a non-regular

structure, therefore the hydroxyl groups attached to these silica atoms are not all

equivalent in their adsorption characteristics. However, it has been shown that particle

size or radius of curvature of the surface may be a more important factor in the surface

adsorption than the differences between the amorphous and ordinary states of silica.

Surface impurities can also greatly modify the surface properties [147].

The first reported DNA solid phase extraction on a microchip was by Christel [21] with

a DRIE silicon chip having a reaction chamber coated with silica containing 20µm

diameter pillars, 200µm high at a pitch of 34µm for increased surface area. Extraction

efficiencies of around 50% were demonstrated with bacterial lambda DNA as the target.

Since then similar in-channel silica coated pillars have been demonstrated by others

[131].

Microdevices made of borofloat glass were investigated by Wolfe et al for incorporating

silica matrices for the solid phase extraction of λ-phage DNA and recovery suitable for

PCR amplification [148]. The silica matrices investigated were silica beads, sol-gels,

and a combination of the two.

The 5% glycerol/water solution of silica beads were packed in the microchips by

vacuum. The extraction efficiencies achieved were up to 80% but highly irreproducible

between runs and on different chips. A possible cause of this may have been the change

in packing densities during operation. Observations showed that as the chips were

operated the beads packed more tightly, thus increasing the backpressure, reducing

flow, and allowing for different flow paths. In some cases continued use destroyed the

chips.

To avoid these bead retention and packing problems two base catalysed sol-gel methods

were investigated in capillary columns. The better extraction efficiencies were obtained

with the more hydrophilic gel, however, neither performed as well as the chip with only

beads.

32

The immobilization of the beads in a solgel network resulted in better reproducibility

than with the sol-gel networks alone. This gel and bead based strategy was then used to

extract 500bp lambda and human genomic DNA [149]. The optimization of the load

buffer pH and flow rate found that for faster extraction times, lowering the pH increased

binding and resulted in faster absorption. Extraction efficiencies were not given,

success was determined by off-chip PCR amplification and subsequent identification

with capillary electrophoresis. This work was then extended by the same group to

incorporate on-chip PCR amplification [150] and CE detection [17]. The Human DNA

was found to have lower extraction efficiency than lambda DNA, possibly due to

inefficient elution of the adsorbed DNA and not the inefficient retention of the DNA on

silica as previously reported [151]. However, more recent work by Wen et al. showed

that the binding capacity of C18 silica for DNA was reduced by protein absorption

when using whole blood. A protein capture bed prior to DNA extraction was shown to

improve the recovery of DNA [152].

There have been few reports of polymer based microfluidic devices employing the SPE

method to extract human genomic DNA from whole blood. However, polymer based

microfluidic devices have been reported using SPE for other applications. Examples of

these include the use of solid phases trapped by magnetic fields [112],[153] or

immobilized in a UV-polymerised monolith [154].

2.3.5 Biochip Summary

Microfluidics offers advantages of reduced reagent requirements, speed, and integration

with the potential to drastically reduce current costs of clinical diagnosis and move them

out of the centralised laboratories to the point-of-care. Towards this goal microdevices

have been fabricated for the various stages of sample preparation, amplification and

detection.

Various on-chip detection strategies have been investigated, with capillary

electrophoresis being the most popular, it is however the microarrays that offer higher

throughput SNP analysis. Glass has been the traditionally preferred substrate with recent

33

reports citing polymers highlighting the background fluorescence as a problem for

fluorescent detection.

PCR is the industry standard amplification technique and there has been extensive work

in miniaturizing devices in silicon with fewer recent publications on polymer

integration. These new reports of polymer devices have raised the importance of surface

interactions for biocompatibility.

A sample preparation technique is necessary to remove the many known PCR inhibitors

in whole blood before amplification. Two on-chip approaches have been taken for DNA

sample preparation; leukocyte filtration and solid phase extraction. In both cases an

integrated filtration method is commonly used. There are few reports of polymer

microfiltration devices for extracting human genomic DNA from whole blood.

2.4 Conclusion

The motivation behind this research is to develop techniques for the commercially

viable production of disposable biochips for genetic screening. For this reason the

literature review was broken up into two main areas; microfluidic fabrication and

biochip devices.

Early reports of microfluidic devices used silicon and glass substrates and were

fabricated by methods traditional from the semiconductor industry. More recent

advances investigate polymers as a cheaper alternative. The structuring, aligning and

bonding of polymer films offers a technique for forming complex three dimensional

devices with traditional two dimensional machining approaches. This technique is also

compatible with continuous manufacturing; a manufacturing approach that promises the

low-cost mass production of microfluidic devices. However, it is necesasry to develop

techniques compatible with web-based systems for microstructuring these continuous

polymer films.

A promising technique for microfluidic structuring is laser machining by frequency

tripled Nd:YAG lasers. This technique offers speed improvements in comparison to the

more traditional excimer machining systems, and improved feature size formation in

34

comparison to IR lasers. Their solid state operation also offers advantages compared

with femtosecond and excimer laser systems in terms of cost of ownership, smaller

footprint, no toxic gas requirements, and less maintenance.

PET is investigated as the substrate in this thesis as it is readily processed on web

systems in the printing and packaging industries and has been shown suitable for

microfluidic fabrication by UV laser machining. However, there is little literature

reporting the ablation of PET using frequency tripled Nd:YAG lasers. Although the

individual photon energies of the frequency tripled Nd:YAG lasers are not high enough

for direct electronic excitation, it remains to be seen how these higher frequency lasers

perform in PET for micromachining microfluidic channels.

For the development of a genetic screening biochip, the three processes of sample

preparation, amplification, and detection are required. Microarray analysis promises

highly parallel detection and recent microfluidic reports investigating polymer

substrates have highlighted problems with surface chemistry and background

fluorescence. Polymer microfluidic devices integrating PCR amplification have been

investigated, raising issues of the surface biocompatibility. Sample preparation

techniques have been less developed with few reported cases of filtration utilising

polymer substrates. With the increased focus on disposable microdevices there is a need

to develop reliable, reproducible and cost effective fabrication methods to integrate

filtration into polymer microdevices. Laser machining of polymer films has been

shown to be a suitable method to fabricate complex microstructures in polymers with

sub-micron accuracy and represents an opportunity for both filter and channel

fabrication.

In conclusion, there is a need for sample preparation in polymer devices suitable for

integration with PCR and microarray analysis, and fabricated in a manner compatible

with web-based manufacturing techniques. Towards this goal PET is investigated in this

thesis as a suitable material for the microfluidic fabrication of a filtration and DNA

extraction device. Issues relating to methods of laser microstructuring suitable for

continuous manufacturing, and the materials’ compatibility with integration of PCR and

fluorescent microarrays are also investigated.

35

Design of a PET Biochip for DNA extraction

The biochips developed in this thesis are designed to perform DNA isolation from

whole blood samples. The biochips are fabricated from five layers of crystalline PET

film with each layer machined entirely through, and then stacked and bonded together.

An in-line filtration design was chosen due to its ability to be used as a sieve for

leukocytes or as a retaining mechanism for a solid phase without complicating the

fluidic pathway. Alternative cross flow designs were not investigated as they typically

require either a larger effective filter area per unit volume, or multiple passes through

the same filter with suitable transmembrane pressures to achieve high separation

efficiencies.

The devices are fabricated with ports large enough to allow both the connection of

external electrodes for electroosmotic flow (EOF), and tubing interconnects for external

pressure driven pumping. As shown in Figure 5 , each microfluidic device contains two

inlet ports for reagent and sample introduction, a relatively long inlet channel for

packing, a filtration chamber, and a single outlet channel. The filtration chamber was

chosen to provide a 1mm x 2mm filter with no in-chamber membrane support.

Inlet ports

Outlet port

FilterChamber

Figure 5 Microfluidic filter design

36

3. PET Characterisation and Modification

3.1 Introduction

Polymers are commonly used in many products due to their low cost and well-

established methods of processing. The fabrication of the laminated microfluidic

devices discussed in this thesis requires the use of thin film substrates (typically

<250µm). For biochip operation these thin films need to be compatible with the

fabrication techniques as described in chapter 5; be temperature stable up to 100oC for

PCR amplification; not inhibit the biological protocols; and have relatively low

background fluorescence suitable for microarray detection.

Orientated PET films, sometimes referred to as crystalline poly(ethylene terephthalate)

(cPET) films, have been investigated as the substrate for this thesis. These films are

available in a wide variety of thickness (typically from 6-250µm), are cost effictive and

commonly used in the printing and packaging industries. These materials have been

shown to ablate with current laser processing techniques [60], and have suitable thermal

stability, mechanical strength, clarity, chemical inertness and relatively low water

permeability [155]. Examples of microfluidic devices fabricated using thin film cPET

have been shown using Excimer laser machining and sealing by lamination [156],[157].

In this chapter the surface and bulk characteristics of selected cPET films and methods

for their surface modification are investigated and discussed in relation to biochip

operation. The film’s thermal and optical properties are characterised for the biochip

fabrication method described in Chapter 5. Techniques of chemical treatment and UV

modification are detailed, with the results discussed in terms of their effect on the

surface chemistry, fluorescence and biocompatibility.

3.2 Background

3.2.1 PET

PET is a linear polyester and was first formed by the condensation reaction between

terephthalic acid and ethylene glycol, but it is now more commonly made commercially

by the ester interchange of dimethyl terephthalate [158]. The cPET films investigated in

37

this thesis are oriented (or annealed) after extrusion giving a semi-crystalline structure

for improved mechanical strength, transparency and permeability when compared to

their amorphous counterparts. These films are made by extruding PET, followed by

immediately quenching to avoid crystallisation, and then biaxially orientating (i.e. the

film is stretched in two orthogonal directions) above 80oC. The heat stability is further

improved by causing partial crystallisation of the film by heating it to above 200oC

followed by rapidly cooling the film whilst holding it to prevent contraction. Thicker

films, typically for films >100µm, usually only undergo an annealing stage to avoid the

film sagging under its own weight in an orientation process.

Many of these commercially available cPET films contain proprietary additives for

improved handling, UV resistance, and clarity, with some films having surface

treatments for improved binding of inks or adhesives. For the purpose of this thesis

several commercially available films were investigated for comparison, however, the

exact nature of their additives and surface treatments were not known, nor were their

manufacturing processes or handling conditions before they arrived in the laboratory.

3.2.2 Surface Modification

In many applications polymers often have the desired bulk physical and chemical

properties but have unsuitable surface characteristics and hence require surface

modification techniques prior to their use. Many methods have been developed in the

printing and packaging industries to improve these properties including modifying the

surface energy for wettability and enhanced adhesion of adhesives, inks, and other

surface coatings (see Table 1) [159].

Table 1 Surface treatment technologies [159]

Technique Process Types Technology Status

Comments

Abrasion Mechanical Dry or wet blasting, hand or machine sanding

Obsolete Labor-intensive, dirty, applicable only for low production volumes, must deal with residuals.

Solvent cleaning

Physical and Chemical

Wiping, immersion, spraying or vapor degreasing

Obsolete Safety, disposal and environmental concerns (i.e, emissions)

Water-based cleaning

Physical Multistep power wash Contemporary Low environmental systems impact, high volume capacity, and relatively low cost.

Chemical Chemical Immersion, brushing, rinsing, Obsolete Safety issues due to the use of

38

etching with acids or bases

spraying corrosive, toxic materials and hazardous-waste disposal problems.

Chemical primers

Chemical Solution application of poly ethyleneamine, polyurethanes, acrylates, chlorinated polymers, nitrocellulose, or shellac

Mature Requires specific equipment, and different primers are necessary for specific end-use requirements.

Flame treatment

Thermal and chemical

Available for flat films or three-dimensional configurations

Mature Fire hazard, limited to some extent to thermally insensitive materials.

Corona discharge

Electrical and chemical

Available for both conductive and dielectric substrates

Contemporary Applicable primarily to films and webs

Gas plasma Electrical and Chemical

Available for film or three dimensional applications can use ac, dc, or microwave frequency

Contemporary Convenient and cost effective; non toxic materials or disposal issues; can be effective in numerous different configurations

UV and uv/ozone

Electrical and Chemical

For distinct parts in batch systems.

Developmental, Contemporary

Generally only in batch format and requires longer residence times

Evaporated acrylate coatings

Physical and Chemical

Currently for webs and films only Developmental, Contemporary

Still being developed for commercial-scale applications

Fluorination Chemical Short exposure to elemental fluorine can be batch or continuous

Developmental, Contemporary

Specialized equipment required for delivery and monitoring fluorine.

Electrostatic discharge control

Electrical Can be in the form of charge dissipation or charge neutralization

Contemporary Equipment can be simple through complex and expensive, depending on the application

In microfluidics, the small dimensions (1-500µm) and large surface volume ratios

(typically >> 10) enhance the effects seen from the surface and liquid interaction.

Controlling these surface energy interactions is critical for many microfluidic

applications such as sample introduction through capillary force [160], passive valve

operation [161], droplet manipulation [162], electroosmotic flow [163],[164], and

reduced non-specific binding [165]. In applications where molecules are required to be

tethered to the surface, such as oligonucleotide probe or antibody attachment, the

presence of certain surface functional groups can be used to provide anchors for

adsorption and covalent linkage [166],[167],[168].

Corona discharge is readily used in continuous manufacturing of cPET films in the

packaging industry and has the advantages of short exposure times and simple

integration into current manufacturing processes. However, reports suggest corona

treatment has problems with surface roughness and limited surface charge lifetimes

[169].

Chemical etching was initially chosen as a surface modification method as it is both a

relatively fast and simple process to implement for batch based systems, and has been

39

implemented in the printing industry on reel-to-reel systems. UV modification was

investigated as a non-contact process that enables the complex patterning of surfaces via

mask projection. Exposure systems are already available in manufacturing for UV

curing applications. UV treatment has been shown to provide a lower surface

functionalisation density than other plasma techniques and prolonged exposure degrades

the polymer creating a weak boundary layer [170], however it remains to be shown if

the technique is suitable for cPET microfluidic devices. Furthermore, the different

types of cPET films may process differently due to additives within a film having a

significant effect on the surface oxidation and lifetime of the induced surface charge

[169]. For these reasons experiments were performed on the films used for microfluidic

fabrication to characterize their performance in relation to biochip operation.

3.2.3 Oxidation

Polymer oxidation can occur when a polymer is exposed to UV radiation in the presence

of oxygen. This process known as photo-oxidation involves the four main categories of

initiation, propagation, chain branching, and termination [171].

Initiation takes place when chromophores within the polymer, or impurities ( P ) absorb

a photon ( hν ). For hydrocarbon molecules this absorption of UV radiation most likely

results in scission of a C-H bond in the order of tertiary > secondary > primary bonds.

PHhν (O2)

H+P

Propagation involves the free radicals interacting with oxygen forming peroxy-radicals,

which may then abstract a hydrogen from another polymer molecule. Hydrogen

abstraction by free radicals occurs in the order tertiary>secondary>primary C-H bonds

and is independent of the type of attacking radical.

P + O2 POO

+POO PH +P POOH

Chain branching proceeds with the photolysis of the O-O bond to form oxy and

hydroxyl radicals. These may then add to the propagation by further hydrogen

abstraction from the polymer, or react in a number of ways resulting in aldehyde,

ketone, carbonyl, or hydroxyl formation.

40

Termination occurs when the free radicals combine with each other forming more stable

species. Species such as POOP and POOH are metastable and can degrade over time

into their oxy radicals for further reaction.

With the presence of non-hydrocarbon elements in the polymer, or polymer matrix in

the case of some additives, the situation is further complicated by these groups acting as

initiators for further photo-oxidation. Chromophores, such as the carbonyl groups in

PET, can absorb UV light exciting to singlet and triplet states, with the triplet state

being the more reactive and consisting of a bi-radical molecule. In some cases the triplet

states can abstract hydrogen from the polymer, producing alkyl radicals that can react

with oxygen as with initiation. Two common reactions resulting in chain scission from

these two excited carbonyl states are the Norrish Type I and the Norrish Type II

reactions. These Norrish reactions lead to a range of radical species that can terminate

with carboxylic acid end groups or produce carbon dioxide leaving aldehyde terminal

groups.

CH2 C

O

CH2 CH2 C

O

CH2+ CH2 CO CH2+ +hν

Figure 6 Norrish Type I reaction [171]

CHH

CH2 CH2

C

O

CH2+ H2C CHC

OH

H3C C

hν

O

Figure 7 Norrish Type II reaction [171]

These processes were confirmed for cPET by Day et al who reported the primary

photolytic process of cPET degradation from long wavelength UV light (>310nm) to be

photolytic cleavage, or subsequent hydrogen abstraction, followed by radical formation

and subsequent reactions to form –COOH [190].

41

3.2.4 Thermal Degradation

Thermal degradation reactions of polymers can be divided into two categories;

depolymerisation and substituent reactions. Depolymerisation reactions are

characterized by the breaking of the main polymer chain, giving lower molecular weight

products that are similar to the original material. Substituent reactions occur when the

side chains react, producing a polymer where the monomer units are altered [172].

It has been shown that for PET thermal decomposition occurs randomly along the

polymer chain with the major mechanism being alkyl-oxygen scission involving a six-

membered ring transition state producing vinyl and carboxyl end groups, see Figure 8.

Continued degradation allows radical formation followed by hydrogen abstraction,

resulting in many reaction products including anhydride groups, ketones, methane,

acetylene, benzoic acid acetphenone, vinyl benzoate, and anhydride groups; with the

main volatile products comprising terephthalic acid, acetaldehyde and carbon monoxide

[172].

Figure 8 Mechanism of thermal degradation of PET [172]

3.2.5 Fluorescence

Many polymers exhibit phosphorescence or luminescence in the visible region from

chromophores in the monomer unit or from additives like plasticisers and UV

stabilisers. Fluorescence is due to chromophores in the material absorbing photons and

reemitting light from the decay of their excited electrons.

A photon may be absorbed if it has energy greater than or equal to the energy difference

between the molecule’s ground and excited states. The excited electron may then

dissipate this energy by radiative transitions, non-radiative transitions, bimolecular

42

deactivation or dissociation. The major radiative and non radiative processes can be

represented by the Jabowlski diagram in Figure 9. The radiative processes are

collectively known as luminescence and are made up of fluorescence and

phosphorescence. Fluorescence occurs from electron decay to a state of the same

multiplicity and is characteristic of relatively short lifetimes, approx 10-7-10-9s. Whereas

phosphorescence is associated with longer lifetime, approx 10-10-3s, from transitions

between states of different multiplicities [173].

S1

S2

S3

T1

T2

T3

Intersystem crossing

(ISC)

S0

Ground state

Singlet (S) manifold Triplet (T) manifold

Figure 9 Energy level diagram for a typical organic molecule; non-radiative

transitions and radiative transitions [173]

Allen et al compared the fluorescence and luminescence of PET film with the monomer

unit dimethyl terephthalate [92]. Their results showed that the fluorescence spectrums

do not match, see Figure 10 a) suggesting that the emission is not from the monomer

unit itself. They ruled out the excimer state, that had previously been suggested as a

mechanism, as there was a difference in the excitation spectrums which is not possible

where an excimer dissociates in the ground state. Instead they suggested that the

fluorescence is from an associated ground state dimer by analogy with poly(N-

43

vinylcarbazole) and poly-1-and 2-vinylnapthalenes, which showed a similar shift in

emission spectrum from the presence of an associated ground state dimer formed

between adjacent side groups. However, the phosphorescence spectrums of PET and the

monomer unit did match in wavelength (see Figure 10 b) and lifetime, suggesting that

the phosphorescence is due to the dimethyl terephthalate monomer unit.

Figure 10 PET a) Fluorescence and b) Luminescence excitation and emission

spectra [92]

Polymers have had limited use as substrates for microarray analysis partly due to their

relatively high background fluorescence [5]. Microarray detection techniques require

the substrate’s photon emission to be significantly lower than the emissions from the

sample dye near the wavelengths of interest for good signal-to-noise characteristics. As

discussed in Section 2.3.2 some researchers have highlighted the problem of

background fluorescence when using polymer substrates for microfluidic devices [174],

[90], [89]. They did not describe the mechanisms for the fluorescence or methods that

could be employed to overcome this limitation for microfluidic applications. However,

Hawkins et al measured the fluorescence of several materials including a cPET film

over a 3-hour period and observed a photobleaching effect of the material under

continued exposure [91], and suggested that allowing a burn in time to pre-bleach the

material prior to use may improve the signal-to-noise characteristics.


44

The results presented by Allen et al showed that the PET film does not fluoresce above

approximately 470nm, and although its phosphorescence emission spectra does extend

up to approximately 560nm, it requires excitation below 340nm [92]. This suggests that

with the careful choice of excitation and detection optics pure PET film could be used

for fluorescent microarray detection. However, additives such as light stabilisers and

antioxidants exhibit their own characteristic spectrum. One of the most commonly used

pigments, titanium dioxide, has been well characterised showing its anatase crystalline

form to have an excitation peak at 340nm with a strong emission peak at 540nm for low

temperatures [173]. Furthermore, the intensity of this emission has been shown to vary

according to the different surface treatments that the titanium oxide undergoes.

Manufacturing conditions under which polymer films are processed vary between

manufacturer, and even between each production line [175], giving rise to different

amounts of crystalline structure (proximity for inter molecular interactions), and having

different proprietary additives and surface treatments with different chromophores.

These factors made it necessary to characterise the films used in this thesis for the

selected biochip application.

3.2.6 Biocompatibility

The biocompatibility of a material generally describes the degree to which the material

affects a biological process. For in-vitro biochips a major problem is the non-specific

binding of biological molecules to the substrate’s surfaces. The adsorption of large

macromolecules such as DNA and proteins can adversely affect sensor performance.

Adsorption onto a surface can prevent the molecules under test from reaching the

sensor, initiate various undesirable biochemical responses, or reduce the availability of

reactants in solution [176], [177].

Proteins are large macromolecules that have complex geometries with secondary and

tertiary structures formed from intra-molecular forces within the polymer chains. Their

conformational structure and resultant surface chemistry dictates the protein’s

interaction with external bodies and surfaces. Many researchers have investigated

protein binding on surfaces discussing mechanisms in terms of entropy, and ionic,

hydrophobic and van der Waals forces [166]. It is generally understood that the surface

45

energy of the substrate has a major impact on protein adsorption. For very hydrophilic

surfaces the driving force for protein adsorption is considered low. In contrast, the

dehydration at hydrophobic surfaces promotes adsorption resulting in conformational

and therefore functional changes of the protein for strong hydrophobicities.

Double stranded DNA is a linear polymer made from a combination of four bases

(Adenine, Cytosine, Guanine, Thymine) arranged in complementary pairs along a

negatively charged phosphate backbone. Long chain DNA molecules like genomic

DNA have complex 3-D structures from intra-chain molecular forces. These

macromolecules have a constant charge density and have been shown to have a free

solution mobility in standard electrophoresis buffers (TAE, TBE) independent of

molecular weight when larger than ≈400 base pairs [178]. The negatively charged DNA

structure suggests less binding on a negatively charged hydrophilic surface, however,

this is a simplistic approach as it does not take into account the reaction/adsorption of

molecules onto the polymer surface or account for the DNA complexing with other

molecules in solution. Stellwagen et al [178] suggested an observed change in DNA

mobilities between buffers was due to the formation of such complexed molecules.

The biological processes for the application described in this thesis involves DNA

extraction from whole blood, PCR amplification from the extracted material, and

microarray fluorescent imaging from the PCR products.

Microchip based DNA extraction from whole blood commonly employs leukocyte

filtration or solid phase extraction (Section 2.3.4). In both cases a diluted mixture of

blood can be used, reducing the chances of device blockage by decreasing the viscosity.

This decrease in viscosity is further enhanced for a flowing solution where the channel

dimensions are between 6µm and 500µm, due to the Fahraeus effect, in which particles

in solution accumulate along the channel axis [179], thereby decreasing the chance for

cells and other large particles to come into contact with the surface. Upon lysis the DNA

is released from the cell and nucleus and is free to expand into a larger volume.

Investigation is required into the effects of genomic DNA binding onto PET surfaces

from complex solutions such as lysed blood.

46

The polymerase chain reaction requires a careful balance of reagents for successful

amplification [94]. It is suggested that problems with surface absorption altering the

reagent balance will be exaggerated due to the large surface to volume ratios in

microfluidic devices [97]. Enzymes such as Taq Polymerase are large complex and

convoluted molecules consisting of a variety of hydrophilic, hydrophobic and charged

surfaces. Their 3-D morphology is critical for function and adsorption onto a surface

can permanently change the molecule’s morphology inhibiting its proper response.

Although PET devices have been used in medical implants, resulting in protein

adsorption [177] and PCR amplification in wells [110], the effect of high surface to

volume ratios (>>10), inherent in the microfluidic devices for this application, remains

unknown.

3.3 Thermal Analysis

3.3.1 Experimental

3.3.1.1 Thermogravimetric Analysis

Thermogravimetric Analysis (TGA) is a technique used to provide material composition

and thermal stability data by measuring weight changes with temperature. Figure 11

depicts a method for characterising TG curves [180]. Curve A shows no mass change

under the experimental conditions, suggesting that the decomposition temperature has

not been reached. Curve B is typical of the evaporation of volatile components. A single

stage decomposition reaction produces type C curves, whereas multi-stage

decomposition gives type D or when poorly resolved type E curves. Interaction with the

atmosphere, including surface oxidation, can produce type F shapes and may be

followed with a decomposition reaction producing a type G curve.

47

Figure 11 Standard Thermogravimetric Analysis curves [180]

TGA experiments were performed on 5mm2 samples (TypeA=15.336mg,

Type542=13.897mg ) with a TAInstruments 2960 SDT v3.0F TGA-DTA module,

operated from 20oC to 500oC at a ramp rate of 0.3oCs-1 with a sampling rate of 0.5Hz.

3.3.1.2 Modulated Differential Scanning Calorimetry

Differential Scanning Calorimetry (DSC) is a technique that provides quantitative and

qualitative data on physical and chemical changes within the material by measuring the

temperature and heat flow associated with material changes as a function of temperature

and time. Figure 12 shows a DSC scan of PET after quench cooling with a well defined

Tg at approximately 80oC indicating a significant amount of amorphous structure, and a

peak at 150oC showing a change to a more crystalline structure before melting at around

250 C [181].

Modulated DSC is a similar technique that allows greater sensitivity by applying a

sinusoidal temperature oscillation onto the linear DSC heating profile. The resulting

heat profile can then be used to separate the heat flow into its heat capacity-related and

kinetic components.


48

Figure 12 DSC scan of PET after quench cooling [181].

MDSC data was collected with similar samples (TypeA=11.500mg,

Type542=13.000mg) on a TAInstruments 2920 V2.6A standard modulated DSC

module. Temperature rise was set from 20oC to 300oC with a 1oC 0.1Hz modulation at a

0.3oCs-1 ramp rate with a sampling rate of 5Hz.

3.3.2 Results

The results from the Thermogravimetric Analysis (TGA) and Modulated Differential

Scanning Calorimeter (MDSC) of PET type 542 are shown in Figure 13 and Figure 14

respectively.


49

Figure 13 TGA of 100µm T542 cPET sample

The TGA curve, Figure 13, is relatively flat below 250oC indicating that there are no

significant reactions in this material that produce volatile products below this

temperature. Above 250oC a minor amount of volatile products are released before

thermal decomposition occurs at around 400 oC.

50

Figure 14 MDSC of 100µm T542 cPET sample

In MDSC the baseline shifts are caused by changes in sample weight, heating rate, or

the specific heat of the sample. A change in specific heat often occurs after the sample

has gone through a transition such as curing, crystallization or melting. Sample weight

often changes during volatilization or decomposition [182]. The heat flow curve in

Figure 14 has a small Tg transition at approximately 80oC suggesting some of the

material is amorphous, however, there is no cold crystallisation peak on the Nonrev

Heat Flow curve at approx 150oC. The sharp transition around 250oC can be attributed

to the onset of crystalline melt, with no ∆H of crystallisation this indicates that a large

portion of material was already in the crystalline phase prior to heating in the DSC. The

degree of crystallinity, and therefore the presence or absence of the cold crystallisation

peak, is highly dependent on the thermal history of the sample. The impact of the

thermal bonding method used for this thesis on the crystallinity is discussed in Chapter

5.

51

3.4 Surface Modification

3.4.1 Experimental

3.4.1.1 Chemical Modification

Chemical surface modification was performed by placing the samples in a solution of

5M NaOH (Merck, Australia) at 80oC for timed intervals (10s, 30s, 1min, 5min, 10min,

15min, and 30min) followed by rinsing under flowing isopropanol then deionised water.

3.4.1.2 UV Modification

The UV exposure was performed by placing the samples in a custom made light box

with four 14W 185nm mercury halide Lamps (Heraeus). All samples were placed

60mm directly below the lamps for timed intervals (10, 30, 60, 150, 360 mins).

Patterned samples were fabricated by 30min exposure under the UV lamps with a

stainless steel contact mask placed on the top surface of the sample. The stainless steel

masks were fabricated using 50µm stainless steel sheets machined with the 3ω Nd:YAG

laser (as described in Section 4.3.1 using a 10kHz beam at 185µJ) exposing 500µm,

250µm, and 100µm square patterns.

3.4.1.3 XPS

X-ray Photoelectron Spectroscopy (XPS) was used to identify the elements present on

the surface and their valance electron energies, giving an indication of the chemical

environments of the species present.

e-

e-

e-

hνννν

e-X-rays

photoelectrons

Specimen

detector

e-

e-

e-

hνννν

e-X-rays

photoelectrons

Specimen

detector

Figure 15 Diagram of XPS operation

52

The technique operates by directing monochromated x-rays onto the sample, which

excite the molecules present causing electrons to be emitted, see Figure 15. From the

detection of the kinetic energy of these emitted electrons the binding energy can be

calculated according to Equation 1, where KE is the kinetic energy of the electron, BE

is the binding energy of the electron, and φs is the work function of the spectrometer.

Equation 1

sφν − ΒΕ − = h KE

The magnitude of the response is directly proportional to the concentration of the

particular elements present. Allowing the relative atomic concentrations to be calculated

according to Equation 2, with respect to each element i; C(i) is the atomic concentration,

I(i) the integral intensity measured for the photoelectron peak, and ASF(i) is the atomic

sensitivity factor.

