Femtosecond Laser Material Processing for Biomdeical

Home , Fused quartz, Gain (laser)

FEMTOSECOND LASER MATERIAL PROCESSING FOR MICRO-/NANO-SCALE

FABRICATION AND BIOMEDICAL APPLICATIONS

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

The Degree Doctor of Philosophy in Graduate

School of the Ohio State Univeristy

Hae Woon Choi, M.S.

*****

The Ohio State University

2007

Dissertation Committee:

Professor Dave F. Farson, Adviser Approved by

Professor Charles E. Albright

Professor L. James Lee ______

Welding Engineering Graduate Program ABSTRACT

Femtosecond laser ablation has interesting characteristics for micromachining,

notably non-thermal interaction with materials, high peak intensity, precision and

flexibility. In this dissertation, the potential of femtosecond laser ablation for fabrication

of biomedical and electronic devices is studied. In a preliminary background discussion,

some key literature regarding the basic physics and mechanisms that govern ultrafast

laser pulse interaction with conductive materials and dielectric materials are summarized.

In the dissertation work, results from systematic experiments were used characterize laser

ablation of ITO (Indium Tin Oxide), stainless steel (hot embossing applications),

polymers (PMMA, PDMS, PET, and PCL), glass, and fused quartz. Measured parameters

and results include ablation threshold, damage threshold, surface roughness, single- and

multiple-pulse ablation shapes and ablation efficiency. In addition to solid material,

femtosecond laser light interaction with electrospun nano-fiber fiber mesh was analyzed and studied by optical property measurements. Ablation of channels in nano-fiber mesh was studied experimentally. Internal channel fabrication in glass and PMMA polymers

was also demonstrated and studied experimentally. In summary, it is concluded that

femtosecond laser ablation is a useful process for micromachining of materials to produce

ii microfluidic channels commonly needed in biomedical devices such as micro-molecular

magnetic separators and DNA stretching micro arrays. The surface roughness of ablated

materials was found to be the primary issue for femtosecond laser fabrication of microfluid channels. Improved surface quality of channels by surface coating with

HEMA polymer was demonstrated.

iii

Dedicated to God, my wife, my families and teachers

iv ACKNOWLEDGMENTS

There are many people whom I would like to thank for their support of my research at

The Ohio State Univeristy. Edison Welding Institute (EWI) and the Nano Science and

Engineering Center (NSEC) program at OSU have been providing a comfortable research

environment as well as their financial support in my life of graduate study. I would like to thank to my advisor, Dr. Dave Farson, for his invaluable advice, endless patience and support. As committee members, Dr. Stanislav Rokhlin, Dr. Charles Albright, Dr. James

Lee provided me a lot of references and research considerations. I also thank to our lab colleagues, Yong Chae Lim, Jian Chen, and Min Hyun Cho for their support and encouragement.

I would like to thank God for the successful completion of this dissertation. I dedicate this work to my wife, Jin Sun, for all of her love, help and encouragement. This dissertation is also dedicated to my parents, and my lovely brothers and sisters, for providing me with the opportunities and support to achieve my goals. My mentor, Pastor

Jay Chun encouraged me a lot during Ph. D study. Chungnyun boo mokjang family were my greatest supporters in prayer.

This dissertation is a result of much collaboration with other NSEC fellows. In chapter 3, femtosecond laser for metallic materials, I had collaboration with Prof. Allen

v Yi and Prof. James Lee. Lei Li, Chunhe Zhang and Chunmeng were dedicated to getting the results and I appreciate their time and effort. In chapter 4, femtosecond laser for dielectric materials, Prof. Jeff Chalmers, Prof. James Lee, Prof. John Lannutti, and Prof.

Susan Olesik supported me to complete the research. Burr Zimmerman and Jeremy

Steach are the men whom I should specially thank. I also thank to Nick Farrell for his support on characterization of channels and other valuable advice. We enjoyed the research together and shared much information. Jed Johnson, Sarah Drilling, and Ruth Li supported my research by providing PCL fiber and the results of cell growth.

vi VITA

Oct. 16th, 1972 …………………………… Born – Taegu, South Korea

1992 – 1996 ……………………………. BSME, Keimyung University

Taegu, South Korea

2001- 2003 ……………………………. MSME, Univ. Central Florida

Orlando, South Korea

1989 – 1991 ……………………………. Engineer, Samsung Heavy Industries

1995 – 2000 ……………………………. Engineer, Daehyun Tech

2000 – 2003 ……………………………… Eng. Manager, Laser Tech USA

2003 – present …………………………… Ph. D study at Ohio State University

vii PUBLICATIONS

A. Journal Articles

1. H W Choi, D F Farson, M Cho, “A hybrid laser+GMAW process for control of the

bead humping defect”, Welding Journal 85 (8): 174S-179S Aug. 2006

2. D F Farson, H W Choi, S I Rokhlin, “Electrical discharges between platinum

nanoprobe tips and gold films at nanometer gap lengths” Nanotechnology,

17(1):132-139 Jan. 2006.

3. D F Farson, H W Choi, C Lu, and L. James Lee “Femtosecond bulk laser

micromachining of microfluid channels in PMMA” Journal of Laser Applications,

18 (3): 210-215 Aug. 2006

4. H W Choi, D F Farson, J Bovatsek, A Arai, D Ashkenasi, “Direct-write patterning

of ITO film by high pulse repetition rate femtosecond laser ablation” , Applied

Optics 46 (23), Aug. 2007

FIELD OF STUDY

Major: Welding engineering

Minor: Biomedical engineering

viii TABLE OF CONTENTS

ABSTRACT...... ii

ACKNOWLEDGEMENTS...... v

VITA...... vii

LIST OF FIGURES ...... xiii

LIST OF TABLES...... xvii

Chapter 1...... 1

1.1 Introduction...... 1

1.2 Organization of dissertation...... 2

Chapter 2 Femtosecond laser for micro/nano-scale machining...... 8

2.1 Micro-/nano-scale machining technologies ...... 8

2.2 The mechanism of sub-picosecond ablation...... 11

2.3 Femtosecond laser system for experiments ...... 12

2.3.1 Femtosecond laser system...... 12

2.3.2 Beam delivery system...... 16

2.4 References...... 18

ix Chpater 3 Femtosecond laser interaction with metallic materials ...... 21

3.1 ITO ablation...... 21

3.1.1 Introduction...... 21

3.1.2 Experiments ...... 27

3.1.3 Results and Discussion ...... 28

3.1.4 Conclusions and future works...... 49

3.1.5 References...... 51

3.2 Hot embossing ...... 53

3.2.1 Introduction...... 53

3.2.2 Experiment procedure...... 56

3.2.3 Laser ablation...... 61

3.2.4 Ablation and damage thresholds and pattern resolution...... 61

3.2.5 Mold micromachining...... 74

3.2.6 Hot embossing ...... 79

3.2.7 Conclusions and future works...... 85

3.2.8 References...... 87

Chapter 4 Femtosecond laser interaction with dielectric materials ...... 90

4.1 Poly-caprolactone (PCL) machining...... 90

x 4.1.1 Introduction...... 90

4.1.2 Experimental Apparatus and Procedure...... 94

4.1.3 Results...... 98

4.1.4 Conclusions and future works...... 113

4.1.5 References...... 114

4.2 Laser machining of dielectric materials for biomedical applications ...... 116

4.2.1 Introduction...... 116

4.2.2 Analysis of Laser Ablation ...... 120

4.2.3 Experimental Apparatus and Materials...... 123

4.2.4 Ablation Threshold Measurement...... 128

4.2.5 Microchannel Ablation - Procedures and Results...... 130

4.2.6 Surface roughness and HEMA coating process...... 144

4.2.7 Surface roughness characterization...... 147

4.2.8 Conclusions and future works...... 151

4.2.9 References...... 152

4.3 Femtosecond Laser Bulk Micromachining of Microfluid Channels in PMMA...... 156

4.3.1 Introduction...... 156

4.3.2 Concept of gas assisted material removal:...... 160

4.3.3 Experimental...... 162

xi 4.3.4 Results and discussion ...... 164

4.3.5 Conclusions and future work ...... 174

4.3.6 References...... 175

Chapter 5 Measurement of femtosecond laser scattering in ES-PCL nanofiber ...... 177

5.1 Introduction...... 177

5.2 Experiment setup...... 184

5.3 Results...... 190

5.4 Conclusions and future works...... 196

5.5 References...... 198

References...... 201

xii LIST OF FIGURES

Figure 2.1 Schematic of seeding laser source ...... 14

Figure 2.2 Regenerative amplifier ...... 15

Figure 2.3 External beam delivery system...... 17

Figure 3.1 Geometry of overlapping ablation spots ...... 26

Figure 3.2 Single pulse ITO ablation energy-diameter curves...... 29

Figure 3.3 ITO squared ablation diameter vs. fluence...... 30

Figure 3.4 Ablation depth of ITO using different focus optics ...... 31

Figure 3.5 Surface profile of laser ablation of ITO ...... 32

Figure 3.6 Single-pulse ablation threshold calculation of glass substrate ...... 34

Figure 3.7 2 kHz multipulse ablation...... 35

Figure 3.8 100 kHz laser ablation results ...... 36

Figure 3.9 Ablation line width vs. fluence...... 37

Figure 3.10 Variation of line ablation threshold with focus spot overlap ...... 39

Figure 3.11 Geometry of overlapping ablation spots ...... 43

Figure 3.12 Scan speed vs. pulse repetition frequency predicted from the results...... 45

Figure 3.13 Predicted and experimental ablation width versus scanning speed...... 47

Figure 3.14 Steps for fabrication of a polymer microfluid device...... 57

Figure 3.15 Surface profile of polished mold blank ...... 58

Figure 3.16 Block diagram of beam delivery system setup...... 60

xiii Figure 3.17 Single pulse ablation patterns for various pulse energies...... 65

Figure 3.18 Ablation diameter results with regression-fit lines...... 66

Figure 3.19 3D profile of laser ablation (Left E=500nJ, Right E=65nJ)...... 70

Figure 3.20 Ablation volume as a function of laser pulse energy ...... 68

Figure 3.21 Ablation efficiency...... 71

Figure 3.22 Overlap of adjacent focus spot locations...... 73

Figure 3.23 Multipulse overlapped scanning (E=165nJ)...... 77

Figure 3.24 (a) Hot embossing mold on stainless steel, (b) surface profile of mold...... 80

Figure 3.25 Image of hot embossed molecular magnetic separator pattern in PMMA.... 82

Figure 3.26 SEM images of an embossing tool machined ...... 83

Figure 3.27 Theoretical diameter of ablation and damage regions...... 84

Figure 3.28 Ablation profiles for pulses with various peak fluence...... 85

Figure 4.1 Fiber diameter size distribution of ES-PCL used in the experiments ...... 97

Figure 4.2 SEM images of (a) Q-switched laser and (b) femtosecond laser grooves...... 99

Figure 4.3 Width of grooves machined in ES nanofiber meshes...... 102

Figure 4.4 Depth of grooves machined in ES nanofiber meshes...... 103

Figure 4.5 Comparison of grooves ablated in (a) PCL and (b) PET ...... 104

Figure 4.6 Two grooves ablated in PCL nanofiber...... 105

Figure 4.7 (a) Focused and (b) defocused (multiple passes, right) scans s...... 106

Figure 4.8 Groove width squared of nanospun polymer vs. fluence ...... 108

xiv Figure 4.9 (a) transmittance of ES PCL (b) reflectance of nanofiber mesh ...... 110

Figure 4.10 Matrix of microscale structures ablated in PCL nanofiber mesh ...... 112

Figure 4.11 Keldysh parameters for soda lime glass and fused quartz...... 118

Figure 4.12 Definition of focus spot overlap LO...... 122

Figure 4.13 Block diagram of beam delivery system setup...... 124

Figure 4.14 Single pulse ablation diameter (a) pulse energy, (b) fluence...... 129

Figure 4.15 Ablation pattern with respect to laser fluence and scanning speed...... 132

Figure 4.16 Ablation depth as function of pulse energy...... 133

Figure 4.17 Profilometer scan across channels ablated at pulse energy and pitch ...... 134

Figure 4.18 Peak-peak surface roughness versus pulse energy at a pitch of 2.5 µm...... 136

Figure 4.19 Accumulated laser fluence (a) 0%, (b) 25%, (c) 50%, (d) 75%...... 138

Figure 4.20 Accumulated fluence as a fraction of single pulse...... 140

Figure 4.21 Fluence variation vs. overlap...... 142

Figure 4.22 Optical images of the channel quality and shape ...... 143

Figure 4.23 HEMA coating process ...... 146

Figure 4.24 Optical images of HEMA coated channels (before and after) ...... 146

Figure 4.25 Surface roughness data for soda-lime glass ...... 148

Figure 4.26 Surface roughness data for fused quartz...... 148

Figure 4.27 Surface roughness data for HEMA coating...... 149

Figure 4.28 Surface roughness data of soda-lime glass and fused quartz ...... 150

xv Figure 4.29 Concept of gas assisted material removal ...... 160

Figure 4.30 Experimental setup...... 163

Figure 4.31 Channel fabrication without gas assistance...... 165

Figure 4.32 Channel fabrication with gas assistance at different flow rates ...... 167

Figure 4.33 Plan view of channels created with multiple passes...... 169

Figure 4.34 Cross sectional view of channels...... 170

Figure 4.35 Reynolds numbers in the flow channel for 0.65 ml/s, 1.3 ml/s, 2.6 ml/s.... 171

Figure 4.36 Micrograph of internal channels before being filled with a liquid...... 172

Figure 4.37 Flourescense spectra confirming presence of yellow dye in a channel...... 173

Figure 5.1 Light propagation inside and outside of a fiber mesh ...... 180

Figure 5.2 Definition of TE and TM modes ...... 181

Figure5.3 Electro spun fibers (left) and fiber diameter distribution (right)...... 185

Figure 5.4 System setup for transmittance and reflectance measurements ...... 188

Figure 5.5 System setup for transmittance and reflectance measurements ...... 189

Figure 5.6 Output power vs. input power of samples...... 190

Figure 5.7 Transmittance measurement for fiber PCL and solid PCL ...... 191

Figure 5.8 Reflectance measurement...... 192

Figure 5.9 Ablation diameters vs. Fluence ...... 194

Figure 5.10 Increased ablation diameter by multiple scattering...... 195

Figure 5.11 Ablated fiber with femtosecond laser (E=500nJ)...... 196

xvi LIST OF TABLES

Table 2.1 Laser systems in Femtosecond laser...... 13

Table 3.1 Single pulse ablation thresholds and incubation factors...... 41

Table 4.1 Physical properties of PCL and PET ...... 91

Table 4.2 Material properties of soda-lime glass...... 127

Table 4.3 Material properties of fused-quartz...... 128

Table 4.4 Experimental parameter settings...... 131

Table 5.1 Material properties of PCL ...... 186

xvii

CHAPTER 1

INTRODUCTION

1.1 Introduction

The femtosecond laser has opened a new era of micromachining. Its precision, high peak power, flexibility, non-thermal interaction, and non-clean room operation provide many advantages in micromachining for biomedical applications. Femtosecond lasers with pulse durations of ~100fs have been widely used for micromachining because of collateral damage is mostly eliminated in dielectric materials [1] and heat affected zones in metal can be smaller [2]. The main characteristic of femtosecond (or ultrashort) laser pulses is very high peak intensity (> 1016 W/cm2) and rapid deposition of energy into the material. At this high intensity, the onset of optical breakdown that initiates ablation can be more deterministic than stochastic [3, 4] and a wide range of materials can be machined.

Benefits of ultrashort laser pulses arise because the pulse duration τ ~ 150fs is less than the time τei 1–10 ps equilibration between electron and lattice-ions subsystems.

1 In addition, the time τh needed for the electron thermal energy diffusion to reach the

optical penetration depth is several orders of magnitude longer than τ [5]. The short

duration can eliminate the affects of hydrodynamic motion of the matter during laser

irradiation [3]. The short pulse also can prevent the shielding of incoming laser beam by

plasma from the ablated surface, which can maximize absorption efficiency [3].

Due to these advantages, femtosecond laser provides a very flexible way to fabricate devices of nearly all types of materials. Many hard metals or carbides which

would make very rugged and durable embossing molds are not compatible with

micromachining by etching processes. Also, the same superior mechanical properties also

create difficulties for mechanical machining. For example, Ti-Ni alloys undergo severe

strain hardening during mechanical machining, degrading the quality of machined

surfaces and shortening the life of machine tools. In addition, stress-induced martensitic

transformation at the cutting front leads to an undesirable contact interface between the

tool and the workpiece [6]. However, femtosecond laser was proven to be a very effective

processing tool for such hard materials and a wide variety of others as well. The

applications of femtosecond laser ablation that we developed during this research are

discussed in following chapters.

2 1.2 Organization of dissertation

This dissertation is organized into 6 chapters including this chapter. Chapter 2 is prepared to provide an introduction to the fundamentals of femtosecond lasers and the processing systems that were setup for the experiments.

Chapter 3 provide experimental results for the laser micromachining of conducting materials and applications such as indium doped with tin oxide (ITO) film for digital displays and stainless steel for hot embossing of polymer. The first part of

the chapter describes ITO ablation. Laser pulse energy and focus spot size were varied

in single-spot ablation tests and for ablation of linear features with scanned multiple

pulses. The single-pulse ablation threshold of ITO was smaller than that of the glass

substrate so the entire thickness of ITO could be removed in a single pulse or with

multiple intersecting pulses without the possibility of substrate ablation. Linear

features could be created at much higher scanning speeds using a high repetition

frequency (100 kHz) Yb fiber amplified laser as compared to a lower repetition

frequency (2 kHz) laser. An analysis showed that incubation affects lowered ITO

ablation thresholds when pulse frequency was high relative to scanning speed,

contributing to large feasible scanning speeds for high pulse frequency lasers.

The second part of Chapter 3 describes a hot embossing mold fabrication with

femtosecond laser on stainless steel. Femtosecond laser micromachining was used to

fabricate hot embossing molds of a durable material, stainless steel grade 304L. A

systematic study of ablation threshold, ablation efficiency and affect of machining

parameters on mold accuracy was performed. A broad range of laser energy, 13nJ to

3 500nJ, was applied, and the most effective region of ablation was found to be between

65nJ and 135nJ. The ablation threshold for stainless steel was found to be 0.189J/cm2 and

ablation efficiency was estimated to be around 1.2%. For scanned ablation, 75% pulse

overlap and 135nJ laser power was found to be the optimum process with minimum

recasting. A finished mold was used was then used to fabricate a microfluidic micro-

molecular magnetic separator in poly-methyl macryolate (PMMA) polymer.

Chapter 4 describes the laser beam interaction with dielectric materials such

as glass, fused quartz, and electrospun poly-caprolactone (ES PCL) and poly-ethylene

terephthalate (ES PET) fiber mesh. In addition, a novel method to fabricate internal

microfluidic channels by femtosecond laser is presented. The first section describes

PCL and PET ablation. Nano fiber meshes of PCL and PET and PET were structured by ablation of linear grooves with a scanned femtosecond laser. Focus spot size, pulse

energy and scanning speed were varied to determine their affects on groove size and

the characteristics of the electrospun fiber at the edges of these grooves. The

femtosecond laser was seen to be an effective means for flexibly structuring the surface of ES PCL scaffolds. Femtosecond ablation resulted in much more uniformly ablated patterns compared to Q-switched nanosecond pulse laser ablation. Also, the width of the ablated grooves was well-controlled by laser energy and focus spot size although the grooves were significantly larger than the spot size. Also, some melting of fibers was observed at the edges of grooves. These affects were attributed optical radiation and multiple scattering from laser-induced plasma at higher pulse energies and melting of fibers at laser fluences lower than the ablation threshold.

4 In the second part, we describes techniques for microfluidic channel fabrication in soda-lime glass and fused quartz using femtosecond laser ablation and ablation in conjunction with polymer coating for surface roughness improvement. Systematic experiments were done to characterize how process variables (laser fluence, scanning speed, and focus spot overlap, and material properties) affect machining quality for dielectric materials. Laser fluence and focus spot overlap showed the strongest influence on channel depth and roughness. At high fluence, the surface roughness was measured to be between 395nm-731nm RMS. At low fluence, roughness decreased to 100nm-350nm

RMS and showed a greater dependency on overlap. The surface roughness of laser ablation was also dependent on the material properties. For the same laser ablation parameters, soda-lime glass surfaces were smoother than fused quartz. For some applications, especially those using quartz, smoother channels are desired. A hydroxyethyl methacrylate (HEMA) polymer coating was applied and roughness of coated channels was improved to 10-50nm RMS.

The third part of this chapter describes a novel method to fabricate a microfluidic channel inside the transparent material. Internal channels in polymer are widely used in biotechnology applications such as DNA stretching and in devices such as Micro Total

Analysis Systems (μ-TAS) and Lab on Chip (LOC) systems. For manufacturing prototype devices, femtosecond pulsed laser energy has been used to implement a convenient direct write bulk-machining process in glass. In this technique, the laser beam is focused inside of a transparent material, resulting in the ablation of an internal channel.

Initial experiments for internal channel fabrication in poly-methyl methacrylate (PMMA)

5 polymer revealed a significant problem with clogging of channels by debris and rough,

fractured channel walls. In this paper, we describe a new method to fabricate internal

channel in PMMA using femtosecond pulsed laser energy and a gas-assisted material removal concept. Relatively smooth channels with a minimum diameter of 2 μm, a maximum diameter of 20 μm and a maximum length of 10 mm were achieved with this

technique.

Chapter 5 is describing the measurement of optical properties for electrospun

PCL nano fiber mesh. The light absorption in fiber mesh is different from that in solid

material and scattering plays a significant role. Multiscattering theory, modified from

Kubelka-Munk theory, is used to quantify the scattering and absorption coefficients.

Integrating sphere was used to measure the transmittance and reflectance of light in ES

fibers. The ablation threshold is measured to be 0.094 J/cm2 and it is observed that 98%

of laser beam which is absorbed inside the mesh is scattered.

1.3 References

1. B. Stuart, M. Feit, A. Rubenchik, B. Shore, and M. Perry, Laser-induced damage in dielectrics with nanosecond to subpicosecond pulses. Physical Review Letters, 1995. 74(12): p. 2248-2252.

2. S. Nolte, C. Momma, H. Jacobs, A. Tunnermann, B. Chichkov, B. Wellegehausen, and H. Welling, Ablation of metals by ultrashort laser pulses. J. Opt. Soc. Am. B, 1997. 14(10): p. 2716-2722.

6 3. X. Liu, D. Du, and G. Mourou, Laser ablation and micromachining with ultrashort laser pulses. IEEE journal of quantum electronics, 1997. 33(10): p. 1706-1716.

4. D. Du, X. Liu, G. Korn, J. Squier, and G. Mourou, Laser induced breakdown by

impact ionization in SiO2 with pulse widths from 7ns to 150fs. Appl. Phys. Lett., 1994. 64(23): p. 3071-3073.

5. T. Itina, J. Hermann, Ph. Delaporte, and M. Sentis, Modeling of metal ablation induced by ultrashort laser pulses. Proceedings of symposium H on photonic processing of surfaces - Thin films and devices, 2004. E-MRS: p. 453 - 454.

6. H. Huang, H.Z., and Y. Liu, Experimental investigations machiniability of Ni50.6Ti49.4. Smart Materials and Structurs, 2005. 14: p. S297-S301.

CHAPTER 2

FEMTOSECOND LASER FOR MICRO/NANO-SCALE MACHINING

In this chapter, a wide range of micro-/nano-scale fabrication techniques that are available both in clean room and non-clean room environments are overviewed. The femtosecond laser process is compared to other processes and its advantages are discussed based on the review of previous publications. The basic configuration of femtosecond laser system and experiment setups used for this dissertation and are presented.

2.1 Micro-/nano-scale machining technologies

Micro- and nano-scale machining is becoming increasingly important for biomedical and semiconductor applications. One of the most popular fabrication technologies is a process technique typically used in semiconductor fabrication: photolithography. Photolithography transfers a pattern from a mask to a photosensitive material using a radiation source such as mercury lamp or ultra violet (UV) laser. By

8 using a combination of both photolithography and polymer processing, many advanced

micromachining techniques have been developed and used for a variety of applications

[1-3].