Equation 2

∑=

jjASFjI

iASFiIiC

)(/)(

)(/)()(

Chemical binding information is interpreted from the altered binding energies of atoms

bound to atoms with different electronegativities. In a similar manner aromatic “shake-

up” peaks give an indication of the amount of aromaticity and conjugation on a polymer

surface. The information is limited however, as certain functional groups appear with

the same binding energies and therefore cannot be resolved by XPS. For organic

polymers, 95% of the response is from the top 9nm, the bulk of which is from the

surface layers with the number of electrons emitted decreasing exponentially with depth

into the sample. The surface penetration is dependent on the angle between the sample

and the electron detector. Increasing this angle increases the surface sensitivity.

The XPS data was collected on a Kratos AXIS Ultra imaging XPS instrument using a

monochromated Al Kα X-ray source at a power of 1486.6 eV. The energy scale of the

instrument was calibrated using the Au 4f7/2 photoelectron peak at binding energy (EB)

= 83.98 eV. Charge correction was performed using the C 1s photoelectron component

peak corresponding to C-C species at EB = 284.7 eV. As the samples were insulating,

charge neutralisation was performed using low energy electron bombardment.

53

Sample analysis was performed on an area of approximately 700µm x 300µm in a two

step process. Firstly, survey spectra were acquired at an energy of 160eV to identify

surface constituents. Secondly, high resolution elemental region spectra were obtained

with a pass energy of 20 eV to characterise the chemical state or states of the elements

of interest. The spectra were quantified, and corrected by background subtraction, using

the manufacturer’s supplied sensitivity factors with an overall uncertainty estimated to

be ± 10% of the measured value.

The samples were prepared by sonication in a 1:1 mixture of ethanol and water followed

by multiple rinses in isopropanol and deionised water. The modified samples were then

surface treated. All samples were stored in air and analysed by XPS within two days.

3.4.1.4 Contact Angle

The contact angle of a droplet on a surface provides a measure of the surface energy

between a three-phase boundary. The surface energy results from a molecular force

imbalance at the surface which causes an attraction of the surface molecules towards the

interior of the phase. This then provides a tendency for the surface to contract to

minimise its area, thereby minimising the free energy of the system. Therefore when a

droplet is in contact with a surface it forms a shape to minimise the free energy of the

system. This can be represented by the sum in equation (Equation 3) where γ is a

surface or interfacial tension, A is the interfacial area, and lg, sg, sl, refer to liquid-gas,

solid-gas, and solid-liquid interfaces respectively [183].

Equation 3

slslsgsg Α + Α +Α γγγ lglg

On an ideal surface a droplet will form a droplet as depicted in Figure 16. When in

equilibrium the contact angle, θ, is related to the surface energy of the system according

to Equation 4.

54

θγslγsg

γlg

Gas

Solid

Liquid

Figure 16 Contact angle of a liquid droplet on an ideal surface

Equation 4

0coslg =−− θγγγ slsg

On a non ideal surface there are several allowable contact angles due to the combination

of surface energy, roughness and heterogeneity [184]. The most reproducible contact

angles are the largest and smallest, known as the advancing, θa, and receding, θr, contact

angles. The advancing angle provides a measure of the wettability of the low-energy

component of the surface, while the receding angle is more characteristic of the high-

energy component.

Contact angle measurements were determined using the droplet method [183]. A droplet

was placed onto the horizontal surface, via a syringe, allowing the stationary (sessile)

angle to be measured. The advancing contact angle was measured by increasing the

volume of the droplet and reading the largest contact angle made before the droplet

circumference expanded. Similarly the receding angle was measured by reducing the

droplet volume and measuring the smallest contact angle before the circumference

shrank. Imaging was performed using a microscope fitted with a goniometer eyepiece

(Rame-hart, Inc). A micrometer driven syringe allowed controlled dosing of droplets.

Each measurement was the average of at least four independent readings.

3.4.2 Results

3.4.2.1 Chemical Surface Modification

Chemical etching is a common technique used in the printing industry for surface

modification. The precise nature of a polymer surface after chemical exposure is

dependent on the composition of the polymer and etchant solution, the thermal history

55

of the polymer and its time and temperature of exposure. Surface contact angle

measurements were performed on both Mylar Type A and Melinex Type 542 over

timed intervals to assess the optimal exposure time. Only sessile contact angles, a

combination of polar and non-polar components, are shown due to difficulties

encountered in measuring the receding angles, which were too low to measure (<10

degrees) possibly due to the surface roughness.

The PET samples were immersed in 5M sodium hydroxide at 80oC for the time intervals

of 0.5, 1, 5, 15, 30 minutes, and then rinsed in deionised water before analysis. Figure

17 shows that the treated PET samples (Melinex Type 542) had a reproducible contact

angle that reached a plateau at around 16o, whereas the standard PET (Mylar Type A)

initially followed the same trend before fluctuating to a value as high as 34o. Along with

this change a cloudy substance was observed coming off the standard PET, unlike the

modified samples that visually remained unchanged.

Contact Angle vs Exposure Time

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

0.00 5.00 10.00 15.00 20.00 25.00 30.00

Exposure time (mins)

Co

nta

ct

An

gle

(d

eg

)

Melinex Type 542 Mylar Type A

Figure 17 Contact angle measurements of Melinex Type 542 and Mylar Type A

after NaOH exposure

A proposed mechanism for this is that of saponification, or ester hydrolysis in basic

solution, as is depicted in Figure 18. Essentially the nucleophilic addition of a hydroxide

ion to the ester carbonyl group gives a tetrahedral alkoxide intermediate. A carboxylic

acid is then generated from the elimination of an alkoxide ion, which then removes the

acidic hydrogen to yield a carboxylate ion. The Melinex seemed to undergo no further

56

reaction whereas the Mylar sample continued to undergo nucleophilic addition on the

carboxylate ion from another hydroxide to give its carbonate form. Hence the cloudy

substance being discharged from the surface is possibly sodium carbonate, which would

also account for the fluctuating contact angle with the surface build-up and removal of

carbonate ions.

Figure 18 Proposed ester hydrolysis mechanism for surface modification

Table 2 Relative % atomic concentrations of the elements present on the surface of

clean T542 cPET and NaOH treated T542 cPET

Relative % Atomic Concentrations

Sample C O Na

Clean cPET 74.9 25.1 -

NaOH treated cPET 45.1 37.5 16.4

An XPS surface analysis of the Mylar exposed for 10 minutes was performed to help

verify this process. Table 2 Relative % atomic concentrations of the elements present on

the surface of clean T542 cPET and NaOH treated T542 cPET shows the relative

percentage atomic concentrations of the PET sample before and after NaOH treatment.

Following treatment there is a significant reduction in the level of carbon and an

increase in the oxygen concentration. In addition, sodium is present on the treated

surface. Figure 19 shows high resolution C 1s and O 1s region spectra for the

unmodified and modified PET surfaces. The C 1s spectrum for clean PET consists of

four components at 284.7, 286.2, 288.7 and 290.9 eV corresponding to aromatic and

aliphatic carbon species, C-O, O-C=O and aromatic shake-up species respectively 9.

The modified PET C1s spectrum shows an additional component at 289.2 eV attributed

to Na2CO3 species. The presence for carbonate species is further supported by the

additional component at 530.9 eV observed in the O 1s spectrum for the modified PET

[185]. The unfitted peak at higher binding energies in the O 1s spectrum corresponds to

Ester Carbonyl Intermediate Carboxylic acid Carboxylate ion

57

a Na KLL (Auger) peak. The XPS results from the modified PET support the proposed

reaction mechanism suggested above.

Figure 19 High resolution C 1s and O 1s region spectra for the (a) unmodified and

(b) modified Mylar PET surfaces

3.4.2.2 UV Surface Modification

It is well understood that photo-oxidation of a polymer surface under atmospheric

conditions leads to the incorporation of oxygen functional groups that improve the

surface wettability.

a)

b)

58

PET samples were exposed to UV radiation for up to six hours, with subsequent contact

angle measurements taken up to one month apart. Figure 20 shows sessile contact angle

results. The results show a substantial reduction in the contact angle after a 30-minute

treatment, followed by a slight increase for longer exposures. The 30-minute sample

shows a significant contact angle increase after the first day. This difference suggests

the possibility that the reactions had completed to relatively stable intermediates such as

POOP or POOH that degraded after the first day into free radicals that reacted to form

more stable, and less polar, functional groups. Rinsing the UV irradiated substrates in

deionised water after exposure increased their contact angle up to 70% of their original

value, suggesting the removal of lower molecular weight surface particles [186],[187].

Exposure time vs contact angle

0

10

20

30

40

50

60

70

80

90

Sample Exposure Time (min)

Conta

ct A

ngle

(D

egre

es)

Shelf life = < 1 day Shelf life = 14 days Shelf life = 35 days

Figure 20 Sessile contact angle of UV irradiated Melinex Type 542 samples

Table 3 XPS analysis of UV exposed Melinex Type 542

Relative % Atomic Concentrations Ratio

PET Samples Si C O N O/C N/C

Untreated clean Type542 0.0 74.9 25.1 0.0 .335 0

15 30 60 120

59

XPS analysis of the UV exposed samples (0.5 and 6hr exposed Type 542 PET film) was

performed to provide further chemical information of the surface. Samples required

washing before exposure and analysis, to remove the surface silicon present, (6.65%

atomic concentration) possibly due to silicon-based oils used during the manufacturing

of the film [175]. The oxygen-to-carbon ratios for both exposure times are similar, see

Table 4, suggesting the oxygen incorporation has reached saturation within the first half

hour of exposure. However, unlike oxygen, the nitrogen incorporation showed a

significant increase after six hours.

Table 4 XPS analysis peak assignment of UV exposed Melinex Type 542

It is known that UV irradiation under normal atmospheric conditions leads to radical

formation and subsequent oxygen incorporation rather than crosslinking [188]. The

increase in contact angle after rinsing suggests the loss of oxygen rich lower molecular

weight groups and the uncovering of less exposed material from the surface, which

infers the degradation of the surface polymer chains with no significant crosslinking.

Detailed spectral results for the 30-minute exposed sample gave a spectral shift down

3.16eV showing a reduction in the aromatic and aliphatic carbon species, see Table 4.

The C1s2 has a shift of 1.59eV from C1s1 indicating the possible presence of C-O-C,

C-OH, and *C-O-C=O groups, consistent with Norrish Type II reactions resulting from

oxygen incorporation. The oxygen incorporation is further evident from the decrease in

aromatic and aliphatic carbon peak C 1s 1 and increase in the oxygen containing C 1s 2

peak heights.

30min UV treated Type542 0.0 63.8 31.3 4.9 0.49 0.076

6hr UV treated Type542 0.2 61.7 30.8 7.3 .50 .12

Peak Binding Energy

/eV

Peak

Assignment

% Total Carbon

(Theoretical)

% Total Carbon

(30min UV mod)

C 1s 1 281.84 aromatic and

aliphatic C

60 52.98

C 1s 2 283.43 C-O 20 27.18

C 1s 3 285.93 O-C=O 20 19.84

60

It is possible to provide a more detailed analysis of the surface functional groups

through the use of attachment chemistry [168]. This would entail altering the problem of

functional group identification into the determination of element concentration, but this

was considered beyond the scope of the present work.

3.4.2.2.1 UV Patterning

UV patterning allows for a simple approach to define spatially resolved arrays of

surface functional groups for chemical linkages, as occurs with oligonucleotide

coupling in microarray analysis [189], or limiting the surface treatment in regions for

improved biocompatibility or fluid control. To test the patterning resolution, samples

were UV exposed for 30-minutes through a contact mask with 1.0, 0.5, 0.25, and 0.1mm

square patterns. In some cases the mask was up to 1mm from the sample surface. A

solution of dye in deionised water (1:100) was rinsed over the exposed samples. The

samples were tilted (≈45o) to allow excess water to run off leaving the more hydrophilic

areas covered by a meniscus of water. Figure 21 a) shows the meniscus on a 1mm

square, whereas b) and c) show the dried samples for 0.5mm and 0.25mm respectively.

The well defined edges on these samples show a clear change in hydrophobicity at the

surface.

Figure 21 Wetting of UV patterned Melinex Type 542 with a solution of dye in

deionised water (1:100)

As the mask used was not in close contact to the surface, the well defined edges give an

indication that the surface treatment is due significantly to the incident UV photons

rather than the secondary interaction of gaseous radicals. The increase in the rounding

of the corners and inability to retain water droplets on the 0.1mm square patterns was

a) b) c)

1.0mm 0.5mm 0.25mm

61

due to surface tension of the water droplets rather than the exposed pattern resolution.

The well defined corners were observed using fluorescent imaging (see Section 3.5.2).

3.5 Optical characteristics

3.5.1 Experimental

Fluorescent images were acquired on a GenePix 4000B microarray scanner (Axon,

Australia). This is a confocal laser microscope running 532nm and 633nm laser sources

with 557-592nm and 650-690nm optical emission filters. The system has a resolution

down to 5µm per pixel with a variable focal depth of 250µm. Samples used for

background fluorescent images were cleaned before analysis in a similar manner to the

samples in the previous section, however, microfluidic samples were not cleaned again

after the fabrication process to ensure surface chemistries remained unchanged.

The microarray samples were prepared by fabricating channels as per Chapter 5 with

fluorescently labelled probes spotted inside the channels by a custom robotic arrayer

producing approximately 0.6nl droplets. The probes consisted of 5µl of 10µM Cy5

labelled oligonucleotide (20mer MW=6626.7 5’-8CA GGA TGC TAC TCA CTG CGT

–3’) in 3xSSC buffer.

3.5.2 Results

The luminescence of a sample can reduce the effectiveness of microarray imaging and

has been highlighted as a major issue for polymer devices. The problem arises when the

emission and excitation wavelengths of the substrate and signal fluorophores are

similar. The films used in this thesis were tested for the application of microarray

analysis as these films contain additives and have undergone processes that introduce

functional groups other than those from the monomer units.

Figure 22 shows the emission intensity histograms for some of the commercially

available films tested during this thesis on a confocal microarray scanner at the

excitation wavelengths of 532nm and 635nm. The intensities clearly fluctuate greatly

between the different types of PET and at 532nm are typically orders of magnitude

larger than glass. Allen et al [92] has shown that pure PET does not excite above about

62

450nm. Suggesting that either; impurities containing higher wavelength absorbing

chromophores are present in the material; and/or the surface has undergone some

surface reaction or absorption to have these luminescent functional groups present. The

fluctuation between the different PET samples may be attributed to the different

impurities or surface groups present as it is known that these films have different

additives and have undergone different processing treatments. Furthermore, the UV

treatment of the PET T542 shows a significant increase in the fluorescence at both

wavelengths (from 2500 to 5800 fluorescent units at 532nm and from 250 to 950

fluorescent units for the non-treated and treated PET samples respectively), suggesting

the oxidised surface increases the material’s fluorescence at these wavelengths. Possible

oxidation mechanisms leading to fluorescently active groups include oxy radical

substitution of the phenylene ring [190].

63

Figure 22 Relative fluorescence intensities of cPET samples at 532nm and 635nm

as measured on a GenePix 4000B microarray scanner

a)

b)

64

It is also known that the thermal processing of PET can oxidise the material, suggesting

that the materials thermal history may also play an important role in determining the

material’s fluorescence. However, no significant variations in background fluorescence

were observed after thermally bonding the cPET devices, which may be attributed to the

relatively low temperatures of bonding that were used to avoid thermal decomposition

(see Section 5.3.3).

In the context of some microfluidic applications a material’s fluorescence may be

corrected through background subtraction. However, this correction process may be

further complicated if the background signal level is not constant during the analysis

period. Hawkins et al observed a photobleaching effect of the polymer materials and

suggested exposure to pre-bleach the material prior to analysis to reduce the background

signal [91].

Unlike Hawkins et al who normalised the fluorescent signals to the material thickness

and treated the fluorescence as a bulk material effect, it is clear that the surface

functional groups contribute significantly to PET’s fluorescence and must be considered

when using confocal imaging systems. The UV patterned samples in Figure 23 show a

clear increase in intensity of the UV irradiated portions indicating the presence of more

fluorescent chromophores in these regions. As these measurements were taken using a

confocal scanner with a focal depth in the order of approximately ten micron, the

chromophores on the surface contribute significantly to the signal received. The

decrease in background signal from left to right across the image can be attributed to the

sample moving out of the focal plane.

65

Figure 23 Surface fluorescence of Type 542 cPET film after 30 minutes UV

exposure through a non-contact mask

The obvious fluorescent activity of surface groups impacting on device performance

suggests that careful attention is required in manufacturing for the selection of

materials, their storage conditions, and processing parameters. Further investigations are

needed to reduce background fluorescence to allow a wider range of materials and

enable device usage across the entire spectrum.

However, it is known that the excitation and emission of molecular groups vary across

the visible spectrum, so careful selection of the optical system can limit the effect of the

background signal, as shown by the distinct differences in the 532nm and 635nm signal

responses. This is further illustrated by the clearly visible microarray in Figure 24 where

a linear array of 10µM Cy5 labelled oligonucleotide spots were deposited in a

microchannel fabricated from PET T542 and analysed at 635nm. The image was

captured with the instrument gain adjusted to provide a saturated signal at the array of

fluorescent spots. It is clear that the background fluorescence is relatively low in

comparison to the polycarbonate microfluidic devices used by Liu et al [88], who

demonstrated detection of a 20µM Cy5 array in a similar instrument.

0.5mm

66

Figure 24 Microarray analysis of Cy5 labbeled oligonucleotides spotted in a cPET

microchannel

The increase in fluorescence observed on the channel walls in Figure 24 may be

attributed to higher levels of surface oxidation or geometric effects causing imaging

artefacts. Polymer oxidation products have been shown to be luminescent, with possible

chromophores in PET produced from oxy radical substitution of the phenylene ring

[190]. However, the fluorescence is unlikely to be attributed to an increase in surface

oxidation of the channel walls themselves as the XPS analysis from Section 4.4.6

showed a decrease in the ester component and an overall increase in the hydrocarbon

component on the Nd:YAG laser machined surfaces. The inconsistent intensity of the

background fluorescence in Figure 24 suggests the possibility that the varied intensity

may be a measurement artefact attributed to the surface roughness and angle of the

channel walls causing light scattering and thereby altering the amount of light collected

by the confocal imaging system.

3.6 Biocompatibility

3.6.1 Experimental

3.6.1.1 DNA Absorption

DNA absorption experiments were performed at two temperatures (25oC, 95oC) and at

two different pH’s in ionically adjusted buffers (MES pH=7.2, TES PH=9 both

adjusted to 50mM) similar to standard PCR mixes. Due to the minimal volume

requirements of the detector’s cuvette, 200µl was chosen as the test volume.

67

Microfluidic chips were fabricated (as per Chapter 5) using 250µm PET film for a large

surface to volume ratio (16000:1), with their geometries retained within a 2.5x2.5mm

heating zone for the thermocycling instrument (Custom made computer controlled

peltier thermocycler, see Figure 25). Using a 250µm x 250µm channel geometry the

resulting channel walls had a similar surface area for both the laser cut and native PET

surfaces. DNA solutions (50ng/µL) were incubated for 15mins at temperature inside

the biochips then eluted slowly via a syringe pump to minimise shearing forces at the

channel walls (100 µL/min). The DNA absorbance was measured for each aliquot with

a UV/VIS spectrophotometer (Cary50, Varian Australia) on a 260/280nm profile using

the standard Warburg Christen analysis method [191].

Figure 25 Diagram of custom built computer controlled thermocycler

3.6.1.2 Protein Absorption

To observe protein absorption on the surface, enzyme-linked immunosorbent assays

(ELISA) were performed. ELISA is a technique using antibodies or antigens coupled to

an assayed enzyme, thereby providing measurement of the antigen or antibody

concentration. There are many different types of ELISA techniques, and they typically

have five common steps: 1) the surface is coated with antigen; 2) unbound sites are

blocked to prevent false positive results; 3) antibody is added; 4) anti-mouse IgG

68

conjugated to an enzyme is added; and 5) the substrate reacts with the enzyme to

produce a coloured product to give a positive reaction.

The ELISA experiments were performed to establish the relative protein binding to PET

in comparison to the more commonly used microfluidic substrates of glass and

polycarbonate. The mouse protein was chosen as it has been shown to readily bind to

many surfaces and represents a commonly used assay.

The microtitre plate wells were prepared by blocking with 5% skim milk in PBS (1

Oxice Dulbecco A tablet per 100ml) (30mins), washed three times with 0.5% Tween in

PBS, washed three times with PBS, and then rinsed three times with distilled water. The

cPET film samples (Type 542, 7mmx7mmx0.25mm) were washed in distilled water and

placed into the bottom of the microtitre plate wells. The assays were then left for seven

days to allow for outgassing and evaporation.

Antigen (AMA-I 2.5µg/ml, 40µl/well) was placed into the wells and incubated for 60

minutes, washed three times with 0.5% Tween in PBS, washed three times with PBS,

and then rinsed three times with distilled water. Blocking was performed with 5% skim

milk in PBS (30mins) then washed three times with PBS, and rinsed three times with

distilled water. Antibody (5G8 Mab 10µg/ml, 40µl/ml) was then added and incubated

for 60 minutes before another cycle of rinsing with Tween, Oxoide, and water. Finally

the labelled antibody (SAM – Horseradish peroxidase) was added before a final rinse

with the Oxice and water.

3.6.1.3 PCR

A standard PCR mix was used incorporating 1µL forward and reverse primers (25µM),

1µL dNTP (10mM), 5uL Buffer (initially containing 1.5mM Mg2Cl2) 0.1uL Taq

polymerase (2 units µl-1), and made up to reaction volume with deionised water. PCR

experiments in standard microwell format were performed in a commercial thermocyler

(Corbett Research PalmCycler) and the microfluidic chip-based assays were performed

in a custom built instrument allowing the chip to be sandwiched between computer

control thermoelectric elements (see Figure 25). Cycling times were typically 30

69

seconds at 94oC to denature, 30 seconds at 60oC to anneal, and 30 seconds at 72oC for

extension, and repeated 35 times, followed by a final 2 minute extension cycle.

3.6.2 Results

Due to the amount of prior literature on polymer biocompatibility for PCR, including

methods for fabricating devices for this purpose [240], the author felt it unnecessary to

provide an exhaustive experimental exploration on the subject. Rather, an investigation

into possible inhibition of PCR by the cPET-film microchips processed by methods in

this thesis was performed.

Initial experiments to establish the compatibility of the PET film for PCR were

conducted using a typical 50uL PCR reaction mixture, using human genomic DNA

(<50kD), in 96-well plates with 4mm square PET film samples placed inside the

reaction chambers. The results indicated a substantial inhibition for unmodified PET

(Lane C) and complete inhibition for UV (Lane B) modified PET when compared to the

controls (Lane A&D), as shown in Figure 26. With the addition of BSA into the PCR

mixture, the unmodified PET sample (Lane F) provides a stronger response than with

PET alone (Lane C), however, the BSA with the UV-modified PET (Lane D) sample is

still completely inhibited.

Figure 26 Electrophoresis results of PCR amplifications with PET samples

It is generally understood that the mechanisms of PCR inhibition on reaction surfaces

involves the surface adsorption of the PCR reagents [97]. Possible candidates for

absorption include the large macromolecules of DNA and Taq polymerase.

A B C D E F

70

Through its phosphate backbone DNA is known to bind strongly under chaotropic salt

conditions to silica [192]. Similarly DNA has also been shown to bind to carboxyl

groups [193]. Although PET is a substantially different material than silica there are

many surface oxygen groups present, and these films contain unknown additives such as

silicones and titanium oxides [175]. To check if the genomic DNA (<50kbp) binds to

the surface, experiments were performed at two temperature ranges (25, 95 oC) for two

different pH’s (7.2, 9) approximating the typical ranges for PCR. Experiments were

performed with flow rates less than 100 µL/min to minimize shear rates during loading

and washing, and with incubation times of >1hr. No significant amount of DNA was

observed binding in the chip at either temperature with the standard deviation in

measured DNA concentration before and after passing through the chip measured at less

than +/-2ng/µL.

A complex study of protein surface interactions was considered beyond the scope of this

research, however, ELISA assays were conducted to gain a general insight into the

relative amount of protein interactions with the PET film. The results were normalised

to a positive control with a response of 1.00 ± 0.118 with PET giving 0.763 ± 0.041, and

Polycarbonate, which showed less inhibition than PET, gave a response of 0.876 ±

0.104 indicating less binding. Typical PCR inhibition results show lower inhibition for

polycarbonate suggesting there may be a relationship between inhibition and protein

binding.

With surface adsorption as a possible cause of inhibition then the use of facilitating

agents in the buffer may overcome the PCR inhibition by preferentially binding to the

surface. Methods generally known to be successful for reducing non-specific binding

include the use of BSA, Skim Milk, and PEG in solution to preferentially bind to the

surface. PCR experiments were performed in microchannel devices with surface to

volume ratios >100. The results shown in Figure 27 indicate an improved signal

response for the samples containing these facilitating agents (BSA, Lanes B&G; Skim

Milk, Lanes C&H; PEG, Lanes D&I) over the sample without a facilitating agent

(Lanes A&F). These results are consistent with the inference that surface adsorption is

likely the major cause of inhibition for these PET based microchips.

71

Figure 27 PCR assays in cPET chambers with facilitating agents (BSA, Lanes

B&G; Skim Milk, Lanes C&H; PEG, Lanes D&I; controls, Lanes A&F)

3.7 Summary

The surface and bulk characteristics of the polyester films investigated in this chapter

were characterised and discussed in relation to biochip operation. The thermal stability

of the film was analysed, and the techniques of chemical treatment and UV modification

used for fabricating the microfluidic devices in Chapter 5 were detailed. The results

showed that, after surface treatment, there was a significant increase in the surface

oxidation of the films, which caused an increase in the films background fluorescence

and protein binding. This can be problematic for microarray detection and PCR

amplification. However, it was also shown that the increase in surface oxidation could

be controlled spatially through the use of masked UV patterning, thereby avoiding the

increase in surface oxidation in critical areas.

A B C D E F G H I J

72

4. 3ω Nd:YAG Laser Machining of PET

4.1 Introduction

Laser fabrication of polymer microfluidic devices has been typically performed by UV

Excimer systems due to their ability to fabricate anisotropic features in the sub-micron

domain [35;36]. The use of IR CO2 lasers has also been reported for microfluidic

fabrication in polymers [51;53]. These infrared (IR) systems operate at faster processing

speeds than Excimer laser systems and have a lower cost of ownership. However, IR

lasers have much larger heat affected areas than Excimer systems and typically can

only produce features greater than 100µm. Diode pumped, Q-switched, solid-state

lasers, have become available in the last 10 years with higher harmonic modes of

operation for micromachining purposes. Although these high power (355nm, 30ns at 2

GW/cm2) direct-write UV laser systems have greater energy and smaller spot sizes than

their IR counterparts it remains to be shown if they are suitable for polymer microfluidic

fabrication.

In this chapter frequency tripled Nd:YAG laser cutting of cPET film is characterised for

microfluidic fabrication. Comparisons are made with the well-known Excimer and CO2

laser techniques and the results are discussed in terms of the laser’s operation,

mechanism of ablation and impact on microfluidic performance. Commercially

available PET films often contain proprietary additives for improved physical and

thermal performance. Furthermore, films that are marketed under the same name can be

subject to different manufacturing conditions which in turn can affect the physical

characteristics of the material. Due to these potential differences in film properties, the

cutting of these materials by 355nm Nd:YAG laser was characterised using 248nm

Excimer and 10.6 µm CO2 lasers to provide a comparison with the literature.

4.2 Background

4.2.1 Laser Irradiation of Materials

Light emitted by laser (light amplification by stimulated emission of radiation)

processes is coherent, monochromatic and of low divergence. For machining purposes

73

these properties simplify beam delivery to a material surface where material can be

removed with an appropriately shaped beam, pulse size, energy and wavelength.

Lasers used as cutting tools offer advantages over conventional mechanical machining

for some materials through:

� Non-contact processing – avoiding some contamination issues and tool wear, and

allowing for processing in controlled environments;

� Improved cut quality - direct removal of material through ablation processes reduces

the heat affected zone;

� High spatial control - the ablation process can be optimised to remove specific

volumes of material; and

� Reduction in tooling costs – there is no wear of a tooling head only maintenance of

the laser system. Additionally the beam can be shaped as with photolithographic

imaging systems.

Laser Ablation

The absorption of laser radiation by a material can occur when an incident photon

impinges on a molecule with a resonant frequency corresponding to the photon’s

energy. The absorption of a photon by a polyatomic polymer can induce either

electronic modes of excitation or ion vibrations (known as phonons) depending on the

incident photon energy level. Electronic excitation typically occurs with exposure in

the UV spectrum where the photon energies are of similar magnitude to the bond

dissociation energies of the polymers. In comparison, molecular vibrations have their

natural harmonic frequencies in the infrared spectrum due to their much greater mass

[194].

In either of the cases of electronic excitation or ion vibration, the absorption of a

sufficient number of photons in a given area by a material can lead to ablation. The

process of photon absorption is followed by either localised or delocalised energy

transfer mechanisms. Delocalised energy transfers lead to harmonic vibrational

excitation that may translate to thermal processes and relaxation to the ground state,

whereas localised non-radiative transitions are faster and tend to dominate for

desorption and ablation processes. Laser desorption is the loss of an atom or molecule

from a material due to photon excitation. Desorption occurs when the absorbed photon

74

energy is sufficient for bond breakage and the energy remains localised long enough to

sever the molecular bond. In contrast, laser ablation occurs when a sufficient number of

bonds are broken in the surrounding material to induce a volume expansion causing the

relatively large-scale ejection of material. This volume expansion produces a plume of

ejected material that is well characterised by plasma and gas dynamics [195].

Where these relaxation processes have resulted in a delocalised transfer of energy,

subsequent photons may cause electronic or molecular excitation if they occur before

relaxation to the ground state. Therefore, the density of electronic or vibrational

excitation plays a major role in determining the response of the material to the laser

irradiation, both in terms of the accumulative effects of single photon excitations in a

unit area and the occurrence of multi-photon processes. The probability of finding a

photon in a unit area is proportional to the incident flux. Assuming Poisson statistics the

probability of k photons being simultaneously in a cubic cell of length L, refractive

index n, with a laser intensity I and wavelength λ, is given by

Equation 5

)1(!!

<<∝≅= −mI

k

me

k

mP

kk

mk

k

Where m is the mean number of photons in the designated volume according to

Equation 6

2

32.

/.

/)).(.(

int

hc

nILL

nc

L

hc

Iareadurationpulse

photonperenergy

ensitylaserm ===

λ

The laser beam intensity, or irradiance, measures the laser power per unit area and

translates it into a probability per unit time for creating a reaction product. The most

commonly used term to describe laser light is fluence - the energy density per unit area.

As the material interaction is dependent on both the pulse energy and pulse duration,

care must be taken when comparing quoted fluence values.

In many cases the beam energy is not evenly distributed across the beam profile, such as

with the TEM00 mode or Gaussian distributed beam. In the case of the Gaussian laser

beam the intensity is greatest near the optical axis of the beam and can be represented

by Equation 7, where Io is the intensity on the axis (r=0) and w is the beam radius. With

75

the peak power given by L'Hôpital's rule where the peak energy at the optical axis of the

beam is calculated as twice the average fluence (see Equation 8).

Equation 7

I(r,w) = Io exp(-2r2/w2)

Equation 8

P (0,w) = 2 Io / π w2

In some cases the energy distribution of the laser beam may also be altered when the

beam interacts with the material due to thermal induced focusing or defocusing effects.

However, most of the models developed for polymer laser ablation assume a consistent

beam profile that interacts with the material, and assumes that Beers law applies. Beers

law states that when a polymer with a high absorption coefficient for a particular

wavelength is exposed to laser radiation then absorption of the radiation follows a

logarithmic decay into the bulk according to Equation 9.

Equation 9

It = Io 10-αl

Where Io and It are the intensities of the beam before and after transmission through a

thickness l, and α is the absorption coefficient of the material. This is further

complicated for inhomogeneous materials where the refractive index, and therefore the

absorption coefficient, varies within the material. Where the material inhomogeneity is

greater than the irradiated wavelength the optical appearance of the material is altered

by light scattering, which results in greater absorption [194].

If the incident fluence of the incoming energy exceeds the material’s ablation threshold

then a depth of the material is ablated. If the absorption depth is greater than the

ablation depth then a portion of irradiated material, that may affect the ablation depth of

the next pulse, remains as shown in Figure 28. Therefore the first few pulses may have

slightly different etch depths compared to subsequent pulses [196]. The difference

between the effect of the first few pulses of a laser beam on a surface and the effect of

subsequent pulses can occur from a change in the optical properties caused from the

first few pulses [194]. These altered absorption properties can occur due to changes in

the surface topology, or from:

76

� Heat production causing changes in the density or electronic characteristics of

the material

� The optical generation of free carriers by interband transitions or impact

ionisation modifying the absorption coefficient

� Intense beams where the generated electric field distorts the material’s electron

orbitals or whole molecules.

Figure 28 Polymer surface indicating absorption and ablation

depths from a laser pulse [197]

A typical graph indicating etch rate after ablation as a function of fluence for a 248nm

Excimer laser beam incident onto a PET surface is shown in Figure 29 [198] The graph

clearly shows a logarithmic relationship up to approximately 1000mJ/cm2 which is in

agreement with the general absorption mechanism outlined above. For high fluences

the shielding of the laser beam by ejected ablation fragments has been shown to

contribute to the reduction in the expected ablation depth [199].

77

Figure 29 Etch depth versus fluence for 248nm Excimer laser

ablation of PET [198]

Polymer Ablation Mechanisms

Polymer ablation mechanisms are typically classified as being photothermal,

photochemical, or a combination thereof.

Ablation by electronic excitation giving rise to bond dissociation (commonly referred to

as photochemical ablation) is the most desirable for laser micromachining of micron

sized features. When the incident laser beam has photon energies sufficient to directly

sever critical bonds within a substance, and the fluence is above the materials ablation

threshold, then the result is a rapid volume expansion of material. This rapid expansion

of material carries away the ablated material and surplus energy ensuring very little

thermal damage to the exposed area (see Figure 30).

78

Figure 30 Laser irradiation through a projection mask causing

polymer ablation [199].

UV laser ablation of polymers was first reported in 1982 by Kawamura et al [41] and

Srinivasan et al [200] and is usually described by one or more models in terms of

photothermal or photochemical ablation. The exact mechanism or combination of

processes of UV polymer ablation is a controversial issue [199], although it is generally

understood that it is a combination of these mechanisms that cause ablation depending

on the irradiating wavelength and the optical absorption properties of the target

material.

UV ablation of polymers has been demonstrated to occur as a layer by layer process

[199]. For pulsed lasers the extent of the heat affected zone has been given by Equation

10 where D is the thermal diffusivity, and τ is the pulse duration. This equation assumes

that the irradiated material is ablated in a characteristically photochemical process, thus

ensuring that the bulk of the material irradiated above the ablation threshold is removed

in the timeframe of the pulse duration. This has been demonstrated for a predominantly

photochemical process where ablation has been shown to proceed throughout pulse

duration [44].


79

Equation 10

x=2(Dτ)1/2

Photothermal ablation can be a combination of processes that translate the optical

energy into heat. These thermal processes tend to dominate when photon energies of

the incident light are insufficient to directly break the bonds of the irradiated material.

Optical energy is translated into lattice vibrations (phonons), and by way of

contributions from multiple photons, sufficient energy is provided to the target to eject

material. Material ablated in this process typically exhibits signs of thermal damage in

the form of conduction, reflow and redeposited droplets.

Allmen discusses the adsorption of laser energy into heat as a process involving three

steps [194].

1. The excited particles undergo rapid (<<10-15 sec) spatial and temporal

randomisation prior to the atom collision time.

2. Energy equipartation, which involves a large number of elementary collisions

and intermediate states with numerous energy transfer mechanisms. These

mechanisms each have their own related time constant, giving an average overall

time constant of approximately 10-11 seconds [201] depending on the material.

3. Heat flow from the localised laser irradiation.

The temperature gradient for an isotropic solid near the region of irradiation is given by

Equation 11, where φ is the heat flux, k is the thermal conductivity, T is the temperature

and z is the distance [202].

Equation 11

oZ

odz

dTkz

−=)(φ

In one dimension this can be expressed as Equation 12, where Cp is the specific heat, V

is the volume, ρ is the density, and t is the time.

Equation 12

dt

dT

V

C

dz

dTk

dz

d p=

or

t

T

z

T

∂

∂=

∂

∂

κ1

2

2

80

with the thermal diffusivity pC

k

ρκ = .

For a moving pulsed heat point source it has been shown that if Equation 13 is satisfied

then the heat source can be considered stationary. Furthermore, after each pulse of the

heat source the irradiated surface begins to cool. If the pulse repetition rate meets the

criteria of Equation 14 then there will be no accumulation of heat between pulses [203].

Equation 13

τ = ro/10uo

Equation 14

f < κ/30ro2

where τ is the pulse duration, ro is the beam radius, uo is the beam translation speed, and

f is the frequency of the laser pulses.

Ablation Threshold

Both photochemical and photothermal mechanisms have been used to describe the

requirement to meet a certain energy threshold before ablation occurs. From a purely

photochemical perspective, a certain investment of energy is required to sever enough

bonds locally within the substance to result in the rapid volume expansion of material

[194]. In contrast, models for thermal ablation have been used to describe the

requirement for a certain investment of energy in order to reach an ablation temperature.

Subsequently, the combination of delocalised transfer of energy and further absorbed

photons may cause molecular excitation leading to ablation.

This ablation threshold is generally considered to be the point at which a detectable

amount of material is ejected from the surface, changing the surface topography and

measurable by profilometry. It has been shown with mass sensitive techniques [187]

and high speed photography [204] that there exists another lower threshold at which

point gases are lost from the surface. For the purposes of this thesis ablation is referred

to as the removal of material leaving an etched surface measurable by optical

microscope (feature>1µm).

81

4.3 Experimental

4.3.1 Laser Machining

CO2 Laser

The medium in CO2 lasers is a mixture of carbon dioxide, nitrogen, and generally

helium. The excited CO2 molecules produce light by relaxation from an excited

vibrational state. This results in either a 10.6µm photon for the relaxation from an

asymmetric to symmetric stretching state, or a 9.6µm photon for the relaxation to a

bending state. The nitrogen helps excite the CO2 molecules to their upper state due to

the similarity between their lowest vibrational state and the CO2’s asymmetric

stretching mode. This allows the molecules to absorb enough energy to excite the CO2

molecules by resonant collisions. The helium acts as a buffer to improve heat transfer

and helps the CO2 molecules drop from their lower laser levels to the ground state.

The laser used for these experiments was a sealed tube 10.6µm continuous wave CO2

laser (Model 48-2, Synrad Inc.) with an FH series marking head (Fenix, Synrad Inc.)

that contained x and y axis mirrors mounted on separate galvanometer scanners, see

Figure 31. The output lens focused the beam to a spot size of 290µm with a focal length

of 200mm giving a depth of field of +/-2.5mm.

Figure 31 CO2 laser x-y scanning stage

The energy calibration and channel etching experiments were performed at a scan speed

of 400mm/sec, with the output energy adjusted by modulating the duty cycle of the

laser’s RF drive circuit. The calibration curve measured for the CO2 laser’s power

output versus duty cycle is shown in Figure 32.

L

X-

Y-

L

Laser

X axis scanner

Y axis scanner

Work piece

82

Figure 32 Typical CO2 laser output vs duty cycle of RF drive circuit

Excimer Laser

An Excimer laser utilises the stimulated emission from excited rare gas halide

molecules by:

KrF* + hv => Kr + F + 2hv (248nm)

These excited gas molecules are formed by electrical discharge in a sealed cavity

containing the pressurised gas. The Excimer formation process is relatively complex,

however, two reactions are particularly important. The Excimer molecules may form by

the recombination of the ionised gas products via the equation:

Kr+ + F- → KrF*

The other reaction process occurs when the rare gas metastable product reacts with

formed halogen gas by:

Kr* + F2 → KrF* + F

The Excimer laser system utilised in these experiments was an Exitech S8000 system

using a Lambda Physik LPX210I laser source with an integrated mask and sample x and

y stages for image projection lithography, see Figure 33 [205]. The system illuminates a

chrome-on-quartz mask with an homogenized 10mm square beam with an intensity

deviation of approximately 5% RMS. A 10x projection lens with a NA of 0.3 was used

giving diffraction limited resolution of 0.8µm in a 1mm2 area.

83

Figure 33 Excimer laser beam delivery system

The laser system was operated in constant energy mode giving an energy output based

on the attenuator setting, with the results from a typical calibration using an 8mm

circular mask shown in Figure 34. Channels were fabricated by imaging through a

1mm2 mask pattern onto a scanning work-piece at a pulse repetition rate of 10Hz.

1.57 1.571.53 1.51

1.35

1.11

0.85

0.58

0.42

0.260.22

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

1.80

0 0.2 0.4 0.6 0.8 1

Attenuator Setting

Flu

en

ce M

ult

ipli

er

Figure 34 Typical Excimer laser fluence vs attenuator calibration curve

Plane of uniformity

5% RMS Deviation, 10x10 mm beam size

Mask e.g. Chrome-on-quartz

Lens Array Homogenizer

6x6 (36 Element) Double Array Homogenizer

Laser Beam

248 nm (KrF)

Projection Lens

10x , 0.3 NA, 0.8µm resolution, 1mm field

Workpiece motion

200 x 200mm, 0.1µm resolution

84

Nd:YAG Laser

The Nd:YAG laser is a solid state diode pumped laser. The solid-state refers to the class

of solid dielectric gain medium, a yttrium aluminium garnet crystal doped with

neodymium. Neodynium has inner electron energy levels with energies of 1.56-2.32 eV

that can absorb visible and infrared wavelength light, and due to these energy levels

being inner shell transitions they are largely unaffected by the material in which the ions

are placed. A portion of this energy is converted into heat as the electrons, excited by

these energies, drop by non-radiative transitions to a metastable state 1.38 eV above the

ground state. In this metastable state there is no way for the atom to lose energy by a

non-radiative transition and since all the radiative transitions have a low probability of

occurring spontaneously in the absence of a stimulating field, then the neodymium

atoms can stay in this state for a long time. This allows for a population inversion to

build up between the metastable and lower states. When stimulated emission does

occur, an electron can drop from this metastable state to a lower one emitting a photon,

whereby the lower states are depopulated by thermal transitions which cause further

heating of the bulk material.

These thermal effects cause efficiency and optical problems not only with optical

alignment but also by affecting the refractive index of the gain medium. In rod-shaped

gain material the heat builds up in the core, and with cooling provided on the outside a

temperature gradient is produced within the medium that changes the material’s

refractive index. This can be partially overcome by using a rectangular or slab-shaped

crystal that allows for total internal reflection of the laser beam, giving rise to a more

even temperature distribution.

The Coherent AVIATM Nd:YAG system used is a frequency tripled diode pumped solid

state Q-switched laser. Typically these lasers operate by transferring light from the

pump laser diodes through fibre optics to the crystal (see Figure 35) to avoid

transferring the heat energy generated from the laser diode system to the laser optics

setup. The Dichroic mirrors are used to pass the pump light and reflect the laser

emission,and the Q-switch acts as a shutter allowing for a controlled pulsed output.

85

Figure 35 3ω Nd:YAG laser optical configuration

The frequency tripled (3ω) Nd:YAG laser used for this research (AVIA 3.0W, Coherent

Inc, USA) has been incorporated into a computer controlled 2-D laser cutting system

(Lasertec, The Netherlands). The Q-switched, pulsed output of the laser generates a

pulse duration of ~30ns, a pulse energy of up to ~230µJ, and for the purpose of this

work a pulse repetition frequency of 1-10kHz. The normalised beam profile output

from the laser is shown in Figure 36. An f-theta lens, with a focal length of 160mm, was

used to focus the TEM00 beam to a 27µm diameter spot at the work piece. Computer

Aided Design (CAD) drawings of the sample to be machined, in either DXF or HPGL

format, were downloaded to the control software. This information was used to control

the temperature compensated x-y galvo-scanner and the synchronised firing of the laser

(see Figure 31). The beam velocity at the work piece is determined by the scanner

parameters and was set to 0.11 m/s. This combination of parameters results in a pulse

being fired every 11µm or a shot overlap equivalent to 2.45 shots per area. All

experiments were performed with the top surface of the substrate in focus.

86

Figure 36 3ω Nd:YAG laser beam profile

When operated in burst mode the Q-switched laser has a thermal equalization time

associated with rapid switch-on and switch-off. This causes early pulses in a burst to be

significantly lower in energy and requires a ramping up in amplitude until a steady state

is achieved (Figure 37). The effect is very pronounced when the laser is externally

triggered to produce bursts of pulses of different duration and different intervals

between bursts, as is common for cutting out vector patterns for microfluidic circuits,

see Chapter 5. The AVIA's ThermEQ™ automatic compensation mechanism is used to

try to ensure all pulses within a burst are equal in energy.

Figure 37 AVIATM

3ω Nd:YAG laser system’s ThermoEQ pulse stabilisation

Compensation for the thermal lensing effect, common in these diode pumped lasers,

was provided by the systems inbuilt ThermaTrackTM function. Optimisation of this

value was performed prior to experimentation and it was found that, after the inbuilt

Norm

alis

ed

Beam

Inte

nsity

87

optimisation procedure, manual adjustment of the ThermaTrackTM could improve the

output energy by up to another 10%. Figure 38 shows the variation of the calibration

curves on different days for the systems optimised current versus output energy, as

measured from the output of the laser. Series A-C were each taken a week apart and

repeated again a month later for series D-F. The change in curve shape from

calibrations a week apart are indicative of the ageing of the gain medium crystal. A

change in the gain crystal site was a probable reason for the altered curve shape taken

from the measurements months apart.

Energy Ouput versus Current

0

50

100

150

200

250

40 50 60 70 80 90 100

Current (%)

Energ

y O

up

ut (u

J) A

B

C

D

E

F

Figure 38 3ω Nd:YAG laser energy versus current calibration taken one week

apart (A-C) and repeated one month later (D-F)

The output energy of the 3ω Nd:YAG laser beam reduces with increased frequency, as

shown by the energy vs frequency calibration curve of Figure 39.

88

Figure 39 Typical 3ω Nd:YAG laser frequency versus energy

4.3.2 Imaging and Profile Measurements

Cut profile characterisation and sample visualisation was performed with a BX60

Olympus optical microscope having x2.5, x5, x10, x20, and x50 objectives. Detailed

images were taken with scanning electron (SEM) and confocal microscopes. The SEM

analyses were performed using a JEOL JSM35 scanning electron microscope equipped

with a backscatter electron detector at 15kV with 100µA load current. These samples

were prepared with an SEM gold coating unit (Polarin Equipment LTD) at a pressure of

0.1 torr with a current of 25mA for 2mins. Height measurements were made using the

Olympus OLS1100 He:Ne laser scanning confocal microscope with x5, x10, x20 and

x50 objectives.

4.3.3 Surface Chemistry Analysis

XPS

XPS analysis was performed in a similar manner to the description in Chapter 3.

However, these samples were prepared by cutting 350µm PET with the 3ω Nd:YAG

laser setup as described earlier and affixed to the XPS sample holder with double-sided

adhesive tape. Cleaned 2mmx10mm samples mounted on their sideswere used to

analyse the cut edge. Cleaning involved sonication in a 1:1 mixture of ethanol and water

Energ

y (

µJ)

Energ

y O

utp

ut

(µJ)

89

followed by multiple rinses in isopropanol and deionised water. Samples were stored in

air and analysed by XPS within two days.

4.4 Results

The molecular structure of PET has carbon and oxygen bond dissociation energies of C-

C, 3.61eV; C=C, 6.34eV; C-O, 3.73eV; C=O, 7.70eV [206]. The Excimer laser system

operates at 248nm has a photon energy of around 5.0eV, which suggest that individual

laser photons have sufficient energy to break C-C and C-O bonds. This is in contrast to

the CO2 and 3ω Nd:YAG laser systems with wavelengths of 10.6µm and 355nm, which

have photon energies of approximately 0.12eV and 3.49eV respectively. These lower

photon energies suggest that there is not enough energy for intramolecular bond

breakage by individual photons. Although the mechanisms of photochemical ablation

remains a controversial issue, photothermal processes are well characterized and are

often involved in ablation processes.

Furthermore, multiphoton processes have been suggested to contribute to photochemical

ablation. In the case of the Nd:YAG laser Mansour et al [207] reported direct bond

dissociation of PET by multiphoton processes at 266nm with high fluences (1-40J/cm2).

However, the probability of multiphoton electronic excitation is related to the photon

flux density by Equation 5 and Equation 6, which indicates that the probability of our

355nm Nd:YAG system inducing a 2-photon excitation in a 2nm cell is extremely low

(≈10-8) [195].

In the case of the lower energy CO2 laser, the ablation mechanism has been shown to be

a dominantly photothermal process. It has been further demonstrated that the thermal

damage in PET from CO2 laser ablation processes can be minimised by choosing a

wavelength corresponding to the phonon energies of the diatomic C-O pair [50]. The

results showed that the 9.25µm wavelength produced less thermal damage than the

10.28µm and 9.58µm bands from the same laser.

The individual photon energies of the 355nm Nd:Yag laser are unlikely to induce direct

molecular vibrational excitation since the PET molecules have their natural harmonic

frequencies in the IR spectrum. However, the absorbed electronic excitation energy can

90

undergo energy transformations and lead to vibrational excitation that can translate to

thermal processes. In these cases the timescale for these delocalized energy transfer

mechanisms are associated with the thermal diffusivity of the material (0.746 x10-3

cm2s-1 for cPET).

The following results investigate the 3ω Nd:YAG laser cutting of cPET with

comparisons drawn from the distinctly different CO2 and Excimer laser machining

processes. Studies using CO2 lasers to irradiate different organic polymers have been

reported and the results provide an important comparison to UV ablation mechanisms

[60]. For example Dyer et al. [50] ablated PET using a 9.17µm CO2 laser, causing C-O

stretching vibrations, and has shown a relationship between etch depth and fluence

similar to that of UV ablation, although not demonstrating a well defined ablation

threshold.

The results are described with reference to the unit of fluence (J/cm2) to describe the

laser beam energy. As the material interaction is dependent on the pulse energy,

duration, and residual heat in the material, care must be taken when comparing quoted

fluence values. Unless explicitly stated the fluence is calculated by dividing the total

energy profile of the beam by the beam area. The exception is where the peak energy is

used in relation to the Gaussian laser beams of the 3ω Nd:Yag and CO2 lasers. In this

instance the energy at the optical axis is calculated using L'Hôpital's rule giving a peak

energy of twice the average fluence.

4.4.1 3ω Nd:YAG Laser Output Reproducibility

For microfluidic fabrication the 3ω Nd:YAG laser required burst mode operation to

scan the vector patterns. The etching reproducibility between pulses was notably poorest

at 3.5 J/cm2, see Figure 40, with hole diameters varying between 0 and 10µm.

91

Figure 40 3ω Nd:YAG etch reproducibility at 3.5J/cm2 on a 10kHz scanning vector

The manufacturer’s specification of less than 5% pulse-to-pulse stability suggests a

variation of less than 0.175 J/cm2, at 3.5 J/cm2 this should still be well above the

ablation threshold shown in Section 4.4.2. Furthermore, the ablation threshold

measurements were taken from single pulse experiments. For improved pulse stability

these pulse reproducibility results were taken near the end of a 10cm vector scan to

allow time for ThermoEQTM stabilisation of the laser pulses. These results suggest that

either pulse to pulse energy deviation is greater than 5% RMS, or there are significant

changes in the material homogeneity over areas greater than a few micron, or a

combination thereof.

Although material homogeneity may be a contributing factor, it has been demonstrated

that surface microstructures induced by Excimer laser ablation correlate with the

inherent material stresses caused from the orientation process [60]. The surface

structures formed from the resolidified heat affected zones are typically relatively

uniform with a periodicity of less than 2um, suggesting the films crystallinity may not

be the dominating factor.

4.4.2 Ablation Threshold

The cPET samples required a fluence of at least 2.0 ± 0.3 J/cm2 (peak energy 4.0 J/cm2)

in a single pulse to etch the surface reproducibly. Figure 41 shows an ablated surface

having a 2µm hole with and a 7µm visible heat affected zone. Ablation was observed at

lower energies (1.3 J/cm2), however, the results were inconsistent, possibly due to the

poor pulse-to-pulse etch reproducibility as discussed in the previous section. These

observed ablation thresholds in cPET are of a similar order of magnitude to those found

30µm

92

in the literature for CO2 (0.5-1.0J/cm2) [50], and an order of magnitude greater than the

ablation thresholds found using Excimer lasers (≈30mJ/cm2) .

Figure 41 Image of heat affected zone on 100µm PET just above ablation threshold

Multiple pulses at 1Hz in the same location produced no noticeable difference from the

single shot ablation threshold. In contrast, three successive pulses fired at a rate of

10kHz required consistently less energy to produce ablation (1.2 ± 0.3 J/cm2) ,

suggesting that the presence of residual heat in the immediate area from the previous

pulses contributed to the ablation of the material. These photothermal contributions

between pulses which reduce the ablation threshold are in agreement with the concept

of a thermal investment required to establish a boundary layer, and the need for an

ablation temperature to be reached [208].

With regards to photochemical ablation, Sutcliffe et al describes the ablation threshold

as the point below which photo-fragmentation is negligible, and where the threshold

flux (wavelength dependent) is constant for a given material [209]. This is in contrast to

the observations for the 3ω Nd:YAG laser which showed a significant reduction in the

ablation threshold with successive pulses at 10kHz, and a greater flux threshold than

exhibited by Excimer machining. This possibly indicates that photochemical ablation is

not the dominant mechanism involved in Nd:YAG machining at 355nm. However, it

should be noted that Burns et al have shown a dependence of photochemical etch depth

and ablation threshold of weakly absorbing polymers for repetition rates up to a few

hundred hertz [210].

10µm

93

4.4.3 Etch Rate vs Fluence

A comparison of the etch rates for each laser is observed for the same cPET film sample

because the cutting performance of each laser is more than a function of its energy

density. The cutting performance is also related to the chemical structure of the polymer

and therefore its absorption properties at the specific wavelengths (see Section 4.2.1).

As the first few pulses from the laser beam onto a surface are different from subsequent

pulses [194], and larger errors are associated with measuring small profiles resultant

from single pulses, the etch rate of the material was measured as the average rate over

many pulses. The beam was operated in a scanning pattern similar to the methods used

for microfluidic channel fabrication (Chapter 5).

The relationship between the etch rate and fluence for the cPET samples using the

Excimer laser firing 128 pulses at 10Hz is shown in Figure 42. This trend is in

agreement with literature and indicates the approximately linear relationship between

etch depth and number of shots for energies above the ablation threshold and below 1

J/cm-2. For deeper microstructures the ablated products are emitted from the site in a

1D expansion, normal to the surface and back in the direction of the incoming laser

beam, which shields the target surface from the laser energy [211]. As higher fluences

are used more material is present in the plume, therefore resulting in higher losses and a

reduced average etch rate per pulse. Shallower structures allow the beam to expand in

3-D, creating a far less dense plume and consequently less shielding of the target

surface. Therefore, shallower structures usually result in an observed higher average

etch-rate-per-pulse.

94

0.35

0.4

0.45

0.5

0.55

0.2 0.4 0.6 0.8 1 1.2

Etch Rate [um/pulse]

Fluence [Jcm-2]

Figure 42 Etch rate versus fluence for the 248nm Excimer laser ablation of cPET

using 128 pulses.

In comparison to the Excimer system the CO2 laser is well known to have a purely

photothermal mechanism [50]. Although the etch rate is much higher than the Excimer

laser, the response is linear, see Figure 43, and consistent with the findings of other

researchers [50;51;53]. At the timescales involved for thermal ablation (200ns) [50] and

the laser scanning speed of 400mm/sec, the continuous beam can be considered to be

stationary. The thermal decomposition products from IR laser ablation of PET are

known to strongly absorb in the IR region. However, the results show a linear response

suggesting the percentage of energy shielded by the plume is negligible, or that different

mechanisms of cutting are occurring. The formation of a plasma from the IR absorption

by the ablation products has been known to contribute to cutting, however, in this case it

is considered unlikely as the 25W 10.6µm CW CO2 laser produces a peak fluence of

less than 75kW/cm2, which is considered insufficient energy for plasma formation

[194].

Etc

h R

ate

(µ

m/p

ass)

95

Figure 43 Etch rate versus fluence for the CO2 laser ablation of cPET.

The 3ω Nd:YAG laser was operated with a constant firing rate of 10kHz using a

scanning pattern with vectors having greater than one second cutting intervals. This is

typical of the scenario used for fabricating the microfluidic devices [212]. The large

time interval ensured any thermal effects and plume interaction would be due to the

overlapping from successive pulses in a single scan rather than from successive scans.

As the pulse width (30ns) is much less than the time taken to traverse the beam width

(245µs) the beam can be considered to be stationary. The scan speed of 0.11m/s and

firing rate of 10kHz gave a shot overlap of approximately 2.45. The linear response of

Figure 44 suggests that the ablated plume interaction does not alter the etch rate at these

energy levels despite the high etch rates in comparison to the Excimer laser’s response.

This is consistent with the relative absorption at these wavelengths by the native cPET

film, which has a photon transmission of much less than 0.05% at 248nm and greater

than 60% at 355nm.

96

0

5

10

15

20

25

30

0.0 5.0 10.0 15.0 20.0 25.0 30.0

Fluence (J/cm2)

Etc

h r

ate

(um

/pass)

Figure 44 Etch rate versus fluence for the 3ω Nd:YAG laser ablation of cPET.

It has been argued that a purely thermal model can be used to explain both photothermal

and photochemical laser ablation [208]. Experimental data indicating the change in etch

rate when approaching a cooled boundary fits well with basic heat conduction theory

and suggests the requirement of a thermal energy contribution in the lattice to prevent

bond recombination after photon absorption. From a photochemical perspective, Beer’s

law suggests that photon absorption remains constant at a particular depth including

when a boundary is approached. An explanation of the reduction in etch rate from a

photochemical perspective may possibly lie with the need for a critical mass of broken

bonds to induce the volume expansion seen during ablation. As the material depth

reduces and approaches the boundary, then the total number of photons absorbed in the

material is also reduced, requiring more energy to remove the same amount of material

than compared to etch rates that are not near the boundary. A distinction between the

two mechanisms can be made in the time domain where the thermal contribution to

successive pulses are characteristic of the material conductivity, unlike the

photochemical mechanism that can undergo changes on a timescale orders of magnitude

faster.

According to Equation 14 for pulse repetition rates above approximately 13Hz there

will be an accumulation of heat in the material from successive pulses (κ= 0.746x10-3

97

cm2s-1). The results of Table 5 show the size of the ablated area for 10 stationary pulses

at various frequencies. The design of the laser system limited the frequency setup to

below 1kHz, except where manual triggering was achieved at 1Hz. The results clearly

show an increase in the ablated volumes at 1kHz and 10kHz compared with manual

triggering. For a dominantly photothermal mechanism the 10kHz frequency would be

expected to have a higher residual heat between shots and therefore more material

ablated. However, it was shown in Figure 39 that the laser’s output energy decreased

with increasing frequency, which became pronounced at 100kHz with the energy

dropping below the ablation threshold. These results suggest that there is a thermal

contribution between successive pulses contributing to the amount of material ablated.

Table 5 Ablation hole size for increasing frequency at 8.7J/cm-2

Frequency Hole depth (µµµµm) Hole diameter (µµµµm)

Hz µµµµ σσσσ µµµµ σσσσ

1 21.6 2.8 26 7

1k 30.6 10.4 29 4.3

10k 30.6 5.1 32 2.5

100k - - - -

The results also showed a decrease in the etch depth per pulse from successive scans of

the laser beam. Figure 45 illustrates the decreasing etch rate per pulse for 1, 2, 5, and 10

successive beam scans from the Nd:YAG laser. The explanations provided in the

literature for the 3-D heat conduction of the laser pulse fit well with this observation

[203]. Prior to ablation the surface is relatively smooth and heat conduction is

essentially one dimensional. Upon further irradiation the surface becomes structured

allowing heat flow in three dimensions. As the hole deepens further, heat conduction

occurs on the sidewalls. For an infinitely thick substrate this would be expected to reach

a steady state, however, the substrates were cut through before this occurred.

98

Figure 45 Energy Level versus etch depth per pulse for the 3ω Nd:YAG

laser ablation of cPET (Energy levels 1= 2.5, 2=10, 3=20, 4=28 J/cm-2

)

4.4.4 Cut Quality

Although the etch rates and scanning speeds of the CO2 and 3ω Nd:YAG lasers are far

greater than the Excimer laser, the quality of the resultant cut for micromachining

purposes differs greatly. The minimum feature resolution is dependent on the beam size

and the heat affected zone from the beam and material interaction. It is generally

understood that a dominantly photochemical interaction will give cleanly ablated

structures in PET, whereas a dominantly photothermal interaction will induce relatively

large heat affected zones compared with our feature sizes.