A different type of lithography used to fabricate micro- and nano- scale polymeric

devices from a master mold by photolithography is called soft lithography. Soft lithography is an alternative method to silicon-based micromachining and a non- photolithographic strategy based on self assembly that uses replica molding of nontraditional elastomeric materials and bimolecules to fabricate stamps. It provides a convenient, effective, and low-cost method for patterning [4-8]. However, it still requires multiple steps from the development of appropriate masters to a working device and residual chemicals remaining after the process that can be toxic if implanted in a

biomedical device [1]. Several different techniques are collectively described as soft

lithography: Near-Field Phase Shift Lithography, Replica Molding, Micromolding in

Capillaries, Microtransfer Molding, Solvent-assisted Microcontact Molding (SAMIM),

and Microcontact Printing. Other techniques used include micromachining of silicon for

microelectricalmechanical systems (MEMS) and embossing of thermoplastic with

patterned quartz [9]

Instead of a lithographic approach, the atomic force microscope (AFM) has also

been used for nano-scale machining. The reported advantage for AFM is that the

machining scale can be as small as the size of the AFM probe tip. The integration of

conventional and AFM-based lithographic techniques brings the resolution of fabrication

technology to a nanometer scale [10, 11]. The AFM with silicon or diamond tip etching

9 or micromilling machine can provide a way to fabricate sub-micron features by

scratching the surface of a soft material with its sharp tip. The range of materials that

have been studied for soft materials, such as polymers to relatively hard layers such as

oxide thin films [1, 10-14]. This approach provides a flexible way to fabricate

complicated geometry in non-clean room environment. However, several drawback to

AFM etching is that it is a slow process to create sub-micron features by scratching the

surface of the material with a nano-sized pyramidal tip, the AFM tip wearing, and the

problem of fabrication area needs to be solved [1]. In addition to the above process, Ion milling and reactive ion etching are another choice of micro-/nano-scale fabrication [15,

16]

As a non-contact and non-clean method, a variety of laser systems such as

Nd:YAG, CO2, Excimer, and copper vapor lasers have been used [17-23]. Due to the

high absorption properties of the materials, especially in dielectric materials, UV (Ultra

Violet) laser such as Excimer lasers have been used as a micromachining tool on wide

bandgap dielectric materials, such as CaF2, BaF2, sapphire, and quartz for optical components such as gratings and lens. With the photochemical interaction of the laser with target material, it can eventually achieve very fine feature sizes[19, 20, 22, 24].

However, corrosive gas handling and UV radiation damages the optics made for the UV pulsed laser is still an issue and less desirable tool for industrial applications[25].

10 2.2 The mechanism of sub-picosecond ablation

Femtosecond lasers have been used for material processing for less collateral

damage in dielectric materials [26] and reduced heat affected zones in metal [27]. The

major benefits of an ultrashort laser pulse include its ability to produce very high peak

intensity and rapid deposition of energy into the material. It is well known fact that the

principle characteristic of femtosecond pulses, which makes process advantageous for

metal treatment, is that the duration of pulse is shorter than the time τei 1–10 ps for

equilibration between electron and lattice-ions subsystems. In addition, the time τh needed for the electron heat diffusion to reach the optical penetration depth is several orders of magnitude longer than τ, the duration of femtosecond pulse [28].

By having a short duration, this can eliminate the hydrodynamic motion of the matter during laser irradiation and no fluid dynamics is involved during the laser and matter interaction [25]. The theoretical study of the transient evolution of the distribution function of the electron gas in a metal during and after irradiation with a subpicosecond laser pulse of moderate intensity has been studied by many prior researchers [29-32].

Moreover, the femtosecond laser can also prevent the plasma shielding of incoming laser beam by ablated material which can lead to maximum absorption efficiency [25].

11 2.3 Femtosecond laser system for experiments

2.3.1 Femtosecond laser system

In our experiments, commercial femtosecond laser systems, CPA 2110 (Clark-

MXR), CPA 2161 (Clark-MXR), and FCPA μJewel D-400 (IMRA America) were used as femtosecond laser energy sources. The first one has 775nm central wavelength, pulse duration of 150 fs, pulse repetition frequency of 2kHz and an average power of 1.6 W.

The system has a single-mode erbium (Er) fiber oscillator to provide a seed pulses for chirped pulse amplification in a Ti:Al2O3 –based regenerative amplifier. The second

system is similar to the first but has higher pulse repetition rate, adjustable from 3-6kHz.

The third system is based on an Er fiber oscillator with pulses amplified in a ytterbium

(Yb) fiber-based amplifier followed by chirped pulse amplification in a Yb fiber-based

amplifier. This yields output at a wavelength of 1045 nm, pulse duration less than 500fs

and a higher repetition frequency of 100kHz and average power of 0.2W The system

setups are slightly different depending on experiments, and details will be explained at

each section.

The CPA system consists of four lasers including a diode laser for Er fiber laser pumping source, SErF fiber laser which is a frequency doubled fiber seeding laser, ORC-

1000 laser for Ti:Al2O3 laser pumping, and Ti:Al2O3 laser for regenerative amplifier. The

frequency and power are listed in Table 2.1

12 Laser Wavelength (nm) Power Pulse Energy Nd:YAG 532 < 65W < 25mJ

Ti:Al2O3 775 < 5W < 5mJ ErF Fiber 1550 50mW < 10nJ SErF Fiber 775 < 20mW < 1nJ

Table 2.1 Laser systems in Femtosecond laser [33]

SErF is based on the fiber ring “Stretched Pulse” laser. It is a unidirectional,

polarization rotation additively pulse mode-locked (APM) fiber laser which uses Erbium

doped fiber as the gain medium. The pump source for the laser is a solid-state fiber-

coupled laser diode operating at approximately 980 nm. The SErF contains a laser diode

and associated control electronics, an active fiber ring laser, some bulk optics used for polarization control, output coupling and wavelength control, a compressor to eliminate residual dispersion of the output pulses, and a temperature-stabilized periodically poled lithium niobate (PPLN) frequency doubler. The schematic of the seed pulse laser, SErF is depicted in Figure 2.1.

13 Output

Collimator Collimator

WP BRF Isolator WP

Erbium doped fiber

Diode Laser WDM

Figure 2.1 Schematic of seeding laser source [33] (WP: λ/4 waveplates, BRF: birefringent filter, WDM: wavelength-division multiplexing)

The seed pulse from the frequency doubled SErF laser (λ=775nm) is stretched to protect optics during pulse amplification. The stretched seed pulse is then transferred to regenerative amplifier to amplify the laser intensity. The optimum injection or cavity dumping time is controlled by electronic devices (DT-505, Clark-MXR) and is very critical for the power and pulse duration. The injected seed laser is amplified through

Ti:Al2O3 cavity which is pumped by a frequency doubled Nd:YAG laser (λ=532nm)

through dichroic through mirror shown in Figure 2.2. After multiple passes of

amplification, typically 5 times, the polarization of a Pockels cell is changed to eject the

amplified beam to the compressor.

Figure 2.2 Regenerative amplifier (HR: high reflective mirror, FR: Faraday rotator, PC: Pockels cell, DM: Dichroic mirror, RF: Radiofrequency) [33]

The amplified laser beam then travels through compression optics where the width of pulse is compressed to back to the sub-picosecond region, typically less than

150fs.

Due to limitations of components used for optical pulse amplification, CPA femtosecond laser pulse repetition rates remained around several kHz for some time, which in turn limited the maximum micromachining speed to inches/sec. This has been the main drawback of the process. However, recent development of technologies enabled design of systems with pulse repetition rates in the MHz regime, which is comparable

15 acoustooptically Q-switched pulsed Nd:YAG laser systems. The advanced technology decreased optical alignment requirements, so laser maintenance is minimized.

2.3.2Beam delivery system

The gated pulse train produced by the laser systems has constant average power of 2.6W at 3kHz / 6kHz setting and 1.6W at 1kHz/2kHz setting. Since the typical range of power for micromachining is less than 5mW, a great amount of attenuation is required.

As shown in Fig. 2.3, a series of polarizing beam splitters (PBSs) were installed to attenuate the power for micromachining. The direction of polarization is horizontal which is parallel to the optical table and set by the manufacturer. During the process, the horizontally polarized laser beam is attenuated thorough the combination of thin-film polarizing beam splitters (PBSs) and a λ/2 wave plates.

Figure 2.3 External beam delivery system

The PBS regulates the transmission and reflection rate based on the polarization direction which can be adjusted by a λ/2 wave plate and incidence angle of laser beam.

The diameter of raw beam was measured to be 5mm and the beam quality, M2, was measured by using autocorrelator to be 1.3 where a perfect Gaussian beam is M2=1. The laser beam can be turned on or off using an external high speed mechanical shutter with

4ms shutter control time, (LS055, NMlaser Inc) and is finally delivered to the target material on a high precision XY stage (Parker-Hannifin) with 0.5μm resolution and

50mm travel distance. The focusing optics is mounted on Z axis with 0.5μm resolution and 25mm travel distance. A 50x infinity corrected microscope objective lens with

17 numerical aperture of NA=0.42 (50x M Plan Apo NIR, Mitutotyo) was used for fine

focusing.

For the consistency and easy focusing purpose, a coaxial vision system was installed and one can visually locate the material at the focus within +/- 1μm range where focal depth is 1.6μm for this selected optics.

2.4 References

1. C. Aguilar, Y. Lu, S. Mao, and S. Chen, Direct micro-patterning of biodegradable polymers using ultraviolet and femtosecond lasers. Biomaterials, 2005. 26: p. 7642-7649.

2. J. Narashimhan, a.I.P., Polymer embossing tools for rapid prototyping of plastic microfluidic devices. J. Micromech. Microeng, 2004. 14: p. 96-103.

3. F. Tay, J.K., F. Watt, and W. Choong, A novel micro-machining method for the fabrication of thick-film SU-8 embedded micro-channels. J. Micromech. Microeng, 2001. 11: p. 27-32.

4. M. Unger, H.C., T. Thorsen, A. Scherer, and S. Quake, Monolithic microfabricated valves and pumps by multilayer soft lithography. Science, 2000. 288: p. 113-116.

5. Y. Xia, and G. Whitesides, Soft lithography. Annu. Rev. Mater. Sci., 1998. 28: p. 153-184.

6. J. McDonald, D.D., J. Anderson, D. Chiu, H. Wu, O. Schueller, and G. Whitesides, Fabrication of microfluidic systems in PDMS. Electrophoresis, 2000. 21: p. 27-40.

7. R. Kane, S. Takayama, E. Ostuni, D. Ingber, and G. Whitesides, Patterning proteins and cells using soft lithography. Biomaterials, 1999. 20: p. 2363-2376.

8. G. Whitesides, E.O., S. Takayama, X. Jiang, and D. Ingber, Soft lithography in biology and biochemistry. Annu. Rev. of Biomedical Engineering, 2001. 3: p. 335-373.

18 9. G. Whitesides, Soft Lithography. 1998, WTEC hyper-librarian.

10. T. Fang, C. Weng, and J. Chang, Machining characterization of the nano- lithography process using AFM. Nanotechnology, 2000. 11: p. 181-187.

11. Y. Kim, and C. Lieber, Machining Oxide Thine Films with an AFM: Pattern and Object Formation on the Nanometer Scale. Science, 1992. 257(5068): p. 375-377.

12. J. Park, D.L., N. Takano, and N. Morita, Diamond tip cantilever for micr/nano machining based on AFM. Materials Science Forum, 2006. 505: p. 79-84.

13. B. Bhushan, and V. Koinkar, Nanoindentation hardness measurements using AFM. Appl. Phys. Lett., 1994. 64(13): p. 1653-1655.

14. Z. Hu, S. Zhang, and X. Zheng, Three dimensional micromachining based on AFM. Key engineering materials, 2006. 315: p. 800-804.

15. E. Chanson, S.P., J. Poate, J. Borland, M. Current, T. Rubia, D. Eaglesham, O. Holland, M. Law, C. Magee, J. Melngailis, and A. Tasch, Ion beams in silicon processing and characterization. J. Appl. Phys., 1997. 81(10): p. 6513-6561.

16. I. Adesida, A.M., E. Andideh, M. Khan, D. Oslen, and J. Kuznia, Reactive ion etching of GaN in silicon tetrachloride plasmas. Appl. Phys. Lett., 1993. 63(20): p. 2777-2779.

17. H. Klank, J.K., and O. Geschke, CO2 laser micomachining and back end processing for rapid production of PMMA based microfluidic systems. Lab on a chip, 2002. 2(4): p. 242-246.

18. B. Seddon, Y. Shao, J. Fost, and H. Girault, The application of excimer-laser micromachining for the fabrication of disc microelectrodes. Electrochimica Acta, 1994. 39(6): p. 783-791.

19. J. Ihlemann, a.B.W.-R., Excimer laser micro machining of inorganic dielectrics. Applied surface science, 1996. 106: p. 282-286.

20. G. Kopitkovas, T.L., C. David, A. Wokaun, and J. Gobrecht, Surface micromachining of UV transparent materials. Thin solid films, 2004. 453: p. 31- 35.

21. A. Glover, E. Illy, and J. Piper, High speed UV micro machining of polymers with frequency doubled copper vaport lasers. IEEE journal of quantum electronic, 1995. 1(3): p. 830-836.

19 22. K. Naessens, H.O., P. Daele, and R. Baets, Flexible fabrication of microlenses in polymer layers with excimer laser ablation. Applied surface science, 2003. 208- 209: p. 159-164.

23. E. Illy, D.B., M. Withford, and J. Piper, Enhanced polymer ablation rates using high-repetition-rate ultraviolet lasers. IEEE journal of quantum electronic, 1999. 5(6): p. 1543-1548.

24. H. Endert, R.P., and D. Basting, Excimer laser: A new tool for precision micromachining. Optical and Quantum electronics, 1995. 27(12): p. 1319-1335.

25. X. Liu, D. Du, and G. Mourou, Laser ablation and micromachining with ultrashort laser pulses. IEEE journal of quantum electronics, 1997. 33(10): p. 1706-1716.

26. B. Stuart, M. Feit, A. Rubenchik, B. Shore, and M. Perry, Laser-induced damage in dielectrics with nanosecond to subpicosecond pulses. Physical Review Letters, 1995. 74(12): p. 2248-2252.

27. S. Nolte, C. Momma, H. Jacobs, A. Tunnermann, B. Chichkov, B. Wellegehausen, and H. Welling, Ablation of metals by ultrashort laser pulses. J. Opt. Soc. Am. B, 1997. 14(10): p. 2716-2722.

28. T. Itina, J. Hermann, Ph. Delaporte, and M. Sentis, Modeling of metal ablation induced by ultrashort laser pulses. Proceedings of symposium H on photonic processing of surfaces - Thin films and devices, 2004. E-MRS: p. 453 - 454.

29. B. Rethfeld, A. Kaiser, M. Vicanek, and G. Simon, Ultrafast dynamics of nonequilibrium electrons in metals under femtosecond laser irradiation. Physical Review B, 2002. 65: p. 214303-1 - 214303-11.

30. F. Vidal, T.J., S. Laville, O. Barthelemy, M. Chaker, B. LeDegoff, J. Margot, and M. Sabsabi, Critical-point phase separation in laser ablation of conductors. Physical Review Letters, 2001. 86(12): p. 2573 -2576.

31. N. Bulgakova, Possibility of rarefaction shock wave under short pulse laser ablation of solids. Physical Review E, 1999. 60(4): p. R3498 - R3500.

32. E. Gamaly, A.R., B. Luther-Davies, and V. Tikhonchuk, Ablation of solids by femtosecond lasers: Ablation mechnasim and ablation thresholds for metals and dielectrics. Physics of plasma, 2002. 9(3): p. 949-957.

33. Clark-MXR, CPA 2110 User manual. 2 ed. 2004: Clark-MXR.

CHAPTER 3

FEMTOSECOND LASER INTERACTION WITH METALLIC MATERIALS

In this chapter, femtosecond laser micromachining of conductive materials is

presented. The first part describes thin film ablation with femtosecond laser. Indium

oxide dope with tin oxide (ITO) is conductive transparent metallic layer which is used in display applications such as liquid crystal display (LCD), organic light-emitting diode

(OLED), and other flat panel display application. The second part of this chapter describes a fabrication of a hot embossing mold for microfluidic channel applications.

3.1 ITO ablation

3.1.1 Introduction

Indium oxide doped with tin oxide (ITO) provides high electrical conductivity

and transparency in the visible and near IR (infrared) wavelengths. Thus, it is widely used

as a transparent electrode for the fabrication of liquid crystal displays (LCD’s) and organic light emitting diode displays (OLED’s), photovoltaic devices and other optical

21 applications [1]. Its optical properties, determined by composition, deposition process and parameters, are important for applications and also affect laser processing [2].

During device fabrication, ITO is applied as a continuous film and is then locally removed to form conductors, device interconnections and other elements. Direct-write laser ablation removal of ITO from glass or other underlying materials is sometimes used, in lieu of conventional photolithography, for device manufacturing and is also needed for repair applications. Nd:YAG and Nd:YVO4 (fundamental and frequency-converted) [3,4]

and excimer [5] lasers have been used for ablation. In general, short pulse laser ablation offers intriguing possibilities for many microelectronics fabrication tasks [6,7,8]. Unique and generally desirable ablation characteristics are observed when irradiance is large and pulse length is shorter than the picosecond-scale times required for transfer of absorbed

energy from electrons (either initially free or released from bound states by multiphoton

absorption or tunneling) to the material lattice [9]. Consequently, thermally-induced

defects, which are often difficult to avoid with longer laser pulses, can be minimized in

the remaining material.

The femtosecond laser ablation has been analyzed [10] and results for

femtosecond- and nanosecond-pulsed laser ablation of ITO and multilayer transparent

films have been reported [11]. Results pertinent to this investigation are summarized

here. When illuminated with laser pulses of varying fluence, many materials are ablated

only when fluence exceeds a distinct threshold level. To analyze the size of ablation

spots, one may consider single pulse removal of a material with ablation threshold

fluence Fth by a beam with Gaussian radial fluence profile having a maximum greater

22 than the ablation threshold fluence. Then, the radius at which the laser beam fluence equals the ablation threshold fluence can be found by solving the equation

2 2 th 0 (−= 2exp wrFF 0 ) (3.1.1)

where w0 is Gaussian beam radius and F0 is maximum (or peak) fluence, related to pulse energy Ep by

2 0 = p /2 πwEF 0 . (3.1.2)

Similarly

2 = thth /2 πwEF 0 , (3.1.3)

where Eth is the pulse energy corresponding to ablation threshold fluence. Re-arranging

Eq. (3.1.1) to solve for the diameter D of the area near the center of the beam from which material is removed results in the relation

2 2 2 = 0 ( 0 th ) = 0 [ ( 0 ) − lnln2/ln2 (FFwFFwD th )] (3.1.4)

Experimentally, single pulse ablation spots with measured diameter D are produced by varying pulse energy over a range of values while focus spot radius is

23 2 maintained at a known, fixed value. A semi-logarithmic plot of D versus ln()F0 is created and the ablation threshold fluence is found by extrapolating the linear curve resulting from this plot to zero diameter where th = FF 0 . Since calculated focus spot sizes are subject to inaccuracy, an effective focus spot radius for experiment can be calculated directly from the ablation data [12]. Using Eq. (3.1.2) and Eq. (3.1.3) in Eq. (3.1.4) yields

2 2 = 0 ( 0 /ln2 EEwD th ). (3.1.5)

Thus, the effective radius of the laser focus spot needed for fluence calculations can be calculated from the slope of a plot of experimental ablation diameter-squared and pulse energy data.

At fluences lower than the ablation threshold, it is common to speak of damage threshold fluence Fd above which visible damage (but not removal) of the surface results.

If d 0 << FFF th , experimental damage diameters can be measured and Fd can be found by a similar procedure as for Fth .

If N > 1 laser pulses of the same energy are incident on the same surface location, it has been found that ablation occurs for pulse energy less than that corresponding to the ablation threshold fluence for N = 1 and that larger N results in more decrease. For at least some materials, the ablation threshold fluence for large N is only marginally larger the damage threshold fluence, which does not vary with the number of pulses [11]. The decrease of the ablation threshold for multiple pulses, termed incubation [13], is

24 attributed to increased absorption due to accumulation of damage or effects (i.e. color center formation) from individual pulses. The affect of incubation on ablation threshold can be quantified by a relationship of the form [10]

ξ −1 th = th )1()( NFNF (3.1.6)

where N is number of pulses incident on a given location and ξ is called the incubation factor.

When ablation is carried out with a pulsed laser having pulse repetition frequency f, focused to a Gaussian spot radius w0 , scanned at a linear velocity s, it is customary to refer to the ratio of the length along the scan centerline of the intersection of focus spot areas of successive pulses to the Gaussian focus spot diameter as the focus spot overlap, denoted Od. As shown in the geometry illustrated in Fig. 3.1, Od can be calculated as

⎛ s ⎞ Od ⎜1−= ⎟ (3.1.7) ⎝ df ⎠

Figure 3.1 Geometry of overlapping ablation spots showing calculation of theoretical ablation line, = 2wd 0 .

For scanned ablation, one may also speak of the overlap of ablated spots, calculated by the same formula but substituting D, the diameter of a spot ablated by a single pulse for the laser focus spot diameter.

The recent development of high repetition frequency amplified femtosecond lasers [14,15] presents interesting possibilities for direct-write thin film patterning by ablation of linear features with a single scanned focused laser spot. The high repetition frequency implies that, even at relatively high scanning speeds, a given location along the scan path can be irradiated with a large number of laser pulses, a condition where ablation may occur at a pulse fluence (and hence, pulse energy) smaller than that required for ablation with a single pulse. In this paper, we first study single-pulse ablation of ITO

26 and glass substrate materials. The results are then used to analyze and compare the ablation of lines in ITO using femtosecond lasers with pulse repetition rates of 2 kHz and

100 kHz. Finally, line ablation results are extrapolated to predict what possible processing rates with lasers having higher pulse repetition frequency.

3.1.2 Experiments

The material samples used in all of the experiments consisted of ITO film with a nominal thickness of 150 nm on 2.8 mm-thick soda-lime glass substrate. ITO ablation experiments were performed at atmospheric conditions using the beam from a Ti:Al2O3 regenerative amplifier operating at a wavelength of 775 nm, pulse duration of 150 fs and a repetition frequency of 2 kHz (Clark-MXR CPA 2110). The lower repetition frequency made study of single-pulse ablation convenient with this laser. Output power was attenuated with thin-film polarizing beam splitters to pulse energies of 50 nJ to 3 μJ. The beam had a diameter of 5 mm, an M2 quality parameter of 1.3 and was focused with a

25.4 mm achromatic lens (NA = 0.10) and a 10x microscope objective (NA = 0.25). The linear optical polarization axis was normal to the scan direction.

Other experiments used an Yb fiber-based chirped pulse amplified laser (IMRA

America FCPA μJewel D-400) operating at a wavelength of 1045 nm, pulse duration less than 500 fs and a higher repetition frequency of 100 kHz. Pulse energies ranged from 60 nJ to 930 nJ for the experiment. The beam had a diameter of 5 mm, an M2 quality

27 parameter of <1.5 and was focused with a 20x (NA = 0.28) infinity-corrected objective.

Larger spots were produced by defocusing the beam. The beam polarization was converted from linear to circular with a quarter-wave plate.

Processing scan speeds ranged from 1.5 to 200 mm/s and single spot ablation performance was assessed from study of non-intersecting ablation spots produced with the lower pulse repetition frequency laser. The depth and width of ablated spots and lines were measured from profiles obtained with an Atomic Force Microscope (AFM, Quesant

Q-Scope 250) and with a stylus profilometer (Dektak IIA, Veeco) as well as scanning electron and optical microscope images.

3.1.3 Results and Discussion

Single-pulse and line ablation results for the 2 kHz laser and line ablation results for the 100 kHz laser are discussed separately below. The single-pulse ablation results quantify the difference in ablation thresholds of ITO film and glass substrate and the depth of film removed by a single pulses of given fluence. These values are used in the discussion of line ablation results obtained with both lasers.

3.1.3.1 Single pulse ablation diameter and depth

Fig. 3.2 shows the single spot ablation diameters produced by the Ti:Al2O3 laser with different pulse energies and focus optics.

28 120.0

100.0 ) 2

m 80.0 y = 56.54Ln(x) + 830.1 μ (

2 NA 0.1 60.0 NA 0.25 y = 14.21Ln(x) + 232.51 40.0 Diameter 20.0

0.0 1.00E-08 1.00E-07 1.00E-06 1.00E-05 Energy (J)

Figure 3. 2 Single pulse ITO ablation energy-diameter curves produced using different focusing optics with fitted logarithmic curves.

The effective laser focus spot diameters calculated from the slope of these curves were 22.14 = 5.3 μm for the 0.25 NA optic and 24.56 =10.6 μm for the 0.1 NA optic. The ablation fluence was calculated using these effective laser focus spot diameters and fluence vs. ablation diameter-squared plots were then extrapolated to zero ablation spot diameter to obtain ITO ablation thresholds of 0.7 J/cm2 for the 0.25 NA optic (Fig.

3.3) and 0.95 J/cm2 for the smaller laser focus of the 0.1 NA optic.

29 )

2 40.00 m

μ 35.00 ( 2 30.00 25.00 20.00 15.00 10.00 5.00 y = 15.964Ln(x) + 2.6234

Ablation Diameter Ablation 0.00 1.0 10.0 Fluence (J/cm2)

Figure 3.3 ITO squared ablation diameter vs. fluence for 0.25 NA optic showing curve fit 2 parameters that correspond to Fth = 0.7 J/cm .

It is noted that other investigators have also found variations in ablation threshold fluence with focus spot size. One investigator [16] has hypothesized that as the beam waist decreases, Fth has to be increased to compensate for larger net energy dissipated away from the beam center by the hot electron diffusion. A subsequent focused, systematic study will be needed to clarify this dependence of ablation threshold on irradiation area.

30 200

160

120

40 0.25 NA Ablation depth (nm) depth Ablation 0.1 NA 0 0.0 1.0 2.0 3.0 4.0 Laser fluence (J/cm2)

Figure 3.4 Ablation depth of ITO using different focus optics

Fig. 3.4 shows the ablation depth with respect to laser fluence with the 0.25 NA and 0.1 NA focus lenses. With both lenses, fluence above about 1.2 J/cm2 completely removed the 150 nm layer of ITO and the ablation depth was relatively constant as fluence was further increased. For the 0.10 NA optic, substrate ablation increased ablation depth for fluence above 2.5 J/cm2.

Figure 3.5 Surface profile of laser ablation of ITO (2kHz, 0.1 NA lens, peak fluence F0 = 2.5 J/cm2 ; scanning speed s = 20 mm/s)

Fig. 3.5 shows a typical profile of the ITO surface after laser ablation at a scanning speed that produced approximately 25% pulse overlap. There was no evidence of substrate ablation at the spot overlap area and the lines produced electrically-separated

ITO regions. Ridges of re-deposited ITO material were located along the edges of the ablated line. The re-deposited material height was between 20 nm and 100 nm relative to

32 the un-ablated ITO surface and minor re-deposited material was noted further from the ablated line.

The ablation characteristics of the glass substrate are also important for this research. The single-pulse ablation results shown in Fig. 3.6 were produced with the 0.1

NA lens. The glass ablation threshold is calculated as ( ) = 0.262.1258.8exp J/cm2. This provides a comfortable margin for removing ITO in single pulses without ablating the glass substrate. It is smaller than the fluence at which glass ablation was observed in Fig.

3.4 because ITO and glass were both ablated by a single pulse in those experiments.

14.00 ) 2 12.00 (um

2 10.00 8.00 6.00 4.00 2.00

Ablation diameter y = 12.617Ln(x) - 8.5827 0.00 1.0 10.0 Fluence (J/cm2)

Figure 3.6 Single-pulse ablation threshold calculation of glass substrate material showing an ablation threshold of 2.0 J/cm2, well above the ITO ablation threshold of 0.7 J/cm2.

Summarizing, the single pulse ablation results illustrate that lines can be ablated in ITO by a multiple-pulse process or a single-pulse process. For example, from Fig. 3.4

2 at a laser fluence of 0.4 J/cm and with the 0.1 NA optic (Ep = 0.3 μJ), the ablation depth was approximately 40nm. Thus, approximately 3 pulses could completely remove the

ITO without a possibility of damaging the glass substrate. For ablating lines, three-pulse ablation implies at least 67% pulse overlap would be required to ensure that three pulses

34 are incident on all locations along the scan centerline. Fig. 3.7 shows an example of multi-pulse ablation at low energy fluence (the slightly non-linear ablation pattern corresponds to motion system stepper motor resolution). At a fluence of 1.1 J/cm2 or higher, a single pulse will ensure that ablation depth is 150 nm but the laser pulse energy must be at least 0.8 μJ for this processing approach.

Figure 3.7 2 kHz multipulse ablation with 0.25 NA lens (F = 0.5 J/cm2, s = 1.55 mm/s)

35 3.1.3.2 Line Ablation Results

Examples of processing results obtained with the 100 kHz fiber-based laser at a variety of pulse energies and scanning speeds with focus spot size 2w0 = d0 = 3.6 μm are shown in Fig. 3.8. The results show both complete and partial ablation of lines in the ITO with larger ablated line widths at higher pulse energies. Pulse energies of 80 nJ and above produced lines having electrical separation at the scan speed of 100 mm/s while 250 nJ was required to obtain electrical separation at 200 mm/s.

Figure 3.8 100 kHz laser ablation results for focus spot diameter d0 = 3.6 μm, scanning speed s = 100 mm/s (top) and 200 mm/s (bottom) and varying Ep showing minimum pulse energy for electrical separation: ~80 nJ (top), ~250 nJ (bottom)

36 The edges of lines ablated with smaller spot size and higher scanning speed were more irregular than those obtained with the larger spot size and lower speed and some ablation of the glass substrate is noted at the higher pulse energy (310 nJ) at 100 mm/s.

The corresponding fluence is 3.0 J/cm2, equal to the fluence where glass ablation was observed in single pulse ablation tests.

The 100 kHz line ablation results were analyzed to determine the ablation threshold fluence for difference spot sizes and scanning speeds with the fiber-based laser.

The calculations were done as for single-pulse thresholds but with ablated line width used instead of ablated spot diameter. The data corresponding to a sample calculation for laser focus spot diameter d0 = 3.6 μm and scanning speed s = 50 mm/s is shown in Fig. 3.9.

18 )

2 16 y = 6.5001Ln(x) + 6.7366 m 14 μ ( 2 12 10 8 6 4

Ablation width width Ablation 2 0 0.1 1 10 Fluence (J/cm2)

Figure 3.9 Ablation line width vs. fluence for f = 100 kHz, s = 50 mm/s, d0 = 3.6 μm. The data closely follows the expected logarithmic trend.

37 2 The corresponding ablation threshold is Fth = (− ) = 35.05.6/74.6exp J/cm .

Then, an example line width-fluence relationship for d0 = 3.6 μm and s = 50 mm/s is

2 2 = 0 ln2 ( 0 FFwD th ) = (F0 35.0ln48.6 ) . (3.1.8)

The results of all tests with varying laser focus spot size, pulse energy and pulse overlap are displayed in Fig. 3.10. The threshold ablation fluence values range from 0.23

J/cm2 to 0.63 J/cm2 depending on pulse overlap. This variation of ablation threshold can be explained by the affects of incubation.

38 ) 0.7 2 0.6 0.5 0.4 11 um 0.3 0.2 6.3 um

0.1 3.6 um

Threshold Fluence (J/cm Fluence Threshold 0 40% 50% 60% 70% 80% 90% 100% Focus Spot Overlap (%)

Figure 3.10 Variation of line ablation threshold with focus spot overlap at various laser focus spot diameters showing decrease in threshold at high overlap due to incubation.

At low pulse overlap, the ablation threshold approaches the single pulse ablation

2 threshold fluence of 0.7 J/cm obtained using the Ti:Al2O3 laser whereas at high pulse overlap, the ablation thresholds approach the ITO damage threshold fluence of ~0.25

J/cm2 reported elsewhere [11].

The affect of incubation on ablation threshold data shown Fig. 3.10 was analyzed using the relations summarized above, modified to account for the linear scanning of the beam. When a pulsed laser focus spot is scanned linearly, multiple pulses may impinge

39 on points along the scan path so incubation may be used to analyze the scanned ablation results. A convenient measure of this affect is a computation of the number of pulses N incident on a single point along the scan line. Again referring to the Fig. 3.1, the number of pulses incident on a single point along the centerline of a linear scan can be approximated as

d 1 N == . (3.1.9) 1− Ofs d

Since the beam energy extends beyond the Gaussian radius assumed in the calculations, it is appropriate to use whole numbers rather than integers in this calculation. Also, the number of pulses incident on points laterally displaced from the scan centerline decreases as the cosine of the lateral distance, falling to a single pulse as distance approaches the beam radius. Pulse numbers calculated from Eq. (3.1.9) were used to calculate the pulse ablation thresholds and incubation factors corresponding to the high frequency scanned laser ablation results of Fig. 3.10; the results are displayed in

Table 3.1.

Focus spot Calculated single pulse Incubation

diameter, d0 ablation threshold, Fth(1) factor, ξ

μm J/cm2

3.6 0.76 0.6469

6.3 0.71 0.779

11 0.82 0.7745

Table 3.1 Single pulse ablation thresholds and incubation factors for lines ablated at with different focus spot sizes.

The calculated single ablation thresholds range from 0.71J/cm2to 0.82J/cm2 and are comparable to those obtained for single pulse ablation with the Ti:Al2O3 laser, which were 0.7 J/cm2 and 0.95 J/cm2 . The thresholds and incubation factors obtained at different spot sizes are slightly different, as was also the case with the single pulse experiments.

The data displayed in Fig.3.10 and the results for ablated line diameter versus fluence allow the maximum ITO ablation speeds for higher repetition frequency lasers to be assessed with the affects of incubation included. For fixed pulse energy, incubation lowers the ablation threshold at higher pulse overlap, which makes the ablated lines wider and also increases the overlap of ablated spots, increasing line quality. A straightforward

41 calculation allows prediction of the ablation performance of higher repetition frequency lasers. Briefly, one begins with data from one curve of Fig. 3.10, corresponding to a focus optic of interest. For example, the data corresponding to d0 = 3.6 μm may be used.

Then, pulse repetition frequency f and focus spot size w0 corresponding to a laser and focusing NA of the optic of interest and a maximum ablation line width, D, and fractional line width variation, Δ D , are chosen as application requirements. It is noted that, by the geometry of overlapping ablation spots illustrated in Fig. 3.11,

2 ()11 Δ =−− Dfs . (3.1.10)

Δ/2 2 ⎛ sD ⎞ 1− ⎜ ⎟ D/2 2 ⎝ Df ⎠

s/f

Figure 3.11 Geometry of overlapping ablation spots showing calculation of theoretical ablation line ripple as a function of ablation spot diameter and processing parameters.

Thus, the scanning speed s and all other ablation parameters can be calculated.

Focus spot overlap can be found by Eq. (3.1.7) and a corresponding ablation threshold for

ITO is found from the focus spot overlap vs. ablation threshold relationship from data plotted in Fig. 3.10. If the fluence required by Eq. (3.1.4) cannot be produced by the laser or is less than the ablation threshold, then the focus spot size must to be adjusted and calculations repeated.

To illustrate the procedure and the correctness of the calculations, consider f =

100 kHz, d0 = 3.6 μm and ablation line width D = 2.75 μm and line width variation of 5

%. Then, s = 86 mm/s from Eq. (3.1.10) and Od = 76% from Eq. (3.1.7). An ablation

43 2 2 threshold of 0.48 J/cm is found from Fig. 3.10, fluence F0 = 1.54 J/cm from Eq. (3.1.4) and Ep = 78 nJ from Eq. (3.1.2). For comparison, an image of a line ablated at s = 100 mm/s , Ep = 80 nJ and Od = 72% is shown in Fig. 3.8. The measured width of 2.6 μm and estimated width variation of 5% are in reasonable correspondence with the predicted values.

One goal of the analysis is extrapolation of the data obtained from the experiments to estimate ablation performance at higher pulse repetition frequencies. A plot of scanning speed versus pulse repetition frequency for a focus spot diameter of d0 =

3.6 μm from such a calculation is shown in Fig. 3.12.

44 1.00E+01

1.00E+00 94% 67%

1.00E-01 Scan speed (m/s)

1.00E-02 1.E+05 1.E+06 1.E+07 Pulse repetition rate (Hz)

Figure 3.12 Scan speed vs. pulse repetition frequency predicted from the results for pulse repetition frequency f = 100 kHz, Ep = 100 nJ and d0 = 3.6 μm for ablated spot overlap of 94% and 67%.

From this figure, a frequency of 5 MHz would be expected to yield reasonable line quality at scanning speeds from 1.0 m/s to 5.0 m/s, corresponding to pulse overlaps of 64% to 97%. Similarly, at 10 MHz pulse repetition frequency, scanning speeds ranging from 2.0 to10.0 m/s are predicted. A plot of data from the same analysis for a focus spot diameter of 6.3 μm produces the same curve as the 3.6 μm focus spot size. However, the ablation line widths and fluences for the two spot sizes are quite different. For example the line width and fluence of the 65% overlap curve are 2.5 μm and 0.65 J/cm2 with 6.3

45 μm focus spot diameter. Line width and fluence of the 67% overlap curve are 3 μm and

1.96 J/cm2 with the 3.6 μm focus spot. To illustrate this affect, the variation of line width with scanning speed for the 6.3 μm optic and Ep = 185 nJ is shown in Fig. 3.13. Two available experimental data points, which are well-matched to the calculated values, are also plotted. The reason for the variation of line width with overlap is variation of ablation threshold due to incubation.

46 6

m) 5.5 μ

5 6.3 Calc. 6.3 Exp 4.5

4 Ablation line width (

3.5 0 0.05 0.1 0.15 0.2 0.25 Scanning speed (m/s)

Figure 3.13 Predicted and experimental ablation width versus scanning speed for d0 = 6.3 μm and Ep = 185 nJ showing the effect of lower ablation threshold at the lower speeds. The experimental point at 0.025 m/s has a pulse overlap of 96% and an ablation threshold

It is interesting that the highest speed occurs at the lowest pulse overlap, indicating that benefits of lower ablation threshold produced by incubation from highly overlapping pulses do not overcome the speed penalty incurred to high pulse overlap.

However, the lower ablation threshold due to incubation could be used to advantage if the pulse energy of the available laser were limited. For a pulse energy of only 20 nJ, corresponding to a fluence of 0.4 J/cm2, single-pulse ablation would not be possible with this laser as the threshold fluence is approximately 0.7 J/cm2. However, allowing for

47 considerable pulse overlap with a 10 MHz laser, the lower threshold due to incubation would allow ablation of quality lines at high speeds: 85% pulse overlap at 2.5 m/s in this case. One effect that may limit the maximum pulse repetition frequency for effective laser ablation is related to plasma formation. As pulse repetition frequency is increased, it is foreseeable that interference from a non-extinguishing plasma above the material surface would eventually set an upper limit on achievable ablation scanning speed

[17,18]. With regard to the lowering of the ablation threshold by incubation, it is also noted that the ablation thresholds achieved in this work were on the order of 0.25 J/cm2, which are the same as the damage thresholds reported by Ashkenasi et al. [11] mentioned earlier in this report. Thus, it would not be expected that much further decrease in ablation threshold could be achieved by higher pulse overlap.

One result is noticed when performing calculations of scan speed vs. pulse energy for given ablated spot overlap. Essentially, while an applied fluence above the threshold fluence (including incubation) is necessary for ablating a scan line, once this requirement has been met, the pulse energy does not influence the ablation scanning speed. This can be seen by rearranging Eq. (3.1.7) to obtain an expression for scanning speed.

= ⋅ ⋅ − Odfs d )1( (3.1.11)

Noting that Eq. (3.1.7) also applies when ablated spot overlap is used instead of focus spot overlap, an alternative equation for scanning speed is

= ⋅ ⋅ − ODfs D )1( (3.1.12) where OD is ablation spot diameter overlap. Further, defining c' as the slope of the scanning speed vs. pulse repetition rate graph so that = ′fcs and noting that this slope

48 equals the distance traveled during one pulse repetition period (the “pitch”), it is clear that scanning speed does not depend on energy once either ablated spot overlap or focus spot overlap has been fixed.

3.1.4 Conclusions and future works

Experimental results for the femtosecond laser ablation of ITO films from glass at different pulse repetition rates were analyzed. The single pulse ablation threshold of ITO was substantially less than glass, so single pulses with F = 1.0 J/cm2 – 2.5 J/cm2 could completely remove the ITO without substrate damage. Taking advantage of the additive ablation of several overlapping low energy pulses, ablation of lines at 1.5 mm/s was demonstrated with the 2 kHz system at Ep = 300 nJ. Lines were ablated in ITO with the

100 kHz pulse repetition frequency laser. Due to the highly overlapping pulses, femtosecond laser pulse ablation was effective for patterning ITO film at relatively lower pulse energies and higher scanning speeds. Ablation of good quality lines at the s = 100 mm/s, ds = 3.6 μm and Ep = 80 nJ was demonstrated. Analysis of the ablation performance of higher pulse repetition frequency lasers was used to predict performance of higher pulse repetition frequency lasers.

The feasibility study showed that the femtosecond laser can be used for ITO ablation, but low productivity or low scanning speed can be still an issue for production.

Recently, lasers with higher pulse rates (> 1MHz) and sufficient pulse energy for micromachining have been developed. They have much potential for use in

49 micromachining applications. The motion system may be the limitation for machining with these high pulse rate systems, but high speed scanners would be expected to solve this issue. The scanner application needs more study in future work for assessment of scanning mirrors damage thresholds and focus spot diameter limitations.

50 3.1.5 References

1. B.J. Luff, J.S. Wilkinson and G. Perrone, “Indium tin oxide overlayered waveguides for sensor applications,” Appl. Opt. 36, 7066-7072 (1997).

2. R. Bel Hadj Tahar, T. Ban, Y. Ohya and Yasutaka Takahashi, “Tin doped indium oxide thin films: Electrical properties,” J. App. Phys. 83, 2631-2645 (1998).

3. O. Yavas and M. Takai, “Effect of substrate absorption on the efficiency of laser patterning of indium tin oxide thin films,” J. Appl. Phys. 85, 4207-4212 (1999)

4. Y.H. Tak, C.N. Kim, M.S. Kim, K.B. Kim, M.H. Lee and S.T. Kim, “Novel patterning method using Nd : YAG and Nd : YVO4 lasers for organic light emitting diodes,” Synthetic Metals 138, 497-500 (2003).

5. C. Molpeceres , S. Lauzurica , J.L. Ocana, J.J. Gandia, L. Urbina and J. Carabe, “Microprocessing of ITO and a-Si thin films using ns laser sources,” J. Micromech. Microeng. 15, 1271-1278 (2005).

6. M. Giridhar, K. Seong, A. Schulzgen, P. Khulbe, N. Peyghambarian and M. Mansuripur, “Femtosecond pulsed laser micromachining of glass substrates with application to microfluidic devices,” Appl. Opt. 43, 4584-4589 (2004).

7. C.H. Fan, J.P. Longtin, “Modeling optical breakdown in dielectrics during ultrafast laser processing,” Appl. Opt. 40, 3124-3131 (2001).

8. M. Elbandrawy and M. Gupta, “Optical characteristics of femtosecond laser micromachined periodic structures in Si <100>,” Appl. Opt. 45, 7137-7143 (2006).

9. R. Stoian, D. Ashkenasi, A. Rosenfeld and EEB Campbell, “Coulomb explosion in ultrashort pulsed laser ablation of Al2O3,” Phys. Rev. B 62, 13167- 13173 (2000).

10. J. Krüger and W. Kautek, “Ultrashort Pulse Laser Interaction with Dielectrics and Polymers,” Adv Polymer Sci 168, 247–289 (2004).

51 11. D. Ashkenasi, G. Muller, A. Rosenfeld, R. Stoian, I.V. Hertel, N.M. Bulgakova and E.E.B. Campbell, “Fundamentals and advantages of ultrafast micro-structuring of transparent materials,” Appl. Phys. A-Mater. Sci. Processing. 77, 223-228 (2003).

12. A. Ben-Yakar and R.L. Byer, “Femtosecond laser ablation properties of borosilicate glass,” J. Appl. Phys. 96, 5316-5323 (2004).

13. D. Ashkenasi, M. Lorenz, R. Stoian and A.Rosenfeld, “Surface damage threshold and structuring of dielectrics using femtosecond laser pulses: the role of incubation,” Appl. Surface Sci. 150, 101-106 (1999).

14. J.R. Buckley, F.W. Wise, F.O. Ilday and T. Sosnowski, “Femtosecond fiber lasers with pulse energies above 10 nJ,” Opt. Lett. 30, 1888-1890 (2005).

15. L. Shah, Z.L. Liu, I. Hartl , G. Imeshev, G.C. Cho and M.E. Fermann, “High energy femtosecond Yb cubicon fiber amplifier,” Opt. Express 13, 4717-4722 (2005).

16. B. Fisette and M. Meunier, “Three-dimensional microfabrication inside photosensitive glasses by femtosecond laser,” JLMN-J. Laser Micro/Nanoengineering, 1, 7-11 (2006).

17. B. Salle, O. Gobert, P. Meynadier, M. Perdrix, G. Petite and A. Semerok, Femtosecond and picosecond laser microablation: ablation efficiency and laser microplasma expansion,” Appl. Phys. A – Mat. Sci. Processing 69, S381-S383 (1999).

18. S.M. Klimentov, T.V. Kononenko, P.A. Pivovarov, S.V. Garnov, V.I. Konov, A.M. Prokhorov, D. Breitling and F. Dausinger, “The role of plasma in ablation of materials by ultrashort laser pulses,” Quant. Elec. 31, 378-382 (2001).

52 3.2 Hot embossing

3.2.1 Introduction

Micro- and nanoscale fluid flow channels in polymer materials are needed for fabrication of many biomedical devices. Polymer materials are advantageous for such devices because they offer a wide range of material properties including good biocompatibility [1], low conductivity for electrokinetic pumping or electrophoretic separation, low material cost for high-volume mass fabrication of disposable devices and replication methods for mass production [2]. There are a number of options for fabrication of microfluid channels in polymer. Some surface micromachining processes, notably plasma etching [3, 4] and direct-write laser micromachining [5-7] have been used to fabricate prototype or small lots of polymer microfluidic devices. However, mass production techniques such as embossing or injection molding are best suited for large scale, economical polymer manufacturing.

Of the mass production processes that are capable of high-rate manufacturing of polymer microfluidic products, hot embossing [8] is more commonly used for the finest features. In hot embossing, a tool impresses a pattern into the surface of a polymer that is softened by heating above its glass transition temperature. The hot embossing method has been used to fabricate inexpensive, precise polymer microstructures for optical components or microfluidic devices [2, 8-10]. The tool can be made from a variety of materials including silicon, polymer and metals. Relative to the surface micromachining

53 processes mentioned previously, hot embossing offers much higher productivity,

comparable feature sizes and lower cost for forming precision shapes in polymer. The

hot embossing process is compatible with a variety of polymers with a wide range of

material properties and surface chemistries for specific applications.

Hot embossing tools may be created with a variety of processes, the choice of which depending on the material, precision and productivity requirements. For silicon

and metal films, many micromachining processes commonly used in electronics

fabrication may be applied for creating microscale and nanoscale features. A

disadvantage of the silicon tools is a tendency to break during demolding, which is not as

problematical with the more durable metal mold materials [11]. Also, lithographic

patterning requires a costly clean-room environment and needs potentially-costly

multiple fabrication steps [12, 13]. Laser and electron beam lithography and focused ion

beam (FIB) milling techniques have high resolution for generating microscale features as

well as nanoscale ones in durable metal tool materials. However, FIB has low throughput

even considering that the manufacturing cost of the tool will be used to manufacture

many polymer parts. LIGA, a process that combines E-Beam lithography and

electroplating to create metal parts with microscale features has also been applied for

embossing tool fabrication but E-Beam lithography apparatus is expensive. CNC machining is used for creating larger features in a wide range of metals, but is not useful

for submicron feature sizes.

It is well known fact [14] that the femtosecond lasers with typical pulse durations

of about 150fs can ablate material with significantly lower thermal effect on the residual

54 material than nanosecond lasers with pulse width ranges from 10 to 100’s of nanoseconds.