The difference in dominant ablation mechanisms can be clearly seen in the cutting

operations of the Excimer compared with the other two lasers, Figure 46 and Figure 47.

The cross sections, as depicted in Figure 46, show similar thermal effects for both the

CO2 and Nd:YAG systems whereas the Excimer shows comparatively cleanly ablated

structures with negligible heat affected zones. Cutting to 120µm depth with the CO2

laser required a) 2 passes at 72.6 J/cm2 and b) 1 pass with 29.5J/cm2; the Nd:YAG

system required c) 26 passes using a 5.1J/cm2 beam energy, and d) 6 passes using a 25.5

J/cm2 beam energy for 110µm channel depths; the Excimer was operated at e) 256shots

at 1.0 J/cm2, and f) 128 shots at 2.0 J/cm2.

(µm

)

99

Figure 46 Example cross sections illustrating the effect of the wall

angle and thermal damage produced by a-b) CO2, c-d) 3ωωωω Nd:YAG,

and e-f) Excimer laser machining of cPET.

4.4.4.1 Heat Affected Zone

Although the cut depths in Figure 46 and Figure 47 are similar, the processing times

between the laser systems and between different energy levels of the same laser, are

considerable due to the number of pulses required. The cut quality is also significantly

changed at different energy levels. Reflow of material, the presence of bubbles, charred

material, and a wider cut are all consequences of the increased thermal contribution at

high power inputs. The CO2 and 3ω Nd:YAG lasers produced larger wall angles for

increasing energy with relatively large heat effected zones, up to 115µm for the CO2

a) b)

c) d)

e) f)

100µm 100µm

100µm 100µm

100µm 100µm

100

and 50µm for the 3ω Nd:YAG, with the ridge around the perimeter of the cuts measured

as large as 25µm high, Figure 46 and Figure 47.

101

Figure 47 A comparison of the typical cut quality between a) 248nm Excimer laser

ablation, b) 355nm frequency tripled Nd:YAG laser ablation, and c) CO2 laser

ablated channels to 100µµµµm depth in 250µµµµm thick PET films.

30

µµ µµm

1

00

µµ µµm

1

80

µµ µµm

a)

b)

c)

102

In contrast, the wall angle for Excimer ablation, Figure 46 and Figure 47, was found to

decrease with an increase in fluence, as has been shown previously [213]. This is due to

increasing beam energy density overcoming the effects of diffraction losses caused by

the edges of the structures. These losses tend to lower the incident fluence and therefore

the etch rates in the region of the sidewalls. Once sloping sidewalls are initiated then the

taper serves to increase the surface area of the ablation site and the fluence is further

lowered.

Furthermore, the debris from the cutting processes differs substantially between the

Excimer and other two laser processes. The Excimer results exhibit no sign of thermal

reflow or the presence of bubbles, consistent with the CO2 and Nd:YAG laser results.

The Excimer system produced very well defined edges with no lip and very fine

redeposited particles that were typically less than 1µm in size. The surfaces were

smoothly etched with little sign of thermal damage.

For a dominantly photochemical process the bulk of the absorbed energy is expected to

be ejected from the material with the ablation debris. Therefore, according to Equation

10, a 20ns laser pulse is expected to produce a heat affected region of approximately

80nm in cPET. Clearly the Excimer induces a heat affected area with a similar order of

magnitude (<<1µm), whereas the CO2 and Nd:YAG lasers effect a much larger region

(>10µm) indicating that a significant fraction of the irradiated energy remains behind in

the material after ablation.

The thermal effects of the Nd:YAG laser beam interaction with the cPET film is further

evidenced by the increase in etch depth with higher frequency pulses, as discussed in

the previous section with reference to Table 5. It was also noted that the increased pulse

repetition altered the heat affected zone, as is evidenced by Figure 48. The lower

frequency pulses show smaller structures with more material deposited in the vicinity of

the hole, whereas, at 10kHz more material was ablated in smaller fragments. This trend

has been shown for other laser processes [199]. The heat affected zone within the bulk

remains relatively unchanged by the change in frequency, with a deviation of 20% of

the heat affected zone common at all frequencies.

103

Figure 48 Thermal effect of frequency variation of 10 shots at 27J/cm2 at a) 1Hz b)

1kHz and c) 10kHz using the 3ω Nd:YAG laser ablation of cPET

When drilling entirely through the substrate a smaller heat affected zone was observed

inside the bulk material than for shallower structures. Thermal effects from increasing

the number of shots and cutting directly through the substrate are shown in Figure 49

for 3.5J/cm2 pulses at 10kHz for a) 10, b) 50, and, c) 500 shots (cut entirely through

layer). Figure 49 c) shows an increase in re-solidified material around the lip

corresponding to the greater volume removed but also shows little sign of the heat

affected zone within the bulk compared to the other two depths. This suggests the

removal of material before significant heat transfer takes place.

Figure 49 Heat affected zones after 3ω Nd:YAG laser ablation of cPET at 3.5J/cm2

10kHz for a) 10, b) 50, and, c) 500 shots (cut entirely through layer)

Masking of the outer regions of the beam was performed to 1) establish if the large heat

affected zones (≈200µm) were a result of the heat transfer from the lower energy

density at the edges of the Gaussian beam or 2) to establish if the higher energy density

portion of the beam was sufficient to ablate the bulk of the irradiated material and

thereby reduce the heat transfer. Figure 50 presents typical images from masked

irradiated surfaces. It was observed that the cutting performance of a masked beam was

similar to using an unmasked beam at lower energy levels. This indicates that the lower

flux portions of the beam contribute significantly to the ablation process with a

dominantly photothermal mechanism.

a) b) c)

100µm 100µm 100µm

a) b) c)

200µm 200µm 200µm

104

Figure 50 Typical images of drilling through a 100 µm PET substrate with

reducing energy distributions using masked 3ω Nd:YAG laser beams a) 28J/cm2

40 shots, b) 60 shots at 28 J/cm2 with 8mm mask), c) 500 shots at 28 J/cm

2 with

4mm mask.

These results suggest that the heat affected zone seen within the bulk of the substrate, at

10-25µm below the surface, is due the heat diffusion of the ablation front after the last

pulse.

4.4.4.2 Fluence Versus Cut Quality

To create channel geometries suitable for microfluidics it was necessary to use the 3ω

Nd:YAG laser in a vector scanning format. The effects of single laser beam scans and

10 laser beam scans on 100µm cPET (T542) at 10kHz for different fluences are shown

in Figure 51 and Figure 52. Single beam scans leave a regular pattern of widening

debris from the flow of molten polymer, which are averaged with multiple scans

providing a more uniform edge. Table 6 summarises their respective dimensions.

a) b) c)

100µm 100µm 100µm

105

Figure 51 Cut quality on cPET after single 3ω Nd:YAG laser beam pass at 10kHz

with fluences of a) 4.4 J/cm2, b) 8.7 J/cm

2, c) 18 J/cm

2, d) 26 J/cm

2, e) 33 J/cm

2

100µm

a) b)

c) d)

e)

100µm

100µm 100µm

100µm

106

Figure 52 Cut quality on cPET after multiple (10) 3ω Nd:YAG laser beam passes

at 10kHz with 30sec b/w passes and at the fluences of a) 4.4 J/cm2, b) 8.7 J/cm

2, c)

18 J/cm2, d) 26 J/cm

2, d) 33 J/cm

2

100µm

a) b)

c) d)

e)

100µm

100µm 100µm

100µm

107

Table 6 Cut dimensions for 100µm thick cPET film machined by 3ω Nd:YAG laser

with increasing fluence

Single Scanning Beam Pass 10 Scanning Beam Passes

Fluence

(J/cm2)

Depth

(µµµµm)

Width

(µµµµm)

Lip

(µµµµm)

Depth

(µµµµm)

Width

(µµµµm)

Lip

(µµµµm)

4.4 1±1 3±3 - 30±5 15±3 12±3

8.7 7±3 10±5 7±3 54±4 22±4 15±3

18 17±3 7±3 16±6 85±5 32±3 22±3

26 25±5 25±5 22±7 * 45±2 35±5

33 23±4 45±5 35±5 * 40±2 45±5

*substrate cut entirely through

As expected for a dominantly photothermal process the results indicate a clear trend in

increasing heat affected zone with increasing beam fluence. Lower fluences can be used

in conjuction with multiple beam scans to achieve deeper structures with a greatly

reduced heat affected zone, however the trade-off is machining time.

4.4.4.3 Cut Width Versus Material Thickness

The cut width from multiple beam scans varied according to the material thickness.

Table 7 expresses the values obtained for 12, 30, 100, and 350 µm thick substrates with

the same machining parameters. The 350µm compressed value illustrates an example of

what happens to the channel width after thermal compression bonding. The presence of

the resolidified material from the cut forming ridges along the cut edge is a strong

indication of the thermal effects from the laser, as are the wider entrance widths and the

evidence of bubbling at the interface. Furthermore, the thin 12µm cPET layer required a

liquid film (isopropanol) to both hold it flat to the workpiece chuck and act as a heat

sink to achieve channel widths close to that of the diameter of the beam, otherwise cut

width varied between 80-140µm.

108

Table 7 Dimensions of laser machined cuts using different thickness substrates

Sample Thickness

(µµµµm)

Entrance Channel

Width (µµµµm)

Exit Channel

Width (µµµµm)

12 27 ± 2 27* ± 2

30 35 ± 5 30 ± 5

100 40 ± 5 30 ± 5

350 70 ± 5 30 ± 5

350 (compressed) 30 ± 5 10 ± 5

4.4.4.4 Edge Quality

The confocal microscope cross-sectional information depicted in Figure 53 shows the

ridge dimensions, indicating that these volumes represent only a fraction of the total

ejected volume that was vaporised in the beams path and subsequently collected by the

system’s extraction vacuum filter. Ultrasonic cleaning and washing in isopropanol and

water removed most of the smaller pieces of ejected material that had redeposited

around the cut leaving only the larger fragments behind, shown in Figure 54.

Figure 53 Confocal image of lip profile for 10-350µµµµm thickness cPET films

machined by the 3ω Nd:YAG laser at a fluence of 18 J/cm2

109

Figure 54 Scanning electron microscope image of 350µµµµm PET machined

using the 3ω Nd:YAG laser at a fluence of 18 J/cm2

4.4.5 Raster Scanning

To create 3-sided channels that are wider than a beam width it was necessary to scan the

beam across the channel in a raster profile, similar to the technique used by Jensen et al

for CO2 machining [54]. The laser was operated at 10kHz scanning back and forth on a

250µm vector. Energy profiles were varied (2 to 30J/cm2) with increasing distances

between beam scans (10µm to 100µm). Figure 55 shows SEM pictures of typical results

from a) low and b) high energy combinations. Lower energies (<14J/cm2) in

conjunction with a 20µm offset scan pattern tended to cause localised melting and

reflow without vaporisation and bubble formation. This allowed material to be shifted

out of the beams path, whereas, higher energies induced vaporisation causing a

roughened surface with poor control of structuring.

Confocal

scan

100µµµµm

Larger re-deposited

debris

110

Figure 55 Structures induced in 100µµµµm cPET film by raster scanning of the 3ω

Nd:YAG laser beam at a) 10 J/cm2 and b) 30 J/cm2

4.4.6 Surface Chemistry

The impact of a change in surface chemistry can dramatically affect the performance of

microfluidic devices. A change in hydrophobicity alters the electrophoretic mobilities,

the non-specific binding of molecules to the surface, and the capillary flow performance

(see chapter 6).

In relation to microfluidics the surface chemistry of a channel includes the ablated

surface and the redeposited debris. Surface morphology changes due to the melting,

reflow and solidification are expected to change crystallinity properties of the cut

surface. The relatively fast solidification of the melt upon contact with the bulk allows

for fast cooling giving rise to a more amorphous structure than the bulk orientated film.

A lower softening point would be expected in these areas with the possibility of shorter

molecular weight groups from the thermal degradation processes. Crystallinity changes

on laser cut surfaces has been demonstrated for both IR [60] and Excimer laser

machining [198] resulting in more amorphous structures that impact bonding as

described in Chapter 5.

The chemical structure of cPET is shown in Figure 56. Table 8 shows the relative %

atomic concentration of carbon and oxygen elements present on the surface of clean

cPET, Excimer ablated cPET and YAG cut cPET as determined by XPS. The results

from the unmodified cPET are in agreement with the literature [214], and in both laser

a) b)

250µm 250µm

111

cutting cases the process reduces the amount of oxygen and increases the amount of

carbon present on the cut surface. After laser cutting there appears to be an increase in

the hydrocarbon component and decrease in the ester component for both processes, but

only Excimer machining exhibits a decrease in the C1s2 percentage. This trend

confirms the loss of oxygen from the surface. This decrease in the O/C ratio can be

attributed to the loss of small gaseous products such as CO and CO2, which has been

reported for both thermal and photochemical ablation [215]. These results are in

agreement with the trends observed by Lazare [215] and Watanabe [216], although

these authors performed their laser exposures using 193nm irradiation in air, and at

248nm in argon, respectively. Analyses of the Excimer machined samples were

performed on structures produced by static ablation so that the surfaces were relatively

free of re-deposited carbon. When using the mask-dragging technique to create

channels carbon is continually redeposited behind the beam as it travels along the cut

direction. Several cleaning processes are available to reduce this effect, including a

final pass with the Excimer laser beam with a few shots per area exposure.

Figure 56: Chemical structure of Poly(Ethylene Terepthalate)

Table 8 Distribution of carbon species on uncut and 3ω Nd:YAG laser machined

cPET as determined by XPS

Peak Binding

Energy /eV

(measured)

Peak

Assignment

% Total

Carbon

(Uncut)

% Total

Carbon

(Excimer cut)

% Total

Carbon

(YAG cut)

C 1s 1 284.7 Aromatic and

aliphatic C

62 81 68

C 1s 2 286.2 C-O 22 10 22

C 1s 3 288.7 O-C=O 16 9 10

112

Lazare et al suggested that the type of chemical modification resultant from UV

irradiation of a cPET surface was dependent on the power of the radiation source. This

inference was based on their observations that a much lower energy of 193nm laser

radiation was required to reduce the O/C ratio in comparison to the much higher 185nm

lamp radiation required to modify the surface. Furthermore, the 185nm-modified

surface remained stable after washing in acetone whereas the low molecular weight

material formed from the laser ablation disappeared after washing. This suggests that

although there has been a trend with these two UV laser processes in reducing the O/C

ratio the mechanisms of ablation and resultant surface chemistry might be very

different.

In contrast to Excimer laser ablation the photothermal mechanism of CO2 laser

irradiation has been shown to oxidise the cut surface [217]. Dadsetan et al investigated

the surface properties of cPET after irradiation with a TEA CO2 laser of wavelengths 9-

11µm at 0.4-2.0 J/cm2. Although the surface morphology changed according to the

filtered wavelengths, the overall photodegradation mechanism proposed from the laser

exposure was the dissociation of ester bonds and radical formation on the surface. These

radicals reacted in air to form stable peroxides that would decompose under further

irradiation to give various oxidised groups such as carboxylic acid, hydroxyl, carbonyl

and aldehydes.

Dyer et al measured the gaseous decomposition products of ablation using a similar

TEA CO2 laser setup [50]. Their observed products were similar to those formed by

pyrolysis with the major molecules ablated being PET, CO, CO2, CH4, C2H2, C2H4,

C6H6, C4H4, and CH3CHO. The respective yields of the different molecules were found

to be dependent on the fluence. For energy levels just above the ablation threshold the

molecules produced were predominantly involatile high molecular weight PET species,

and as the fluence increased, further fragmentation occurred producing the smaller more

volatile species with a substantial fraction being CO and CO2.

Unlike the photothermal process reported by Dadsetan et al for the CO2 laser ablation,

and the long wavelength UV photodegradation process proposed by Day et al [190], the

XPS results (Table 8) from the Nd:YAG laser cutting did not show an increase in the

surface oxidation along the cut edge. The decrease in the measured O/C ratio caused by

113

the decrease in the ester component suggests another mechanism. It is known that the

carbonyl groups in PET can absorb UV light and be excited to singlet and triplet states

resulting in chain scission, carbon dioxide production, and termination with aldehyde

groups [171]. This suggests the possibility that photochemical mechanisms are involved

in this ablation process.

4.5 Summary

A process for 3ω Nd:YAG laser machining of cPET film was characterised for the

application of biochip fabrication. The results show that thermal processes contribute

significantly to the ablation of PET with the 3ω Nd:YAG laser process. The machined

cuts were more representative of the well known thermal CO2 laser process rather than

the dominantly photo-oxidative process of the Excimer laser. However, the ability of

these lasers to provide a relatively high fluence (~30 J/cm2) over a small beam area

(typically around a 25µm diameter spot) enabled a finer machining resolution than their

CO2 counterparts. This high energy density also enables a faster rate of machining

compared to Excimer based systems, and although the ultimate resolution is not as high

(30µm beam size compared to the Excimer that has 0.8µm diffraction limited optics), it

is shown in the subsequent chapters to be more than adequate for microfluidic channel

formation. Although the 3ω Nd:YAG laser process is unsuitable for the fabrication of

filters with dimensions of a few micron, the results for excimer machining show cleanly

ablated structures with negligible heat affected zones. Consequently the excimer laser

system is combined with lithographic beam shaping for filter fabrication in the

following chapter.

114

5. Biochip Fabrication

The purpose of this chapter is firstly to establish a fabrication method for the

construction of an entirely polymer based microfluidic filtration device, using the

materials and processing techniques described in Chapters 3 and 4, and secondly to

fabricate a microfluidic filtration device that can be applied to sample preparation for

genomic-based diagnostics. In particular, the microfluidic filtration device will be

applied to the preparation of DNA from whole blood for subsequent amplification by

the polymerase chain reaction (PCR). It is well understood that whole blood inhibits the

PCR process [218]. Two methods commonly applied in laboratories to remove these

inhibitors involve either the isolation of white blood cells or solid phase extraction of

the DNA. The filtration biochips used in this study are designed to accommodate either

white blood cell isolation or the trapping of microparticles for solid phase extraction.

A layered approach to microfluidic fabrication is used with the commercially available

cPET films. The method of 3ω Nd:YAG laser processing of cPET, discussed in Chapter

4, is applied to microchannel formation, and the filter membranes are fabricated by

perforating cPET film using 248nm Excimer laser ablation. The devices are bonded

together by thermal diffusion and in some cases UV surface modification is performed

after laser machining and prior to assembly to alter the bonding, electroosmotic flow,

and capillary flow characteristics for the device performance characterisation as

discussed Chapter 6.

5.1 Background

5.1.1 Bonding Mechanisms

In general, mechanisms of adhesion can be categorised into six different theoretical

models [219];

1. Mechanical Interlocking

2. Electronic theory

3. Theory of boundary layers and interphases

4. Adsorption theory

5. Diffusion theory

6. Chemical bonding theory

115

The mechanical interlocking model uses the interlocking of the adhesive into cavities,

pores, or surface roughness of the substrate to provide the joining strength.

Electronic theory suggests that there is an electronic transfer mechanism between the

adhesive and substrate that contributes significantly to the bond strength. The

equalisation of the different electronic band Fermi levels at the surfaces induces the

formation of an electric double layer at this interface. Although this double layer exists

in many cases it is unclear whether it is the cause or the consequence of high bond

strength in these systems.

The theory of boundary layers looks at the formed interfacial layer between the

substrate and the adhesive, and suggests that interfacial failure is a contributing factor to

bond strength. This layer may extend from the molecular level up to a few micron and is

a function of many physical, physiochemical, and chemical phenomena.

Adsorption theory suggests that if an intimate contact between the adhesive and

substrate occurs then the adhesion is due to the intermolecular and inter-atomic forces

established at the interface. The most common of these forces are those of van der

Waals and Lewis acid-base interactions, with the magnitude of these forces being

related to the surface free energies. Molecular diffusion and wetting are means of

obtaining good absorption of a polymer at the interface, these allow the molecules to

come into close proximity enabling intermolecular and interatomic forces to be

established. For a description of wetting see Chapter 3.

Diffusion theory of adhesion suggests that the polymer macromolecules are mobile and

mutually soluble enough to diffuse across the interface and create an interphase region.

This theory was first proposed by Voyutskii [220] and then further developed by the

reptation model of de Gennes [221] who accounted for the polymer chains being

constrained in a complex matrix that allowed for only movement in a curve-linear

direction, or snakelike motion.

Chemical bonding across the adhesive-polymer interface has the potential to form a

much greater bond strength than with the absorption methods. These primary

interactions have a much greater bond strength, such as a covalent bond energy which is

116

typically in the order of 100-1000kJ/mol, whereas the secondary interactions of

hydrogen bonds and van der Waals forces do not exceed 50kJ/mol [219].

Adhesion is a complex process and several adhesion mechanisms can be involved

simultaneously, however, it is generally assumed that the adsorption theory defines the

main mechanism in most situations. It allows for the formation of an intimate contact

and the development of physical forces at the interface, which can then lead to the

subsequent inter-diffusion and chemical bonding mechanisms to further enhance the

adhesion.

The bonding and sealing of microfluidic channels provides particular challenges over

their macro-scale counterparts due to the tight dimensional constraints around the join.

The channels need to be fully sealed and yet the bonding mechanism must not interfere

with its dimensional integrity.

Solely mechanical methods of attachment have been demonstrated for the joining of

microfluidic interconnects [222] providing a localised, liquid-tight connection, however,

it is very difficult to implement for complex microfluidic geometries or for capping

planar channels. Other techniques like microwave, ultrasonic and laser welding show

promise but are beyond the scope of this dissertation.

The coating of an oriented film with a polymer having a lower softening point can

provide a heat sealable layer by a method of mechanical interlocking on a much smaller

scale. Since the oriented film is not heated to its softening point the tendency to shrink

is avoided. This method has the additional advantage that the oriented film gives

support during the heat scaling operation, provided that the heat seal temperature of the

coating is far enough below the softening point of the oriented film substrate. Rossier et

al demonstrated the sealing of cPET microfluidic devices by thermal lamination [156].

However, when this technique of lamination is used for the bonding of machined films,

the issues of material compatibility are further complicated from the difference in

material properties of the orientated substrates and sealing layers. Similarly the use of

adhesives, either liquid or pressure sensitive tapes, requires changing the surface

chemistry in the region of the bond.

117

For this application important considerations in choosing a particular bonding method

include; the surface homogeneity, biocompatibility, surface energy for electroosmotic

flow, leachable molecules and other surface interactions that may affect the device’s

performance. For these reasons thermal diffusion bonding was chosen for this study

because it retained the material homogeneity of the device, thereby simplifying the

liquid-solid interface for the devices.

5.2 Experimental

5.2.1 Channel Fabrication

Channels were constructed from layers of laser micromachined cPET film (Melinex®

Type 542, DuPont Australia). Chapter 3 describes the Nd:YAG laser and

micromachining process. Channel dimensions ranged from 100µm to 250µm in width

and depth, and 3-6cms in length except for filter chambers with a length and width of

5mm and 2mm respectively.

5.2.2 Filter Fabrication

5.2.2.1 Filtering force simulations

Material stress simulations were performed using I-DEASTM with the following material

properties defining the cPET membrane: density of 1400 kg/m3; Young’s Modulus of

2.76 GPa; and Poisson’s ratio 0.38. The dimensions were scaled by a factor of 100:1

due to the dimensional limitations of the software package. The evaluated forces assume

that stretching occurs within the elastic limit of the material. Therefore, the analysis of

stress allows the corresponding scaling of the resultant forces to be in proportion to the

square of the linear dimension (i.e. 10000:1).

2;; LFAFA

F∝∝=σ

Where σ is the stress, F is the force, A is the area and, L is the length. The basic

modelled variables were wall angles of 0o, 7o, 13o and 18o with a pore size of 3.5µm

using 5µm silica spheres. Simulations were carried out to show the force required to

pass a silica particle through excimer laser machined pores with these wall angles.

Figure 57 illustrates one of the geometric layouts used in these simulations.

118

Figure 57 Sphere embedded in tapering hole at position of maximum interference

5.2.2.2 Filter fabrication

Filters were fabricated by perforating 12µm cPET film (Melinex® Type 542, DuPont

Australia) with an Excimer laser beam projected through a chrome-on-quartz mask

containing an array of circular holes. Section 4.3.1 describes the laser and imaging

system configuration, and Table 9 lists the array dimensions used on the mask. The

mask and work-piece stages were held stationary during ablation and the laser was

typically operated in constant energy mode at 1 J/cm2 and 2 J/cm2, with a pulse

frequency of 10Hz. Alignment of the machined area onto the membrane was performed

with the aid of alignment marks machined into the film during the Nd:YAG machining

process.

Table 9 Commonly used mask patterns for 10x10mm pore arrays

Approximations for membrane strength have been applied in the literature for

perforated films. For materials where the intrinsic tensile stress is much smaller than the

# Hole

size

(µm)

Pitch

(µm)

(# of

holes)1/2

# Hole

size

(µm)

Pitch

(µm)

(# of

holes)1/2

1 25 40 250 7 35 55 181

2 25 65 153 8 35 75 133

3 25 90 111 9 35 100 100

4 30 50 200 10 40 60 166

5 30 75 133 11 40 75 133

6 30 100 100 12 40 120 83

119

yield stress σyield, then the pressure pmax that an unperforated membrane can withstand

before rupturing can be estimated by Equation 15 [23]. Where l is the distance between

the membrane supports, h is the thickness of the membrane and E is Young’s modulus.

For polymers this calculated pressure is lower than the actual maximum pressure

because once the material’s yield stress has been reached the material will plastically

deform at the places of maximum stress, thereby redistributing the stress over a larger

area. Perforations and other machining artefacts may also increase membrane

deformation and distribute the stress under load.

Equation 15

2/1

2/3

max 58.0lE

hp

yieldσ=

5.2.3 Bonding and Sealing

Prior to bonding the samples were cleaned by sonication in a 1:1 mixture of ethanol and

water followed by multiple rinses in isopropanol and deionised water.

5.2.3.1 Diffusion Bonding

Thermal diffusion bonding was performed at 180oC in a similar procedure to the

embossing process (Section 2.2.2.3) except no master template was used and the heating

cycle times were up to 2 hours.

5.2.3.2 Surface Functionalised Bonding

Surface functionalised bonding was achieved by UV irradiating (see Section 3.4) the

samples for 10 minutes before bonding. Bonding was achieved in a similar manner to

the Diffusion Bonding except the cycle times were typically reduced to 5 minutes.

5.3 Results

Devices were fabricated from five layers of cPET film stacked and bonded together,

their separate layers are shown in Figure 58. The first layer incorporates the inlet and

outlet ports that are cut entirely through the layer. The second and forth layers are the

channel layers, which are sealed by the first and third, and third and fourth layers

respectively. The third layer provides fluid communication between the channel layers

120

by an array of perforated holes forming the filter membrane. The outlet ports pass

entirely through the first, second and third layers to connect with the fourth layer. The

two holes in adjacent corners on each layer are provided for alignment pins.

Figure 58 Microfluidic chip layers

5.3.1 Channel Fabrication

Two channel fabrication techniques were employed. The first 3-sided technique creates

a channel by etching partway through the substrate, as shown in Figure 59 a). This

method exposes three sides of the channel to the laser machining process. The second 2-

sided technique etches completely through the substrate creating a gasket that can then

be sealed on either side, see Figure 59 b). This method exposes two of the channel walls

to the laser machining process.

Figure 59 a) Three and b) Two Sided channel formation

121

For the gasket or 2-sided technique, where the channel geometries were larger than a

single beam pass, the outer edge of the channel pattern was machined and the resulting

insert removed. The edge quality produced from a 18J/cm2 scanning beam was

considered an adequate trade-off to achieve a relatively fast etch rate (see Section

4.4.4.2). In contrast the 3-sided channels required the beam to be scanned in a raster

profile with an energy of 10J/cm2 to produce smooth shaped surfaces.

5.3.2 Filter Fabrication

A 12µm cPET film was chosen for the filter layer due to its commercial availability in

bulk quantities, compatibility with the other device layers, clarity and rigidity. The

design requirements for the filter included the application needs, laser machining

limitations, and material strength constraints.

The membranes are designed to filter WBC directly from whole blood and to retain

5µm silica particles for solid phase extraction of DNA from lysed whole blood, as

described in detail in Chapter 6. Briefly, the filter needs to retain WBC that are

relatively rigid and roughly spherical, typically ranging in size from 10-20µm, and

remove proteins and RBC. The RBC form the larger of the two major inhibitors. They

are highly deformable and discoidal in shape with a diameter of approximately 8µm and

width of approximately 1µm. Wilding et al. found that their microfluidic weir filters

with 3.5µm gaps in 250µm channels provide the best separation efficiency [223].

The filters in this dissertation were fabricated with pores typically having exit

dimensions between 1-4µm. For a fixed pore size the membrane flow resistance is

reduced by increasing the porosity, however, the filter backpressure and the material

strength limits the pore density. As the pore density increases the membrane’s tensile

strength reduces, lowering the pressure at which the filter will rupture. According to

Equation 15 the maximum pressure the unperforated membrane can withstand is 10psi.

Assuming the approximation applied by Kuiper et al [23] for polymer microseives being

an order of magnitude greater, and a reduction in membrane strength being proportional

to the reduced cross sectional area, then a 52% reduction in cross sectional area (3.5µm

122

hole, 10µm pitch, 12µm membrane thickness, 13o wall angle) would reduce the

maximum pressure the perforated membrane can withstand to approximately 52psi. The

pore arrays were initially fabricated with a square pattern then a staggered format was

used to increase the pore density whilst maximising the membrane strength.

5.3.2.1 Effect of Pore Wall Angle

The Excimer laser machining process produces a wall angle relative to the fluence of

the laser beam. Wall angles of between 7o and 18o degrees were achieved with

machining energies between 1J/cm2 and 2J/cm2 on the Exitech S8000 laser system

described in Section 4.3.1. The effect of this artefact on the ability for a pore to retain a

microparticle under pressure was investigated by simulation using the I-DEAS™

software package. The maximum stress condition for passing a 5µm sphere through a

3.5µm hole with zero degree wall angle is shown in Figure 60.