Studies have shown that the heat affect zone (HAZ) width with femtosecond laser can be a few hundred times lower than for nanosecond pulses [15]. An alternative laser used for direct-write ablation of metals is the excimer laser. It is capable of making microstructures with feature sizes on the order of 1 μm and applicable for polymers, metals, and ceramics [16], However, excimer laser ablation still has some issues of heat input and is more favorable for materials with low thermal conductivity and low melting temperature, such as glass[17]. Many metals and carbides which would make very rugged, durable embossing molds are not compatible with deposition and micromachining processes often used in microfabrication. Also, the same superior mechanical properties also create difficulties for mechanical machining; stress-induced martensitic transformation at the cutting front leads to an undesirable contact interface between the tool and the workpiece [18]. However, femtosecond laser micromaching can make micro- and nanoscale features without material limitations [19].

The combination of hot embossing of polymer materials with tools flexibly micromachined by femtosecond laser ablation is seen as a potentially advantageous process for mass production of biomedical devices with microscale and sub-microscale features. The objectives of this research were to characterize capabilities of femtosecond laser micromachining for creating embossing tools of stainless steel. A series of experimental investigations were carried out to analyze the effect of ablation process parameters on achievable feature size of 50μm (minimum 8μm), surface roughness of

285nm (RMS), free standing trench angle of 36 degrees, and material removal rate

55 measured to be is 0.88 μm3/pulse at 500nJ which can remove material at 5,652 μm3/s with 6kHz pulse repetition rate.

3.2.2 Experiment procedure

Steps for polymer microfluid device fabrication illustrated in Fig. 3.14 including mold fabrication, hot embossing and cover bonding. This investigation focused on fabrication of hot embossing molds in polished AISI 304L stainless steel blanks by femtosecond laser micromachining.

Figure 3.14 Steps for fabrication of a polymer microfluid device by the hot embossing process: (a) precision polishing of mold blank; (b) femtosecond laser machining of mold; (c) hot embossing; (d) assembly of device with top cover.

The mold material was selected as one typically used for durable, long-life hot- embossing tools [20]. Since the final embossed patterns replicate the features on the mold, the roughness of the mold material directly affects to the roughness of the microfluid passages. Thus, the surface of the mold blank was prepared by precision polishing (Ultra

Tec MULTIPOL 8) using 3μm aluminum oxide. The surface profile was measured using an optical profilometer (Veeco Wyko DMEMS NT3300) with an array size of 736 by 480 points and resolution of 1.65μm, covering an area 1.21mm x 0.92mm.

An image of the optical profilometer results characterizing the polished surface is shown in Fig.3.15. The surface height data was processed to remove the tilt of the planar surface. The average surface roughness Ra was 12.37 nm and the root mean square

57 surface roughness Rq was 16.43nm. The difference between the maximum and minimum

surface heights in the measuring area Rt was 355.80nm. A more reliable value of the maximum variation of surface height can be found as the difference of the averages of the

10 highest profile points and the 10 lowest profile points. This value is denoted Rz, and its value was 233.10nm.

Figure 3.15 Surface profile of polished mold blank as measured by optical interferometry.

58 The polished mold blanks were patterned with a laser micromachining system

comprised of a Ti:Al2O3 femtosecond laser (Clark-MXR CPA2110) and a submicron

resolution motion system. The laser has a central wavelength λ = 775nm, adjustable pulse

repetition rate fp = 2kHz and 3kHz, measured full width half maximum (FWHM) pulse

width Tp = 150fs and average output power of 2.6W. The irradiance distribution of the

laser output beam is nominally Gaussian. The 1/e2 beam diameter and quality factor were

measured to be 5mm and M2 = 1.3. The laser beam was delivered to the mold surface for

the machining process by beam delivery optics; a detailed optical schematic of the laser

micromachining system is shown in Fig. 3.16 Since the typical range of power for direct- write micromachining is less than 5mW, the laser output power was attenuated by a combination of two thin-film polarizing beam splitters (PBS) and a λ/2 wave plate

polarizing rotation optics. The polarization angle of the linearly polarized beam emitted

by the laser is horizontal (parallel to the optical table) and is rotated by the λ/2 wave

plate. The amount of power transmitted by the second PBS varies depending on the angle

of the polarization axis of the incoming light with respect to the optical axes of the beam

splitter [21]. Laser power was switched by a high speed mechanical shutter with 4ms

control time (NMlaser LS055) in coordination with the part motion. The attenuated laser

beam was focused onto the mold surface by an infinity corrected long working distance

microscope objective lens with numerical aperture NA = 0.42 (Mitutoyo M Plan Apo

NIR 50x). A coaxial vision system was integrated into the laser beam delivery optics to

provide a capability for accurate focusing. The focus location was adjusted by the

motorized z axis and a manual micrometer by visual inspection of the coaxial image

59 generated by a CCD camera-based vision system. It was estimated that the vision system

allowed laser beam focus to be positioned relative to the material surface with an accuracy of +/- 1μm, comparable to the 1.6μm focal depth of the 50x objective.

Figure 3.16 Block diagram of beam delivery system setup

60 The mold blank was fixtured on an XY stage (Parker Daedl MX80) with motion range of 50mm. The focusing optic was manipulated in the Z coordinate axis by a linear stage with a motion range of 25mm. The axes were driven by linear motors controlled by a computer numerical control (CNC) system (Parker Compumotor ACR9000) and their positioning accuracy was measured by optical encoders with resolution of 0.5μm.

3.2.3 Laser ablation

To determine appropriate laser micromachining process variables, it is helpful to determine the material ablation threshold, defined as the minimum laser energy density that can ablate the material. The relations used in the analysis of laser ablation are briefly summarized and the results of experiments that were conducted to determine ablation threshold and suitable laser machining process variables are discussed below.

3.2.4 Ablation and damage thresholds and pattern resolution

The minimum diameter of a focused laser beam, and hence the maximum fluence, occurs at the focus waist. For the multimode laser beam generated by the laser used in these experiments, the diffraction-limited minimum focus spot sized d may be estimated as

61 2 λ Md 2 ⋅⋅= (3.2.1) π NA where NA is numerical aperture, M2 is beam quality factor and λ is the wavelength of the

laser radiation.

For the laser beam and focusing optic used in this work, the calculated minimum focus

spot diameter is d = 1.6μm. However, as will be discussed below, when performing experiments to measure the material ablation threshold, it is more accurate to calculate the focus spot diameter from the experimental data. The radius at which the laser beam fluence equals the ablation threshold fluence Fth can be found by solving the Gaussian

intensity distribution equation

2 2 th 0 (−= 2exp wrFF 0 ) (3.2.2)

where w0 is Gaussian beam radius (half the Gaussian beam diameter) and F0 is maximum

(or peak) fluence, related to pulse energy Ep by

2 0 = p /2 πwEF 0 . (3.2.3)

From Eq. (3.2.3), one may also write

62 2 = thth /2 πwEF 0 , (3.2.4)

where Eth is the pulse energy corresponding to ablation threshold fluence. Re-arranging

Eq. (3.2.2) to solve for the diameter D of the area near the center of the beam from which material is ablated results in the relation

2 2 2 = 0 ()0 th = 2/ln2 0 [ln( 0 ) − ln(FFwFFwD th )] (3.2.5)

Single pulse ablation spots were produced by scanning the focus spot over the flat surface at a sufficient speed that the spots were separated. Pulse energy was varied over a range of values while focus spot radius was maintained at a known, fixed value. A semi-

2 logarithmic plot of D versus ln(F0 ) was created and the ablation threshold fluence was found by extrapolating the linear curve resulting from this plot to zero diameter where th = FF 0 . The effective focus spot radius corresponding to ablation experiments can be calculated directly from the ablation data [16]. Using Eq. (3.2.3) and Eq. (3.2.4) in

Eq. (3.2.5) yields

2 2 = 0 ( 0 /ln2 EEwD th ) (3.2.6)

Thus, the effective radius of the laser focus spot needed for subsequent calculations can be computed from the slope m of a semi-logarithmic plot of D2 vs. laser pulse energy by

m ω = . (3.2.7) 0 2

At fluences lower than the ablation threshold, it is common to speak of damage threshold fluence Fd above which visible damage (but not removal) of the surface results. If Fd <

F0 < Fth, experimental damage diameters can be measured and Fd can be found by a similar procedure as for Fth.

To determine ablation threshold, linear ablation scans were conducted with laser energy set to values in the range from 20nJ to 750nJ and a scanning speed of 30mm/s, selected to ensure separation of adjacent ablation spots. The ablation patterns were imaged by scanning electron microscope (SEM) at an acceleration voltage of 20kV. The average diameter from all spots in a scan row was calculated. Examples of typical ablation spot images for each of the powers used in the experiment are presented in Fig.

3.17. The outer diameter of each ablation spot was measured and squared for use in Eq.

(3.2.6) and the resulting plot of squared ablation spot diameter versus pulse energy and laser fluence are shown in Fig. 3.18.

Figure 3.17 Single pulse ablation patterns for various pulse energies

(a)

(b)

Figure 3.18 Ablation diameter results with regression-fit lines: (a) squared ablation spot diameter vs. peak energy; (b) squared ablation spot diameter vs. laser fluence

66 Extensive melting, expulsion of melt and deep drilling at the center of ablation spots were observed at high laser pulse energy ( > 375nJ). Such phenomena are characteristic of high energy femtosecond laser ablation of metal where ablated material forms persistent plasma which radiates energy to the processed surface and causes the observed effects [22]. Conversely, only slight ablation with practically no melting was observed at lower laser energies (< 50nJ). The ripple patterns observed for low fluence ablation had a direction identical to the laser polarization and their period is comparable to the laser wavelength, consistent with the observations by other investigators [23]. The ripple patterns were indistinct because, unlike some of the earlier work, the ablation spots were all due to a single laser pulse so there was no opportunity for the ripples to grow by self-induced interference of scattered and incoming laser radiation.

A linear equation was regression fit to the semi-logarithmic plot of squared ablation diameter vs. pulse energy data. The regression fit line is plotted with the experimental data in Fig. 3.18(a) and its equation is

2 6.02 - )Ln(E2.83 D = p )Ln(E2.83 - 6.02 (3.2.8)

2 where D is expressed in μm and Ep in nJ. The slope of the linear function is 2.83μm /nJ, so the corresponding focus spot radius and diameter calculated from the ablation data are w0 = 1.2μm and d = 2.4μm. It is noted that the focus spot diameter calculated from the ablation data is comparable to the diffraction-limited focus spot diameter of 1.6μm calculated from the laser beam characteristics and focusing optic parameters using Eq.

67 (3.2.1). All fluences in this work were calculated using this spot size calculated from the ablation data. The equation of the regression-fit line shown in Fig. 3.18(b) is

4.71 Ln(F)2.83 D2 = Ln(F)2.83 + 4.71 (3.2.9)

The X-intercepts calculated from the linear functions correspond to an ablation

2 threshold energy of Eth = 4.71nJ and an ablation threshold of Fth = 0.19J/cm . The calculated ablation threshold is comparable to the value of 0.21J/cm2 found in previous research work for a similar material, 316L stainless steel [23]. Other literature sources show that the ablation threshold of common metals is in the range of 0.1-0.2 J/cm2 [24].

It should be also noted that the ablation threshold differs for single pulse and multiple pulse ablation experiments. The latter situation occurs during machining of lines or areas with a scanned focus spot where the scan speed is slow relative to the pulse repetition frequency. Then, multiple laser pulses impinge on any selected point within the machined area. The single-shot ablation threshold is typically higher than the multi-shot threshold due to the incubation effect. The significance of incubation was first noted in the context of dielectric material ablation, however, it is reported that the incubation effect is also significant for metals [23]. For the process parameters used in most of the machining in this research (described in more detail below), the laser focus spot overlap was 75% so a maximum of sixteen pulses impinged on any location within a machined area. The relationship used to predict the ablation threshold at a location with impingement of N pulses, denoted th (NF ) , is

S −1 th ()= th (1)NFNF (3.2.10)

68 where S is an incubation threshold. Using Eq. (3.2.10) with the single-pulse ablation threshold calculated above and an incubation coefficient for AISI 316L stainless steel

2 [22], an estimated sixteen-pulse ablation threshold is Fth (16)= 0.16 J/cm .

Most ablation spots in Fig. 3.17 have clearly visible damaged areas surrounding the ablation crater. The diameters of these damaged regions was estimated from the SEM images and the same curve fitting procedure used to determine the ablation threshold was repeated to determine a damage threshold. The final result of this calculation was a damage threshold of 0.09 J/cm2.

To study the efficiency of laser ablation, the volume of single pulse ablation craters was measured. To make this measurement, three-dimensional profiles of ablation spots were acquired with an Atomic Force Microscope (AFM, Digital Instruments

Nanoscope). In Fig. 3.19, profile scans are compared to SEM pictures of the same ablation spot from Fig 3.17. Due to axisymmetry of the laser beam fluence distribution, the actual depth profile of the ablation cavity was also approximately axisymmetric so the volume of the ablation cavity was estimated by integration based on the AFM profile scan [25]. The measured ablation volume is presented in Fig. 3.20 and a monotonic increase with pulse energy is observed.

Figure 3.19 3D profile of laser ablation (Left E=500nJ, Right E=65nJ)

Figure 3.20 Ablation volume as a function of laser pulse energy

The ablation efficiency is a possible criterion for optimizing the laser machining process. Assuming that femtosecond laser ablation removes all material from the ablation crater by evaporation, one can estimate the efficiency of material removal from the volume data. A basic assumption inherent in such a calculation is that all material within the ablation volume just attains energy equal to the evaporation enthalpy at the boiling

71 temperature. The efficiency then is computed as the ratio of this energy to the laser pulse energy. A formula for ablation efficiency η based on this concept can be written as

ρ mp 0)(*[ +−++− LTTCLTTC vmvpm ])(* η = ∗V (3.2.11) Ep

where C p is average specific heat of the material, ρ is average density of material, V is ablation volume, Tv is evaporation temperature (assumed equal to boiling), T is melting temperature, T0 is room temperature, Ep is pulse energy and Lm and Lv are latent heats for melting and evaporation phase changes. There are several sources of error in such a simple efficiency calculation. One is due to the likelihood that much of the ablated material has considerably more energy (thermal and kinetic) than that assumed. This is clearly the case at high pulse energy, where plasma is visible above the ablation area.

Conversely, to the extent that melted material is ejected by vapor pressure. The images in Fig. 3.17 confirm that melt ejection was significant for experiments at the medium and high range of laser energies. Ejected molten material could contain less energy than evaporated material since it only needs to be heated above the melting temperature to be ejected.

The efficiency estimated by Eq. (3.2.11) is plotted as a function of pulse energy in

Fig. 3.21. A maximum efficiency of 2% is predicted at relatively low pulse energy of

65nJ. It is observed that ablation efficiency is lower at low laser energy. This is a

72 reasonable result since at lower pulse energy; the fluence of most of the laser focus spot area is below the ablation threshold. Thus, surface damage rather than ablation is the result over much of the focus spot area. Lower ablation efficiency was also found at the higher energies. This is likely because the energy of the ablated material is significantly above the evaporation enthalpy. The peak ablation efficiency region was found to be in the range from 35nJ ~ 165nJ. From Fig. 3.17, the ablation patterns produced at these energies were relatively distinct with minimal recast material and surface damage around the rim of the ablation crater.

2.50%

2.00%

1.50%

Efficiency 1.00%

0.50%

0.00% 0 100 200 300 400 500 600

Energy (nJ)

Figure 3.21 Ablation efficiency

73 3.2.5 Mold micromachining

Mold machining required the removal of material from areas much larger than the laser focus spot diameter. Also, machining depths were required to be larger than the single-pulse ablation depth for reasonable pulse energies. To achieve the required machining depth, ablation was carried out in multiple layers with each ablation layer consisting of multiple, overlapped curvilinear machining passes. For each scan pass, the length of intersection of individual ablation spots in the scan direction (the ablation spot

“over-lap”) was adjusted to produce uniform material removal. Consecutive scan passes were also over-lapped in the lateral direction by a distance equal to the overlap of ablation spots within the scan passes.

The overlap between individual ablation spots along a scan line depends on the scanning speed and laser fluence. Because of the latter dependence, it is most convenient to specify the machining process in terms of overlap of focus spots rather than ablation spots. Thus, the geometry of a typical scanned ablation pass is shown in Fig. 3.22 where the scan speed s, laser focus spot diameter d and overlap Ld are shown.

Figure 3.22 Overlap of adjacent focus spot locations corresponding to consecutive laser pulses during a scanned ablation pass illustrating a focus spot overlap of approximately 33%

The pitch p shown in this sketch is the scan distance covered in one pulse repletion period, expressed as

s p = . (3.2.12) f

The length of intersection of adjacent focus spots corresponding to two consecutive pulses is d – p, so the overlap as a percentage of focus spot diameter is

75 − pd L = × %100 . (3.2.13) d d

Assuming known focus spot diameter and pulse repetition frequency, Eqs.

(3.2.12) and (3.2.13) can be combined and the result re-arranged to yield an expression for the scanning speed that corresponds to a desired focus spot overlap:

⎡ ⎛ Ld ⎞⎤ s ⎢1−= ⎜ ⎟⎥ ⋅⋅ fd . (3.2.14) ⎣ ⎝100⎠⎦

Scan speeds to produce focus spot overlaps of 25%, 50%, 75%, and 90% were calculated as 5.4 mm/s, 3.6 mm/s, 1.8 mm/s and 0.7 mm/s and each speed was used with pulse energies of 165nJ, 250 nJ, 330nJ, and 500 nJ in a series of single line scan ablation tests.

SEM images were examined to determine the smoothness, width and width variation of the ablated lines. Examples of lines ablated at 50% and 75% overlap and pulse energy

165nJ are shown in Fig. 3.23.

Figure 3.23 Multipulse overlapped scanning (E=165nJ)

Of the energies investigated for multi-pass machining, the smoothest most consistent area ablation results were obtained at this lowest value, although the material removal rate was also lowest. Also, the best compromise between channel quality and ablation speed was judged to be produced for the 75% focus spot overlap. From the slope of the data in Fig. 3.20, the material removal rate is estimated as

V = 0017.0 ⋅ Ep (3.2.15)

77 3 where V is ablation volume (μm /pulse) and Ep is laser pulse energy (nJ). Most of the machining needed for mold fabrication required removal of areas of material with width larger than that removed by a single scan pass. It was found that area ablation done using the energy and overlap mentioned above and adjacent scan passes spaced for a focus spot overlap of 75% produced satisfactory results.

We note that the smoothest and most precise femtosecond laser ablation was obtained for relatively low pulse energies in our work. It is notable that other investigators (previously cited references) have generally found this to be the case for a variety of materials. A recent trend in ultrafast pulsed laser development has been toward systems with low pulse energy but high pulse repetition rate. Ablation of SiC at a pulse repletion frequency of 50MHz has been demonstrated [26]. If 50MHz and the same pulse energy, spot size and focus spot overlap of 75% were used for the present application, the maximum linear scanning speed would 90m/s.

As channels were being machined during this investigation, it was noted that the ablated material tended to re-deposit onto the channel surface. Examples of such debris can be seen in the channels depicted in Fig. 3.2.23. This effect also needs to be minimized in order to obtain the best quality machined surface. In our experiments, it was found that providing a cross-flow of argon gas at 39x10-6 m3/s (5 ft3/hr) from a small nozzle helped to avoid this problem.

78 3.2.6 Hot embossing

The developed parameters were used to fabricate a prototype microfluid channel mold for a molecular magnetic separation device. The designed channel width was 75μm and as shown in Fig. 3.24(a) and 3.24(b), the width of the channel feature on the mold was the approximately the same dimension at its base. A total of 5 layers of ablation were required to machine mold features tall enough to produce a channel depth of 10μm and the taper of the channel feature resulted in a channel width of only 25μm at the top of the mold feature. The taper of the channel feature can be observed in the surface profile in

Fig. 3.24 (b).

100μm

(a)

(b)

Figure 3.24 (a). Hot embossing mold on stainless steel, (b) surface profile of mold

80 The hot embossing experiment was conducted on custom-made equipment. The system consisted of an electromechanical microtester (Instron 5848), which controlled the embossing speed and force. A sample of PMMA (McMaster Carr) (0.75in x 0.75in x

1/16in) was put on top of bottom mold and the embossing mold fabricated by laser ablation was placed on top of the PMMA. Before the embossing, both the mold and the

PMMA were heated up to 130°C. The embossing mold was then impressed onto the

PMMA surface with an embossing force of 25lb. Both the force and the temperature were held constant for 5 minutes, after which the system was cooled down to 80°C. Finally, the mold was removed and the embossed PMMA sample was separated from it.

The magnetic separation device fluid channel pattern was replicated on PMMA by hot embossing process; the finished device is depicted in Fig. 3.25. The bottom of the molded channel was smooth, with RMS roughness of 20 nm, approximately equal to that of the polished mold surface. The PMMA surface adjacent to the channel had RMS roughness of 285nm. The profile of this polymer surface had approximately the same roughness as the laser-ablated mold surface and was the minimum obtainable for the chosen laser ablation parameters. However, this roughness is not important for the function of the finished microfluid device since it was covered by the top-lid of the device, attached with UV curable glue.

Figure 3.25 Image of a hot embossed molecular magnetic separator pattern in PMMA

Other laser machining experiments were done to determine the minimum channel width that could be machined into a mold. The laser ablation parameters remained unchanged from those used to produce the mold in Fig. 3.25, but the motion system program was modified to produce a thinner channel. The height of the channel features was set at 12μm. As shown in Fig. 3.26, a mold for a minimum channel width of 8μm was achievable. Attempts to produce a mold for a channel width of 6μm resulted in a feature with a somewhat irregular top surface.

Figure 3.26 SEM images of an embossing tool machined to create 8μm wide channel on AISI 304L stainless steel. The depth of the well and channel features is 12μm.

The achievable mold feature sizes are related to the minimum ablation spot size, which are determined by the ablation parameters being used. A common technique to ablate features smaller than the diffraction-limited focus spot size is to use laser energies that generate peak fluence only marginally larger than the ablation threshold. As shown in Fig. 3.27, by careful selection of laser energy, one can produce ablation spots with diameter smaller than the laser beam spot size. All ablation spots for which power is lower than 38 nJ have diameters smaller than the focus spot diameter, 2.4μm. More laser machining parameter development using these low energy is planned. It is noted that the material removal rate for such low energies estimated from Eq. (3.2.15) is also small, but

83 restricting use of low energy to machining portions of the mold where the finest features are needed would minimize the decreased productivity.

Figure 3.27 Theoretical diameter of ablation and damage regions for a pulse fluence distribution with maximum fluence of 0.59 J/cm2, an ablation threshold of 0.19 J/cm2 and a damage threshold of 0.09 J/cm2.

It is also noted that ablation diameter at higher fluence is wider than theoretically predicted value which can be calculated from the ablation threshold value. As explained before, outer irradiated area where laser fluence is still higher than that of ablation threshold was wasted to heat the material. At lower fluence, the ablation pattern is more

84 stochastically distributed rather than deterministic behavior. The ablation quality is again very clean in mid fluence range as shown in Fig. 3.28.

(a) (b) (c)a

Figure 3.28 Ablation profiles for pulses with various peak fluence: (a) Ep= 13nJ, fp=0.58 2 2 2 J/cm ; (b) Ep= 165nJ, fp =7.3 J/cm ; (c) Ep= 500nJ, fp=22 J/cm .

3.2.7 Conclusions and future works

Femtosecond laser micromachining was used to fabricate hot embossing molds for microscale fluid channel fabrication. Single pulse experiments with a broad range of laser pulse energies from 13nJ to 500nJ were conducted to determine the focus spot size and ablation threshold. The threshold ablation fluence for the AISI 304L mold material was calculated to be 0.19 J/cm2. Based on data from scanned ablation experiments, pulse energy of 250nJ with 75% ablation spot overlap was used to fabricate a hot embossing

85 mold for molecular magnetic separator. Molds for creating channels with a minimum width of to 8μm were achieved. The corresponding ablation depth was measured as

12μm, yielding a maximum depth/width aspect ratio of 1.5. An embossing mold with channel width of 75μm was used with the hot embossing process to produce a microfluid device of PMMA polymer. The characterization of flow properties such as flow rate, pressure gradient, flow profile across the channel and turbulence remains as a future work.