Figure 60 Stress distribution for a 0.35 mm pore stretched to 0.5 mm

Force simulation results for pores having exit holes of 3.5µm with wall angles of 0°, 7°,

13° and 18° are shown in Table 10. These force simulations show that a change in wall

123

angle between 7° and 18° does not significantly alter the force required to push the

sphere through the orifice.

Table 10 FEA results of pressure versus wall angle to

force a 5 µm sphere through a pore having a 3.5 µm

exit hole diameter.

Wall Angle

(deg.)

Pressure

(psi)

0 107.2

7 18.3

13 16.9

18 15.9

The significant increase in stretching force required to pass the sphere through a pore

with zero degree wall angle, and the small difference in force required for the tapered

pores, can be attributed to the amount of material required to stretch under the force

conditions. The straight hole displays the largest force as the stretch condition was

applied to the total membrane thickness, whereas the stretch was only applied locally

for cases with wall angles.

Furthermore, the work or effort will also vary with changes in wall angle. The smaller

the wall angle the longer the given force needs to act against the resistance of the

sidewalls to force the sphere through. For a more accurate indication of force required,

the friction coefficient corresponding to the surfaces in contact (i.e. silica and PET) and

the fluid medium in which the spheres are suspended, need to be taken into account.

5.3.2.2 Laser Machining

To achieve pore dimensions under 10 micron the cPET films were fabricated using the

excimer laser ablation. Due to the small feature sizes being machined the quality of the

ablated film was highly dependent on the membrane’s placement within the focal point

of the laser beam. The NA of 0.3 gives a depth of field less than 3µm at 248nm. Other

problems that can influence the machining quality at these resolutions include: lens

aberrations; beam path changes due to interactions with the plume; feedback effects on

124

the axis stages introducing oscillations; and material movement from the ablation

process.

It was found that the height of the sample would vary out of the focal plane between

samples and at different locations on the same sample. To overcome some of these focal

problems a focusing procedure was performed for each sample at a proximal location to

the filter area. A thin coating of isopropanol was also required to provide thermal and

mechanical contact between the film and laser work-piece during the machining

process. This coating improved the machining quality, provided bubble formation did

not move parts of the sample out of focus.

Examples of rectangular pore filters fabricated by Excimer laser machining are shown

in Figure 61 to Figure 64. It was found that there was a significant decrease in

membrane strength when maintaining a length to width ratio of greater than two, and a

relatively high porosity (>10%). Consequently these rectangular pore designs were

considered unsuitable for this filtration application and circular pores filters were

investigated.

Figure 61 Excimer laser machined cPET membrane with 15 x 30 µm rectangular

pores

50µµµµm

125


pores


pores

50µµµµm

25µµµµm

126


pores

The circular filters were designed with 1-4µm exit holes with pitches down to 7.5µm,

with the successfully fabricated membranes having porosities of up to approximately

65%. Reducing the pitch further resulted in the inlet of the pores overlapping and

thinning of the membrane. This resulted in a significant weakening of the membrane to

the point of failure when handled. This suggests that the approximation applied for

calculating the membrane strength is not accurate under these conditions. A contributing

factor to the weakening of the membrane may be due to the machining processes

altering the materials intrinsic stresses locally and causing machining artefacts that

enhance crack propagation.

Typical examples of filters fabricated in 12µm PET film with square and offset pore

arrays are shown in Figure 65 and Figure 66 respectively. Figure 65 shows a square

pore array with 8µm entrance holes and 1µm exit holes. The arrays were fabricated with

porosities of approximately 50% to avoid rupturing the membrane at high operating

pressures (see Section 5.3.2). To maintain the membrane strength whilst slightly

increasing the porosity an offset pore array design was chosen. Figure 66 shows an

offset pore array with 8µm entrance holes and 1-2µm exit holes.

20µµµµm

127

Figure 65 Excimer laser machined filter membrane with a square array of pores

having a) 8µm entrance holes and b) 1µm exit holes

25µµµµm

a) b)

25µµµµm

128

Figure 66 Excimer laser machined filter membrane with an offset pore array of

pores having a) 8µm entrance holes and b) 2µm exit holes

Experimentally, a deviation of greater than 10µm from the focal plane would

significantly alter the pattern geometry, producing pattern overlap and thinning of the

membrane, and resulting in unusable filters. The results, shown in Figure 67,

demonstrate this change in pore size and distribution with a change in focus. Figure 67

a) depicts the structures formed when the membranes are in focus, and Figure 67 b) and

c) show changes in focus of 20µm and 30µm respectively.

50µµµµm

50µµµµm

50µµµµm

a)

b)

25µµµµm

129

Figure 67 Excimer laser machined filters fabricated a) in focus b) 20µm out of

focus and c) 30µm out of focus

a)

b)

c)

50µµµµm

50µµµµm

50µµµµm

130

5.3.3 Bonding and Sealing

It was considered that the thermal bonding process would not significantly affect the

PET surface chemistry since the Thermo Gravimetric analysis shown in Chapter 4

showed no significant thermal degradation of PET below 200oC. Despite carrying out

the bonding process at elevated pressures, the channel voids prevented an increase in

pressure inside the channel region.

Prior to bonding the device layers were cleaned with isopropanol in an ultrasonic bath

removing most of the smaller pieces of ejected material that had redeposited around the

cut, and leaving only the larger fragments behind, as shown in Figure 68 a). During

bonding the debris and ridges formed from the laser cutting process, which were as

large as 20µm high, were evenly compressed back into the bulk giving steeper wall

angles and a more even surface (Figure 68 b). The combination of the pressure

distribution and the reduced softening point of the debris and ridge material would be

expected to contribute to the redistribution of material around the cut during this

bonding process.

Figure 68 Scanning electron microscope image of 3ω Nd:YAG laser cut 350µm

thick PET film before a), and after b) bonding

The sealing of oriented films presents particular difficulties due to the film attempting to

return to its original unstretched state when heated. Highly oriented films are also liable

to crystallise further during temperature cycling by nucleation from the existing

100µm Formed ridge 100µm

Larger re-deposited debris Compressed ridge a) b)

131

crystalline structure. From the thermal analysis of Section 3.3 it was shown that the

glass transition temperature of the cPET was approximately 80oC and that crystalline

melt occurs around 250oC, with thermal degradation not occurring until well above

300oC. For bonding purposes the higher temperatures may increase the particle mobility

for enhanced inter-substrate diffusion, whilst maintaining the temperature below the

crystalline melt point may help to retain the orientated film’s bulk characteristics.

However, the longer the cPET is held at temperatures between Tg and Tm the greater

the growth of crystallisation. It is also known that polyester film will hydrolize and

become brittle under conditions of high temperatures and humidity, and has been shown

to reduce the tensile strength of the material by approx 25% after 250 hours exposure to

steam at atmospheric pressure [175]. Consequently, to facilitate drying of the samples

the bonding was performed under vacuum conditions (-95 kPa) with the samples raised

to temperature at a rate of less than 10 oC/min.

It was observed that bonding trials above 200oC resulted in brittle devices with their

structures delaminating or fracturing when handled, suggesting a reduction in tensile

strength far greater than would be expected from hydrolysis at atmospheric pressure. It

was also observed that quench cooling the samples, at atmospheric pressure, to room

temperature immediately after the bonding cycle caused the cPET to become brittle and

fragile. Furthermore, it was found that significant increases in pressures (>100bar) also

produced brittle non-bonded films characteristic of highly crystalline material. These

observations are consistent with the findings of Wei et al who showed that the

temperature of crystallisation for cPET is lower for faster cooling rates and lower

pressures, and that higher rates of crystallisation in cPET are obtained with an increase

in pressure [224]. Therefore, it is understood that optimum bonding for these orientated

cPET films by diffusion requires a careful balance of the temperature, pressure, bonding

time, and cooling rate to minimise further crystallisation.

The thermal bonding parameters of pressure, temperature and time were varied both

during the bonding and cooling processes according to Table 11. Typically suitable

bonds were achieved over 60 minutes at 80bar and 180oC, although longer timeframes

would tend to produce a more robust device. Lower applied pressures (<60bar) typically

resulted in uneven pressure distribution producing only partially bonded devices. This

132

was due to slightly uneven bonding plates requiring a minimum pressure to apply an

even force distribution onto the sample.

Table 11 Bonding parameters

Parameter Range Units

Pressure 50-100 Bar

Temperature 160-220 Celsius

Time 20-180 Minutes

Cooling rate 5-200 oC/min

Examples of poorly bonded and well bonded substrates with the same layered

construction are shown in the cross section images of the devices in Figure 69 a) and b),

respectively. The poorly bonded substrates in Figure 69 a) were cooled at 200 oC/min at

atmospheric pressure and clearly show the delamination at two interfaces after the cross

sectioning process, and all the layers subsequently delaminated upon further handling.

In contrast the device in Figure 69 b) was cooled at 5oC/min under 80bar pressure and

showed no signs of delamination during use. The striations and rough edges seen at the

edges of the channel are artefacts from using the laser in the cross sectioning process.

Figure 69 Channel cross sections of 3ω Nd:YAG laser machined channels in cPET

film illustrating a) poorly and b) well bonded devices

Figure 70 shows cross sections of channels fabricated using the 2-sided and 3-sided

techniques using Excimer and of 3ω Nd:YAG laser machining. Clearly channels have

been formed where the machining process has entirely removed material for both

techniques. However, the two-sided machining of 3ω Nd:YAG process that produced a

delaminations

100µm 100µm

a) b)

133

smooth cross section prior to bonding, see Figure 70 d) insert, did not entirely remove

the material from the machined area resulting in material reflow upon bonding and

channel blockage (Figure 70 d). The two sided Excimer laser machining, that clearly

ablated material, produced a well defined rectangular channel cross section (Figure 70

b).

Figure 70 Sealed cPET fluidic channels showing excimer cut a) two

sided and b) three sided, and 3ω Nd:YAG cut c) two sided and d)

three sided channels with the insert showing the 3 sided channel prior

to bonding.

It was found that the bonds between surfaces on layers adjacent to channel voids failed.

This was attributed to the deformation into the microstructures providing pressure relief

at these points during the bonding process. Therefore a multistep bonding method was

used to ensure that the membrane layer was fully sealed above the exit channel. For

these devices, layers three to five were first bonded together before layers one and two

were added. Figure 71 a) shows the result of a single step bond process and Figure 71 b)

shows the result of well bonded surfaces from the multistep procedure. The striations

100µµµµm

a) b)

100µµµµm

c) d) 100µµµµm

120µµµµm

134

and rough edges seen at the edges of the channel are artefacts from using the laser in the

cross sectioning process.

Figure 71 Channel cross section of 3ω Nd:YAG laser

machined channels in cPET film bonded by the a) single step

procedure, and b) two step procedure

As seen in Figure 71 b), deformation of the upper and lower surfaces into the channel

occurred during the one step bonding procedures. Deformation of this type was a factor

of the applied pressure, dwell time, the width of the channel, and thickness of the

spanning layers.

To reduce bonding cycle times and channel deformation, and to overcome problems of

delamination due to brittle devices, a bonding technique using UV modified surfaces

was trialed. A similar method was employed by Truckenmuler et al. [79] who bonded

PMMA surfaces at lower temperatures by UV exposure of the substrates surfaces to

reduce the materials interfacial Tg. Although these oxidised surfaces can lead to

interference with the analytical/biological processes during device operation (Section

3.6), from Section 3.4.2 it was shown that UV surface modification can be limited to

patterned areas, thereby limiting exposure to the bonded areas.

It is well known that prolonged UV exposure causes surface molecules to break into

smaller molecular weight species. These smaller molecules are not bound as strongly as

the rest of the bulk material and may be more easily removed from the surface, see

Section 3.4, weakening the adhesion of the interfacial layer. It was found that devices

50µµµµm 50µµµµm

Membrane

layers

a) b)

135

with UV exposures greater than 30 minutes tended to delaminate easily or provide no

initial bond. UV treated surfaces were therefore limited to 30 min UV exposures to

provided surface oxidation and limit the extent of molecular degradation.

Attempts to peel apart the bonded substrates generally resulted in substrate failure prior

to delamination. Figure 72 shows a 250x100µm cross section of a UV bonded channel

with the dotted white lines indicating the bonded interfaces. The striations and rough

edges seen at the edges of the channel are artefacts from using the laser in the cross

sectioning process.

Figure 72 Cross section of an of 3ω Nd:YAG laser machined channel in cPET

bonded using the UV thermal bonding method

The UV exposed surfaces were bonded in 5 minutes at 160oC in a similar experimental

setup to the diffusion bonding experiments. Although the UV exposure produced lower

molecular weight groups on the surface the time scale of the bonding process was far

less than the purely diffusion based bonding experiments. These short bonding cycles

introduced the possibility that molecular interdiffusion was no longer the main

mechanism of adhesion. It is likely that the applied temperature and pressure allow the

surfaces to come into close proximity, where other mechanisms of bonding occur.

Mechanisms of bonding include chemical bonding via Van Der Walls interactions

between the oxidised surfaces and or mechanical interlocking from the flow of the UV

exposed surfaces due to a lowering of the interfacial Tg as suggested by Truckenmuller

et al. [79]. It was considered unlikely that covalent bonding between substrates would

100µm

136

occur from metastable species as substrates that were left at atmosphere for one week

still bonded under these conditions.

Figure 73 depicts a typical test microchip fabricated with the aforementioned machining

and bonding processes. The microchip was assembled from five layers of PET film

stacked and bonded together; their separate layers are shown in Figure 58. The

microchannels and through holes were machined by the Nd:YAG laser, and the filter

perforations(inset) were machined with the Excimer laser, prior to UV surface

modification and thermal bonding.

5.4 Summary

Microfluidic filtration devices were successfully fabricated from five layers of cPET

film using laser machining, surface treatment and thermal bonding processes. The filter

membranes were fabricated by perforating 12µm cPET films using Excimer laser

lithography. The 3ω Nd:YAG laser machining enabled rapid prototyping of complex

fluidic circuits in gasket mode. The raised edges and debris formed from the machining

process was compressed back into the bulk during bonding. Diffusion bonding of the

cPET film was found to require a careful balance of temperature, pressure, and time to

5mm

Figure 73 A filtration microchip fabricated from 3ω Nd:YAG laser machining

(insert shows Excimer laser machined filter membrane)

137

minimise substrate crystallisation. A faster UV modified bonding procedure was found

to provide stronger bonds at lower temperatures with faster cycling times.

138

6. Performance Evaluation of Sample Preparation

Biochip

Sample preparation is a key requirement for the successful integration of microfluidic

systems and their use as point of care devices. Typically, for genomic based analysis

from whole blood, samples require extensive possessing to remove inhibitory

substances for subsequent analysis. Furthermore the microfluidic integration of the

previously separate sample preparation, amplification, and detection stages imposes

restrictions on the materials and fabrication methods used. As discussed in the previous

chapters, surface chemistry plays a critical role in capillary and electroosmotic flow,

non-specific binding, and background fluorescence.

This chapter evaluates the performance of biochips fabricated by the methods discussed

in the previous chapters. Performance is firstly established in terms of the flow

characteristics of the devices, particularly for capillary, electroosmotic and pressure

driven flow. The filter chip performance is then discussed with relation to sample

preparation involving: blood flow in microchannels; particle retention for both silica

particles and leukocytes; and DNA extraction performed from whole blood.

6.1 Background

6.1.1 Filtration

Filtration is an operation where particles suspended in solution are separated by passing

the solution through a porous medium. The particles may be retained on the filter’s

surface or within its voids depending on the filter geometry and the particle size. The

filter may also be arranged in line where the filter is arranged perpendicular to the fluid

flow, or in a cross-flow format where the fluid flow is across the membrane surface.

To provide a filtration device capable of both white blood cell and solid phase

extraction an in-line configuration was chosen. An important issue with in-line filtration

is the development of a cake layer trapping unwanted materials and increasing the

backpressure. Methods used to help minimise these problems include back-flushing to

139

physically dislodge trapped material, and surface modification to reduce surface

adsorption [225].

6.1.2 Capillary Flow

The surface energy plays an important role when initially wetting microchannels or

when air bubbles are introduced into the wetted system. The forces generated by a fluid

meniscus are a result of the interfacial tensions and can be represented by free energy

equation (Equation 3). The contact angle of a liquid was described in Chapter 3 with

relation to characterising the surface energy of a material.

The resultant pressure drop across a liquid/gas interface in a capillary can be

represented by equating the energy reduction of a fluid rising in a capillary with the

work performed in raising the liquid column [226]. In rectangular microchannels the

pressure produced by capillary action can be represented by Equation 16, where ∆p is

the pressure difference, γla is the surface tension of the liquid air interface, and w and h

are the channel width and height.

Equation 16

+−

+=

−+=∆

2211

1111cos2

))((2

hwhwwh

whp la

slsa θγγγ

From this equation it can be clearly seen that the force applied to a fluid meniscus is

dependent on the channel dimensions and surface energy. Therefore capillary flow can

be controlled by both channel geometry and surface energy changes [227]. Generally

for small channels, if the channel is more hydrophilic then cosθ approaches one and the

channels will fill by capillary pressure. For surfaces with contact angles greater than 90o

then an applied pressure is required to pass fluid through the same channels.

In microfluidic systems employing capillary flow with a finite input reservoir, the input

reservoir’s fluid meniscus applies a force on the fluid system based on the Laplace

Equation 16. For reservoirs with input geometries where the fluid meniscus reduces

with the reduction in volume, as is the case with a droplet on a flat surface, there is an

increase in the pressure applied to the fluid from the meniscus as the radius of the input

droplet reduces [228].

140

The pressure required for initial wetting or passing a bubble through a filter depends on

the geometry of the membrane and the contact angle generated at the solid-liquid-air

interface. This membrane backpressure is related to the maximum pore size and can be

calculated from the membrane bubble point method [229]. The bubble point method

allows the characterisation of the maximum pore size in a membrane by measuring the

pressure required to pass the air/liquid interface through the pores. The relationship

between pressure and pore radius is then provided using the Young’s equation and can

be represented as Equation 17 where r is the pore radius, P is the pressure, θ is the

liquid-solid contact angle, and γ is the surface tension of the liquid.

Equation 17

P = 2 γ cosθ / r

6.1.3 Electroosmotic Flow

Electroosmotic flow has played an important role in capillary electrophoresis systems

and more recently in the development of µTAS pumping systems. Electroosmotic flow

(EOF), electroosmosis or electrooendoosmosis are terms used to describe the movement

of an electrolyte solution in contact with a solid surface due to an applied tangential

electric field and the presence of fixed surface charges [230]. This movement occurs

from the migration of the electrical double layer formed at the solid/liquid interface. In

appropriately sized capillaries the adjacent fluid bulk is moved through viscous force

interactions along with the charged layer. The direction of movement is dependent on

the charge of the double layer that is formed from the fixed surface charges attracting

oppositely charged ions in solution. The resultant electroosmotic velocity, νeo, and

electroosmotic mobilities, µeo, are given by;

νeo = µeoE

µeo = εζ / 4πη

where E is the applied field and ζ is the zeta potential; η is the buffer viscosity; and ε is

the buffer dielectric constant. For particular ionic species the resultant electroosmotic

flow is the sum of the electroosmotic and electrophoretic µep, mobilities due to the ionic

charge attraction along the applied field, and is given by

ν = ( µeo + µep ) E

141

6.1.4 Pressure Driven Flow

For pressure driven laminar flow it has been shown that the frictional forces of

rectangular microchannels are similar to circular capillaries. Therefore the pressure drop

through a rectangular microchannel can be approximated by using the Hagen-Poiseuille

relationship assuming cylindrical channels, Equation 18, where L is the length of the

pipe, η the dynamic viscosity, u the average velocity and d is the effective tube diameter

[226].

Equation 18

∆P = (32ηµL)/(d2)

For filter membranes where the pore diameters are significantly smaller than the tube

length the Hagen-Poiseulle relationship can be related to the entire membrane by

factoring the number of pores per unit area (N), as shown in Equation 19.

Equation 19

∆P = (128ηµL)/(Nπd4)

For channels packed with rigid particles the pressure drop associated with the packed

material is given by Equation 20; where k is the permeability given by k=α3/(150(1-α)2,

with the interparticle porosity α.

Equation 20

∆P = (ηµL) / (kd2)

6.1.4.1 Blood Flow Effects in Channels

Blood is a viscous fluid mixture consisting of plasma and cells. Proteins represent

about 7-8 wt% of the plasma, the bulk of which are albumin, globulin, and fibrinogen,

which plays a major role in the blood clotting process. The cellular component of blood

consists of three main cell types (erythrocytes, leukocytes, thrombocytes), the most

abundant of which are the red blood cells (RBCs) or erythrocytes comprising about 95%

of the cellular component of blood. The red blood cells can form stacked coin-like

structures called rouleaux, which can clump together to form larger RBC structures

142

called aggregates. Both rouleaux and aggregates break apart under conditions of

increased blood flow or higher shear rates. At lower shear rates, reduced average

velocity of less than 1/sec, red blood cell aggregates start to form with a size

comparable to that of the tube diameter.

Therefore, blood is a non-Newtonian fluid that at low shear rates has an apparent

viscosity that is quite high due to the presence of rouleaux and aggregates. However, at

shear rates above about 100/sec, only individual cells exist, and blood behaves as if it

were a Newtonian fluid. In small capillaries or microchannels the blood flow is

Newtonian near the vessel wall where the wall shear rate is significantly higher than

100/sec, and as one gets closer to the centre line of the vessel the shear rate approaches

zero and blood exhibits non-Newtonian behaviour.

For Newtonian fluids the apparent shear rate, γω, is given by Equation 21 where Q is the

volumetric flow rate, R is the effective capillary radius, and Vaverage is the average

velocity.

Equation 21

R

V

R

Q Average443

==π

γ ω

Tube flow of blood at high shear rates (>100/sec) shows two anomalous effects that

involve the tube diameter. These are the Fahraeus effect (Figure 74) and the Fahraeus-

Lindquist effect [231],[179].

Figure 74 The Fahraeus effect [231]

The RBCs tend to accumulate along the tube axis forming a thin cell-free layer along

the tube wall. The axial accumulation of RBCs, in combination with the higher fluid

velocity, maintains the RBC balance (HF = HD) even though the hematocrit in the tube

is reduced. The thin cell-free layer along the tube wall is called the plasma layer; see


143

Figure 75. The thickness of the plasma layer depends on the tube diameter and the

hematocrit and is typically of the order of several microns.

Figure 75 Axial accumulation of red blood cells [231]

Because of the Fahraeus effect, it is found that as the tube diameter decreases below

about 500 microns, the viscosity of blood also decreases. The decreased viscosity is a

direct result of the decrease in the tube hematocrit. The reduction in the viscosity of the

blood is known as the Fahraeus-Lindquist effect [231],[179].

6.1.5 Sample Preparation

6.1.5.1 DNA isolation

Key to the successful isolation of DNA for PCR is the removal of red blood cells

containing haemoglobin which can form porphyrin compounds that inhibit the

amplification process [94]. Other relevant PCR inhibiting substances include ethanol,

isopropanol, sodium chloride, EDTA and heparin. Commonly employed techniques for

isolating DNA for PCR include Leukocyte filtration and Solid Phase extraction [218].

The difficulty with whole blood filtration lies within the elasticity and structural

strength of the biological cells. The Leukocytes or white blood cells (WBC), are highly

deformable and range in diameter from 6-8µm for the smaller lymphocytes to 12-20µm

for the monocytes, with the most abundant being the neutrophils that comprise

approximately 60-75% of the WBC present and are around 10-15µm in size. The

remaining largest constituents in whole blood are the erythrocytes or Red blood cells,

which have a relatively rigid discoidal shape 8µm in diameter and 2µm thick. Large

mechanical and osmotic stresses during filtration can cause cell rupture, allowing loss of

retentate and addition of further inhibiting substances to the filter medium [229].


144

The silica-guanidiniumthiocyanate solid phase extraction procedure has been used

routinely by many laboratories for the purification of DNA from biological specimens.

The technique involves the shielding of the negatively charged surface with a high ionic

strength buffer to reduce the electrostatic repulsion between the silica surface and DNA.

The chaotropic salt then dehydrates the surface thereby promoting DNA adsorption via

hydrogen bonding. It is known that the buffer properties have a significant effect on the

binding mechanism. For the purposes of this thesis reagents commercially available in

an extraction kit [232] have been used, although it is understood that they have not been

optimised for the microchip method.

6.2 Experimental

6.2.1 Capillary and Blood flow

Samples were fabricated by the gasket or 2-sided technique described in Chapter 5

using Nd:Yag laser machining and thermal bonding. Surface treatments were performed

by 30 minute UV exposure as described in Chapter 4. Images were captured on a

conventional microscope (BX-60 Olympus) interfaced to a CCD camera with x5, x10,

x20, and x50 objectives. Aqueous samples (72 dynes/cm) were introduced for capillary

flow by pipette and custom computer-driven syringe pumps provided flow rates for

blood flow observations (see Figure 76).


The electroosmotic flow velocity was measured by the current monitoring technique

developed by Huang [233]. A single channel was filled via pipette with a 25mM

phosphate buffer, then the reservoir at one end was emptied and replaced with a 50mM

solution of the same buffer. A voltage of 150V/cm was applied by a high-voltage power

supply HP5r (Applied Kilovolts, UK) between the reservoirs. A custom computer

interface controlled the voltage while current measurements were taken across a 10k

Ohm resistor with a custom instrumentation amplifier connected to a data acquisition

card AT-MIO-16E-10 (National Instruments, Australia). The current was monitored

until it reached a plateau giving the time taken for the channel to fill with the less

concentrated solution. Dividing this time by the length of the channel gave the

electroosmotic velocity.

145


Flow based experiments were performed using custom, computer controlled dual

syringe pumps and pressure sensors (0-100 psi) configured as shown in Figure 76. A

conventional microscope (BX-60 Olympus) interfaced to a CCD camera with x5, x10,

x20, and x50 objectives was used for imaging. The 5µm silica particles were purchased

from Bangs Laboratories and the 25-75µm Silica from Agilent (AccuBond). Whole

peripheral blood was collected by finger prick of a healthy Caucasian male and

solutions of methylene blue (0.005%), triton x (1%) and phosphate buffer were all

prepared in deionised water.

Figure 76 Custom computer controlled pumping apparatus


Whole blood of a healthy Caucasian male was collected by radial arterial puncture using

a heparinised syringe, and used within 24 hours. The microchip extractions were

performed using the syringe pump as shown in Figure 76 with a modified protocol from

the QIAamp DNA mini kit (Qiagen, USA). Sample preparation involved mixing 40µL

of whole blood with 10µL Protease, 100µL Buffer Al, and 60µL PBS and incubating

for 20 minutes at 56oC before mixing with 100µL ethanol. The sample solution was

divided into four and loaded into the syringe tubing and pumped through the microchip,

followed by two successive washing steps with 150µL of AW1 and AW2 buffers, and

elution with three successive rinses of 100µL buffer AE. Sample loading, rinsing, and

elution was performed at flow rates of 10µL/min. Elutions were concentrated to 20 µL

by evaporative centrifugation.

146

The isolated DNA were quantified using 1uL Picogreen dsDNA quantification reagent

(Molecular Probes, The Netherlands) in 20µl of the DNA elution and analysed by a

Fusion Plate reader (Packard Biosciences, USA) absorption spectrometer. Figure 77

shows the calibration curve produced by serial dilution of a 10ng/µL DNA sample.

Figure 77 Picogreen dsDNA Calibration Curve

DNA purity was evaluated by observing the ratio of the absorbance of the sample at

260nm and 280nm using spectrophotometer (Cary50, Varian Australia). The

A260/A280 was first described by Warburg and Christensen and is recognised as a

common means to establish the purity of nucleic acids [191]. The samples were

measured by inserting the Cary50 optical probe attachment into the microcentrifuge

tubes containing the samples. Prior to each reading, the probe was washed and a

baseline was measured in distilled water.

PCR was performed with a reaction mix containing 1µL forward and reverse primers

(25µM), 1µL dNTP (10mM), 4uL Buffer (initially containing 1.5-2.5mM Mg2Cl2)

0.1uL Taq polymerase (5 units µl-1), and 20µL diluted sample volumes. 0.4µL MgCl

147

was added to experiments using an adjusted salt mixture. The experiments were carried

out in standard microwell format using a commercially available thermocyler (Corbett

Research PalmCycler). Cycling times were typically 30 seconds at 94oC to denature, 30

seconds at 60oC to anneal, and 30 seconds at 72oC for extension, and repeated 40 times,

followed by a final 2 minute extension cycle. Electrophoresis was performed at

200V/cm with 2µL sample and loading dye.

6.3 Results

6.3.1 Capillary Flow

Capillary flow and passive valving techniques were investigated for the 2-sided

microchannel Nd:YAG laser machining process. A problem for filling microchannels

by capillary forces arises when uneven channel walls alter the angle of the advancing

liquid interface. The contact angles are determined by the surface energy at the liquid-

air-solid interface, however, it is the channel geometry that determines the fluid flow for

a given surface energy. Therefore it is important in microfluidic systems using capillary

flow to control both the surface energy and the channel geometry.

The contact angle for the unmodified PET and surface modified PET was determined in

Chapter 4. PET is naturally hydrophilic with an advancing contact angle of

approximately 75o which reduces to approximately 40o when UV exposed for 30

minutes. These contact angles suggest that the changes in the wall angle required to stop

capillary wicking for the untreated and treated materials would be significantly

different. By applying Equation 16 the minimum wall angles calculated to stop an

advancing capillary meniscus with infinite rear reservoir is approximately 21o and 68o

for the treated and untreated surfaces respectively.

Table 12 Summary of capillary flow experimental results

Angle (a) Untreated UV Treated

30 √ √

45 √ √

60 √ √

148

90 X √

120 X √

135 X X

The Nd:YAG laser machining process was shown to increase the carbon component on

the cut surface and introduce a surface roughness (refer to Section 4.4) of less than ±5

µm. For rectangular channels with the cross section dimensions of 100µm x 250 µm the

effect of these machining artefacts did not inhibit or provide inconsistent capillary flow.

The angles shown in the experimental results of Table 12 represent the wall angle (a) as

depicted in Figure 78 where two channels of different widths are joined. It is clear that

the surface roughness from the machined edges does not stop capillary wicking, and that

the more hydrophilic the surface then the greater the geometric angle required to stop

the capillary flow. These results show that the Nd:YAG machined channels are suitable

for microfluidic flow under capillary forces. The larger angles required to stop capillary

flow in comparison to the calculated values from Equation 16 suggests a possible

increase in the surface roughness and surface charge on the machined walls, which is

consistent with the results from Chapter 4. Furthermore the meniscus of the fluid at the

inlet port imparts a force onto the fluid that would be expected to increase the required

geometric angle to stop the capillary flow.

Figure 78 Capillary flow experimental wall angle (a) between channels x and y.

Avoiding dead volumes is an important consideration when using changes in channel

geometry to stop capillary flow. The larger the corner angle inside a channel, then the

greater the chance of producing a dead volume under flow conditions. In cases where

UV surface treatment is used to increase the capillary force, then patterning techniques

(as shown in Section 3.4.2) can be employed to keep the passive valve areas more

149

hydrophobic. Ensuring these areas are more hydrophobic reduces the required

geometric angles and therefore the chances for dead volumes.

6.3.1.1 Membrane Bubble Point

The filter’s bubble point pressures were measured to understand fluid flow

characteristics during initial wetting and the passage of bubbles within the

microdevices. The native PET membranes stopped fluid flow, whereas the surface

treated membranes allowed capillary wicking to continue through the filter. The

unmodified PET membranes with pore dimensions of 4µm gave bubble point pressures

of 2.6 ± 0.46 psi, which is below the calculated value of 9.8 psi from Equation 19. This

discrepancy between the measured and calculated values may be attributed to a

combination of factors, such as: the pore exits having slightly rounded geometries and

varying in size; a variation of surface tension; and the bubble point pressure method

being limited to measuring the largest pore hole present. Figure 79 is a graph of a

typical bubble point measurement showing the build-up and then sudden drop in

pressure when the liquid is forced through the membrane.

Bubble Point Pressure Measurement

0

20

40

60

80

100

120

1 1001 2001 3001 4001 5001 6001

Time (arb. Units)

Re

lati

ve P

ressu

re (

arb

. U

nits

)

Figure 79 Graph of pressure differential across a filter membrane (100x100 array

of 3.5µm pores) during initial wetting showing the point of maximum pressure

(bubble point) required to force liquid entirely through the membrane.

150


The electrosmotic flow velocities were measured for the untreated and chemically

modified samples. In both cases the direction of flow was from the anode to the

cathode, confirming the presence of negatively charged surface groups. The unmodified

PET gave an electroosmotic mobility of 0.76 x 10-4 ± 0.12 x 10-4 cm2/Vs and after a five

minute NaOH treatment this was increased to 1.2 x 10-4 ± 0.14 x 10-4 cm2/Vs. These

values indicate that YAG laser cut channels in unmodified PET give a comparable

result to embossed polycarbonate, as demonstrated by Liu et al who achieved mobilities

of 0.7 x 10-4 cm2/Vs [234]. Likewise, the three hour UV treatment used by Liu et al

produced a similar result to the five minute base hydrolysis treatment with an increase

in mobility to approximately 1.25 x 10-4 cm2/Vs.


The pressure drop across a membrane filter with cylindrical pores is related to flow rate

according to Equation 19. Figure 80 illustrates the measured and theoretical pressure

drops for 12µm thick membrane filters having 3.5µm cylindrical pores. The predicted

values assume a rigid membrane with cylindrical pores whereas the experimental

samples were fabricated in cPET with conical pore structures. The deviation of the

experimental result from the theoretical curve may be attributed to the widening of the

pore geometry by membrane deflection with an increased flow rate, giving rise to an

effective average radius increase from 3.5µm to 4µm at 500µL/sec.

151

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 100 200 300 400 500

Flow Rate (mm^3/s)

Pre

ssu

re (

psi)

Theory (psi)

Experimental (psi)

Figure 80 Backpressure profile of filter membrane with 100x100 array of 3.5µm

pores

6.3.3.1 Pressure Driven Blood Flow in Microchannels

In practice blood samples are often diluted upon or prior to introduction into

microfluidic devices thereby reducing the fluids viscosity and tendency to aggregate and

block the channels. Observation of the diluted blood flow within the microchannels

showed an accumulation of the cellular components of the blood along the axial

direction of the microchannels, as depicted in Figure 81. This type of flow behaviour

has been well characterised in capillaries for whole blood and described in terms of both

Newtonian and non-Newtonian behaviour [179]. The development of this plasma, or

cell free, layer along the walls of these microchannels reduces the overall viscosity of

the solution and is known as the Fahraeus-Lindquist effect.

152

Figure 81 Axial accumulation of blood cells in 250µm wide 100µm deep

microchannels fabricated by 3ω Nd:YAG laser machining

To avoid the cells settling out of solution and the formation of rouleaux and aggregate

structures for the channel in Figure 81, it was found that flow rates needed to be kept

above approximately 0.1 µL/sec for whole blood. This flow gives a shear rate of

approximately 180/sec in these microchannels, which is of a similar order of magnitude

as those described by Fournier to maintain Newtonian flow of whole blood in capillaries

[231]. The dilution of the sample was found to enhance Fahraeus effect by increasing

the cell free layer. In 400µm x 100µm channels, dilutions of 1:5 and 1:10 provided an

increase in the cell free layer from 8µm for whole blood to 75µm. It was also observed

that the distinctive axial accumulation of cells was lost for shear rates above

approximately 1000/sec.

The Fahraeus effect can be used to improve microfluidic performance in some

microfluidic devices. The causes of blockages, contamination and non-specific binding

caused by uneven wall geometries from the machining process can be minimised for

particulates in solution by maintaining the axial proximity of the particles with high

shear rates. However, depending on the microchip operation, contamination of fluids

and reduced mixing rates may need to be considered due to small dead volumes created

on the machined surfaces. For the Nd:YAG laser machined microchannels described in

50µµµµm

153

this dissertation, the ±5µm wall deviation provided areas of greater adhesion for cells

under low flow conditions where the Fahraeus effect was not observed.


6.3.4.1 Particle Retention

The filters were designed to retain 5µm silica particles for the packing of the

microchannels for solid phase extraction, see Chapter 5 for details. The silica

microspheres were introduced by pipetting an aqueous solution of 10% beads and 1%

Triton X into the device and applying a vacuum (10psi) to the channel outlet. Under this

vacuum the beads took approximately 30 minutes to pack the column tightly, leaving

only a small trailing amount of loosely packed particles, as can be seen in Figure 6a)

which shows a 225µm x 100µm packed channel. Increasing the applied pressure speeds

up the packing process, however, too large a force can cause the microspheres to deform

the membrane and pass through the filter. It was found that pressures significantly

greater than 10 psi allowed the spheres to pass through the membrane at low sphere

concentrations, 1mg/ml. This pressure is of a similar magnitude when compared with

the simulated results of 15.9 psi for a 3.5µm exit hole and 18 degree wall angle (Section

5.2.2). When bead concentrations were increased to 50mg/ml then the bulk of the

spheres were still trapped by filter pressures up to 30 psi, likely due to a keystone effect

where multiple spheres block a single pore. However, at these higher bead

concentrations, care was required to avoid channel blocking at the inlet. Loss of beads

through the filter was not a problem for larger bead to pore ratios (>1.5).

Upon introduction the beads initially filled the length of the channel and then

concentrated at the filter end during the 30 minute packing process. Figure 82 shows a

250µm wide and 100µm deep microchannel packed with 5µm silica spheres. Figure 82

a) shows a channel during the packing process before all the spheres have been tightly

packed. Figure 82 b) shows a close up view of the end of a tightly packed channel with

the cover layer removed. The spheres form an arc at the end of the column due to the

surface tension from the packing process.

154

Figure 82 Particle retention of 5µµµµm silica microspheres in a a) tightly

packed sealed channel, and b) an image of packed microspheres in a

channel with the capping layer removed

Under flow conditions the packed column of 5µm spheres provides a pressure drop

according to Equation 20. To avoid device failure from delamination the channels were

packed with spheres up to a distance of 10mm, ensuring the pressure drop was below

50psi. This assumes a uniform random packing orientation giving rise to an effecting

porosity of 0.40 [235].

6.3.4.2 Lymphocyte Filtration

Lymphocyte filtration was performed in microchips containing filter elements with exit

pores of 4µm. A mix of whole blood 2µl, in 20µl of 0.005% methylene blue (for

nucleus staining) was injected into the devices at 1µL/min, followed by washing with

200µl PBS. The clear red blood cells and smaller stained white blood cells passed easily

through the membrane leaving a cake of mostly stained leukocytes. Figure 83 shows a

filter membrane a) before and b) after lymphocyte filtration. For the unmodified PET

membrane, the flow rate was greatly reduced when the filter clogged, and back-flushing

met with only limited success.


a) b)

155

Figure 83 Membrane filter with 8µm entrance pores a) before filtration, and b)

after filtration of methylene blue stained white blood cells

Modifying the membrane with the one minute chemical treatment before the device was

assembled produced a much more hydrophilic membrane that had less irreversible

binding of particles to its surface. The lower amount of irreversibly bound particles

allowed some back-flushing to occur on the first reverse wash, however, the bulk of the

cells remained adhered to the filter surface after the first flushing cycle. This may be

attributed to the removal of some of the functional groups on the modified filter surfaces

during the first backwash, which causes a decrease of the surface energy as the new

unmodified surfaces are exposed (refer to Section 3.4.2).

The formation of the cake layer increased the backpressure substantially (>50 psi), and

in some instances to the point of device failure through delamination. After multiple

attempts at back-flushing, RBCs were still readily observed on the filter surface and

within the bulk of the cake layer. As the RBCs have been shown to inhibit PCR down to

the 1% level [94] it was decided that leukocyte filtration would not be used for DNA

extraction in these devices. Further work on reduced sample volumes could provide

adequate filtration, however, cPET inhibition of PCR was demonstrated in Section

3.6.1.3 suggesting possible difficulties in performing PCR from a low DNA count.

Other filter configurations such as cross-flow designs [23] may also provide successful

leukocyte isolation in a cPET film microchip.

6.3.4.3 DNA Extraction

DNA extractions were initially performed with 5µm silica spheres using the silica-

guanidiniumthiocyanate solid-phase extraction technique. The introduction of whole

blood into the microdevices packed with 5µm silica particles produced a cake layer


a) b)

156

immediately before the silica particles, resulting in an increase in backpressure and

channel blockage. Back-flushing and pressure pulsing by reversing the syringe pumps

met with only limited success, enabling the sample to penetrate further into the silica

matrix before blocking. A hundred-fold dilution of the sample improved loading

however, debris build-up in the silica matrix was still evident. Furthermore it was

observed that the packing density increased with operation of the devices, thereby

increasing backpressure and decreasing flow rate, a suggested cause of reduced

extraction efficiencies [19].

In order to reduce the backpressure of subsequent devices the channels were packed

with the larger irregular silica particles (25-75µm). Although the lower packing density

of these particles provided an overall reduced surface area for binding, noting that the

54 Amstron pores of the material are too small to allow genomic DNA entry, the

increase in channel size and reduction in tortuosity allowed the blood introduction and

washing stages to occur without blockage. The larger irregular particles also provided a

packed structure that was relatively stable in comparison to the 5µm particles.

Sample loading, rinsing and elution steps were performed at flow rates less than 5

µL/min to ensure the shear forces on the bound DNA did not overcome their binding

energy to the surface [18]. The amount of silica particles packed in each microchannel

varied by up to 25%. To account for this deviation the extraction results were

normalised to the channel with the largest volume of packed silica, the assumption is

that the available surface area for binding would be substantially similar.

Whole blood extractions were performed with 10µL of blood, providing a possible 300-

600ng of DNA assuming 5k-10k cells/µL and 6pg of genomic DNA per cell [232]. The

normalised total from three successive elutions of DNA, was extracted from six separate

microchips. These results shown in Figure 84, give a normalised average of 42.3 ±

4.9ng of DNA which equates to an extraction efficiency of approximately 7-14%.

Results are comparable to Christel et al [21] and Kim et al [236] who used similar load

and elution volumes. Christel et al achieved extraction efficiencies of 10% from

bacteriophage lambda DNA in a silicon etched device, and Kim et al achieved 5%

efficiency from purified DNA in a photosensitive-glass etched device.

157

Total DNA from Extractions (Normalised)

0

10

20

30

40

50

1 2 3 4 5 6

Extraction Chip

DN

A (

ng

)

Figure 84 Solid phase DNA extractions from six different microchips packed with

25-75µm irregularly shaped silica particles

The maximum binding capacity of glass has been reported to be approximately

40ng/cm2 [237]. Assuming the maximum binding capacity of the particles has been

reached with the large excess of sample [21] then approximately 1cm2 of surface area is

available for DNA binding. However, it has been demonstrated that extraction

efficiencies are reduced for whole blood in comparison to purified and plasmid DNA

[18], suggesting a greater available surface area.

The volume of DNA eluted after washing is an important factor for the integration of a

microfluidic solid-phase extraction with DNA amplification. Figure 85 shows the

results of three successive elutions taken from six separate devices and illustrates that

the bulk of the DNA is removed in the first 100µl elution. The values of successive

elutions are shown in Table 13. On average approximately 82% ± 4.6% of the total

DNA collected was eluted in the first fraction, 13% ± 4.9% from the second, and 4.7%

± 0.7% from the final fraction. The elution profile is consistent with the results reported

by Cady et al [131] for the extraction of DNA from Ecoli cells on a silicon chip, where

80% of the DNA was eluted from the first 100µl. Other researchers have shown higher

extraction efficiencies and similar elution profiles using smaller volumes, typically in

the order of 10µL, however, the loading DNA volumes are typically an order of

158

magnitude smaller, well below the binding capacity of their microfluidic devices [16],

[18], [17].

Figure 85 Amount of DNA present in multiple elutions from six solid phase

extraction microchips

Table 13 Normalised DNA elution data

Normalised values

# EL1 EL2 EL3 Total

1 31.53 8.15 1.86 41.53

2 37.48 4.58 2.50 44.56

3 28.88 6.57 1.61 37.07

4 41.55 3.09 2.47 47.11

5 30.25 4.52 1.27 36.03

6 39.71 5.23 2.30 47.24

Average 34.90 5.36 2.00 42.26

Std. dev. 5.35 1.77 0.51 4.90

The purity of the first elutions were 1.27 ± 0.04 using the ratio of the absorbance at

260/280nm [191]. Given that a solution of relatively pure DNA would be expected to

159

return a value of approximately 1.8, the low ratio of absorbance of the elution suggests

that the eluted samples are contaminated. When considering just proteins as

contaminants this value is lower than expected. Less contamination was expected as the

experiment protocol involved two wash stages of 150µL each, and Tian et al had used a

similar porous solid phase and showed that although the proteins were more difficult to

desorb than DNA nearly all the material was removed within the first 30µL [16]. Other

contaminants such as phenols, RNA, and similar aromatic substances can also effect the

absorbance ratio.

Achieving PCR amplification from the eluted DNA proved to be more difficult than

expected. To avoid the surface compatibility issues discussed in Section 3.6.2, the PCR

experiments were performed in standard 96 well plates. Direct PCR amplification from

20µL of the eluted samples gave no detectable results after electrophoresis.

The samples were then prepared by spin evaporation and rehydrated in 20µL distilled

water. The spin evaporation was performed to concentrate the sample and remove

volatile contaminants that might inhibit the PCR reaction. One such inhibitory agent

was the ethanol present in the wash buffer, which is removed from the silica matrix in

the Qiagen protocol by centrifugation prior to elution. It is known that ethanol inhibits

PCR in quantities >1% [238].

Initial electrophoresis experiments on the extracted samples showed no detectable

results. For subsequent extractions an altered washing procedure involving backflushing

(approx 10µl every 50µl of wash buffer) was incorporated. The MgCl2 concentration

was increased in an effort to improve the amplification by lowering the specificity

[146]. It is known that the Mg2+-ion binds tightly to the phosphate sugar backbone of

nucleotides and nucleic acids, and that generally variations of the Mg2+ concentration (<

4mM) are known to inversely alter the specificity.

It can be seen from the electrophoresis gel image in Figure 86 that PCR amplification

was successful on samples A through to F. The lower intensity of the samples, with

respect to the control in lane G, suggests that the samples had either a lower starting

amount of DNA, or that there were contaminants present inhibiting the PCR assay.

160

Figure 86 Gel electrophoresis result after PCR amplification of eluted

DNA from a cPET extraction microchips with a silica particle solid phase

6.4 Summary

The biochips fabricated using the methods developed in the previous chapters were

investigated for microfluidic operation with regards to a sample preparation biochip.

The key findings showed that:

• the increased surface oxidation from the surface modification procedures of the

preceding chapters provided a significant increase in capillary force and electro-

osmotic mobility;

• the diluted blood samples showed a significantly greater axial accumulation of cells

for shear rates between 180-1000/sec;

• the larger irregularly shaped silica particles provided a more stable matrix under

pressure driven conditions; and

• cake layer formation and the subsequent pressure increase proved to be an issue

during leukocyte filtration, with back-flushing achieving only limited success for

one flush cycle when using the surface modified membranes.

DNA extraction was performed successfully using the irregularly shaped silica particles.

The first elution contained 80% of the total eluted DNA, however, issues of

contamination were observed. PCR amplification was achieved from the eluted samples

with the addition of extra washing and evaporation stages.

A B C D E F G

161

7. Conclusions

The research undertaken in this dissertation presents details of the development of a set

of techniques for biochip fabrication using crystalline poly(ethylene terephthalate) film.

The processes and key findings of these investigations are broadly summarised and

discussed in the following results summary. To assist in the replication of the biochips

and to highlight areas requiring further development, the practical findings of this

research are then presented as a set of process guidelines. Finally, the outcome of the

overall body of work is concluded by comparison with the original objectives.

7.1 Results Summary

7.1.1 PET Characterisation and Modification

Many commercially available cPET films contain proprietary additives for improved

handling, UV resistance and clarity. Additionally some films have surface treatments

for improved binding of inks or adhesives. The exact nature of the additives and surface

treatments of the films investigated in this thesis were not known to the author prior to

their arrival in the laboratory. Neither were the manufacturing processes or handling

conditions. Therefore, the surface and bulk characteristics were characterised and

discussed in relation to biochip fabrication and operation. A thermal analysis and

techniques of chemical treatment and UV modification of the cPET Type 542 film,

relevant to the fabrication of the microfluidic devices in the subsequent chapters, were

detailed with the results discussed in terms of their effect on the surface chemistry,

fluorescence and biocompatibility.

Thermal analysis

A thermal analysis was performed on the samples to ascertain aspects of their behaviour

during the thermal bonding process described in chapter 5. The TGA and MDSC results

showed that the cPET film had a glass transition temperature of approximately 80o, a

high degree of crystallinity with a crystalline melt point at 250oC, and no significant

decomposition below 350oC.This implies that the film should remain stable during the

thermal cycling process of PCR, and that it is important to keep bonding temperatures

well below 250oC to retain the structural integrity of the devices’ microstructures.

162

Surface Modification

Surface modification is commonly employed in the fabrication of microfluidic devices

to improve device bonding, fluid flow in channels, and the binding of surface species

for sensing applications.

A method of chemical surface modification involving the cPET film exposure to sodium

hydroxide was implemented. This technique gave an optimum exposure at 1 minute

giving a consistent contact angle of less than 20o after washing. The results indicated

that the reaction proceeds by saponiofication, where the cPET undergoes ester

hydrolysis to form carboxylic acid end groups. The crystalline film was shown to be

more stable than the amorphous PET as the surface of the amorphous material etched

during exposure. XPS analysis confirmed that the base treatment provides a clear

increase in the oxygen components on each of the PET surfaces, with the addition of

carbonate species for the amorphous film.

For functionalising microfluidic channels the UV irradiation method achieves a far less

effective increase in surface energy than the chemical etching process. Although the

results showed contact angles of approximately 35o after 60 minutes exposure, the

contact angle typically increased to approximately 50o after washing suggesting the

removal of lower molecular weight species from the surface. However, the UV surface

modification enabled the use of a simple masked-based surface patterning technique to

limit surface exposure. A pattern resolution of less than 0.1mm was obtained using a

non-contact mask. The inference from the use of a non-contact mask suggested that the

surface modification was due to direct UV exposure rather than a secondary reaction

with gaseous radicals. The XPS analysis confirmed the clear increase in the oxygen to

carbon ratio on the exposed cPET surfaces, consistent with photolytic cleavage.

Fluorescence

The luminescence of a sample can reduce the effectiveness of microarray imaging and

has been highlighted as a major issue for polymer devices [87], [88], [89;90]. The films

used in this thesis were tested for the application of microarray analysis as it was not

known whether the films contained additives or had undergone processes that introduce

functional groups other than those from the monomer units.

163

For the different types of cPET film trialled, it was observed for the two tested

wavelengths of 532nm and 632nm that the magnitude of the fluorescence intensities

fluctuated up to an order of magnitude. For each of these cPET films the intensities at

632nm were substantially larger than at 532nm, and were typically orders of magnitude

larger than glass. The work of Allen et al on the excitation and emission spectra of pure

PET suggests that the observed fluorescence at 532nm is not due to the native PET

material, and therefore other surface chromophores are present. The significant increase

in fluorescence observed after UV exposure confirmed the importance of surface

chromophores contributing to the fluorescence signal. However, no significant changes

in the fluorescence intensity was observed at these wavelengths after thermal bonding

indicating that the relatively low temperatures used for thermal bonding were not

causing any significant amounts of thermal degradation at the surface.

The background fluorescence of cPET film can be minimised by choosing a film with

low surface oxidation. The obvious fluorescent activity of surface groups impacting on

device performance suggests that careful attention is required during manufacturing for

the selection of materials, their storage conditions, and processing parameters. Further

investigations are needed to reduce background fluorescence to allow the use of a wider

range of materials and enable device usage across the entire visible spectrum. However,

it is known that the excitation and emission of molecular groups varies across the visible

spectrum. Therefore, careful selection of the optical system can limit the effect of the

background signal, as shown by the distinct differences in the 532nm and 635nm signal

responses. This is further illustrated by the clearly visible microarray and low

background fluorescence in Figure 24 where a linear array of 10µM Cy5 labelled spots

were deposited in a microchannel fabricated from PET T542 and analysed at 635nm.

Biocompatibility

It was found that the cPET film inhibited the PCR reaction using surface to volume

conditions similar to standard microwell plate PCR assays. As the surface adsorption of

DNA and protein have been identified as the two most probable causes of PCR

inhibition by surface adsorption, the biocompatibility of the cPET films was tested in

relation to these two types of compounds. It was observed that only the protein bound

164

significantly to the surface, suggesting that the inhibition is likely due to protein rather

than DNA surface adsorption.

It was shown that the UV exposed samples significantly inhibited the PCR, which is

consistent with an increase in surface energy causing stronger protein binding.

However, as shown in Section 3.4.2.2, patterning techniques can be used to limit UV

exposure. By limiting UV exposure to the surface regions involved in the PCR process,

problems with inhibition can be minimised. Furthermore, facilitating reagents that bind

preferentially to the surface (such as BSA, skim milk and PEG) can be used in the

reaction mixture to improve the PCR amplification.

It is clear from these results that the large surface to volume ratios of microfluidic

devices will have a great impact on the PCR process. Although the inhibition can be

reduced by controlling the surface energy and using reagents in the reaction mixtures

that will bind to the surface preferentially, further investigations on the surface binding

of the PCR reagents and the alteration of the reagent concentrations by surface

adsorption is required to provide a more detailed understanding.

7.1.2 3ω Nd:YAG Laser Machining of cPET

It is clear from the results that thermal processes contribute significantly to the ablation

of PET with the 3ω Nd:YAG laser. The experimental observations showed

contributions from residual heat during machining for frequencies >1 kHz resulting in

an increased etch rate and lowering of the ablation threshold. Furthermore the 3ω

Nd:YAG laser machined cuts were more representative of the well known thermal CO2

laser process rather than the dominantly photo-oxidative process of the Excimer laser.

This was evidenced by the heat affected areas showing melting, reflow and evidence of

bubbles in a much larger area than would be expected from a purely photochemical

mechanism. However, the decrease in the C1s3 peak of the XPS results is indicative of

the dissociation of ester bonds and the release of low molecular weight species that is

typical of photo-oxidation processes in PET. This indicates the possibility of photo-

chemical mechanisms occurring within the ablation process at the fluences used for

micromachining these microfluidic devices.

165

With regards to micromachining, the frequency multiplication process of the 3ω

Nd:YAG laser results in a considerable loss of energy. However at 3ω (λ=355nm) these

lasers have the advantage of being able to produce a high fluence (~30 J/cm2) over a

small beam area (typically around a 25µm diameter spot). This high energy density

enables a faster rate of machining compared to Excimer based systems, and although the

ultimate resolution is not as high (30µm beam size compared to the Excimer that has

0.8µm diffraction limited optics), it is more than adequate for many microfluidic

applications.

When using the 3ω Nd:YAG laser to machine the cPET films, a single shot ablation

threshold was not well defined due to an inconsistent etch performance at fluences

below 2.0 J/cm2. Although material ablation was achieved consistently above this

threshold, a variation in etch rate was evident when machining the cPET films using

vectors at a fluence of 3.5 J/cm2. The ablation threshold measurements were performed

with single pulses whereas these vector measurements were operated with the system

scanning in burst mode, suggesting the possibility that the ThermEQ™ of the 3ω

Nd:YAG laser system had not compensated sufficiently between these two modes.

To create channel geometries suitable for microfluidics it was necessary to use the 3ω

Nd:YAG laser in a vector scanning format. It is important that the heat affected zone is

minimised to improve feature resolution, and that the cut quality of these structures

provides a consistent surface for capillary flow, minimises dead volumes, and

minimises structures that may impede fluid or particulate flow.

Generally, it was found that multiple laser passes at lower fluences provided a

significantly reduced heat affected zone at the expense of machining time. For example

a single beam scan at 33J/cm2 produced a cut of approximately 23µm deep having an

entrance 45µm wide with a lip width of 35µm, whereas 10 successive beam scans at

4.4J/cm2 produced a cut of approximately 30µm deep having an entrance 15µm wide

with a12µm lip width. For the fabrication of microfluidic devices using 100µm thick

cPET film, 15 beam scans using a fluence of 18 J/cm2 gave a reasonable trade-off

between etch rate and edge quality. At these relatively high fluencies the effects of the

pulse to pulse deviations were considered negligible.

166

When entirely cutting through a substrate it was observed that the width and lip size of

the cut entrance increased with an increase in cutting energy, either with an increasing

fluence or a constant fluence with increasing number of beam scans. In contrast the cut

exit remained of a size consistent with the beam diameter. During raster scanning it was

evident that low beam energies (<14J/cm2) induced localised melting and reflow with

smoothly structured surfaces. However higher beam energies induced vaporisation

causing roughened surfaces with poor control of structuring.

In comparison to the Excimer machining process, the 3ω Nd:YAG laser cutting resulted

in only a slight decrease in the surface O/C ratio with the reduction of the ester

component. Although this reduction in the ester component would still represent a

reduction in the surface energy compared to the native cPET surfaces, it is expected that

the resultant increase in contact angle would be counteracted by the machined surface

roughness.

7.1.3 Biochip Fabrication

This chapter investigated a fabrication method for the construction of microfluidic

filtration devices using the materials and processing techniques discussed in the

previous two chapters. The fabrication method used a layered approach of cPET film,

where the channels were formed by 3w Nd:YAG laser machining and the filter

membranes were fabricated by Excimer laser patterning. The machined films were then

surface treated, and then stacked and bonded together by a thermal bonding process. A

summary of the key results are as follows:

3ω Nd:YAG Laser machining

• Channels were formed in 100µm cPET film using scanned vector patterns

according to the parameters established in Chapter 4 (10 kHz beam with

fluences of 18 J/cm2 and 10 J/cm2 for the 2-sided and 3-sided channels

respectively).

• It was observed that the raised edges and debris formed from the 3ω Nd:YAG

machining process was compressed back into the bulk of the layer upon thermal

bonding. This material reflow reduced the cut entrance width and therefore

improved the channel geometry by providing an even cut width consistent with

167

the beam diameter for the 2-sided channel geometries. However, where the films

were partially etched to create 3-sided channels by raster scanning, then this

material reflow entirely blocked the channel structures during the bonding

process.

Excimer Laser Patterning

• The 12µm cPET films were successfully perforated by Excimer laser

lithography using a 10Hz pulsed beam at 1.0 J/cm2.

• Porous membranes were produced successfully with exit holes down to 1µm and

porosities to 65%, increasing the porosity further resulted in poor membrane

yield due to overlapping of the inlet pores significantly weakening the

membrane structure. Based on the trade-off between membrane strength and

porosity the filtration microchips were fabricated using an offset array of

circular pores giving porosities of approximately 50%.

• The Excimer laser machining resulted in wall angles greater than 7 degrees, with

the simulation results showing a significant reduction in the force required (from

107psi down to <18psi) to pass a 5µm sphere through a 3.5µm pore in

comparison to straight pore geometries.


• Surface modification was performed after machining to allow exposure of the

channel walls along the cut edge.

• Optimum modification of cPET film by Saponification was achieved by 1

minute exposure to 5M NaOH at 50oC.

• Optimum modification of cPET film by Ultra-Violet was performed after 30min

exposure to lamp radiation.

Bonding

• Diffusion bonding of cPET film required a careful balance of the temperature,

pressure, and time to minimise substrate crystallisation. Suitable bonds were

achieved at 180oC and 80 bar for 1 hour followed by a 5oC/min cooling rate.

• A multi-step bonding procedure was necessary to avoid poorly bonded

membrane surfaces adjacent to substrate voids.

• In comparison to the thermal diffusion bonding process for unmodified cPET,

the UV surface modified bonding procedure provided stronger bonds at lower

temperatures (160oC) with faster cycling times (5 minutes). As discussed in

168

chapter 3 the fluorescence and biocompatibility issues associated with the UV

exposure can be overcome by surface patterning.

7.1.4 Performance of Sample Preparation Biochips

The biochips fabricated using the methods developed in the previous chapters were

investigated for microfluidic operation with regards to a sample preparation biochip.