The surface roughness vs. flow behavior of internally machined channels is also an interested topic to be explored. The molecule magnetic particle separator is designed for the particle size around 2μm. However, much smaller particles (~ 50nm) will be tested in the future and corresponding channel size needs to be developed.

86 3.2.8 References

1. H. Becker, and U. Heim, Hot embossing as a method for the fabrication of polymer high aspect ratio structures. Sensors and Actuators, 2000. 83: p. 130-135.

2. C. Choi, Fabrication of optical waveguides in thermosetting polymers using hot embossing. J. Micromech. Microeng, 2004. 14: p. 945-949.

3. A. Holland, C. Cher, G. Rosengarten, and D. Simon, Plasma etching of thermoset polymer films for microchannels. Smart Mater. Struct., 2006. 15: p. S104-S111.

4. J. Rossier, C. Vollet, A. Carnal, G. Lagger, V. Gobry, H. Girault, P. Michel, and F. Reymond, Plasma etched polymer microelectrochemical systems. Lab on a Chip, 2002. 2(3): p. 145-150.

5. M. Roberts, J. Rossier, P. Bercier, and H. Girault, UV laser machined polymer substrates for the development of microdiagnostic systems. Anal. Chem., 1997. 69: p. 2035-2042.

6. D. Farson, H. Choi, C. Lu, and L. Lee, Femtosecond laser bulk micromachining of microfluid channels in poly(methylmethacrylate). Journal of Laser Applications, 2006. 18(3): p. 210-215.

7. J. McDonald, V. Mistry, K. Ray, and S. Yalisove, Femtosecond pulsed laser direct write production of nano- and microfluidic channels. Appl. Phys. Lett., 2006. 88(183113): p. 1-3.

8. J. Mecomber, D. Hurd, and P. Limbach, Enhanced machining of micron-scale features in microchip molding masters by CNC milling. International Journal of Machine Tools & Manufacture, 2005. 45: p. 1542 - 1550.

9. G. Lee, C. Hung, B. Ke, G. Huang, and B. Hwei, Micromachined pre-focused 1XN flow switches for continuous sample injection. J. Micromech. Microeng, 2001. 11: p. 567-573.

10. J. Zhang, K. Tan, G. Hong, L. Yang, and H. Gong, Polymerization optimization of SU-8 photoresist and its applications in microfluidic systems and MEMS. J. Micromech. Microeng, 2001. 11: p. 20-26.

11. M. Esch, S. Kapur, G. Irizarry, and V. Genova, Influence of master fabrication techniques on the characteristics of embossed microfluidic channels. Lab on a Chip, 2003. 3(2): p. 121-127.

87 12. H. Cao, Z.Yu, J. Wang, J. Tegenfeldt, R. Austin, E. Chen, W. Wu, and S. Chou, Fabrication of 10nm enclosed nanofluidic channels. Appl. Phys. Lett., 2002. 81(1): p. 174-176.

13. B. Jo, L. Lerberghe, K. Motsegood, and D. Beebe, Three-dimensional Micro- channel fabrication in PDMS elastomer. Journal of microelectromechanical systems, 2000. 9(1): p. 76 - 81.

14. B. Chichkov, C. Momma, S. Nolte, F. Alvensleben, and A. Tunnermann, Femtosecond , picosecond and nanosecond laser ablation of solids. Applied Physics A: Materails Science & Processing, 1996. 63(2): p. 109-115.

15. S. Valette, R. Harzic, N. Huot, E. Audouard, and R. Fortunier, 2D Calculations of the thermal effects due to femtosecond laser-metal interaction. Applied surface science, 2005. 247: p. 238-242.

16. H. Endert, R. Patzel, and D. Basting, Excimer laser: A new tool for precision micromachining. Optical and Quantum electronics, 1995. 27(12): p. 1319-1335.

17. A. Tseng, Y. Chen, and K. Ma, Fabrication of high-aspect-ratio microstructures using excimer laser. Optics and Lasers in Engineering, 2004. 41: p. 827-847.

18. H. Huang, H. Zheng, and Y. Liu, Experimental investigations machiniability of Ni50.6Ti49.4. Smart Materials and Structurs, 2005. 14: p. S297-S301.

19. C. Li, D. Johnson, and R. Kovacevic, Modeling of waterjet guided laser grooving of silicon. International Journal of Machine Tools & Manufacture, 2003. 43: p. 925 - 936.

20. Z. Chen, Y. Gao, J. Lin, R. Su, and Y.Xie, Vacuum assisted thermal bonding of plastic capillary electrophoresis microchip imprinted with stainless steel template. Journal of Chromatography, 2004. A(1038): p. 239-245.

21. Saleh, and Teich, Fundamentals of photonics. 1999: Wiley.

22. S. Klimentov, T. Kononenko, P. Pivovarov, S. Garnov, V. Konov, A. Prokhorov, D. Breitling, and F. Dausinger, The role of plasma in ablation of materials by ultrashort laser pulses. Quantum Electronics, 2001. 31(5): p. 378-382.

23. P. Mannion, J. Magee, E. Coyne, G. O'Connor, and T. Glynn, The effect of damage accumulation behabiour on ablation thresholds and damage morphology in ultrafast laser micro-machining of common metals in air. Applied surface science, 2004. 223: p. 275-287.

24. M. Shirk, and P. Molian, A review of ultrashort pulsed laser ablation of materials. Journal of Laser Applications, 1998. 10(1): p. 18-28.

88 25. A. Ben-Yakar, and R. Byer, Femtosecond laser ablation properties of borosilicate glass. J. Appl. Phys., 2004. 96: p. 5316 - 5323.

26. M. Farsari, G. Filippidis, S. Zoppel, G. Reider, and Fotakis, Efficient femtosecond laser micromachining of bulk 3C-SiC. J. Micromech. Microeng, 2005. 15: p. 1786-1789.

CHAPTER 4

FEMTOSECOND LASER INTERACTION WITH DIELECTRIC MATERIALS

In this chapter, femtosecond laser interaction with dielectric materials is presented.

The first part is describing the ablation of electron spun poly-caprolactone and polyethylene terephthalate fiber by femtosecond laser. Its characteristics of light scattering in

PCL and PET is also described. The second part of this chapter is introducing a fabrication of microfluidic channels in soda-lime glass and fused quartz is described with polymer coating.

4.1 Poly-caprolactone (PCL) machining

4.1.1 Introduction

Electrospun nanofiber mesh [1-4] is widely used as a tissue engineering scaffold to support cellular in-growth and proliferation for generation of biological tissues [5-9].

However, the randomly-deposited structure of electrospun fiber does not allow for spatial control of cells necessary to produce complex organs in vivo. Cells typically grow only

on the exterior surface of electrospun fiber following random seeding techniques [2].

Direct-write femtosecond laser ablation is a flexible, rapid structuring technique that has the potential to alter this via precise control of surface topography to produce specific microenvironments that can influence cellular growth patterns. This article describes the development of procedures for the ablation of linear and shaped grooves and cavity arrays that are to be used in subsequent tissue growth studies.

PCL and PET are commonly used in biomedical applications. PCL is a FDA- approved biodegradable polyester. It has been used in biomedical applications in drug delivery devices, sutures and adhesion barriers and is being investigated as a scaffold for tissue engineering. It has a relatively low melting temperature (Tm ≈ 60°C). PET is

thermoplastic resin of the polyester family used biomedically under the trade name

Dacron® as a vascular graft. It has higher transition temperatures than PCL (e.g. Tm ≈

260°C). Specific polymers properties are listed in table 4.1; in this work, PET mainly provides a material comparison to PCL.

Polymer Melting Glass Thermal Heat Bulk ES Solid temp. trans. conductivity Capacity density mesh density 3 4 Tm (C) Temp. W/(m K) kJ/(kg K) g/cm density g/cm 3 Tg (C) g/cm PCL 58 ~ 63 −65 ~ 0.07* 1.3 1.14 0.24 1.2* −60 PET 130-137 75 0.13-0.15 1.3 1.3-1.4 - * starch * blend PLC/P CL copoly mer

Table 4.1 Physical properties of PCL and PET

No prior reports have been found regarding laser ablation of ES fiber meshes but

laser structuring of PCL films and membranes has been investigated. Thin films of

biodegradable polymeric materials, poly(ε-caprolactone) have been micro-patterned using a Ti-sapphire femtosecond pulsed laser and an ArF excimer UV laser in ambient conditions [10]. Characterization of femtosecond laser ablated PCL by hydrolytic degradation tests showed that the laser irradiation had little affect. The permeability of ultra-thin PCL produced by bi-axial stretching was enhanced by drilling using a femtosecond laser; the surface was rendered more hydrophilic (which typically enhances cellular adhesion) by excimer laser ablation texturing [11].

The 775 nm, 150 fs ultrafast pulsed laser was considered promising for electrospun fiber structuring because ablation with the short, high irradiance pulses

typically has minimal affects on the remaining material. The photon energy (1.6 eV) is

too small for photochemical ablation [12], but amplified femtosecond pulses can

nonetheless be strongly absorbed and produce clean ablation in many materials.

When energetic femtosecond laser pulses are incident on dielectric polymers, it is generally considered that nonlinear processes associated with the irradiance generate free electrons which are then responsible for increased absorption and consequent rapid heating, ionization, charge separation and removal of the absorbing material [12, 13].

Also, the thermal conduction length corresponding to the femtosecond laser pulse duration is generally small, so absorbed energy is not significantly diffused. It has been shown by a number of investigations that femtosecond pulsed laser ablation has fewer

affects on the remaining polymer surface than nanosecond pulsed ablation at comparable

wavelengths [14-17].

Electrospun meshes have some similarities to photonic bandgap crystals

consisting of periodically-spaced wavelength-sized structural units that have been widely

studied in recent years. In the case of ES fiber meshes, fiber diameters are also on the

order of the laser wavelength, but the sizes and locations are quite random. Thus, optical

interactions likely have similarities to that encountered in characterization of human

muscle tissue where radiation is highly scattered [18, 19].

This paper describes microscale laser structuring of electrospun (ES) polycaprolactone (PCL) and polyethylene terephthalate (PET) meshes. The fluence

required for ablation as well as the affect of fluence and focus spot scanning speed on the dimensions of ablated grooves are determined. The fluences are compared to estimates of threshold fluence for ablation of the bulk materials. Key issues for the ablation development include preserving electrospun structure post-ablation, determining practical limits on lateral resolution and depth of ablated structures and establishing net ablation

rates and process productivity. The structure of ES fiber remaining on the walls of

femtosecond laser-ablated grooves in PCL and the regularity of machined grooves are

compared to those of grooves produced by nanosecond laser ablation.

The fundamentals of femtosecond laser ablation have been summarized elsewhere

[13]. For our purposes we can consider single pulse removal of a material with threshold

ablation fluence Fth by a beam having a Gaussian radial fluence profile

2 2 0 (−= 2exp)( wrFrF 0 ) (4.1.1)

where w0 is Gaussian beam radius and F0 is maximum (or peak) fluence, related to pulse energy Ep as

2 0 = p /2 πwEF 0 . (4.1.2)

it is noted that 0 = 2FF av . If 0 > FF th , the ablation threshold fluence, the diameter D of the area near the center of the beam from which material is removed is then

2 2 = 0 (0 /ln2 FFwD th ). (4.1.3)

2 Fth can be found by curve fitting a semi-logarithmic plot of D measured for various F0 and calculating the intercept at zero fluence.

4.1.2 Experimental Apparatus and Procedure

Frequency doubled pulses from a mode-locked erbium-doped fiber laser intensified in a Ti:Al2O3 regenerative amplifier laser (CPA, Clark-MXR CPA2100) were used for this experiment. The maximum output power of the laser was Pav = 1.6 W, the pulse duration was Tp = 150 fs and the pulse repetition frequency was fp = 2 kHz. The

output power was adjusted in two stages. The majority of the attenuation was at the first thin-film polarizing beam splitter, where only a couple percent of the optical power passes through to a ½ waveplate. The beam polarization was adjusted by rotation of this optical element, allowing the additional attenuation at the second polarizing beam splitter to be adjusted. The attenuated laser beam was delivered by a beam delivery mirror train through a mechanical shutter and focused on the material. A 25 mm focal length achromatic lens (NA = 0.1), a 10x microscope objective (NA=0.25) and a 20x infinity- corrected microscope objective (also NA=0.25 with the 5 mm beam diameter) were the focusing optics used in the femtosecond laser experiments. To accurately measure processing power before experiments, a power meter was inserted after the focusing lens.

Prior to performing the experiments, the M2 beam quality parameter was measured with a

CCD camera and beam characterization software (Spiricon). The beam quality was M2 =

1.2 in the vertical Y direction and M2 = 1.3 in the horizontal X direction. The calculated minimum focus spot diameter for the femtosecond laser with 0.1 NA lens is approximately 6.4 μm and that of the 0.25 NA objectives is 2.6 μm. Primarily for purposes of comparison, a Q-switched Nd:YAG laser (TN-50, Spectra Physics) with Tp

= 120 ns at a pulse repetition frequency fp = 10kHz, maximum average power Pmax = 50

W was also used. The beam diameter of this laser was db = 1.9 mm and the quality factor was M2 = 15. The numerical aperture when the focusing with the 25 mm achromatic lens was NA = 0.037 and the estimated diffraction-limited spot diameter was d0 = 271 μm. All ablation of linear grooves was done at a scanning speed of 20 mm/s.

Electrospun PCL was produced using a 12 wt% solution of PCL (Mw 65,000,

Aldrich, St. Louis, MO) dissolved in acetone (Mallinckroff Chemicals, Phillipsburg, NJ) and pumped through a 20 gauge stainless steel blunt tip needle at a flow rate of 24 ml/hr under an applied voltage of 24 kV. The tip-to-substrate distance was 16 cm. The fiber diameter ranged from 0.2 μm to more than 2 μm. The distribution of fiber sizes is given in fig.4.1. The ES PET was prepared using 12wt% PET (Mw 18,000, Aldrich, St. Louis,

MO) in a 50/50 solution of trifluoroacetic acid/dichloromethane. Electrospinning voltage was 22kV, material flow rate was 18ml/hr and needle tip-substrate distance was 18cm.

The mechanical and electrical properties of PCL and PET (such as dielectric values and mechanical viscosity) affect the electrospinning process so the electrospun fiber diameter and fiber density are different. The density of the resulting mesh and other properties of the polymers are summarized in table 4.1.

9 8 7 6 5 Frequency 4 3 Frequency 2 1 0 1 2 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 More Diameter (μm)

Figure 4.1 Fiber diameter size distribution of ES-PCL meshes used in the experiments.

To better understand the penetration of laser and plume energy into the ES PCL, the directional hemispherical transmittance of different thicknesses of electrospun and solid PCL were measured using an integrating sphere (IS236A, Thorlabs, Newton, NJ).

The diameter of the samples used for these experiments was about 20 mm (compared to the unfocused beam diameter of 5 mm).

4.1.3 Results

Linear grooves produced by scanning the focused femtosecond and nanosecond laser beams over the surface of electrospun PCL are shown in fig. 4.2 as an initial comparison that established the benefits of the high intensity of pulses for ES nanofiber ablation. The power of both lasers and the Nd:YAG pulse frequency were varied in a series of experiments to determine feasible values for consistent ablation of microscale grooves. The average power of the femtosecond laser beam was varied between 5 mW and 200 mW while that of the Q-switched laser power was varied between 3W and 10 W; no ablation was found at less than 3W. The distance between individual pulses on the mesh surface was 10 μm for the femtosecond laser and 2 μm for the Q-switched laser.

Thus, the femtosecond laser pulses were not overlapped (spot diameter was only 6.4 μm) while the Q-switched laser pulse overlap was 99.3 % of the larger focus spot diameter and higher pulse frequency. The higher overlap of Nd:YAG laser with high frequency was caused by the limitation of linear motion system scanning speed.

(a) (b)

(c)

Figure 4.2 SEM images of (a) Q-switched laser and (b) femtosecond laser grooves. The width of the femtosecond laser groove is approximately 35 μm. (c) Q-switched laser power was varied from 3 W to 10 W, but no power that produced consistent ablation was determined

SEM images were used to measure the width of femtosecond-ablated grooves and to characterize the affects of groove ablation on the fibers remaining on the groove walls.

It was evident from these initial results the femtosecond laser ablated a relatively uniform groove with widths and depths that varied predictably with average power. Grooves

produced by the femtosecond laser had minimal melting of the remaining fiber at the edges of the ablation area. In contrast, the width of the groove produced by the Q- switched laser varied unpredictably from practically 0 mm to almost 1 mm. As shown in fig. 4.2(a), when it did ablate material, the Q-switched laser induced a high degree of fiber melting on the walls and bottom of the ablation area causing a high loss of the electrospun structure. At other times, in spite of the higher average power, practically no ablation was observed as shown in fig. 4.2(c). The power was varied between 3W and

10W and frequency changed between 5 kHz and 50 kHz, but no significant improvement in consistency of grooves size was found.

The difference in laser machining results obtained with the Q-switched laser and femtosecond lasers is attributed to the affect of variations in the electrospun mesh density on the different ablation mechanisms for the two lasers. As mentioned in the introduction, the high intensity femtosecond laser pulses input energy to the target material over a short time span and diffusion by thermal conduction is commensurately small. Thus, ionization and ablation tends to occur if the input fluence is over a distinct threshold magnitude. The energy input by lower intensity Nd:YAG laser pulses is converted to thermal energy in the material, which then melts and evaporates depending on the maximum temperature that is achieved. It seems likely that the unstable ablation behavior of the PCL nanofiber for Nd:YAG laser processing is related a transient increase of laser energy absorption associated with fiber melting.

To determine ablation properties electrospun mesh for femtosecond laser irradiation, grooves were created at different values of fluence by scanning with a focus

100

spot for varying laser pulse energy. The experiments were conducted with 300 μm thick meshes of PCL and PET on a glass substrate coated with 150 nm of transparent conductive film (ITO, Indium Tin Oxide) to allow electrospinning. The 0.1 NA focus lens was used for the ablation and scanning speed was 20 mm/s. The groove width and depth were measured for both materials and their variation with pulse energy are plotted in fig. 4.3 and 4.4. The widths of grooves formed are significantly wider than the focus spot size of 6.4 μm. It is also observed that the width and depth of grooves in PCL are somewhat larger than those formed in PET at the same laser fluence (examples are shown in fig. 4.5). The material removal rate for PCL mesh was calculated from the groove width, depth and scanning speeds shown in figures 4.3 and 4.4 to be in the range of 0.1 to

1 mm3/s. At the same femtosecond laser ablation fluence and scan speed, the volume removal rate of PET was from 1.5 to 3 times smaller. The difference in the material removal rate for the two polymers is at least partly attributed to differences in thermal properties.

101

300 ESPCL m) 250 μ ESPET 200 150 100 50 Channel width ( width Channel 0 0 10203040 Pulse Energy (μJ)

Figure 4.3 Width of grooves machined in ES nanofiber meshes at various pulse energies using the 0.1 NA optic and speed of 20 mm/s. The groove width is generally much larger than the focus spot size of 6.4 μm. Error bars indicate error in estimating groove dimension

102

300 ESPCL m) 250 μ ESPET 200 150 100 50 Channel depth ( depth Channel 0 0 5 10 15 20 Pulse Energy (μJ)

Figure 4.4 Depth of grooves machined in ES nanofiber meshes at various pulse energies using the 0.1 NA optic and speed of 20 mm/s showing that PCL grooves were slightly deeper. Error bars indicate error is estimating groove dimension.

103

(a) (b)

Figure 4.5 Comparison of grooves ablated in (a) PCL and (b) PET

Some melting of fibers at the edges of the machined grooves was produced by all of the tests described above. A comparison of two grooves ablated in PCL at relatively low pulse energies is shown in fig. 4. 6. It is obvious that melting and fusing of groove wall fibers is increased with pulse energy and that some melting of the mesh fibers was evident even when the surface was not deeply structured by the ablation. Several instances of thinned fibers and fusion of overlapped fibers are seen in this example. At least some melting similar appears to be difficult to avoid with the femtosecond laser used in these tests. The large grooves produced at high pulse energy may be attributed to melting and ablation of fibers beyond the radius of laser beam impingement by optical radiation from the plasma formed by interaction of laser radiation with fibers. Such a plasma is easily visible at higher fluence conditions and is discussed below. Melting of

104

fibers is also consistent with a conclusion that that femtosecond pulses tended to melt rather than ablate the nanofibers when fluence dropped below the ablation threshold. This affect was explored further in experiments discussed below. In any case, cell growth tests currently being carried out indicate that the minimal melting incurred when structuring nanofiber meshes with femtosecond ablation has no detrimental affect on the functionality of the mesh as a tissue scaffold.

(a) (b) (c)

Figure 4.6 Two grooves ablated in PCL nanofiber with the 0.1 NA optic at a speed of 20 mm/s and relatively low energy. (a) pulse energy = 1.5 μJ, fluence = 9.3 J/cm2); (b,c) pulse energy = 0.75 μJ, fluence = 4.6 J/cm2).

The affect of low-fluence femtosecond laser pulses was studied further by varying the distance from the 0.1 NA focus lens to the mesh surface during ablation. Such variations might be expected to occur in practice due to variations in mesh thickness and

105

errors in setting lens position. For producing defocused patterns, the focal point location was set 150 μm above the mesh surface, the laser power was 6mW and multiple scans were spaced 30 μm apart. As with all tests, scan speed remained at 20 mm/s. The calculated spot diameter of the defocused beam was 30.7 μm, roughly five times that of the focused beam. The corresponding laser fluence was 0.3 J/cm2. For comparison, grooves were ablated with a focused beam with power of 5 mW and fluence of 7.7 J/cm2.

(a) (b)

Figure 4.7 (a) Focused and (b) defocused (multiple passes, right) scans showing that the lower fluence due to beam defocusing melted rather than ablated the mesh.

As shown in fig. 4.7, irradiation with the defocused laser beam produced mostly undesirable melting instead of ablation of the fibers. In contrast, the higher fluence focused laser beam laser fluence in this test was larger than that the single pass groove was about 65 μm wide and minimal fiber melting was observed on the walls. However,

106

observed widths of grooves ablated in nanofiber mesh are not consistent with the usual ablation analysis outlined in the introduction where the ablation diameter varies logarithmically with the fluence. Based on the observation of a visible plasma during the femtosecond laser machining, it is believed that heating and melting of fibers by energy radiated from this plasma may be the contribute to wider-than-expected grooves. The formation of plasma above solid surfaces being machined by femtosecond laser radiation is a well-known limitation on the achievable precision of femtosecond laser ablation in air at ambient conditions and it has been well characterized by other researchers. The plasma expands rapidly shortly after the femtosecond laser pulse and is observed to persist for many 10’s of nanoseconds [20, 21]. Given the relatively low thermal conductivity, heat capacity and melting temperature of the electrospun PCL, it is expected that radiation from the plasma would be more effective for increasing the width and depth of grooves in it than in PET, in accordance with the experimental results.

The variation of the squared values of ablated groove width with fluence was analyzed to quantify the ablation threshold. This procedure is based on an assumption that the groove width and the ablation diameter of a single spot are equal. It has been shown in other thin film ablation work by the authors [22] that the width of with lines ablated in thin film with femtosecond pulsed lasers still followed the expected logarithmic trend. As shown in fig. 4.8, the ablation curves for electrospun PCL and PET mesh are non-logarithmic.

107

5.E+04 ES PCL 4.E+04 ES PET ) 2

m 3.E+04 μ ( 2 2.E+04 2 Fth ~ 4 J/cm Width 1.E+04

0.E+00 1.0 10.0 100.0 Laser fluence (J/cm2)

Figure 4.8 Groove width squared of nanospun polymer vs. fluence using the 0.1 NA optic showing non-logarithmic behavior. The PCL ablation fluence threshold suggested by the lower end of the fluence curve is larger than the expected bulk value.

The slope of the curves is still decreasing at the lower two values plotted for PCL, so the threshold fluence value of 4 J/cm2 was obtained by extrapolating these points find the x-axis intercept. In contrast, the expected ablation threshold value for bulk PCL [12] is less than 1 J/cm2. The difference is explained when one considers the large surface area of the fibers within a volume defined by the laser focus spot diameter and extinction depth, as discussed below. A detailed analysis of the ablation properties of polymer

108

nanofiber meshes for femtosecond laser pulses is the subject of current research by the authors.

To better understand the laser ablation characteristics of electrospun PCL, its absorption properties were measured. The results of directional hemispherical transmittance tests of electrospun and solid PCL are plotted in fig. 4.9. Since it is not a very highly scattering material when solid, the absorption coefficient for the solid PCL was approximated directly from this curve as 0.0013 μm-1, corresponding to an absorption length of about 770 μm. The decrease of mesh transmittance with thickness is well approximated by an exponential function and the extinction coefficient is about

0.0085 μm-1, corresponding to an extinction depth of approximately 120 μm. Thus, the fiber mesh extinction coefficient is much larger than the solid material absorption

= 3 coefficient, in spite of the fact that the measured mesh density was ρm 0.24 g/cm , about

20% that of the solid material density. If extinction were primarily due to absorption, the extinction coefficient for low-density nanofiber mesh would be expected to be less than that of the solid material. Thus, the large mesh extinction confirms that scattering rather than absorption is a dominant factor in the attenuation of laser radiation through the mesh.

109

120%

100%

80%

60% Fiber Solid 40% Transmittance (%) Transmittance 20%

0% 0 100 200 300 Thickness (μm)

(a)

100%

80%

60%

40% Reflectance (%) 20%

0% 0 100 200 300 Thickness (μm)

(b)

Figure 4.9 (a) Hemispherical transmittance of ES PCL nanofiber mesh and solid PCL and (b) reflectance of nanofiber mesh for an unfocused femtosecond laser beam. Error bars indicate estimated +/- 10 μm accuracy in mesh thickness measurement.

110

The groove depths produced by laser ablation were on the order of the measured fiber mesh extinction length of 120 μm, as would be expected. The groove widths are also of the same order as the extinction length. This is consistent with a hypothesis that radiated plume light is influential in determining the size of ablated features, since the lateral penetration of radiation would also be expected to be limited by scattering.

However, such a conclusion needs to be confirmed by measurements of extinction length is a direction parallel to the mesh surface since the mesh appears to be rather non- isotropic in the SEM images.

It is also interesting to revisit the ablation threshold estimated for the nanofiber mesh in view of the measured characteristic extinction depth of laser light. A typical fiber spacing of 2 μm is estimated from the fiber diameter distribution shown in fig. 4.1 and the fractional density of nanofiber mesh as compared to solid material. Given the that the incoming light is scattered multiple times within the extinction length of 120 nm, the fiber mesh before being absorbed, it is very plausible that the material surface area illuminated by a femtosecond laser pulse is much larger than the focus spot area and thus, the ablation threshold determined for fiber mesh is corresponding larger for the fiber mesh than for solid material. A detailed analysis of ablation of nanofiber mesh by femtosecond laser pulses is currently being carried out by the authors.

After the study described above, shaped grooves were ablated in the surface of electrospun PCL in preparation for specific tissue growth studies. As shown in fig.

4.10(a), laser ablation was used to form a 1 mm grid pattern with a microstructured cavity at the center of each grid cell. Ablation parameters were the same as those used to ablate

111

the straight groove in fig. 4.6. In fig. 4.10(b), a magnified view of a cavity being used for a cell culture test and a cluster of cells can be observed with growth constrained to the bottom of the cavity.

(a) (b)

Figure 4.10 (a) Matrix of microscale structures ablated in PCL nanofiber mesh and (b) magnified view of a laser-machined cavity in PCL nanofiber mesh (right) being used in a cell culture test.

112

4.1.4 Conclusions and future works

The femtosecond laser was effectively used for ablation of meshes of electrospun

PCL and PET polymers. Laser ablation resulted in uniform ablation grooves with width and depth controlled by femtosecond laser pulse fluence at constant scanning speed. It was observed that the maximum width of the grooves was generally much larger than the laser focus spot size. While some melting of fibers was observed on the edges of grooves ablated by the femtosecond lasers, the melting was minimal compared to that observed for grooves ablated with a Q-switched laser. The large widths of ablated grooves were attributed to material removal by optical radiation from laser-induced plasma. The extinction coefficient of PCL nanofiber meshes was measured and found to be high relative to the absorption coefficient of the bulk material. This indicated that scattering was the dominant mechanism for extinction, a fact that was used to explain the observation that the estimated ablation thresholds of the fiber materials were somewhat larger than the bulk materials.

Understanding the affects of multiple scattering of laser light inside the ES PCL is essential to understand the complete analysis of PCL fiber with femtosecond laser.

Preliminary results by a computer simulation of electromagnetic wave propagation inside the PCL fiber showed the wide scattering or light and further study to control this affect needs to be done as a future work.

113

4.1.5 References

1. P. Baumgarten, Electrostatic spining of acrylic microfibres. J. Colloid Interface Sci., 1971. 36: p. 71-79.

2. J. Lannutti, D.Reneker, T. Ma, D. Tomasko, and D. Farson, Electrospinning for tissue engineering scaffolds. Material science and engineering C, 2007. 27: p. 504-509.

3. D. Reneker, and I. Chun, Nanometre diameter fibres of polymer, produced by electrospinning. Nanotechnology, 1996. 7: p. 216-223.

4. K. Lee, H. Kim, M. Khil, Y. Ra, and D. Lee, Characterization of nano-structured poly( e-caprolactone) nonwoven mats via electrospinning. Polymer, 2003. 44: p. 1287-1294.

5. S. Kidoaki, I. Kwon, and T. Matsuda, Mesoscopic spatial designs of nano- and microfiber meshes for tissue-engineering matrix and scaffold based on newly devised multilayering and mixing electrospinning techniques. Biomaterials, 2005. 26(37-46).

6. W. Li, C. Laurencin,E. Caterson, R. Tuan, and F. Ko, Electrospun nanofibrous structure: A novel scaffold for tissue engineering. Biomedical Mat. Res., 2002. 60: p. 613-621.

7. H. Yoshimoto, Y. Shin, H. Terai, and J. Vacanti, A biodegradable nanofiber scaffold by electrospinning and its potential for bone tissue engineering. Biomaterials, 2003. 24: p. 2077-2082.

8. C. Xu, R. Inai, M. Kotaki, and S. Ramakrishna, Aligned biodegradable nanotibrous structure: a potential scaffold for blood vessel engineering. Biomaterials, 2004. 25: p. 877-886.

9. B. Min, G. Lee, S. Kim, Y. Nam, T. Lee, and W. Park, Electrospinning of silk fibroin nanofibers and its effect on the adhesion and spreading of normal human keratinocytes and fibroblasts in vitro. Biomaterials, 2004. 25(1289-1297).

10. C. Aguilar, Y. Lu, S. Mao, and S. Chen, Direct micro-patterning of biodegradable polymers using ultraviolet and femtosecond lasers. Biomaterials, 2005. 26: p. 7642-7649.

114

11. K. Tiaw, S. Goh, M. Hong, Z. Wang, B. Lan, and S. Teoh, Laser surface modification of poly(epsilon-caprolactone) (PCL) membrane for tissue engineering applications. Biomaterials, 2005. 26: p. 763-769.

12. T. Lippert, Interaction of photons with polymers: From surface modification to ablation. Plasma process polym., 2005. 2: p. 525-546.

13. J. Kruger, and W. Kautek, Ultrashort pulse laser interaction with dielectrics and polymers. Advances in polymer science, 2004. 168: p. 247-289.

14. S. Kuper, and M. Stuke, Femtosecond UV excimer laser ablation. Appl. Phys. B: Photophysics Laser Chemistry, 1987. 44: p. 199-204.

15. S. Kuper, and M. Stuke, Ablation of Polytetrafluoroethylene (Teflon) With Femtosecond UV Excimer Laser Pulses. Appl. Phys. Lett., 1989. 54: p. 4-6.

16. M. Womack, M. Vendan, and P. Molian, Femtosecond pulsed laser ablation and deposition of thin films of polytetrafluoroethylene. Appl. Surface Sci., 2004(221).

17. S. Baudach, J. Bonse, J. Kruger, and W. Kautek, Ultrashort pulse laser ablation of polycarbonate and polymethylmethacrylate. Appl. Surface Sci., 2000. 154: p. 555-560.

18. H. Dehghani, B. Brooksby, K. Vishwanath, B. Pogue, and D. Paulse, The effects of internal refractive index variation in near-infrared optical tomography: a finite element modeling approach. Phys. Med. Biol., 2003. 48: p. 2713-2727.

19. P. Waterman, Scattering, absorption, and extinction by thin fibers. J. Opt. Soc. Am. A, 2005. 22(11): p. 2430-2441.

20. B. Salle, O. Gobert, P. Meynadier, M. Perdrix, G. Petite, and A. Semerok, Femtosecond and picosecond laser microablation: ablation efficiency and laser microplasma expansion. Appl. Phys. A - Mater. Sci. Processing, 1999. 69: p. S381-S383.

21. S. Kilmentov, T. Kononenko, P. Pivovarov, S. Garnov, V. Konov, A. Prokhorov, D. Breitling, and F. Dausinger, The role of plasma in ablation of materials by ultrashort laser pulses. Quant. Electronics, 2001. 31: p. 378-382.

22. H. Choi, D. Farson, J. Bovatsek, A. Arai, and D. Ashkenasi, Direct-write patterning of ITO film by high pulse repetition rate femtosecond laser ablation. Applied Optics, 2007.

115 4.2 Laser machining of dielectric materials for biomedical

applications

4.2.1 Introduction

Glass dielectric materials have been widely used in many biomedical device

applications due to their desirable thermal properties and chemical stability [1-3].

Microfluidic architectures require precise fabrication control and chemical wet etching,

focused ion beam (FIB), reactive ion etching (RIE) and plasma processes have been used

as precision glass machining processes in clean room environments [4, 5]. Laser ablation with a scanned focus spot (direct-write ablation) offers important advantages over these

other material removal processes including flexibility and speed [6, 7]. It also does not require a clean room environment.

Laser ablation by pulses with duration on the sub-picosecond and femtosecond time scales can remove materials with lower residual thermal effects than nanosecond- long Q-switched laser pulses. Part of the reason for this low heat-input machining is the short time for conduction of energy from the laser interaction site by thermal diffusion.

Due to this and other ablation characteristics, femtosecond laser machining has been

found to produce superior results to conventional laser micromachining [1, 3, 7-9]. As a

non-clean room process, it also provides a flexible way to fabricate 3-dimensional

patterns by varying the beam scanning speed during ablation while clean room process

requires highly controlled environment and maintenance cost.

Due to limitations of regenerative amplifiers which require the amplification time,

116 electro-optics device control timing and etc., the pulse repetition rate of femtosecond

lasers was limited to several kHz. However, recent development of a high repetition rate,

high energy femtosecond lasers provides the ability to perform ablation with pulse rates

in the MHz range.

The mechanisms of laser-induced damage to dielectric material in the ultrashort

pulse regime have been studied by many researchers [10-12]. In transparent dielectric

materials, linear absorption of incident laser light is small. The mechanism of dielectric

material breakdown starts with the excitation of valence-band electrons by the incident

laser. The non-thermal component of the mechanism is associated with emission of

electrons which leads to softening and instability of chemical bonds. Then, positive ions

are ejected in a process which is known as impulsive coulomb explosion (ICE) [13-15].

For the initial ionization, high energy intensity is required. The peak power in

femtosecond laser ablation reaches intensities where multiphoton ionization (MPI) and/or

tunnelling ionization (TI) can create high electron densities and localized plasma

formation by avalanche multiplication leads to permanent damage and ablation results

[11].

The Keldysh number can be used to differentiate whether the photoionization is by MPI or TI [16-19] and can be calculated as Eq. (6) [20],

ω ε Emc ω γ = 0 g = 2E e FI g (4.2.1)

where ω is the laser frequency, I is the laser intensity, m and e are the mass and charge of an electron , c is the speed of light, n is the refractive index of material, Eg is the band 117 gap of the material, ε0 is the permittivity of the free space, and F is the electric field

strength. Multiphoton ionization occurs when Keldysh number is γ > 1, while tunnelling ionization occurs when γ < 1 [17, 21] Both soda-lime glass and fused silica are transparent in the visual spectrum [22], but the fused quartz has higher thermal stability and bandgap energy (Eg). The bandgap energies of soda-lime glass and fused quartz are

5eV and 8.9eV, respectively [23-25]. With laser intensity and material properties, we

calculated the Keldysh number to study the initial breakdown mechanism of dielectrics.

In Fig. 4.11, the Keldysh number is plotted with respect to applied laser intensities.

Figure 4.11 Keldysh parameters for soda lime glass and fused quartz. Shaded area represents the experiment region.

118

The laser intensity used for ablation is between 1.06 x 1014 W/cm2 and 3.5 x 1013

W/cm2 so the Keldysh number was below 1 for both soda-lime and fused quartz materials

at the upper end of the ablation range but above 1 at the lower end of the machining range.

Thus, we can conclude that the initial breakdown of dielectric material in this experiment

was be due to tunnelling ionization for some experiments but multiphoton ionization for

others, which provides an opportunity for an interesting comparison on the materials

affects for these two cases.

With the damage occurring at these nonlinear absorption mechanisms, the

absorptivity will increase as surface morphology and color center formation of the

material changes. An “incubation” term is used to explain the effectiveness of lowering

the ablation or damage threshold [26, 27]. The decrease of the ablation threshold is attributed to increase absorption due to accumulation of damage or defects from

individual pulses. In our previous work, at low pulse overlap, the ablation threshold of

ITO approached the single pulse ablation threshold fluence of 0.7 J/cm2 whereas at high

pulse overlap, the ablation thresholds approach the ITO damage threshold fluence of

~0.25 J/cm2 reported elsewhere [28].

One drawback of ultrafast pulsed laser ablation is a tendency to produce rougher

machined surfaces compared with non-pulsed processes [29] due to the removal of

discrete volumes of material. The surface roughness can be reduced if lower laser fluence

is used but at the expense of reduced ablation rate. In this work, we study the effect of laser focus variables on surface roughness channels on soda lime glass and fused quartz 119 and surface smoothening process by applying a poly-hydroxyethyl methacrylate (HEMA) coating for the reducing of the surface RMS roughness.

4.2.2 Analysis of Laser Ablation

Results needed for analysis of laser ablation are summarized below for convenience [14, 28]. It has been found that many materials have a distinct power density threshold for ablation by femtosecond laser pulses. Single pulse ablation experiments using known pulse energies allow this ablation threshold fluence Fth to be determined. A Gaussian radial fluence profile F(r) for an axisymmetric focused laser spot is

⎛ − 2r 2 ⎞ = FrF exp)( ⎜ ⎟ (4.2.2) 0 ⎜ 2 ⎟ ⎝ w0 ⎠

where w0 is Gaussian beam radius and F0 is maximum fluence. F0 is related to pulse energy Ep as

2E F = p 0 π 2 (4.2.3) w0

> If F0 Fth , the ablation threshold fluence, the diameter D of the area near the centre of the beam from which material is removed becomes

120 ⎛ F ⎞ 2 = 2 ⎜ 0 ⎟ wD 0 ln2 ⎜ ⎟ (4.2.4) ⎝ Fth ⎠

2 A semi-logarithmic plot of D versus F0 allows Fth to be calculated as the x-axis

intercept of a linear curve fit to the data. The slope of this plot allows calculation of the

effective radius of the laser focus spot. If there exists a threshold fluence, Fd, for visible

> damage, but not for removal of material and 0 FF d , then Fd can be found by a similar

procedure.

For scanned ablation, pitch p is defined as the distance between the centers of

adjacent pulses. Pitch was varied for different experiments but longitudinal and

horizontal pitches were always set equal. The horizontal pitch (corresponding to Y-axis direction, Fig. 4.12) was controlled by the motion control program. The longitudinal pitch

(corresponding to the X-axis or scan direction, Fig. 4.12) was adjusted by setting the

scanning speed s according to the relation

= s p (4.2.5) f p

where fp is laser pulse repetition frequency.

121

Figure 4.12 Definition of focus spot overlap LO.

The distance of intersection between the focus spots of adjacent laser pulses is

referred to as the focus spot overlap. The focus spot overlap for adjacent pulses when the

spot is scanned at a constant speed is illustrated in Fig. 4.12. The scanning speed, pulse

repletion rate, focus spot diameter, and focus spot overlap are related each other. From

the geometry, focus spot overlap LO can be expressed as

−= O pdL (4.2.6)

where d is focus spot diameter. Combining Eq.s (4.2.5) and (4.2.6) results in

−= s O dL (4.2.7) f p

If N > 1 laser pulses are incident on a location, both damage and ablation threshold fluences are less than for N = 1 and there is progressively greater reduction in thresholds with increasing N. This effect, termed incubation [27], is due to accumulation

122 of damage from individual pulses. The affect of incubation on ablation threshold can be

quantified by a relationship of the form

= ξ −1 th th )1()( NFNF (4.2.8)

where N is number of pulses incident on a given location and ξ is called the incubation

factor.

4.2.3 Experimental Apparatus and Materials

Our experiments used commercial Ti:Al2O3 femtosecond laser systems (CPA

2110 and CPA 2161, Clark-MXR) The two laser systems had identical characteristics

except as noted. The emitted laser beams had a central wavelength of 775 nm, pulse duration of 150 fs and pulse repetition frequencies of 2 kHz (CPA 2110) and 3 kHz (CPA

2161). Fig. 4.13 contains a detailed diagram of the optical beam delivery system.

123

Figure 4.13 Block diagram of beam delivery system setup (PBS: polarizing beam splitter, SHT: shutter, BD: beam dump)

The un-attenuated output of the lasers had pulse energy of over 0.5mJ, too large

for precision machining. The pulse energy was attenuated to the range from 50 nJ to 3.75

μJ by thin-film polarizing beam splitters (PBS) and a λ/2 waveplate. The laser beam was linearly polarized with polarization axis generally maintained normal to the scan direction. Some experiments were done to study the effect of polarization on the laser ablation. For these experiments, the beam polarization was rotated with a λ/2 waveplat so its axis was normal to the travel direction. Pulse energy ranged from 2.0 μJ to 3.75 μJ for channel fabrication and 667nJ for surface roughness and coating experiments. To set the

124 pulse energy for the experiments, laser output power was measured at the output of the focus optic with a power meter and the power divided by the pulse repetition frequency.

Thus, pulse energy of 667 nJ corresponds to average power of 2mW at 3kHz pulse repetition frequency.

The beam had a diameter of 5 mm, an M2 quality parameter of 1.3 and was

focused with a 20x microscope objective (M Plan Apo NIR 20X, Mitutoyo) and a 50x

microscope objective (M Plan Apo 50X NIR, Mitutoyo) as noted. The NA of the 20X

objective was 0.40 and the NA of the 50X objective was 0.42. Focus spot radius and can

be calculated by the relation

λ = 1 2 0 Mw (4.2.9) π N / A

The diffraction limited focus spot diameter d0 (twice the radius) was calculated to

be 1.6 μm for the 20x objective and 1.5 µm for the 50X objective lens. The propagation

of higher-order Gaussian beams characterized by a beam quality factor can be modelled

by the relation

2 ⎛ 2λzM ⎞ wzw 1)( += ⎜ ⎟ 0 ⎜ π 2 ⎟ (4.2.10) ⎝ w0 ⎠

125

where M2 is the beam quality factor, z is propagation distance from focus location, and λ is wavelength [30]. For the channel machining experiments, the focus lens was moved away from the material surface by 12.5μm from the focus location to yield a focus spot

diameter of 10µm. This spot size was used for the subsequent channel machining

experiments.

The laser system used a precision linear motor-driven rectilinear motion system

with 0.5μm encoder resolution on each axis to move the sample relative to the focus

point (X,Y axes) and to position the focus optic relative to the sample surface(Z axis).

The system was equipped with a coaxial vision system for adjusting the focus of the laser

beam relative to the material surface.

Laser ablation was done on glass and quartz substrates. Soda-lime glass substrates

(12-544-1, Fisher Scientific) had a nominal thickness of 500μm and fused quartz

substrates (GE 124, Technical Glass Products) had a nominal thickness of 1.6mm

(1/16”). The soda-lime glass was selected because it is widely used for biological

applications at affordable price. Fused-quartz is expensive relative to soda-lime glass but

is advantageous for some applications because it has higher softening temperature

(1683°C vs. 494°C). These and other material properties are listed in Tables 4.2 and 4.3

for soda-lime glass and fused quartz, respectively.

126 Supplier/Model Fisher Scientific Inc./ 12-544-1

Thickness 500μm

Softening Temp 720C

Density 2.48 g/cm3

Light transmission (VIS) 91.50%

Refractive index 1.517 @ 546.07nm

Poisson's ratio 0.2

SiO2-72%, NaO-14.3%, CaO-6.4% MgO- Chemical composition 4.3%

Table 4.2 Materia l properties of soda-lime glass [31]

127

Supplier/Model Technical Glass Products Inc./ GE 124

Thickness 1.6mm (1/16”)

Softening Temp 1683C

Density 2.2g/cm3

Light transmission (VIS) 93.0%

Refractive index 1.4585

Poisson's ratio 0.17

Impurities (ppm) Al<14, Ti<1.1, Zr<0.8, Ca<0.4, Li<0.6, F<0.6

Table 4.3 Material properties of fused-quartz [32]

4.2.4 Ablation Threshold Measurement

Single pulse ablation features were generated with the 20X focusing objective lens while scanning at 30 mm/s, the selected speed was high enough to separate the ablation cavities produced by individual pulses. The average diameters of the ablation cavities were measured from microscope images and the effective focus spot diameter and ablation threshold and were calculated. From the plot of squared ablation diameter versus energy shown in the Fig. 4.14(a), the slope of a line fitted to the data was 8.01.

2 From Eq. (4.2.4), the slope equals 2w0 and the effective focus spot diameter is 4µm. This

diameter was used to convert pulse energy to fluence and the X intercept of the squared

ablation diameter versus fluence plot in Fig. 4.14(b) was used to calculate ablation

128 threshold. The X intercept is -7.076 and the corresponding ablation threshold is 2.42

J/cm2. This measured value is in range with that found in other literature, 1.7J/cm2 –

4.0J/cm2 depending on the types of glass [33, 34].

14.00

12.00 y = 8.0093Ln(x) + 15.079 ) 2 10.00 (mm

2 8.00

6.00

4.00

Ablation width Ablation 2.00

0.00 0.10 1.00 Energy (μJ)

(a) 14.00

12.00 ) 2

m 10.00 μ ( 2 8.00

6.00 y = 8.0093Ln(x) - 7.076 4.00 Ablation width Ablation 2.00

0.00 1.0 10.0 100.0 Fluence(J/cm2)

(b)

Figure 4.14 Single pulse ablation diameter as a function of (a) pulse energy, (b) fluence.

129

4.2.5 Microchannel Ablation - Procedures and Results

Initial ablation experiments used the 2kHz pulse repetition rate and 20X focusing objective and spanned a wide range of laser fluences and scanning speeds. Cracking and irregular glass surfaces were observed at the higher fluences, possibly due to thermal radiation from plasma visible at the ablation spot. It was also noted during the initial experiments that some ablated material re-deposited in channels. To decrease re- deposition, argon gas was flowed over the ablation site at a rate of 39 x 10-6 m3/s to convect ablated vapor and debris away from the channel before condensation or re- deposition could occur. Based on feasible parameter ranges determined from these preliminary experiments, an experimental matrix was designed.

130

Sample ID Ep (uJ) Pitch (um)

#1 3.75 2.5 #2 3.75 5.0 #3 3.75 7.5 #4 3.75 10.0 #5 3.00 2.5 #6 3.00 5.0 #7 3.00 7.5 #8 3.00 10.0 #9 2.50 2.5 #10 2.50 5.0 #11 2.50 7.5 #12 2.50 10.0 #13 2.00 2.5 #14 2.00 5.0 #15 2.00 7.5 #16 2.00 10.0

Table 4.4 Experimental parameter settings

As shown in table 4.4, ablation experiments were performed at 16 different

parameter combinations. Laser pulse energy (Ep) was varied between 2.0 μJ and 3.75μJ and pulse pitch (p) from 2.5 μm to 10 μm. These pitches corresponds to scan speed (s) from 5 mm/s to 20 mm/s and the pitch in the lateral direction (normal to scan direction) was set equal to the scan direction pitch. A channel width of 200 μm was targeted for these experiments and the number of adjacent scan passes was adjusted to achieve the

131 target channel width in accordance with the pitch for that experiment. Since ablation

depth varied with pulse energy, the depth of ablation also varied for different

experimental settings. To achieve reasonable channel depths, each experimental

parameter combination was repeated to ablate 5 layers of material in the depth direction.