The key findings are summarised below.

a) A consistent and controlled capillary flow was achieved through the use of channel

geometry for both standard and UV treated devices. The surface roughness and

reduction in surface energy observed from the 3ω Nd:YAG machining process did

not stop or cause significant uneven capillary flow. This indicates that similar

treated or untreated laser-machined channels would be suitable for fluid introduction

by capillary forces.

b) For both standard and surface treated devices the Electro-osmotic flow mobilities

were measured and were found to be comparable to devices in the literature, 0.76 x

10-4 ± 0.12 x 10-4 cm2/Vs and 1.2 x 10-4 ± 0.14 x 10-4 cm2/Vs, respectively. In both

the treated and untreated cases the electrosmotic flow was measured in the direction

of the anode to the cathode, confirming the presence of negatively charged surface

species.

c) The unmodified PET membranes with pore dimensions of 3.5µm gave bubble point

pressures of 2.6 ± 0.46 psi. A sufficiently low value to avoid stopping the capillary

flow or cause device failure from bubble trapping under pressure driven conditions.

d) Under pressure-driven flow conditions a non-linear increase in back pressure was

observed suggesting the membrane was deflecting at very high flow rates

(>100uL/sec), resulting in an increased pore size from 3.5µm to 4µm at

500µl/second.

e) Blood flow effects under pressure-driven conditions were characterised showing an

improved Newtonian flow and Fahraeus-Lindquist effects for diluted samples. A

dilution of 1:5 gave rise to an increase in the cell free layer from 8µm for whole

blood to 75µm for the diluted samples. It was also observed that this axial

accumulation of blood cells was lost at shear rates of less than 180/second or greater

than 1000/second.

169

f) Particle retention was achieved for both 5µm spheres and irregularly shaped silica

particles between 25-75µm. However, it proved to be more difficult to retain

particles where the bead to pore ratio was between 1.0 and 1.5. This loss of filter

performance was attributed to the deformation of the membrane during the packing

process at pressures >10psi.

g) Where the channels were packed with regularly shaped 5µm spheres, it was

necessary to limit the column length to 10mm to avoid increased backpressures

(>50psi) that would lead to device failure. Furthermore, stable backpressures were

not achieved for the spheres as continued device usage increased the backpressure

asymptotically. This was not a problem for the irregularly shaped particles, which

were packed with column lengths of up to 25mm and found to have stable

backpressures (<<50psi) immediately after the packing process.

h) Cake-layer formation and the subsequent pressure increase proved to be an issue

during in-line leukocyte filtration. Relatively solid cake layers were observed bound

to the membrane, which substantially reduced the flow rate and potentially trapped

significant numbers of erythrocytes. Back-flushing met with only limited success for

one flush cycle when using the surface modified membranes.

i) DNA extraction was performed using only the irregularly shaped silica particles.

The use of regular shaped spheres caused problems with cake-layer formation and

packing stability. The results showed DNA recovery at between 7-14% using three

successive elutions, with the first elution containing approximately 82% of the DNA

recovered. The DNA purity was relatively low at 1.27 (Abs.260/280nm) suggesting

significant amounts of contaminates were present.

j) Difficulties of PCR amplification were overcome with extra washing and spin

evaporation stages, suggesting the PCR inhibiting compounds were removed in

these subsequent steps.

7.2 Process Guidelines

The research undertaken in the previous chapters has demonstrated the feasibility of

Biochips fabricated from 355nm laser machined PET for the application of DNA

extraction. The practical findings of this research can be presented as a set of fabrication

guidelines to assist in the replication of the biochips and to highlight areas requiring

further development.

170

7.2.1 Design

An in-line filtration design was chosen due to its ability to be used as a sieve for

leukocytes or as a retaining mechanism for a solid phase without complicating the

fluidic pathway. The design was primarily found to be suitable for solid-phase

extraction due to the formation of a solid cake layer resulting from the trapping of red

blood cells during leukocyte filtration. Future development of a cross flow arrangement

can have the advantage of reduced fouling at the membrane, which in turn can minimise

the effects of the formation of a cake layer. A cross flow filter arrangement however,

typically requires a larger effective filter area per unit volume or multiple passes

through the same filter with suitable transmembrane pressures to achieve high

separation efficiencies.

The 3ω Nd:YAG laser was used to machine channels entirely and partially through

substrate layers for the formation of 2-sided and 3-sided channel structures,

respectively. It was shown that the 2-sided, or gasket, approach was most suitable for

this laser microstructuring. Due to the wall height being defined by the thickness of the

film, this approach allowed greater control of channel dimensions than was possible by

3ω Nd:YAG laser machining alone. In contrast, the 3-sided approach resulted in no

significant material removal due to the low energies required to control structuring. This

caused the channels to block during the bonding process by the compression of the lip

edge.

When considering the type of cPET film to use for these devices, the material’s purity,

manufacturing and thermal history, as well as surface treatment should be considered.

The manufacturing and thermal history has a significant impact on the material’s

crystallinity, which in turn provides the material’s strength and behaviour during

thermal diffusion bonding. The biocompatibility of the materials’ surface with the

amplification process and the material’s fluorescence for microarray analysis must be

considered for this application of DNA sample preparation, and the future integration

with amplification and detection. It is important to choose a material with minimal

surface oxidation to improve biocompatibility and minimise fluorescence. The surface

oxidation can then be limited to controlled areas during fabrication to aid in bonding

and fluid control, whilst not interfering with the amplification or detection stages.

171

7.2.2 Fabrication

Laser Machining

It was shown that 3ω Nd:YAG laser machining of PET is a predominantly photothermal

process. However, the identification of the mechanisms during ablation is complicated

as these mechanisms have been shown to vary according to fluence for both

predominantly photothermal and photochemical processes. Although more information

about the chemistry of both the machined surfaces and the ablation products at different

fluences can help to identify these mechanisms, it has been noted that some of the

ablation products are more volatile than the bulk material and undergo further reaction

after irradiation, thereby making it more difficult to characterise the chemical pathway

during ablation [199].

For microfluidic fabrication it was shown that the machining of the 2-sided channels by

the 3ω Nd:YAG laser was best performed using lower energies and multiple vector

passes to cut entirely through the substrate. To avoid the contribution of residual heat

from multiple beam passes, the time between each pass should remain greater than the

diffusion time of the material. A beam energy of 18J/cm2 at 10kHz gave a reasonable

trade off between speed and cut quality. It was generally found that multiple beam scans

of the 3ω Nd:YAG along the same vector provided a more uniform lip along the cut

edge. This was particularly important as uneven accumulation of the reflowed material

resulted in uneven channel walls after bonding where the lip was compressed back into

the vacated region along the cut edge. It is envisaged that the cutting process can be

further optimised by controlling the laser beam parameters of pulse repetition, duration,

and spot overlap during machining.

The machining of the filter membranes was best performed with the cPET film in

thermal contact to a flat surface, such as a glass microscope slide, with a thin meniscus

of isopropanol. The best performing membranes were produced from a mask pattern

that provided an imaged array of 10µm offset circular pores with 3.5µm exit holes. It

was noted that the tapered wall angles from the Excimer machining process

significantly weakened the membranes and limited the pore density. In future this effect

can be minimised by reducing the pore wall angle through the use of high numerical

aperture laser optics in combination with high beam fluence [213].

172


It was demonstrated that ultra-violet exposure of the cPET surface through a mask

oxidises the surface for improved bonding and fluid flow whilst limiting the exposed

areas to avoid increased fluorescence and protein binding. It was confirmed that too

much exposure degraded the surface layer leading to bond failure and the removal of the

oxidised surface molecules upon washing. Consequently, an exposure of 30 minutes

was considered optimal for the experimental conditions described in this dissertation.

Bonding

The thermal diffusion and UV-surface modified bonding techniques can each provide

adequate bond strengths for this application. However, both methods require sufficient

temperature and applied pressure to ensure all the adjacent surfaces are in direct contact

and that any irregularities are compressed back into the bulk.

The thermal diffusion bonding process required careful control of the temperature,

pressure and cooling rate to avoid an increase in the material crystallinity which can

lead to bond failure. Sufficient bonds were demonstrated using this technique, however,

further work is needed to establish suitable diffusion bonding parameters for

temperatures above 180oC. Adequate control of the cooling rate and high pressure

capability is a key requirement to control the substrate crystallisation process and to

minimise cycle times.

In comparison the UV-surface modified bonding technique provided improved bond

strengths. Further, due to the short bonding times and lower temperature, the cPET film

was not affected by an increase in crystallinity. Additional work that alters the surface

functional groups could be used to control the intermolecular forces and thereby assist

in establishing the exact mechanisms of bonding for this technique.

The use of a multi-step bonding procedure prevented the relatively thin membrane layer

deforming into the channels. In contrast, a single step bonding procedure would usually

result in such a deformation due to the relief of pressure at the surfaces immediately

adjacent to the microstructures. In the multi-step bonding procedure the layers are either

bonded one at a time to build up the complete structure, or the membrane layer is first

173

bonded to all the films on one side before the rest of the films are bonded in a second

step.

The issue of non-bonding at the surfaces adjacent to the channels was only problematic

for the membrane layer as the other layers were sufficiently rigid to retain close contact.

Further work in establishing the aspect ratio of layer thickness versus microstructure

span would provide a design guideline for different film thicknesses and their

structuring limitations to avoid deformation.

Solid Phase Packing

The wall angles produced from the Excimer laser machining of the membranes resulted

in the widening of the pores for pressures greater than 10psi, and loss of the 5µm

spherical particles through 3.5µm pores. Consequently the channels should be packed

with particles significantly larger than the pore diameter. Furthermore, the larger (25-

75µm) and irregularly shaped silica particles produced a more stable packed media than

the spherical particles. Another important consideration during device operation is to

minimise the pressure drop along the packed column. Larger and irregularly shaped

particles produce a more porous media, which results in a lower pressure drop.

7.2.3 Operation

The optimal choice of the solid phase, solution chemistries, and reagent volumes are

key issues in providing a fully integrated DNA based biochip. The extraction protocol

developed in this research used reagents from a commercially available DNA extraction

kit [232]. The reagents were optimised for use with a proprietary silica matrix in a

centrifuge based collection kit, and not for the conditions under which they were used in

this investigation. The best yield obtained from the experiments was 14% using a

sample volume of 40µl. Further development is required to provide optimal conditions

for microchip extraction and to provide conditions suitable for the full integration of the

microdevices with the subsequent stages of amplification and detection.

Generally the dilution of whole blood by reagent addition improves fluid handling

within the microdevices. A more dilute solution reduces the overall viscosity, exhibits a

greater cell free layer near the channel walls, and reduces the chance of blockages from

174

the accumulation of cell debris. However, an important consideration when diluting

solutions for the solid phase extraction method is to ensure the proper ionic strength and

pH of the lysis and binding solution are achieved. This solution is required to rupture

the leucocytes and bind the DNA to the silica. Therefore, to fully integrate the sample

preparation stage onto a microchip platform, the method of reagent introduction and

mixing with the sample is important. Typically two strategies can be employed to

achieve this level of integration. The first involves mixing both the solutions fully prior

to introduction into the solid phase section of the device, as was the case with the

experimental setup used in this investigation. The second strategy, used under flow

conditions, involves mixing the two solutions proportionally to ensure the ionic strength

and concentrations of the resultant solution are not altered detrimentally. Both strategies

require sufficient time for the proper rupture of the leukocytes before binding can take

place.

After DNA binding, the solid phase is rinsed to remove unbound materials that may

interfere with the PCR process. It was observed that backflushing during this rinsing

process significantly improved the removal of cell debris and other contaminants.

Dilution of the sample was also found to reduce the amount of trapped contaminants.

However, an important step in the integration of these solid phase extraction biochips

with subsequent microfluidic processes is the reduction in solution volumes. In high

surface-to-volume ratio microdevices the reagent concentration is dramatically affected

by the surface binding of reagents as volumes are decreased. Therefore, further

understanding of the surface binding of PCR reagents is required for effective volume

reduction.

The integration of DNA extraction with PCR amplification and Microarray detection

introduces further issues associated with fluid handling. Pump and valve components

are required to both control fluid flow and isolate fluid segments during thermal cycling

to avoid vapour loss. Many examples of such pump and valve components have been

presented in the literature, however, many are not suitable due to considerations of

biocompatibility and compatibility with operational pressure, materials, and fabrication

processes.

175

The cPET biochips demonstrated electroosmotic flow characteristics similar to other

polymer based devices. However, there are particular issues associated with the use of

electroosmotic flow in a biochip that uses a solid phase media for DNA extraction. The

electrode placement should ensure that: pH changes occurring at the electrodes do not

adversely impact device performance; the relatively high ionic strength of the binding

solution does not induce bubble formation from Joule heating; and the highly charged

silica matrix surfaces do not induce high shear rates that affect DNA binding. These

issues have been mitigated by various strategies illustrated in the literature including the

use of specific electrode and channel geometries to avoid current passing through the

solid phase area.

176

7.3 Conclusion

The motivation of this study was to provide a method for, and an insight into, some of

the issues associated with biochip fabrication by 3ω Nd:YAG laser micromachining of

PET. With biochip sample preparation focused on as a key stage towards the goal of a

fully integrated microfluidic device for point-of-care DNA testing. The research

question was stated as: “Is 3ω Nd:YAG laser machining of polyethylene terephtalate

(PET) suitable for fabricating a biochip for DNA sample preparation?”.

The research undertaken in the previous chapters has successfully demonstrated the

feasibility of Biochips fabricated from 355nm laser machined PET for the application of

DNA extraction. The results presented have met the research objectives by:

- Characterising the cPET film, this provided the thermal properties for bonding and

highlighted the importance of the film’s surface chemistry and its impact on

biocompatibility and fluorescence;

- Characterising the microstructuring of cPET by direct-write frequency tripled

Neodymium Yttrium Aluminium Garnet (3ω Nd:YAG) laser processing, which

provided the machining parameters for biochip fabrication;

- Developing a fabrication method for the assembly of DNA sample preparation

biochips using laser processing of cPET; and

- Evaluating the performance of these biochips in terms of their fluid flow,

backpressure, filtration, and DNA extraction from whole blood.

This thesis concludes that frequency tripled Nd:YAG laser machining of PET films can

be used to produce components of a biochip for molecular diagnostics, namely

microfluidics and solid phase extraction of DNA. This new laser machining method of

PET film enables the fabrication of microfluidic channels more quickly than excimer

based lasers and with improved feature resolution in comparison to CO2 lasers.

Importantly, the thin film fabrication process can be used for fabricating microfluidic

filtration devices entirely from polymers. It was shown that the detrimental impact of

the fabrication process on device biocompatibility and fluorescence can be removed by

using masking techniques to pattern the oxidised areas.

177

The demonstrated technique of thin film cPET micromachining is compatible with

many reel-to-reel processing methods, and as such represents a flexible and relatively

low-cost method of microfluidic manufacture. The development of a fully integrated

microfluidic production line was considered beyond the scope of this work. However, it

is possible that such a production line could be developed with the further integration of

existing process compatible methods, such as lamination and UV/corona/plasma

treatment. With further developments it is envisaged that completely integrated biochips

for DNA analysis will become readily available.

178

8. Publications

M. Atkin, Patent 2003900793 – Microfluidic Filter, 2003

S. Garst, M Schuenemann, M Atkin, M Solomon, E Harvey, “Fabrication of

multilayered microfluidic 3D polymer packages”. Proceedings of the 55th Electronic

Components & Technology Conference, June 2005.

M. Atkin, K. Poetter, R. Cattrall, and E. Harvey, “A microfabricated polymer filter

device for bio-applications,” Conference Proceedings of SPIE Photonics West,

Micromachining and Microfabrication 2004, San Jose

M. Atkin, E. Mutapcic, J. Hayes, and E. Harvey, “A comparison of 355nm and 248nm

laser ablation of PET for microfluidic applications,” Proceedings of the 1st Pacific

International Conference on Application of Lasers and Optics, April 2004

Matthias Schuenemann, David Thomson, Micah Atkin, Sebastiaan Gars, Abdiraham

Yussuf , Matthew Solomon, Jason Hayes, Erol Harvey, “Packaging of Disposable Chips

for Bioanalytical Applications”, IEEE Electronic Components & Technology

Conference, Nevada, USA 2004

M. Atkin, J. Hayes, N. Brack, K. Poetter, R. Cattrall, and E. Harvey, “Disposable

biochip fabrication for DNA diagnostics,” Proceedings of SPIE, Melbourne, Dec 2002

H. Thissen, J.P. Hayes, B.W. Muir, M. Atkin, E.C. Harvey,” Spatially controlled

surface chemistry by excimer laser ablation of thin films,” Proceedings of SPIE,

Melbourne, Dec 2002

M. Atkin, K. Poetter, R. Cattrall, and E. Harvey, “Improved Techniques for the

Miniaturisation of Bioarrays,” Profiles in Industrial Research-Knowledge and

Innovation 2001 & 2002, IRIS.

179

9. References

1. International Human Genome Sequencing Consortium: Initial sequencing and analysis of the human genome: Nature 409: 860-921, 2001

2. J. Venter, et al.: The Sequence of the Human Genome: Science 291 (5507): 1304-1351, 2001

3. R. Ekins, F. Chu: Microarrays- their origins and applications. Trends Biotechnol. 17: 217-218, 1999.

4. M. Schena: Microarray Biochip Technology. 1-881299-37-6 Eaton Publishing, 2000.

5. S. Shoji, M. Esashi: Microflow devices and systems: J. Micromech. Microeng 4: 157-171, 1994

6. E. Bassous, H. Tuab, L. Kuhn: Ink jet printing nozzle arrays etched in silicon: Applied Physics Letters 31: 135-137, 1977

7. S. Terry, J. Jerman, J. Angell: A gas chromatographic air analyzer fabricated on a silicon wafer: IEEE Trans. Electron Devices 26: 1880-1886, 1979

8. D. Reyes, D. Iossifidis, P. Auroux, A. Manz: Micro Total Analysis Systems. 1. Introduction, Theory, and Technology: Analytical Chemistry 74: 2623-2636, 2002

9. P. Auroux, D. Iossifidis, D. Reyes, A. Manz: Micro Total Analysis Systems. 2. Analytical Standard Operations and Applications: Analytical Chemistry 74: 2637-2652, 2002

10. P. Dittrich, K. Tachikawa, A. Manz.: Micro total analysis systems. Latest advancements and trends: Analytical Chemistry 78 (12): 3887-3907, 2006

11. H. Becker, L. Locascio: Polymer microfluidic devices. Talanta 56: 267-287, 2002.

12. H. Becker, C. Gärtner: Polymer microfabrication methods for microfluidic analytical applications. Electrophoresis 21: 12-26, 2000.

13. A. de Mello: Plastic fantastic (potentials of polymers and plastics as substrate materials in microfluidic applications): Lab on a chip 2 (2): 31N-36N, 2002.

14. Y. Sun, Y. Kwok: Polymeric microfluidic system for DNA analysis: Analytica Chimica Acta 556 (1): 80-96, 2006.

15. J. Landers: Molecular Diagnostics on Electrophoretic Microchips: Analytical Chemistry 75 (12): 2919-2927, 2003.

16. H. Tian, A. Huhmer, J. Landers: Evaluation of silica resins for direct and efficient extraction of DNA from complex biological matrices in a miniaturized format. Analytical Biochemistry 283(2): 175-91, 2000.

17. C. Easley, J. Karlinsey, J. Bienvenue, L. Legendre, M. Roper, S. Feldman, M. Hughes, E. Hewlett, T. Merkel, J. Ferrance, J. Landers: A fully integrated microfluidic genetic analysis system with sample-in-answer-out capability: Proc Natl Acad Sci USA 103 (51): 19272-19277, 2006.

18. M. Breadmore, K. Wolfe, I. Arcibal, W. Leung, D. Dickson, B. Giordano, M. Power, J. Ferrance, S. Feldman, P. Norris, J. Landers: Microchip-Based Purification of DNA from Biological Samples: Analytical Chemistry 75 (8): 1880–1886, 2003.

19. K. Wolfe, M. Breadmore, J. Ferrance, M. Power, J. Conroy, P. Norris, J. Landers: Toward a microchip-based solid-phase extraction method for isolation of nucleic acids. Elecrophoresis 23: 727-733, 2002.

180

20. J. Cheng, L. Kricka, E. Sheldon, P. Wilding: Sample Preparation in Microstructured Devices. Topics in Current Chemistry 194: 215-231, 1998.

21. L. Christel, K. Petersen, W McMillan, G. Kovacs: High Aspect Ratio Silicon Microstructures for Nucleic Acid Extraction. Solid-State Sensors and Actuators Workshop, Transducers research Foundation: 363-366 1998.

22. Y. Liu, C. Rauch: DNA probe attachment on plastic surfaces and micro Fluidic hybridization array channel devices with sample oscillation: Analytical Biochemistry, 317 (1): 76-84, 2003.

23. S. Kuiper: Development and application of microsieves. The Netherlands, University of Twente, Enschede, 2000.

24. S. Terry, J. Jerman, J. Angell: A gas chromatographic air analyzer fabricated on a silicon wafer. IEEE Trans. Electron Devices 26: 1880-6, 1979.

25. P Gravesen, J Branebjerg, O. Jensen: Microfluidics-a review. J. Micromech. Microeng 3: 168-182, 1993.

26. A. Manz, N. Graber, H. Widmer: Miniaturized Total Chemical Analysis Systems: a Novel Concept for Chemical Sensing. Sensors and Actuators, Bl: 244-248, 1990.

27. S. Shoji, M. Esashi: Microflow devices and systems. J. Micromech. Microeng 4: 157-171, 1994.

28. S. Shoji: Micro Total Analysis System (µTAS): Electronics and Communications in Japan: 82 (2): 21-29, 1999.

29. P Gravesen, J Branebjerg, O. Jensen: Microfluidics-a review. J. Micromech. Microeng 3: 168-182, 1993.

30. D. Reyes, D. Iossifidis, P. Auroux, A. Manz: Micro Total Analysis Systems. 1. Introduction, Theory, and Technology: Analytical Chemistry 74: 2623-2636 2002.

31. J. Rossier, F. Reymond, P. Michel: Polymer microfluidic chips for electrochemical and biochemical analyses. Electrophoresis 23: 858-867, 2002.

32. D. Boone, A. Ricco, Z. Fan, H. Tan, H. Hooper, S. Williams: Plastic Advances Microfluidic Devices: Analytical Chemistry 73 (3): 78A-86A, 2002.

33. M. Heckele, W. Schomburg: Review on micro molding of thermopplastic polymers. J. Micromech. Microeng. 14: R1–R14, 2004.

34. G. Fiorini, D. Chiu: Disposable microfluidic devices: Fabrication, function, and application: Biotechniques 38 (3): 429-46, 2005.

35. C. Malek: Laser processing for bio-microfluidics applications (part I): Anal Bioanal Chem 385 (8): 1351-1361, 2006.

36. C. Malek: Laser processing for bio-microfluidics applications (part II): Anal Bioanal Chem 385 (8): 1362-1369, 2006.

37. H. Becker, C. Gartner: Polymer microfabrication technologies for microfluidic systems: Analytical and Bioanalytical Chemistry, 390 (1): 89-111, 2008.

38. X. Zhang, X. Jiang, C. Sun: Micro-Stereolithography of polymeric and ceramic microstructures: Sensors and Actuators A: Physical 77 (2): 149-156, 1999.

39. P. Cooley, D. Wallace, B. Antohe: Applications of Ink-Jet Printing technology to BioMEMS and Microfluidic Systems: Proceedings, SPIE Conference on Microfluidics and BioMEMS 2002.

181

40. C. do Lago, H. da Silva, C. Neves, J. Brito-Neto, J da Silva: A Dry Process for Production of Microfluidic Devices Based on the Lamination of Laser-Printed Polyester Films: Anaytical. Chemistry 75 (15): 3853–3858, 2003.

41. Y. Kawamura, K. Toyoda, S. Namba: Applied Physics Letters 40: 374, 1982.

42. R. Srinivasan, S. Mayne-Banton: Applied Physics Letters 41: 576, 1982.

43. F. Srinivasan: Etching polymer films with ultraviolet laser ulses of long (IO-400 ps) duration: Journal of Applied Physics 72 (4): 1651-1653, 1992.

44. P. Dyer: Excimer Polymer Laser Ablation: 20 Years On: Applied Physics A 77: 167-173, 2003.

45. L. Laude, D. Martinez, Cl. Dicara, Fr. Hanus, K. Kolev: The ablation of polymers under excimer laser irradiation: the physics of the process and the polymer structure: Nuclear Instruments and Methods in Physics Research B 185: 147-155, 2001.

46. M. Roberts, J. Rossier, P. Bercier, H. Girault: UV Laser Machined Polymer Substrates for the Development of Microdiagnostic Systems: Analytical Chemistry 69 (11): 2035-2042, 1997.

47. R. Pethig, J. Burt, A. Parton, N. Rizvi, M. Talary, J.Tame: Development of biofactory-on-a-chip technology using excimer laser micromachining: Journal of Micromechanics and Microengineering 8(2): 57-63, 1998.

48. P. Martin, D. Matson, W. Bennett, D. Hammerstrom: Fabrication of plastic microfluidic components: SPIE conference on Microfluidic Devices and Systems 3515: 172-176, 1998.

49. M. Prins, J. Weekamp, J. Giesbers: Multichannel structures made from micrometre-thick plastic foils: J. Micromech. Microeng. 9: 362-363, 1999.

50. P. Dyer, G. Oldershaw, J. Sidhu: CO2 Laser Ablative Etching of Polyethylene Terephthalate: Applied Physics B: Lasers and Optics 48 (6): 489-493, 1989.

51. H. Klank, J. Kutter, O. Geschke: CO2-laser micromachining and back-end processing for rapid production of PMMA-based microfluidic systems: Lab on a chip 2 (4): 242-246, 2002.

52. J. Cheng, C. Wei, K. Hsu, T. Young: Direct-write laser micromachining and universal surface modification of PMMA for device development. Sensors and Actuators B: Chemical 99 (1):186-196, 2004.

53. M. Bowden, O. Geschke, J. Kutter, D. Diamond: CO2 laser microfabrication of an integrated polymer microfluidic manifold for the determination of phosphorus. Lab Chip. 3(4):221-223, 2003.

54. M. Jensen, M. Noerholm, L. Christensen, O. Geschke: Microstructure fabrication with CO2 laser system: characterization and fabrication of cavities produced by raster scanning of the laser beam: Lab Chip 3: 302-307, 2003.

55. B. Paul, R. Peterson: Microlamination for Microtechnology-Based Energy, Chemical, and Biological Systems: roceedings of the ASME Adanced Energy Systems Division 39: 45-52, 1999.

56. M. Gower: Laser micromachining for manufacturing MEMS devices. Proceedings of SPIE Vol. 4559: 53-59, 2001

57. B. Paul, R. Peterson: Microlamination for Microtechnology-Based Energy, Chemical, and Biological Systems: Proceedings of the ASME Adanced Energy Systems Division 39: 45-52, 1999.

182

58. S. McGinty, G. O'Connor, T. Glynn: A comparative study of channel formation in polymer materials using VUV and UV nano-second laser sources for use in micro-fluidic applications: National Centre for laser applications, NUI-Galway, Ireland, 2005.

59. S. Kuper, M. Stuke: Femtosecond UV Excimer Laser Ablation: Applied Physics B: Laser and Optics 44: 199-204, 1987.

60. T. Lippert, T. Dickinson: Chemical and Spectroscopic Aspects of Polymer Ablation: Special Features and Novel Directions: Chemical Review 103: 453-485, 2003.

61. Y. Zhang, R. Lowe, E. Harvey, P. Hannaford, A. Endo: High aspect-ratio micromachining of polymers with ultrafast laser: Applied Surface Science 186: 345-351, 2002.

62. K. Sugioka, Y. Cheng, K. Midorikawa: Three-dimensional micromachining of glass using femtosecond laser for lab-on-a-chip device: Applied Physics A: Materials Science and Processing 81: 1-10. 2005.

63. K. Yamasaki, S. Juodkazis, S. Matsuo, H. Misawa: Three-dimensional micro-channels in polymers: one-step fabrication: Applied Physics A 77: 371-373, 2003.

64. J. McDonald, M. Chabinyc, S. Metallo, J. Anderson, A. Stroock , G. Whitesides: Prototyping of Microfluidic Devices in Poly(dimethylsiloxane) Using Solid-Object Printing. Analytical Chemistry 74: 1537-1545, 2002.

65. K. Hosokawa, T. Fujii, I. Endo: Handling Picoliter Liquid Samples in a Poly(dimethylsiloxane)-Based Microfluidic Device: Analytical Chemistry (71): 4781-4785, 1999.

66. J. Anderson, C. Daniel, R. Jackman, O. Cherniavskaya, J. McDonald, H. Wu, S. Whitesides, G. Whitesides: Fabrication of Topologically Complex Three-Dimensional Microfluidic Systems in PDMS by Rapid Prototyping: Analytical Chemistry 72: 3158 2000.

67. P. Wilding, L. Kricka, J. Cheng, G. Hvichia, M. Shoffner, P Fortina: Integrated Cell Isolation and Polymerase Chain Reaction Analysis Using Silicon Microfilter Chambers. Analytical Biochemistry 257(2): 95-100, 1998.

68. H. Becker, U. Heim: Polymer high aspect ratio structures fabricated with hot embossing. Proceedings of transducers '99, Sendai, Japan: 1432-1435, 1999.

69. H. Becker, U. Heim: Hot embossing as a method for the fabrication of polymer high aspect ratio structures. Sensors and Actuators 83: 130-135, 2000.

70. G. Alonso-Amigo: Polymer Microfabrication for Microarrays, Microreactors, and Microfluidics: Journal of the Association for Laboratory Automation 5: 96-101, 2001.

71. J. Xu, L. Locascio M. Gaitan, C. Lee: Room-Temperature Imprinting Method for Plastic Microchannel Fabrication: Analytical Chemistry 72: 1930-1933, 2000.

72. V. Piotter, T. Hanemann, R. Ruprecht, J. Hausselt: Microinjection Moulding of Medical Device Components. 2002.

73. R. Mehalso: The Microsystems Road in the USA: mstnews 4: 6-8, 2002.

74. M Schueneman, D. Thompson, M. Atkin, S. Garst, A. Yussuf, M. Solomon, J. Hayes, E. Harvey: Packaging of Disposable Chips for Bioanalytical Applications.: Proceedings of the 55th Electronics Components and Technology Conference [ECTC 05] 603-610, 2005.

75. E. Verpoorte, B. Van Der Schoot, S. Jeanneret, A. Manz, A. Widmer, N. De Rooij: Three-dimensional micro flow manifolds for miniaturized chemical analysis systems: J. Micromech. Microeng 4: 246-256, 1994.

76. H. Dreuth, C. Heiden: Thermoplastic structuring of thin polymer films: Sensors and

183

Actuators 78: 198-204, 1999.