Focus location was adjusted downward by a 5 μm increment after each layer to maintain approximately constant focus spot size.

Sample results from the ablation experiment are shown in the macroscale scanning electron microscope (SEM) images in Fig. 4.15, which compare the overall affects of pitch and laser pulse energy.

Figure 4.15 Ablation pattern with respect to laser fluence and scanning speed

132 As expected, larger laser pulse energy and smaller pitch (larger overlap) ablated the material to a larger depth. As shown in the figures, the channels at the smallest pitch had the smoothest surface and the channels were scanned across the width by using a stylus profilometer (Dektak 3, Veeco) to quantify ablation depth and roughness. Scanned profiles of channels ablated at the low and high energy conditions are shown in Fig. 4.16.

30 25 5 Passes with 2.5um pitch 20 15 10 Depth (um) 5 0 1234 Energy (uJ)

Figure 4.16 Ablation depth as function of pulse energy for channels ablated at scan speed of 5 mm/s (2.5 µm pitch)

133 It is evident that the channel ablated at lower energy is smoother but shallower than the channel ablated at higher energy. More ablation experiments were done at lower energies of 1.5µJ and 1.75µJ to further study the affect of ablation energy on depth and roughness. A plot of microchannel depth as a function of pulse energy is shown in Fig.

4.17.

Sample #13 5 0 -5 0 100 200 300 400 -10 -15

Depth (um) Depth -20 -25 -30 X-direction (um)

(a)

Sample #1

5 0 -5 0 100 200 300 400 -10 -15

Depth (um) -20 -25 -30 X-direction (um)

(b) 134 Sample #11

-5 0 100 200 300 400

m) -10 μ -15

Depth ( Depth -20

-25

-30 X-direction (μm)

(c)

Figure 4.17 Profilometer scan across channels ablated at pulse energy and pitch (overlap) of (a) 2.00µJ, 2.5µm pitch (75% overlap); (b) 3.75µJ, 2.5µm pitch (75% overlap); (c) 7.5µm pitch (25% overlap).

The depth increased nearly linearly with energy over the range tested although.

The peak to peak roughness Rt of the bottom of the channel was also measured from this

scan data and is plotted in Fig. 4.18. The roughness monotonically with energy over the range studied and the decrease was linear in the range from 2.5µJ to 0.667µJ. This observation is notable since the Keldysh parameter crosses over the boundary from tunnelling ionization to multiphoton ionization at approximately 1.5µJ but there is not

any noticeable deviation from the linear trend in roughness at energies corresponding to

135 the different ionization mechanism. From this one could conclude that the ionization

mechanism has no affect on the ablation roughness for the material being used in these

experiments. This is certainly a reasonable observation since one would expect the fact

this ionization occurs at all to be more important than the mechanism by which it occurs.

It is noted that pitch and focus spot overlap are directly related to each other and also to

the number of pulses incident on a point on the material surface within the scan pattern.

3.0 m) μ 2.5 2.0 1.5 1.0 0.5 0.0 RMS ( roughness 0.0 1.0 2.0 3.0 4.0 Energy (μJ)

Figure 4.18 Peak-peak surface roughness versus pulse energy at a pitch of 2.5 µm.

In previous work cited earlier, the decrease of ablation threshold with number of incident pulses measured for a linearly-scanned focus spot and was found to obey the same incubation relationship as for the fixed fluence, fixed location experiment. A situation that is similar to linear ablation occurs when a focus spot is scanned over an area 136 with a fixed pitch. If a point is said to receive a laser pulse if it lies within one Gaussian

radius of the location of the beam center when the pulse occurs, some locations within the

area scan pattern are exposed to energy from more than one pulse. The total amount of

energy received depends on the location and focus spot overlap.

The importance of focus spot overlap and incubation was further analysed for 6 of

the experimental conditions that showed significant and repeatable effects for pitch. Since

the intensity distribution as described in Eq. (4.2.2) is Gaussian, the accumulated fluence

is a function of location and Gaussian distribution function. One can calculate the

accumulated fluence at a given point by the scanned laser focus point as

∞ ∞ 2 −+−− 2 2 yxF ),( = eF i i wyyxx 0 )/))()((2( ∑∑ 0 (4.2.11) ij=−∞ =−∞

th th where i and j are integer numbers, xi and yi are the relative central position of the i and j pulses and w0 is the radius of focused spot. The accumulated fluence as a fraction of the

single pulse peak fluence F/F0, was calculated and results are shown in Fig. 4.19 for 0%,

25%, 50% and 75% focus spot overlaps. The accumulated fluence increased from 1.0 at

0% overlap, 1.04 at 25%, 1.62 at 50% overlap, and 6.28 for 75% overlap. In terms of the

2 2 experiments, for F0 = 2.42J/cm , the total accumulated fluence was as high as 15.2J/cm .

137 -10 1 -5 0.8 0 0.6 5

Y location (um) 0.4

10 0.2 Fluence enhancement -10 -5 0 5 10 0 -20 -10 0 10 20 X location (um) X location (um)

(a) 0% overlap; max accumulated fluence fraction is 1

-10 1.5 -5

0 1

5 Y location (um) 0.5 10 Fluence enhancement -10 -5 0 5 10 0 -20 -10 0 10 20 X location (um) X location (um)

(b) 25% overlap, max accumulated fluence factor: 1.04

138 2 -10 1.5 -5

0 1

5 0.5 Y location (um) Y location 10 Fluence enhancement 0 -20 -10 0 10 20 -10 -5 0 5 10 X location (um) X location (um)

-10 6 -5 4 0

5 2 Y location (um)

10 Fluence enhancement 0 -20 -10 0 10 20 -10 -5 0 5 10 X location (um) X location (um)

(d) 75% overlap, max accumulated fluence factor: 6.28

Figure 4.19 Accumulated laser fluence (a) 0%, (b) 25%, (c) 50%, (d) 75%. The right plot is the cross section of fluence enhancement variation along with Y location =0μm.

139 These plots give an indication of the roughness that may be expected simply due

to intersection of ablation cavities produced by individual laser pulses. Roughness

produced by this mechanism would be expected to have a periodicity similar to that of

the intensity enhancement plots in Fig. 4.19. Indeed, the period of corrugation visible in the left half of the channel profile in Fig. 4.17(c) is close to that shown for the 25% overlap case in Fig. 4.19(b). The period of the minima in both cases is about 7.5µm. The match is not perfect because the phase of pulses if adjacent linear scans are subject to variation because of differences in starting time of the scan motion relative to the laser pulse waveform.

Figure 4.20 Accumulated fluence as a fraction of single pulse peak fluence vs. focus spot overlap. Arrows show 25%, 50%, 75% overlap.

140 The maximum accumulated fluence at points along a linear scan relative to the

pulse fluence is plotted as a function of focus spot overlap in Fig. 4.20. At the 75% overlap condition which produced the smoothest channels, the maximum accumulated energy is over 6 times the pulse energy at some points along in the channel. Although this value should be related to the magnitude of incubation at these points, it is still an open

question as to the effect of this accumulated energy on ablation threshold and further

study is needed.

Another advantage of higher overlap can improve the uniformity of laser fluence.

As shown in Fig. 4.29, there is a laser fluence variation at lower overlaps. The variation can be as Eq. (4.2.12) and results are plotted in Fig. 4.21.

− FF )( δ = max min (4.2.12) Fmax

141

Figure 4.21 Fluence variation vs. overlap

As we can see in Fig. 4.21, the variation of fluence at different locations is almost

negligible once the pulse overlap is kept 60% (less than 0.2%) or higher. So, it is obvious

that the fluence variation at 75% overlap is very negligible.

Separate experiments were conducted to quantify the effect of polarization angles

on the quality of channel ablation. The polarization was changed using a λ/2 waveplate.

The P-polarization is defined as the linear optical polarization axis being aligned with the scanning axis, and S-polarization as the linear optical polarization axis being perpendicular to the scanning axis. It was observed that a P-polarized light laser machining approach was a more effective method for ablation, but S polarization resulted

142 channels with a more square profile as shown in Fig. 4.22. Similar results were also reported during a high aspect ratio trepanned drilling process [35]. Presumably, improved machining occurs with P-polarization because absorption is more effective on the front wall of the channel formed as the beam scans over the material.

Figure 4.22 Optical images of the channel quality and shape with P-polarized beam (left) and S-polarized beam (right) on fused quartz.

143 4.2.6 Surface roughness and HEMA coating process

By the nature of scanned, pulsed laser ablation, material is removed as a series of

partially intersecting cavities so roughness is minimized by combinations of high focus spot overlap and low laser fluence. However, higher focus spot overlap and low laser fluence both decrease the material removal rate of the ablation process. By reducing the surface roughness of channels with a coating polymer, both channel quality and productivity can be optimized. The use of a coating also enhances design flexibility since the coating may be functionalized to change hydrophilicity or to reduce nonspecific binding of proteins.

In this section, we describe the surface roughness improvement achieved with

HEMA coatings. HEMA has been used to bond complex microfluidic devices and for surface modification [36] in devices produced by photolithography. HEMA has good optical properties as well, making it a good selection for surface modifying devices for biological, diagnostic, and other microscale applications that involve optical sensing or photochemistry.

Square 5mm x 5mm areas were prepared by laser ablation with a pulse repetition rate of 3kHz, using the 50X infinity corrected microscope focusing objective lens with a defocused 5μm focus spot diameter. The laser pulse energy was 667nJ which corresponding to 2mW at 3kHz of pulse repetition rate and scanning speed and lateral offset were adjusted to produce 25%, 50% and 75% focus spot overlaps. Some re- deposition of ablated material was noted in these large areas due to incomplete removal by the argon gas flow. Five micron areas were selected randomly for roughness

144 measurement, with smaller (1-2μm) subsets selected within the 5μm area also measured.

Localized roughness much larger or much smaller than the overall average could easily

be found and we selected area which can well represent the entire area of ablation.

To coat channels with HEMA, we followed a procedure similar to that described

by Lai et al. [36]. The procedure is described in Fig. 4.23. Hydroxyethyl methacrylate

(HEMA, monomer) and Phosphine oxide, phenyl bis (2,4,6-trimethyl benzoyl) (Irgacure

819 photoinitiator) were purchased from Aldrich (Milwaukee, WI). After ablating a

channel, a lid was bonded to the channel. Self-adhesive polymer films, thermally bonded

polymer films, polymer films bonded with UV-cured adhesive or unbonded lids all may

be used with this process. HEMA monomer mixed with initiator was then loaded into the

channel by vacuum suction. Next, either positive or negative (vacuum) gas pressure was reapplied until the gas bubble penetrated the HEMA along the entire length of the

channel. By adjusting gas pressure, coating thickness can be adjusted and the detail quantification is not made in this article. Gas pressure was continuously maintained as the

HEMA was cured under high intensity UV radiation resulting in a uniform, smooth coating. The improvement of surface coating is observed by optical microscope shown in

Fig. 4.24.

145

Figure 4.23 HEMA coating process

Figure 4.24 Optical images of HEMA coated channels (before and after)

146 To quantify the surface roughness improvement, we experimented HEMA coating on uncovered surfaces with 5mm x 5mm ablated area with different overlaps. To have uniform spreading of HEMA coating, one can apply spin coating or gas blowing method, and we used gentle gas blow after dropping HEMA on the surface. In this case, the thickness or uniformity control was much more challenging.

4.2.7 Surface roughness characterization

As shown in Fig.s 4.25 and 4.26, the surface roughness of soda-lime glass and fused quartz was measured with Atomic Force Microscope (AFM) by Asylum Inc. Both in AFM and by standard brightfield optical microscopy, individual ablation spots could be seen on the 25% overlap sample. Also in the case of 25% overlap, the depth of the holes could not be directly measured by AFM because the probe tip could not in all cases read the bottom of the hole. The pyramidal geometry observed at the bottom of the ablation holes is an artifact of the tip geometry and is not a read feature of the surface.

The RMS value of surface roughness was calculated within the custom AFM interface provided by Asylum, Inc. developed based on the Igor Pro data analysis platform.

147

Figure 4.25 Surface roughness data for different overlaps. The numbers inside parenthesis is the scanning area. E=667nJ, material: soda-lime glass

Figure 4.26 Surface roughness data (E=667nJ and different overlap) for fused quartz

The 25% overlap data represents the average roughness of areas that include holes; if holes were excluded, much lower values could be measured. The 50% and 75% data contains measurements of 5μm areas as well as a selected of smaller areas to capture the roughness variations within certain localized features. We speculate that the lower roughness for 75% overlap data is due to increased ablation and isotropic re-deposition of

148 some ablated material. AFM scans show blob-like features indicative of re-deposition on all images, with the 75% overlap images showing the most numerous.

Figure 4.27 Surface roughness data (E=667nJ and different overlap) for HEMA coated soda-lime glass. Two vertical scales are provided: The upper row shows near proportional vertical scale (5μm) hile the lower row shows a highly magnified vertical scale (500nm) to see features.

The HEMA coated samples show a uniform, thin coating of HEMA over the surface of soda-lime glass shown in Fig. 4.27. Large scale undulations may represent the contour of the underlying surface, or may be stress in the HEMA due to the high intensity 149 cure. Because the undulations are less noticeable in the 75% sample, it is likely at least some of the undulations are due to the underlying surface. The surface roughness

comparison is summarized in Fig. 4.28. The results displayed show suitable smoothness

for most biological and diagnostic applications.

Figure 4.28 Surface roughness data of soda-lime glass and fused quartz

150 4.2.8 Conclusions and future works

In this paper, we are presenting a non-clean process for dielectric material ablation by using femtosecond laser system. The developed technique was used to fabricate for microfluidic channels on dielectric materials (soda-lime glass and fused quartz) in conjunction with polymer coating for surface roughness modification. A systematic approach was applied to find an optimum window of ablation process parameters. Laser pulse energy was varied between 2.0μJ and 3.75μJ and the roughness of surface was measured to be between 295nm and 731nm RMS depending on both pulse energy and overlap. Surface roughness was improved to around 100 nm when the pulse energy was decreased to 667nJ. HEMA coating was effectively used to decrease the surface roughness to around 10 nm.

The characterization of coating thickness and surface roughness vs. vacuum pressure needs to be further investigated. Coating of carbons or hydrogel on quartz channel is planned to investigate its functionality. Other bio-compatible polymers or functional molecules coating could be used as polymer coating material and further study is needed.

151 4.2.9 References

1. Q.Chen, J. Longtin, S. Tankiewicz, S. Sampath, and R. Gambino, Ultrafast laser micromachining and patterning of thermal spray multilayers for thermopile fabrication. J. Micromech. Microeng, 2004. 14: p. 506-513.

2. J. McDonald, D.D., J. Anderson, D. Chiu, H. Wu, O. Schueller, and G. Whitesides, Fabrication of microfluidic systems in PDMS. Electrophoresis, 2000. 21: p. 27-40.

3. D. Gomez, I.G., I. Lizuain, and M. Ozaita, Femtosecond laser ablation for microfluidics. Optical engineering, 2005. 44(5): p. 051105:1-8.

4. L. Li, T.A., and M. Esashi, Smooth surface glass etching by deep reactive ion etching with SF6 and Xe gases. J. Vac. Sci. Tech (B), 2003. 21(6): p. 2545-2549.

5. C. Anthony, P. Docker, P. Prewett, and K. Jiang, Focused ion beam microfabrication in Foturan(TM) photosensitive glass. J. Micromech. Microeng, 2007. 17: p. 115-119.

6. X. Liu, D. Du, and G. Mourou, Laser ablation and micromachining with ultrashort laser pulses. IEEE journal of quantum electronics, 1997. 33(10): p. 1706-1716.

7. J. Cheng, M.Y., C. Wei, and Y. Chuang, and T. Young, Crack-free direct writing on glass using a low power UV laser in the manufacture of a microfluidic chip. J. Micromech. Microeng, 2005. 15: p. 1147-1156.

8. M. Farsari, G.F., S. Zoppel, G. Reider, and Fotakis, Efficient femtosecond laser micromachining of bulk 3C-SiC. J. Micromech. Microeng, 2005. 15: p. 1786-1789.

9. K. Venkatakrishnan, P.S., and L. Lim, Femtosecond laser ablation of thin films for the fabrication of binary photomasks. J. Micromech. Microeng, 2002. 12: p. 775-779.

152 10. B. Stuart, M. Feit, and S. Herman, Nanosecond to femtosecond laser-induced breakdown in dielectrics. Physical Review B, 1996. 53(4): p. 1749-1761.

11. C. Schaffer, A. Brodeur and E. Mazur, Laser induced breakdown and damage in bulk transparent materials induced by tightly focused femtosecond laser pulses. Meas. Sci. Technol., 2001. 12: p. 1784 - 1794.

12. D. Du, X.L., G. Korn, J. Squier, and G. Mourou, Laser induced breakdown by

impact ionization in SiO2 with pulse widths from 7ns to 150fs. Appl. Phys. Lett., 1994. 64(23): p. 3071-3073.

13. P. Mannion, J.M., E. Coyne, G. O'Connor, and T. Glynn, The effect of damage accumulation behabiour on ablation thresholds and damage morphology in ultrafast laser micro-machining of common metals in air. Applied surface science, 2004. 223: p. 275-287.

14. H. Choi, J.J., J. Nam, D. Farson, J. Lannutti, Structuring electrospun polycaprolactone nanofiber tissue scaffolds by femtosecond laser ablation. Journal of Laser Applications, 2007.

15. E. Gamaly, A.R., B. Luther-Davies, and V. Tikhonchuk, Ablation of solids by femtosecond lasers: Ablation mechnasim and ablation thresholds for metals and dielectrics. Physics of plasma, 2002. 9(3): p. 949-957.

16. G. Gibson, T.L., and C. Rhodes, Tunneling ionization in the multiphoton regime. Physical Review A, 1990. 41(9): p. 5049 - 5052.

17. C. Schaffer, Interaction of femtosecond laser pulses with transparent materials, in Physics. 2001, Harvard University: Cambridge, MA. p. 173.

18. G. Mainfray, a.C.M., Multiphoton ionization of atoms. Reports on progress in physics, 1991. 54(10): p. 1333 - 1372.

19. L. Sudrie, A.C., M. Franco, B. Lamouroux, B. Prade, S. Tzortzakis, and A. Mysyrowicz, Femtosecond laser-induced damage and filamentary propagation in fused silica. Physical Review Letters, 2002. 89(18): p. 186601.

153 20. L. Keldysh, Ionization in the eld of a strong electromagnetic wave. Soviet Physics JETP, 1965. 20: p. 1307-1314.

21. F. Ilkov, J.D.a.S.C., Ionization of atoms in the tunneling regime with experimental - evidence using Hg atoms. Journal of Physics B - Atomic molecular and optical physics, 1992. 25(19): p. 4005 - 4020.

22. J. Kruger, a.W.K., Femtosecond-pulse visible laser processing of transparent materials. Applied surface science, 1996. 96(98): p. 430-438.

23. M. Lenzner, J.K., S. Sartania, Z. Cheng, C. Spielmann, G. Mourou, W. Kautek, and F. Krausz, Femtosecond Optical Breakdown in Dielectrics. Physical Review Letters, 1998. 80(18): p. 4076-4079.

24. V. Bhardwaj, P. Rajeev, P. Corkum, and D. Rayner, Strong field ionization inside transparent solids. J. Phys. B: At. Mol. Opt. Phys., 2006. 39: p. S397-S407.

25. L. Shah, J.T., M. Richardson, and K. Richardson, Femtosecond laser deep hole drilling of silicate glasses in air. Applied surface science, 2001. 183: p. 151-164.

26. A. Rosenfeld, M. Lorenz, R. Stoian, and D. Ashkenasi, Ultrashort laser pulse damage threshold of transparent materials and the role of incubation. Appl. Phys. A, 1999. 69(Suppl.): p. S373-S376.

27. D. Ashkenasi, M.L., R. Stoian, and A. Rosenfeld, Surface damage threshold and structuring of dielectrics using femtosecond laser pulses: the role of incubation. Appl. Surface Sci., 1999. 150: p. 101-106.

28. H. Choi, D.F., J. Bovatsek, A. Arai, and D. Ashkenasi, Direct-write patterning of ITO films by high pulse repetition frequency femtosecond laser ablation. Applied Optics, 2007.

29. D. Yuan, and S. Das, Experimental and theoretical anlysis of direct-write laser micromachining of polymethyl methacrylate by CO2 laser ablation. Journal of Applied Physics, 2007(101): p. 024901-1:6.

154 30. M. Sasnett, Propagation of multimode laser beam-the M2 factor, ed. D. Hall, and P. Jackson. 1989, New York: Adam Hilger. 132-142.

31. Thermofisher scientific, Microslide specification. 2007.

32. Technical Glass Products, Properties of fused-quartz. 2007.

33. A. Ben-Yakar, and R. Byer, Femtosecond laser ablation properties of borosilicate glass. J. Appl. Phys., 2004. 96: p. 5316 - 5323.

34. A. Ben-Yakar, A. Harkin, J. Ashmore, R. Byer, and H. Stone, Thermal and fluid processes of a thin melt zone during femtosecond laser ablation of glass : the formation of rims by single laser pulses. J. Appl. Phys. D, 2007. 40: p. 1447 - 1459.

35. S. Nolte, C. Momma, G. Kamlage, A. Ostendort, C. Fallnich, F. Alvenslebe, and H. Welling, Polarization effects in ultrashort-pulse laser drilling. Applied physics A - materials science & processing, 1999. 68(6): p. 563-567.

36. S. Lai, X. Cao, and L.J. Lee, A Packaging techinque for polymer microfluidic platforms. Anal. Chem., 2004. 76: p. 1175-1183.

155 4.3 Femtosecond Laser Bulk Micromachining of Microfluid

Channels in PMMA

4.3.1 Introduction

Focused amplified femtosecond pulsed laser beam have very high energy densities

that are widely-used for surface microstructuring of many materials. Precision bulk or

internal machining of transparent materials is another interesting capability that is offered

by these short-pulse lasers. When an ultrafast laser pulse is focused inside of a transparent

material, non-linear and multiphoton affects associated with the high energy density can

cause absorption within a small volume of material at the focus [1]. At lower energy

densities, absorbed laser energy may modify the optical or other properties of the material

and such bulk material modification has been used to write waveguides [2, 3], and to store

data inside of material [4]. If the absorbed energy density is high enough, the material

becomes vaporized and ionized by the laser energy and its expansion causes a small pore to form at the interaction site [1, 5]. By translating the beam so that multiple pores are connected, microfluid passages can be machined inside of the material. Being a direct-write process, this technique of fabricating internal channels is useful for device prototyping and low-volume manufacturing.

Creation of internal channels for gas and fluid flow is important for fabrication of bio-medical devices. Channel fabrication with generally-available technologies requires multiple process steps: fabrication of external channels on a substrate as well as a cover plate and subsequent bonding of the cover to the substrate to form closed channels. Using

156 high energy femtosecond lasers pulses, it is possible to directly fabricate internal channels,

eliminating the fabrication steps needed to cover an external channel.

At fluences below the threshold for plasma formation, internal channel fabrication

by femtosecond laser irradiation followed by chemical etching has been recently

developed and fabrication of very clean internal channels was demonstrated. However, this

process is limited to materials whose chemical etching rate can be increased by absorbed

femtosecond laser pulses [6]. At higher fluences, internal micromachining of glass by femtosecond laser pulses has been investigated by several workers [4, 7-9]. Because

polymers are more useful for many biomedical microfluidic device applications [10],

creation of microfluidic channels in polymer by laser bulk machining has also been

studied, although not nearly as extensively [11]. No matter what the material, removal of

debris from the internal channel as it is being machined is a challenge. Remaining particulate or condensed material can clog the channels, preventing the creation of an open passage. In previous reports of internal machining in glass, liquid has been used to to assist in debris removal so as to prevent channel clogging. However, liquid-assisted material removal is complicated since it requires liquid-tight seals and fabricated channel sizes have

been found to be limited [12].

In this paper, an alternative method for the creation of microfluid channels inside of

poly(methyl methacrylate) (PMMA) is described. A gas flow system was configured to

partially-evacuate gas, vapor and fine debris from the channel as it was being created. The use of flowing gas for material removal is more convenient than liquid for practical reasons because it eliminates the need for leak-tight seals.