77. H. Hoffman: Polymicro Newsletter.: 8-12, 2003.

78. J. Findlay, S. Atwood, L. Bergmeyer, J. Chemelli, K. Christy, T. Cummins, W. Donnish, T. Ekeze, J. Falvo, D. Patterson, J. Quenin, J. Shah, D. Sharkey, J. Sutherland, R Sutton, H. Warren, J Wellmen: Automated Closed-Vessel System for in Vitro Diagnostics Based on Polymerase Chain Reaction: Clinical Chemistry 39: 1927-1933, 1993.

79. R. Truckenmuller, Z. Rummler, T. Schaller, W. Schomburg: Low-cost thermoforming of micro fluidic analysis chips: J. Micromech. Microeng. 12: 375-379, 2002.

80. J. Watson, F. Crick: A structure for Deoxyribose Nucleic Acid: Nature 171: 737-738, 1953.

81. J Venter, et al: The Sequence of the Human Genome: Science 291: 1304-1351, 2001.

82. D. Perrett: Capillary electrophoresis in clinical chemistry: Ann Clin Biochem 36: 133-150, 1999.

83. P. Brown, D. Botstein: Exploring the new world of the genome with DNA microarrays. Nature Genetics 21: 33, 1999.

84. J. Cheng, L. Waters, P. Fortina, G. Hvichia, S. Jacobson, J. Ramsey, L. Kricka, P. Wilding: Degenerate oligonucleotide primed-polymerase chain reaction and capillary electrophoretic analysis of human DNA on microchip-based devices. Analytical Biochemistry 257(2): 101-106, 1998.

85. E. Southern, K. Mir, M. Shchepinov: Molecular interactions on microarrays. Nature Genetics 21: 5-9, 1999.

86. M. Seidel, R. Niessner: Automated analytical microarrays: a critical review: J. Micromech. Microeng 5: 57-63, 2008.

87. J. Webster, M. Burns, D. Burke, C. Mastrangelo: Monolithic Capillary Electrophoresis Device with Integrated Fluorescence Detector. Analytical Chemistry 73: 1622-1626, 2001.

88. Y. Liu, C. Rauch, R. Stevens, R. Lenigk, J. Yang, D. Rhine, P. Grodzinski: DNA Amplification and Hybridization Assays in Integrated Plastic Monolithic Devices: Analytical Chemistry 74 (13): 3063-3070, 2002.

89. B. Hock, M. Seifert, K. Kramer: Engineering receptors and antibodies for biosensors. Biosensors & Bioelectronics 17: 239–249, 2002.

90. M. Schena: Microarray Biochip Technology. Eaton Publishing, 1-881299-37-6, 2000.

91. R. Hawkins, P. Yager: Nonlinear decrease of background fluorescence in polymer thin-films - a survey of materials and how they can complicate fluorescence detection in µTAS: Lab Chip 3: 248-252, 2003.

92. N. Allen, J. McKellar: Luminescent Species in Poly(ethylene terephthalate): Makromol. Chem. 179: 1978.

93. K. Mullis, F. Faloona, S.Schart, R. Saiki, G. Horn, H. Erlich: Specifc enzymatic Amplification of DNA in-vitro: The Polymerase Chain Reaction: Nature 324: 163, 1986.

94. H. Erlich: PCR Technology. Principles and Applications for DNA Amplification: 0195098757: Oxford University Press, USA, 1993.

95. C. Zhang, J. Xu, W. Ma, W. Zheng: PCR microfluidic devices for DNA amplification: Biotechnology Advances 24 (3): 243-284, 2006.

184

96. A. Woolley, D. Hadley, P. Landre, A. deMello, R. Mathies, M. Northrup: Functional Integration of PCR Amplification and Capillary Electrophoresis in a Microfabricated DNA Analysis Device. Analytical Chemistry 68(23): 4081, 1996.

97. M. Shoffner, J. Cheng, G. Hvichia, L. Kricka, P. Wilding: Chip PCR. I. Surface passivation of microfabricated silicon–glass chips for PCR: Nucleic Acids Research, 24 (2): 375-379, 1996.

98. J. Cheng, M. Shoffner, G. Hvichia, L. Kricka, P. Wilding: Chip PCR. II. Investigation of different PCR amplification systems in microfabricated silicon–glass chips: Nucleic Acids Research 24 (2): 380-385, 1995

99. E. Schilling, A. Kamholz, P. Yager: Cell Lysis and Protein Extraction in a Microfluidic Device with Detection by a Fluorogenic Enzyme Assay. Analytical Chemistry. 74: 1798-1804, 2002.

100. H. Nakano, K. Matsuda, Y. Masafumi, N. Teruyuki, E. Isao, Y. Tsuneo: High Speed Polymerase Chain Reaction in Constant Flow: Bioscience, Biotechnology, and Biochemistry 58: 349-352, 1994.

101. M. Kopp, A. Mello, A. Manz: Chemical amplification: continuous-flow PCR on a chip. Science 280(5366): 1046-8, 1998.

102. I. Schneegass, R. Brautigam, J. Kohler: Miniaturized flow-through PCR with different template types in a silicon chip thermocycler: Lab on a chip 1: 42-49, 2001.

103. J. Liu, M. Enzelberger, S. Quake: A nanolitre rotary device for polymerase chain reaction: Electrophoresis 23: 1531-1536, 2002.

104. L. Waters, S. Jacobson, N. Kroutchinina, J. Khandurina, R. Foote, J. Ramsey: Multiple sample PCR amplification and electrophoretic analysis on a microchip. Analytical Chemistry 70(24): 5172, 1998.

105. D. Baechi, R. Buser, J. Dual: High-Density Microvalve Arrays for Sample Processing in PCR Chips: Biomedical Microdevices 3: 183-190, 2001.

106. Y. Liu, C. Rauch, R. Stevens, R. Lenigk, J. Yang, D. Rhine, P. Grodzinski: DNA Amplification and Hybridization Assays in Integrated Plastic Monolithic Devices: Analytical Chemistry 74 (13): 3063-3070, 2002.

107. E. Lagally, I. Medintz, R. Mathies: Single-Molecule DNA Amplification and Analysis in an Integrated Microfluidic Device. Analytical Chemistry 73(3): 565, 2001.

108. Y. Shin, K. Cho, S. Lim, S. Chung, S. Park, C. Chung, D. Han, J. Chang: PDMS-based micro PCR chip with Parylene coating: J. Micromech. Microeng. 13: 768-774, 2003.

109. N. Panaro, X. Lou, P. Fortina, J. Kricka, P. Wilding: Surface Effects on PCR Reactions in Multichip Microfluidic Platforms: Biomendical Microdevices 6, 75-80, 2004.

110. Q. Zou, Y. Miao, Y. Chen, U. Sridhar, C. Chong, T. Chai, Y Tie, C. Hui, L. Teh, T. Lim, C. Heng: Micro-assembled multi-chamber thermal cycler for low-cost reaction chip thermal multiplexing: Sensors and Actuators A 102: 114-121, 2002.

111. C. Koh, W. Tan, M. Zhao, A. Ricco, Z. Fan: Integrated Polymerase Chain Reaction, Valving, and Electrophoresis in a Plastic Device for Bacterial Detection Analytical Chemistry 75: 4591-4598, 2003.

112. R. Liu, J. Yang, R. Lenigk, J. Bonanno, P. Grodzinski: Self-contained Fully Integrated Biochip for Sample Preparation, Polymerase Chain Reaction Amplification, and DNA Microarray Detction: Analytical Chemistry 76: 1824-1831, 2004.

113. W. Al-Soud, P. Rådstrom: Purification and Characterization of PCR-Inhibitory

185

Components in Blood Cells. Journal of Clinical Microbiology 39 (2): 485-493, 2001.

114. W. Al-Soud, L. Jonsson, P. Rådstrom: Identification and Characterization of Immunoglobulin G in Blood as a Major Inhibitor of Diagnostic PCR: Journal Of Clinical Microbiology 38 (1): 345-350, 2000.

115. W. Al-Soud, P. Rådstrom: Effects of Amplification Facilitators on Diagnostic PCR in the Presence of Blood, Feces, and Meat: Journal Of Clinical Microbiology 38 (12): 4463-4470, 2000.

116. J. Yang, Y. Liu, C. Rauch, R. Stevens, R. Liu, R. Lenigk, P. Grodzinski: High sensitivity PCR assay in plastic micro reactors: Lab on a Chip 2: 179-187, 2002.

117. H. Tian, A. Huhmer, J. Landers: Evaluation of silica resins for direct and efficient extraction of DNA from complex biological matrices in a miniaturized format: Analytical Biochemistry 283(2): 175-191, 2000.

118. D. Peterson: Solid supports for micro analytical systems: Lab on a chip 5 (2):132-139, 2005.

119. J. Fu, O. Mao, J. Han: Artificial molecular sieves and filters: a new paradigm for biomolecular separation: Trends in Biotechnology 26: 311-320, 2008.

120. X. Chen, D. Cui, C. Liu, H. Cai: Microfluidic Biochip for Blood Cell Lysis: Chinese Journal Of Analytical Chemistry 34 (11):1656-1660, 2006.

121. X. Chen, D. Cui, C. Liu, H. Cai: Fabrication of Solid Phase Extraction DNA Chips Based on Bio-Micro-electron-Mechanical System Technology: Chinese Journal of Analytical Chemistry 34 (3):433-437, 2006.

122. X. Chen, D. Cui, C. Liu: On-line cell lysis and DNA extraction on a microfluidic biochip fabricated by microelectromechanical system technology: Electrophoresis 29 (9):1844-1851, 2008.

123. B. Weigl, P. Yager: Microfluidic Diffusion-Based Separation And Detection. Science 283(5400): 346, 1999.

124. C. Blattert, R Jurischka, A. Schoth, P. Kerth, W. Menz: Separation of blood cells and plasma in microchannel bend structures: Proceedings-SPIE The International Society For Optical Engineering 5591: 143-151, 2004.

125. J. Stuelpnagel, M. Chee, K. Gunderson: The use of microfluidic systems in the detection of target analytes using microsphere arrays. In Illumina, Inc: WO 00/71243 A1 .

126. J. Choi, C. Ahn, S. Bhansali, H. Thurman Henderson: A new magnetic bead-based, filterless bio-separator with planar electromagnet surfaces for integrated bio-detection systems. Sensors and Actuators B 68: 34-39, 2000.

127. Q. Ramadan, V. Samper, D. Poenar, Z. Liang, C. Yu, T. Lim: Simultaneous cell lysis and bead trapping in a continuous flow microfluidic device: Sensors and Actuators B: Chemical 113 (2): 944-955, 2006.

128. C. Karp: Microfluidic filter. In Nanostream, Inc.: US6,811,695 2004.

129. J. Moorthy, D. Beebe: In situ fabricated porous filters for microsystems. Lab Chip, 3 (2): 62-66, 2003.

130. S. Metz, C. Trautmann, A. Bertsch, P.. Renaud: Polyimide microfluidic devices with integrated nanoporous filtration areas manufactured by micromachining and ion track technology: Journal of Micromechanical Microengineering 14: 324-331, 2004.

131. N.Cady, S. Stelick, C. Batt: Nucleic acid purification using microfabricated silicon structures: Biosensors & Bioelectronics 19: 59-66, 2003.

186

132. H. Anderson, W. Wijngaart, P. Enoksson, G. Stemme: Micromachined flow-through filter-chamber for chemical reaction on beads: Sensors and Actuators B 67: 203-208, 2000.

133. H. Sato, T. Kakinuma, J. Go, S. Shoji: In-channel 3-D micromesh structures using maskless multi-angle exposures and their microfilter application: Sensors and Actuators A: Physical 111 (1): 87-92, 2004.

134. Y. Yoon, J. Park, F. Cros, M. Allen: Integrated Vertical Screen Microfilter System Using Inclined SU-8 Structures. IEEE The Sixteenth Annual International Conference on Micro Electro Mechanical Systems: 227 – 230, 2003.

135. L. Leoni, T. Desai: Biomolecular exclusion and diffusion through nanoporous micromachined membranes: Nanotech Conference Proceedings: Biomems and Biomedical: 2001.

136. T. Crowley, V. Pizziconi: Separation of Blood with Planar Microfluidic Devices. Arizona State University, 2000.

137. P. Li, J. Harrison, C. Skinner, R. Oleschuk, L. Shultz-lockyear: Trapping of Bead-Based Reagents within Microfluidic Systems: On-Chip Solid-Phase Extraction and Electrochromatography: Analytical Chemistry 72: 585-590, 2000.

138. B. He, L. Tan, F. Regnier: Microfabricated Filters for Microfluidic Analytical Systems. Analytical Chemistry 71(7): 1464, 1999.

139. M. Ali, T. Wagner, M. Shakoor, P. Molian: Review of laser nanomachining: Journal of Laser Applications 20: 169-184, 2008.

140. J. Cheng, E. Sheldon, D. Wu, A. Uribe, L. Gerrue, J. Carrino, et al: Prepartion and hybridisation analysis of DNA/RNA from E.coli on microfabricated bioelectronic chips. Nature Biotechnol 16: 541-546, 1998.

141. J. Chen, M. Shoffner, G. Hvichia, L. Kricka, P. Wilding: Chip PCR. II. Investigation of different PCR amplification systems in microfabricated silicon–glass chips: Nucleic Acids Research 24: 380-385, 1996.

142. J. Cheng, E. Sheldon, et al: Isolation of cultured cervical carcinoma cells mixed with peripheral blood cells on a bioelectronic chip. Analytical Chemistry 70(11): 2321, 1998.

143. L. Waters, S. Jacobson, N. Kroutchinina, J. Khandurina, R. Foote, J. Ramsey: Microchip device for cell lysis, multiplex PCR amplification, and electrophoretic sizing. Analytical Chemistry 70(1): 158-162, 1998.

144. J. Wen, L. Legendre, J. Bienvenue, J. Landers: Purification of nucleic acids in microfluidic devices: Analytical Chemistry 80 (17): 6472-6479, 2008.

145. B. Vogelstein, D. Gillespie: Preparative and analytical purification of DNA from agarose. Proc.Natl.Acad.Sci USA 76: 615-619, 1979.

146. R. Boom, C. Sol, M. Salimans, C.Jansen, P. Wertheim-van Dillen, J. van der Noordaa: Rapid and simple method for purification of nucleic acids: Journal of Clinical Microbiology 28(3): 495-503, 1990.

147. D. Gerhold, T. Rushmore, C. Caskey: DNA chips: promising toys have become powerful tools. Trends in Biochemical Sciences 24: 168-173, 1999.

148. K. Wolfe, M. Breadmore, J. Ferrance, M. Power, J. Conroy, P. Norris, J. Landers: Toward a microchip-based solid-phase extraction method for isolation of nucleic acids 23: 727-733, 2002.

149. M. Breadmore, K. Wolfe, I. Arcibal, W. Leung, D. Dickson, B. Giordano, M. Power, J.

187

Ferrance, S. Feldman, P. Norris, J. Landers: Microchip-Based Purification of DNA from Biological Samples: Analytical Chemistry 2003, 75 (8): 1880–1886, 2003.

150. L. Legendre, J. Bienvenue, M. Roper, J. Ferrance, J. Landers: A Simple, Valveless Microfluidic Sample Preparation Device for Extraction and Amplification of DNA from Nanoliter-Volume Sample: Analytical Chemistry 78: 1444-1451, 2006.

151. K. Melzak, C. Sherwood, R. Turner, C. Haynes: Driving forces for DNA adsorption to silicca in perchlorate solutions: Journal of Colliod and Interface Science 181, 635-644, 1996.

152. J. Wen, C. Guillo, J. Ferrance, J. Landers: Microfluidic chip-based protein capture from human whole blood using octadecyl (C18) silica beads for nucleic acid analysis from large volume samples: Journal of Chromatography A 1171 (1-2):29-36, 2007.

153. Y. Cho, J. Lee, J. Park, B. Lee, Y. Lee, C. Ko: One-step pathogenic specific DNA extraction from whole blood on a centrifugal microfluidic device: Lab on a chip 7: 565-573, 2007.

154. A. Bhattacharyya, C Klapperich: Thermoplastic Microfluidic Device for On-Chip Purification of Nucleic Acids for Disposable Diagnostics: Analytical Chemistry 78, 788-792, 2006.

155. D. Brooks: PET Packaging Technology. 1-84127-222-1 Sheffield Academic Press, 2002.

156. J. Rossier, M. Roberts, R. Ferrigno, H.Girault: Electrochemical Detection in Polymer Microchannels. Analytical Chemistry 71: 4294, 1999.

157. J.Rossier, A. Schwarz, F. Bianchi, R. Ferrigno, F. Reymond, H. Girault: Microchannels Networks for Electrophoresis Separations. Electrophoresis 20: 727, 1999.

158. D. Miles: Polymer Technology. Chemical Publishing Company.

159. A. Brody: The Wiley Encyclopedia of Packaging Technology. 0-471-063975-5 John Wiley & Sons, Inc, 1997.

160. D. Juncker, H. Schmid, U. Drechsler, H. Wolf, B. Michel, N. Rooij, E. Delamarche: Autonomous Microfluidic Capillary System: Analytical Chemistry 74: 6139-6144, 2002.

161. Y. Feng, Z. Zhou, X. Ye, J. Xiong: Passive valves based on hydrophobic microfluidics: Sensors and Actuators A: Physical, 108 (1): 138-143, 2003.

162. J. Lee, H. Moon, J. Fowler, T. Schoellhammer, C. Kim: Electrowetting and electrowetting-on-dielectric for microscale liquid handling. Sensors and Actuators A 95: 259-268, 2002.

163. S. Barker, M. Tarlov, H. Canavan, J. Hickman, L. Locascio: Plastic Microfluidic Devices Modified with Polyelectrolyte Multilayers. Analytical Chemistry 72(20): 4899, 2000.

164. S. Soper, A. Henry, B. Vaidya, M. Galloway, M. Wabuyele, R. McCarley: Surface modification of polymer-based microfluidic devices: Analytica Chimica Acta 22026: 1-13, 2002.

165. Y. Hanein, Y. Vickie Pan, B. Ratner, D. Denton, K. Bohringer: Micromachining of non-fouling coatings for bio-MEMS applications: Sensors and Actuators B 81: 49-54, 2001.

166. A. Blawas, W. Reichert: Protein patterning. Biomaterials 19595-609, 1998.

167. L. Shamansky, C. Davis, J. Stuart, W. Kuhr: Immobilization and detection of DNA on microfluidic chips: Talanta 55: 909-918, 2001.

168. G. Diaz-Quijada, R. Peytavi, A. Nantel, E. Roy, M. Bergeron, M. Dumoulin, T. Veres: Surface modification of thermoplastics - Towards the plastic biochip for high throughput

188

screening devices: Lab on a chip 7 (7):856-862. 2007.

169. C. Sun, D. Zhang, L. Wadsworth: Corona Treatment of Polyolefin Films - A Review. Advances in Polymer Technology 18(2): 171-180, 1999.

170. J. Friedrich, L. Wigant, W. Unger, A. Lippitz, H. Wittrich: Corona, spark and combined UV and ozone modification of polymer films WeBP23. Surface and Coatings Technology 98: 879-885, 1998.

171. J. Rabek: Polymer photodegradation: Mechanisms and experimental methods. London, Chapman and Hall, 1995.

172. N. Grassie: Polymer Degradation & Stabilisation. Cambridge University Press, 1985.

173. G. Beddard, N. Allen: Emmision Spectroscopy. 1989.

174. J. Webster, M. Burns, D. Burke, C. Mastrangelo: Electrophoresis System With Integrated On-Chip Fluorescence Detection: The Thirteenth Annual International Conference on Micro Electro Mechanical Systems: 306 – 310, 2000

175. Dupont Teijin Films: Private communication.

176. P. Kingshott, H. Griesser: Surfaces that resist bioadhesion. Current Opinion in Solid State and Materials Science 4: 403–412, 1999.

177. B. Kasemo: Biological surface science: Surface Science 500: 656-677, 2002.

178. N. Stellwagen, C. Gelfi, P. Righetti: The Free Solution Mobility of DNA: Biopolymers 42 (6): 687 – 703, 1997.

179. H. Goldsmith, G. Cokelet, P. Gaehtgens: Robin Fahraeus: evolution of his concepts in cardiovascular physiology: American Journal of Physiology - Heart and Circulatory Physiology 257, 1005-H1015, 1989.

180. J. Foreman, S. Saurbrunn, C. Marcozzi: Exploting the Sensitivity of Thermal Analysis Techniques to the Glass Transition. TA-082

181. Thermal Solutions. TS-20A Determination of initial crystallinity by modulated DSC.

182. L. Thomas: TA039. Interpreting Unexpected Events and Transitions in DSC Results.

183. R. Johnson, R. Dettre. Surface and colloid science. New York, John Wiley & Sons, 85-153, 1969.

184. R. Johnson, R. Dettre: Contact angle hysteresis: Adv Chem 43, 112-144, 1964.

185. J. Moulder, W. Stickle, P. Sobol, K. Bomben: Handbook of X-ray Photoelectron Spectroscopy: 0-9627026-2-5, Perkin-Elmer Corporation, 1992

186. S. Lazare, R. Srinivasan: Surface Properties of Poly(ethyiene terephthalate) Fllms Modlfied by Far-Ultraviolet Radiation at 193 nm (Laser) and 185 nm (Low Intensity): J. Phys. Chem. 90: 2124-2131, 1986.

187. S. Lazare, V. Granier: Excimer laser light induced ablation and reactions at polymer surfaces as measured with a quartz-crystal microbalance: Journal of Applied Physics 63: 2110, 1988.

188. M. Day, M. Wiles: Photochemical degradation of poly(ethylene terephthalate). II. Effect of the wavelength and environment on the decomposition process: Journal of Applied Polymer Science 16: 191-202, 1972.

189. M. Schena: DNA Microarrays. Oxford University Press, 1999.

190. M. Day, M. Wiles: Photochemical degradation of poly(ethylene terephthalate). III.

189

Determination of decomposition products and reaction mechanism: Journal of Applied Polymer Science 16: 203-215, 1972.

191. O. Warburg, W. Christian: Isolation and crystallization of enolase: Biochem Z 310, 384-421, 1942.

192. B. Boddinghaus, T. Wichelhaus, V. Brade, T. Bittner: Removal Of Pcr Inhibitors By Silica Membranes: Evaluating the Amplicor Mycobacterium Tuberculosis Kit. Journal of Clinical Microbiology, 39(3750–3752), 2001.

193. T. Hawkins: A Purification and Isolation using Solid-Phase: Nucleaic Acid Research 22: 4543-4544,1994.

194. M. von Allmen: Laser-Beam Interactions with Materials. 3-540-17568-7 Berlin, Springer-Verlag, 1987.

195. J. Miller: Laser ablation and desorption. 0-12-475975-0 London, Academic Press Limited, 1998.

196. R. Srinivasan, K. Casey, B. Braren, M. Yeh: The significance of a fluence threshold for ultraviolet laser ablation and etching of polymers: Journal of Applied Physics 67: 1604, 1990.

197. Y. Chen, H. Zheng, K. Wong, S. Tam: Excimer laser drilling of polymers. Microelectronic Packaging and Laser, SPIE, 1997.

198. J. Rossier, P. Bercier, A. Schwarz, S. Loridant, H. Girault: Topography, Crystallinity and Wettability of Photoablated PET Surfaces. 1999.

199. R. Srinivasan, B. Braren: Ultraviolet Laser Ablation of Organic Polymers: Chemical Review 89 (6): 1303–1316, 1989.

200. R. Srinivasan, W. Leigh: Ablative Photodecomposition: Action of Far-Ultraviolet (193 nm) Laser Radiation on Poly(ethy1ene terephthalate) Films: J. Am. Chem. Soc. 104 (24): 6784–6785, 1982.

201. P. Dyer, J. Sidhu: Excimer laser ablation and thermal coupling efficiency to polymer films: Journal of Applied Physics 57: 1420-1422, 1985.

202. H. Carslaw: Conduction of Heat in Solids. Oxford, Oxford University Press, 1959.

203. W. Steen: Laser Materials Processing. London, Springer-Verlag, 2003.

204. R. Srinivasan: Ablation of polymethyl methacrylate films by pulsed (ns) ultraviolet and infrared (9.17 pm) lasers: A comparative study by ultrafast imaging: J Applied Physics 73: 2743, 1993.

205. P. Rumsby, E. Harvey, D. Thomas, N. Rizvi: Excimer laser patterning of thick and thin films for high density packaging: Proc. SPIE 3184: 176, 1997.

206. P. Atkins: The Elements of Physical Chemistry. Oxford University Press, 1992.

207. N. Mansour, K. Ghaleh: Ablation of polyethylene terephthalate at 266nm: Applied Physics A 74, 63-67, 2002.

208. J. Hayes, H. Thissen, R. Deam, M. Ghantasala: Nanometer resolution excimer laser ablation of thin polymer films. Proceedings of the 1st Pacific International Conference on Applications of Lasers and Optics, 2004.

209. E. Sutcliffe, R. Srinivasan: Dynamics of UV laser ablation of organic polymer surfaces: Journal of Apllied Physics 60, 3315, 1986.

210. F. Burns, S. Cain: The effect of pulse repition rate on laser ablation of polyimide and

190

polymethylmethacrylate-based polymers: Journal of Pyhsics D: Applied Physics 29: 1349-1355, 1996.

211. E. Harvey, J. Hayes: Effect of plume dynamics on the uniformity of excimer laser ablation for microfabrication. Proceedings of SPIE 5276: 92-99, 2004.

212. M. Atkin, J. Hayes, N. Brack, K. Poetter, R. Cattrall, E. Harvey: Disposable biochip fabrication for DNA diagnostics. Proceedings of SPIE 4937: 125-135, 2002.

213. E. Harvey, P. Rumsby, M. Gower, J. Remnant: Microstructuring by Excimer Laser. SPIE - Micromachining and Microfabrication Process Technology 2639: 1995.

214. G. Beamson, D. Briggs: High Resolution XPS of Organic Polymers: The Scienta ESCA300 Database. 0 471 93592 1 West Sussex, England, John Wiley & Sons Ltd, 1992.

215. S. Lazare, R. Srinivasan: Surface properties of poly(ethylene terephthalate) films modified by far-ultraviolet radiation at 193nm (laser) and 185nm (low intensity): J. Phys. Chem. 90: 2124-2131, 1986.

216. H. Watanabe, M. Yamamoto: Laser Ablation of Poly( ethylene terephthalate ): J. Applied Polymer Science 64 (6): 1203 – 1209, 1997.

217. M. Dadsetan, H. Mirzadeh, N. Sharifi: Effect of CO2 laser radiation on the surface properties of polyethylene terephthalate: Radiation Physics and Chemistry 56: 597-604, 1999.

218. W. Al-Soud, P. Radstrom: Purification and Characterization of PCR-Inhibitory Components in Blood Cells. Journal of Clinical Microbiology 39 (2): 485-493, 2001.

219. A. Pizzi: The Handbook of Adhesive Technology. 0824709861, CRC Press, 2003.

220. S. Voyutskii: Autoadhesion and Adhesion of High Polymers. New York, Interscience, 1963.

221. P. Gennes: The formation of polymer-polymer junctions. Amsterdam, Elsiever, pp355-367, 1982.

222. S. Jaffer, B. Grey: Polymer mechanically interlocking structures as interconnects for microfluidic systems: J. Micromech. Microeng. 18, 2008.

223. J. Wilding, L. Kricka, J. Cheng, G. Hvichia, M. Shoffner, P Fortina: Integrated Cell Isolation and Polymerase Chain Reaction Analysis Using Silicon Microfilter Chambers. Analytical Biochemistry 257(2): 95-100, 1998.

224. K. Wei, J. Culculo: Effects on the Compressibioity and Crystallization of Poly(ethylene Terephthalate): Journal of Polymer Science: Polymer Physics Edition 18: 343-359, 1980.

225. P. Belter: Bioseparations: Downstream Processing for Biotechnology. John Wiley & Sons, 1988.

226. M. Madou: Fundamentals of Micrfabrication. CRC Press, 2002.

227. L. Yang, T. Yao, Y. Tai: The marching velocity of the capillary miniscus in a microchannel: Journal of Micromechanics and Microengineering 14: 220-225, 2004.

228. G. Walker, D. Beebe: A passive pumping method for microfluidic devices: Lab on a Chip 2: 131 – 134, 2002

229. J. Howell, V. Sanchez, R. Field: Membranes in bioprocessing: Theory and applications. 0751401498 Chapman & Hall, 1993.

230. A. Weston, P. Brown: HPLC and CE: Principles and Practice. 0121366405 Academic Press, 1997.

191

231. R. Fournier: Basic Transport Phenomena in Biomedical Engineering. Taylor & Francis, 1998.

232. Qiagen: QIAamp 96 Spin Blood Kit Handbook, 1999.

233. X. Huang, M. Gordon, R. Zare: Current-monitoring method for measuring the electroosmotic flow rate in capillary zone electrophoresis: Analytical Chemistry 60: 1837-1838, 1998.

234. Y. Liu, D. Ganser, A. Schneider, R. Liu, P. Grodzinski, N. Kroutchinina: Microfabricated Polycarbonate CE Devices for DNA Analysis: Analytical Chemistry 73: 4196-201, 2001.

235. R. Johnson: The Handbook of Fluid Dynamics. 0849325099 Published by CRC Press, 1998.

236. J. Kim, B. Kim, H. Nam, D. Park, K. Yun: A disposable DNA sample preparation microfluidic for nucleic acid probe assay. The 15th IEEE Conference on Micro Electro Mechanical Systems: 133-136, 2002.

237. B. Vogelstein, D. Gillespie: Preparative and analytical purification of DNA from agrose. Proceedings of Natl. Acad. Sci. 76: 615-619, 1979.

238. Qiagen: The QIAGEN Guide to Template Purification and DNA Sequencing (2nd Edition), 1998.

239 M. Schuenmann and E. Harvey: Polymer-Based Microsystem Techniques: Handbook of

Biochips. Edited by R. Marks, D. Cullen, I. Karube, C. Lowe and H. Weetall. 2007 John Wiley & Sons, Ltd. ISBN 978-0-470-01905-4.

240 C. Zhang, J. Xu, W. Ma, W. Zheng: PCR microfluidic devices for DNA Amplification. Biotechnology Advances 24 (3): 243-284, 2006.

PET biochip fabrication for DNA sample preparation...i PET Biochip Fabrication For DNA Sample Preparation By Micah James Atkin B.Sc.(Hons) B.Elec.Eng.(Hons) A dissertation submitted

Documents