157 The femtosecond laser has unique advantages for material processing over

nanosecond or longer pulsed lasers due to the material interactions associated with the high

intensity that can be achieved when the beam is focused [12, 13]. At high intensity, the

laser pulse interaction in a transparent material is non-linear and self focusing can be

expected to decrease focus size below the un-affected dimensions . Also, when the pulsed laser energy is properly focused inside material, the irradiance incident on the surface of material is not high enough to damage the material while the high irradiance at the focus initiates absorption by generating free electrons via multiphoton ionization and tunneling ionization. Multiphoton processes decrease the effective size of the laser interaction zone

even further since photon density is a very sensitive function of distance from the center of

the focus. Thus, by these affects, the absorption volume is restricted to a small volume and the energy density in the absorbing material becomes so high that the temperature is above the ionization temperature [5]. The generated plasma evaporates the material and high

pressurized gas phase material expands and deforms the focused region creating a void. By

scanning the laser focus, the continued action of this pulsed ablation process can produce

connected voids, which will eventually make internal channel.

In our work, a capability for micro- and nanofluid channel fabrication in PMMA is

desired in chemical and biomedical applications. When a femtosecond laser beam is

focused inside PMMA material, the intense laser energy is absorbed and forms a plasma

inside the PMMA. The induced plasma locally compresses the solid material, resulting in

void formation. Unfortunately, it appears that laser ablation by-products in the internal

158 channel lead to excess plasma pressure and as a result, cracks and rough internal surfaces can be observed in the internal channels.

It is also useful to note that the ultraviolet (UV) laser ablation of PMMA has been extensively researched [14] and some of the results are pertinent to this work. A notable feature of ablation at wavelengths longer than 198 nm is incubation – visible damage does not occur until a number of laser pulses have been applied at the same location. By analysis of decomposition by-products that evolve from the material during the incubation phase, the reaction steps can be inferred. Although composition of the evolved species depends upon wavelength, some amount of volatile methyl-methacryalate (MMA) vapor is produced at both IR and UV wavelengths. The gas phase material can be absorbed by surrounded material or can remain inside the channel. To fabrication high quality channels of maximum length, material removal is one essential task to be solved.

159 4.3.2 Concept of gas assisted material removal:

The concept of gas assisted material removal has been developed based on the

reduced static pressure generated by high velocity fluid flow. In the configuration shown

in Fig. 4.29, a flow channel (2 ~ 5 millimeters in diameter) is first formed.

Figure 4.29 Concept of gas assisted material removal

The laser ablated internal microscale passage is then created starting at the

boundary of this flow channel. To analyze this process, we note that the flow of a

compressible gas can be modeled as incompressible flow if the pressure changes are less than 20% of mean pressure [15]. This assumption is valid in this application since the flow

160 gas is delivered at moderate pressure (345 kPa) and pressure in the fluid channel is on the order of but somewhat less than atmospheric pressure (101 kPa), based on the considerable laminar flow velocity that is developed. Based on incompressible fluid assumption,

Bernoulli's equation can be applied to correlate the pressure and fluid velocity in the channel. The gas phase material then can be effectively removed by pressure differential between a nearby gas flow channel and material ablation spot. The actual pressure development at the laser interaction site during the internal machining process is much more complicated than is suggested by the simple analysis due to generation and expansion of high temperature plasma. It is reasonable to assume that the gas or vapor produced by the ablation quickly reaches ambient temperature induced air expansion, although any remaining over-pressure due to the material vaporization will assist in its removal.

A vacuum to assist in material removal could also be created at the channel entrance using a vacuum pump. Femtosecond laser ablation forms not only gaseous products, but also monomer or oligomer forms of MMA. It is believed that either the static pressure from a flow gas or low pressure created by a vacuum system could be somewhat effective for material removal since particles or gaseous products produced by ablation would be induced to flow toward the lower pressure region in either case. However, oligomer or monomer phase particles may not be as effectively removed by a vacuum pump system. The flowing gas can more effectively evacuate removed material particles once they reach the flow gas channel.

161 4.3.3 Experimental

PMMA (Plaskolite™) plates having thickness of 2 mm were used for the

experiments. PMMA has mechanical properties and biocompatibility that make it useful

for a number of biomedical device applications and its transparency allows it to be

processed by internal machining. Polished steel pins were used in a hot-embossing process

to make smooth-walled gas flow channels. The surfaces of the prepared hot embossed

samples were also scratch-free, so that the laser beam could be focused inside material

without interference from the surface. Compressed Argon was selected as a flow gas

because of its inertness and ready availability. The delivery pressure was maintained at 374

kPa and the flow rate was varied between 0.63 ml/s and 2.6 ml/s by a manual flow control

valve.

Experiments were carried out by varying laser power, gas flow rate and focal spot

scanning speed. The laser used in the work was a Ti:Al2O3 chirp-pulse amplified laser which produced 150 fs pulses at a repetition rate of 2 kHz. The maximum output power of the laser was 1.6 W, corresponding to an average pulse energy of 0.8 mJ. Using two polarizing beam splitters and a rotatable half-wave plate to control the polarization angle of the linearly polarized beam, the pulse energy could be attenuated from this energy to a minimum of 5 nJ per pulse. The beam was focused for processing using a 10x microscope objective with numerical aperture of 0.6. The beam quality factor (M2) of the laser was measured to be 1.5, the wavelength was 775 nm and the nominal beam size was 5 mm.

Thus, the calculated minimum focus spot size is approximately 2.5 μm. To create channels, the pulse energy was varied from 0.25 μJ to 1 μJ and scanning speed was varied

162 from 0.2 mm/s to 1 mm/s. Using the calculated focal spot size, pulse repetition rate and scanning speed, the pulse overlap is calculated to range from 80% to 96%. A photograph of the experimental set up is shown in Fig. 4.30.

Figure 4.30 Experimental setup

To determine the effectiveness of gas flow for improving channel quality, three different conditions were tested. In the first experiments, which were done for comparison, internal channels were machined without venting. The second experiment was done with vent hole but no gas flow. In subsequent experiments, pressurized gas was applied through the gas flow channel at various flow rates.

163 4.3.4 Results and discussion

Channels were first created using a single-pass scanning procedure. Because the

application required wider channels than could be created with this technique, a multiple-pass procedure was also developed. Results from both procedures are described below.

Single pass experiments: It was found that for most effective debris removal,

single-pass channels had to be created by initiating machining at a free surface and then

translating the laser beam focus inward. Even with this procedure, when doing internal

machining at low pulse energies, the channel was easily closed after the focus had passed

by. At higher pulses energies, a continuous channel could be formed. It was observed that

as laser pulse energy decreased, the pore size was correspondingly decreased. To make

sure pores were connected each other to form a channel, the focal spot scanning speed

needed to be adjusted depending on pulse energy. Fig. 4.31 shows representative internal

channels fabricated without venting (Fig. 4.31a) and machined starting from the wall of a

vent hole that had no Argon gas flow (Fig. 4.31b). To prevent clogging of the channels by

debris, the pulse energy used to form these channels was relatively high and the scanning

speed was low.

164

(a) (b)

Figure 4.31 Channel fabrication without gas assistance (a): no vent, (b): vent but no gas flow. Both channels have very rough walls and the former has extensive adjacent cracking. Laser pulse energy was 1μJ and scanning speed was 0.2 mm/s

The surfaces of the channels shown in Fig. 4.31 appear to be quite rough. It is supposed that this effect might be due to violent explosion of combustible methyl methacryalate MMA vapor that results from decomposition of PMMA. In the second experiment, the vent hole was helpful to minimize the cracks. However, there were still a few clearly visible cracks adjacent to the channel, which appeared to be continuous but had very rough walls in the back-lit optical photographs.

165 In subsequent experiments, debris and gas/vapour removal was assisted by argon

with a delivery pressurize of 345 kPa applied through the gas flow channel. The channel shown in Fig. 4.32a was made at a flow rate of 0.65 ml/s (5 ft3/hr) while that in Fig. 4.32b was made at a flow rate of 2.6 ml/s (20 ft3/hr). By comparison of the results with those in

Fig. 4.31, it can be concluded that the channel shape was noticeably improved by the gas

flow, but the results depended on flow rate. Interestingly, increasing the gas flow rate from

0.65 ml/s to 2.6 ml/s degraded channel quality rather than improving it. With the Argon

flow rate near optimum of 1.3 ml/s, the channels appeared to be clear of debris and

relatively uniform for distances of approximately 5 mm to 10 mm from the vent hole. At

longer distances, the channels were more often clogged by debris and their walls became

rougher. This feasible machining length is long enough for some microfluidic channel

applications, although longer channels may be desired for many purposes.

166

Figure 4.32 Channel fabrication with gas assistance at different flow rates. Left: 0.65 ml/s (5 ft3/hr) right: 2.6 ml/s (20 ft3/hr) – Laser pulse energy = 1 μJ and scanning speed = 0.2 mm/s. Channels could be created using pulse energies of 250 nJ to 1.0 μJ and sc

It was considered that that the channel length might be increased by using a reactive flow gas instead of argon. If reactive gas could somehow be introduced to the ablation site,

the etching might be more effective. This situation is analogous to the RIE (Reactive Ion

Etching) process which involves plasma generation with a reactive gas and surface

material interaction [16]. Multiple photons from femtosecond laser irradiation are strong

enough to break carbon-carbon bonds and reactive gas such as oxygen could react with and

facilitate removal of ablated materials. Unfortunately, since the primary purpose of using

the flowing gas is to create low pressure to assist in removal of photo-decomposed material

167 from the ablation site, flow gas concentration at ablation site would be expected to be low

and flow gas reactivity may not be a significant factor.

Multi-pass experiments: The diameter of the single-pass channels was smaller than

desired for some applications. Additional experiments were conducted to modify the

process to produce larger channels. Instead of using a one-dimensional linear scan, a

two-dimensional zigzag pattern was generated. Both pitch and width of zigzag pattern

were set to approximately 15 μm. Processing parameters such as laser pulse energy and scanning speed were fixed while only gas flow rate was changed to determine its influence on channel quality. A photograph of the channels is shown in Fig. 4.33. A typical

maximum feasible channel length with this method was found to be between 5 mm and 10

mm, which is the same as could be achieved with the single-pass machining technique.

168

Figure 4.33 Plan view of channels created with multiple passes– from the left 0.65 ml/s (5 ft3/hr), 1.3 ml/s (10 ft3/hr), 2.6 ml/s (20 ft3/hr), Laser pulse energy = 1μJ and scanning speed = 0.2 mm/s

For a better measurement of channel profiles, the samples were cut and channel cross sections were viewed with a SEM. The samples were cut approximately 5 mm from the inlet of the entrance where the channel was believed to be fully developed. As shown in

Fig. 4.34, 0.65 ml/s gas flow resulted in best channel quality. Although a channel cross section is not displayed, optical photographs showed that the channel was not clearly developed at 2.6 ml/s of flow rate. The channel width varied between 15 μm and 25 μm.

169

Figure 4.34 Cross sectional view of channels (left – 0.65 ml/s, right-1.3 ml/s).

To correlate the channel profile with flow rate, the Reynolds numbers were calculated and plotted in Fig. 4.35. By comparing the Reynolds number and channel profile, we can conclude that laminar flow (Re < 2300) region provided the best effects, presumably because of more effective ventilation.

170 8000

6000

4000

2000

Reynolds number 0 0123 Flow rate (ml/s)

Figure 4.35 Reynolds numbers in the flow channel for 0.65 ml/s, 1.3 ml/s, 2.6 ml/s

To better characterize the shape and roughness of the machined channels, some of the multipass channels were photographed after a fluorescent dye (Flouresin yellow) was flowed through them. Fig. 4.36 shows fluorescence background of the channels under UV light (365nm) using inverted fluorescence microscope (TE-2000S, Nikon).

171

Figure 4.36 Micrograph of internal channels before being filled with a liquid containing fluorescent dye.

The two bright lines indicate that the femtosecond laser machining created open passages, allowing the dye to flow in and change the fluorescence background of the ablated area. Fig. 4.37 show the fluorescence intensity difference before and after the channels were filled with fluorescence dye. It can be seen that the intensity increased, also indicating the channels are open. Although the distribution of dye is relatively uniform, indicating that the channel walls are probably relatively smooth, several pockets of brightly fluorescing trapped dye may reveal surface discontinuities.

172

Before FITC (acid-yellow) is added After fluorescence dye is added:

Figure 4.37 Flourescense spectra confirming presence of yellow dye in a channel.

173 4.3.5 Conclusions and future work

A concept for evacuating ablation by-products during the fabrication of internal

channels has been described and demonstrated. The static pressure of a flowing gas in a

vent channel provided pressure difference for evacuation of small debris and vapour the

channel. Experiments showed that the evacuation of gas from the channel was effective for

improving channel quality. In the experiments, argon gas was used and the venting

technique resulted in visually smooth channels with no obvious cracking. A minimum diameter of 2μm and maximum diameter 20 um were achieved in these experiments. It is believed that the maximum feasible diameter of channels that can be fabricated by this technique was not determined by the experiments. It is also found that laser power, gas flow rate and traveling speed are correlated and need further study. The maximum length of smooth channel was found to be 5 mm ~ 10 mm which is long enough for some microfluidic channel applications.

174 4.3.6 Reference

1. D. Du, X. Liu, G. Korn, J. Squier, and G. Mourou, Laser-induced breakdown by impact ionization in SiO2 with pulse widths from 7ns to 150fs. Appl. Phys. Lett., 1994. 64(23): p. 3071-3073.

2. K. Davis, K. Miura, N. Sugimoto, and K. Hirao, Writing waveguides in glass with a femtosecond laser. Opt. Lett, 1996. 21(21): p. 1729-1731.

3. K. Miura, J. Qiu, H. Inouye, T. Mitsuyu, and K. Hirao, Photowritten optical waveguides in various galsses with ultrashort pulse laser. Appl. Phys. Lett., 1997. 71(23): p. 3329-3331.

4. R. An, Y. Li, Y. Dou, H. Yang, and Q. Gong, Simultaneous multi-microhole drilling of soda-lime glass by water-assisted ablation with femtosecond laser pulses. Optics express, 2005. 13(6): p. 1855-1859.

5. C. Schaffer, A. Brodeur, J. Garcia, and E. Mazur, Micromachining bulk glass by use of femtosecond laser pulses with nanojoule energy. Opt. Lett, 2001. 26(2): p. 93-95.

6. Y. Bellouard, A. Said, M. Dugan, and P. Bado, Fabrication of high-aspect ratio, micro-fluidic channels and tunnels using femtosecond laser pulses and chemical etching. Optics express, 2004. 12(10): p. 2120-2129.

7. D. Hwang, T. Choi, and C. Grigoropoulos, Liquid-assisted femtosecond laser drilling of straight and three-dimensional microchannels in glass. Appl. Phys. A - Matl. Sci. & Processing, 2004. 79(3): p. 605-612.

8. A.Salleo, T. Sands, and F. Genin, Machining of transparent materials using an IR and UV nanosecond pulsed laser. Appl. Phys. A - Mater. Sci. Processing, 2000. 71(6): p. 601-608.

9. K. Ke, E. Hasselbrink, and A. Hunt, Rapidly Prototyped Three-Dimensional Nanofluidic Channel Networks in Glass Substrates. Anal. Chem., 2005. 77(16): p. 5083-5088.

175 10. B. Ratner, A. Hoffmann, F. Schoen, and J. Lemons, Biomaterials science, An Introduction to Materials in Medicine. 1996, San Diego: Academic press.

11. T. Kim, K.C., A. Groisman, D. Kleinfeld, and C. Schaffer, Femtosecond laser-drilled capillary integrated into a microfluidic device. Appl. Phys. Lett., 2005. 86(20): p. 201106.

12. C. Schaffer, Interaction of femtosecond laser pulses with transparent materials, in Applied Physics. 2001, Harvard University: Cambridge.

13. M. Richardson, A. Zoubir, C. Rivero, C. Lopez, L. Petit, and L. Richardson, Femtosecond laser micro-structuring and refractive index modification applied to laser and photonic devices. SPIE, 2003. V: p. 15347-4.

14. T. Lippert, Interaction of photons with polymers: From surface modification to ablation. Plasma process polym., 2005. 2: p. 525-546.

15. C. Geankoplis, Transport Processes and Unit Operations. 3rd ed. 1993, Englewood Clifs: Prentice Hall.

16. C. Zhang, C. Yang, and D. Ding, Deep reactive ion etching of commercial PMMA in O2/CHF3, O2/Ar based discharges. J. Micromech. Microeng, 2004. 14: p. 663-666.

176

CHAPTER 5

MEASUREMENT OF FEMTOSECOND LASER SCATTERING IN ELECTROSPUN POLYCAPROLACTONE NANOFIBER

5.1 Introduction

Poly-caprolactone (PCL) is a Food and Drug Administration approved biodegradable polyester that has been used for many biomedical applications [1]. In tissue engineering, electrospun (ES) nanofiber mesh of poly-ε caprolactone (PCL) [1-4] is widely used as a tissue scaffold to grow cells and to proliferate biological tissues [5-9].

The ideal scaffold requires appropriate structures such as various sizes of porosity, sufficient surface area and a variety of surface chemistries for cell adhesion, growth, migration, and differentiation and a degradation rate that closely matches regeneration rate of the desired natural tissue [3]. To satisfy these requirements, laser ablation is being investigated for producing micro-scale structures on ES-PCL fiber meshes.

In a previous article [10],femtosecond laser ablation was used to produce various structures on ES-PCL tissue scaffold and it was used for cell growth. The ultimate goal of femtosecond laser ablation is to preserve “neotissue” environment for cell growth after

177 structuring the tissue scaffold [3, 10]. Femtosecond laser with a central wavelength of

775 nm and pulse width of 150 fs, was considered to be a promising process for ES fiber

structuring because of its non-thermal interaction, and high peak irradiance for high

precision machining as well as non-clean process. Direct-write femtosecond laser

ablation process can provide a flexible and rapid structuring of tissue scaffold, and has

potential to control surface topography to produce specific microenvironments that can influence cellular growth patterns. It was also noted that the interaction of the laser radiation with ES fiber material was quite different from that usually observed for solid materials. The width of channels ablated by a scanned focus spot (10μm ~ 50μm) was dramatically greater than the focus spot diameter (~ 1.6μm). Based on the results, it was concluded that absorption and scattering of radiation within the fiber mesh was significant and that multiple scattering needed to be further analyzed for a predictive understanding of the ablation process and structures produced by it.

It is noted that absorption and scattering of optical radiation within an ES polymer bears similarity to other previously published analyses for different applications. Some biological tissues have a fibrous structures and their interaction with light has been studied by many researchers [11-16] because it provides much useful information for treatments [17]. The knowledge of the wavelength dependence of scattering is also useful for the determination of structural features of tissue. Photonic crystals consisting of

periodically-spaced near wavelength-sized structural units have been widely studied

recently. The diameters of the ES PCL fibers used in this work are also on the order of

the laser wavelength (775nm). However, the sizes and locations are much more random than is usually considered in the analysis of photonic crystals. 178 As shown in Fig. 5.1, when light irradiates a sample of a material, it may be reflected, absorbed, and/or scattered inside and outside of the samples. The reflectance and transmittance of the sample are relatively easily measured, but differentiation of

absorption and scattering inside of the sample is more difficult. Furthermore, during laser

ablation of ES polymer meshes, interaction with the material is non-linear , so the

absorption of light is exponentially decreases as thick thickness of sample increases. Also,

in the cases of interest for this work, the dimensions of the sample are large relative to the

fiber spacing so incident rays of light are scattered multiple time until they are completely

absorbed by material or escape from the sample. In cases where scattering is an important

component of the material interaction with a material sample, the forward- and backward-directed radiation is much more divergent than the incident beam so measurement of the corresponding power is best done by an integrating sphere to ensures capture of all radiation [13, 15, 16].

179

Figure 5.1 Light propagation inside and outside of a fiber mesh supported by a transparent substrate.

For convenience, a number of the relevant results from analysis of scattering are briefly summarized below. The reader is referred to the cited works for details. Scattering of light from a single particle in a surrounding medium depends on the size of particle and wavelength of light in the medium. Mie theory is applicable for single scattering where wavelength and particle size are comparable[18-20] and are generally valid for homogeneous, isotropic and optically linear materials irradiated by an plane wave of infinite extent. Analytical solutions for scattering from a single cylinder of infinite length are based on the summations of terms involving functions

' ' nn − nn xJymJxJyJ )()()()( tanαn = ' ' (5.2) nn − nn xNymJxNyJ )()()()(

and

180 ' ' nn − nn xJyJxJymJ )()()()( tan βn = ' ' (5.3) nn − nn xNyJxNymJ )()()()(

where αn and βn are phase angle for TE and TM polarization respectively as defined in

Fig. 5.2, Jn is Bessel function of the first kind of order n, and Nn is the second Hankel

function of order n. Dimensionless coordinate positions x and y in (5.2) and (5.3) are

defined as

2πa x = (5.4) λ

2πa ⎛ n1 ⎞ and y = ⎜ ⎟ (5.5) λ ⎝ n0 ⎠

where a is the radius of the cylinder, n1 is refractive index of the cylinder material and n0

is index of the surrounding material.

Figure 5.2 Definition of TE and TM modes (E: electric field, M: magnetic field, k: wave vector)

The extinction efficiencies for single scattering of radiation with TE and TM polarizations can be calculated as 181 2 ∞ Q −extTE = ∑ bn )Re( (5.6) x n −∞=

2 ∞ Q −extTM = ∑ an )Re( (5.7) x n −∞= where

tanαn an = , (5.8) tanαn − i and

tan βn bn = (5.9) tan βn − i

Analysis of multiple scattering requires solutions based on the radiative transfer equation as described in Eq. (5.10) [21]

dI Es ''' μ sa )( IEE ++−= )(),( dIp ΩΩΩΩ (5.10) dz 4π ∫4π

where, μ: the cosine of the angle θ between direction Ω and the surface normal of the sample, I: intensity, p: phase function, Ω’: all direction.

An early solution that is based on a two-flux assumption (radiation is separated into forward- and backward-directed components) within a scattering medium is sometimes referred to a Kubelka-Munk theory. It was originally developed to describe

182 multiple scattering from paint films but also applies in other applications [22-28].

Assuming the case of isotropic scattering, matched boundaries, and diffuse irradiation at the sample surface, the back reflection from an infinitely-thick sample of material R∞ is expressed as [22],

K K 2 K R∞ 1 +−+≈ 2 (5.11) S S 2 S

where S is the K-M absorption coefficient and K is the K-M scattering coefficient.

Even though the two-flux theory is a simple and a popular method for analyzing the optical properties of scattering layers, the assumption of diffuse irradiation does not allows analysis of scattering from a sample illuminated with collimated laser light [27].

Mudgett et al derived a 22 flux approximation and that related the two-flux KM coefficients, K and S, to physical absorption (μa) and scattering coefficients (μs) described in Eq. (5.12).

+− μμ sa )( z )( = 0eIzI (5.12)

The calculations can be as follows [23, 25]

183 K 2 μa ≈ (5.13) − aaS 10 4/)3( μ s

where, a0 and a1 are phase function coefficients, for isotropic material a0 =1 and a1 = 0

Three-flux approximation using the equation of radiative transfer was used to

calculate the scattering and absorption coefficients [23, 29, 30]. Burger et al, generalized

Kubelka-Munk theory using three flux approximation and they came up with an identical

results with Mudgett et al, for diffused irradiation. The following relationship can be

made for collimated or directional illumination [29]

μa K ≈ 27.0 (5.14) μ s S

For analysis of laser ablation, differentiation between scattering coefficient and

absorption coefficient is important and is the subject of the analysis described in this

paper. In this article, procedures of integrating sphere measurements and analysis to estimate absorption and scattering coefficients, determination of ablation threshold for

the fiber material, and widened ablation spot diameter which is believed to be caused by

multiple scattering mechanism of light inside ES PCL fibers.

5.2Experiment setup

184 PCL fibers were electrospun from a 12 wt% solution of PCL (Mw 65,000, Aldrich,

St. Louis, MO) dissolved in acetone (Mallinckroff Chemicals, Phillipsburg, NJ) and

pumped through a 20 gauge stainless steel blunt tip needle at a flow rate of 24 ml/hr

under an applied voltage of 24 kV. Fiber was deposited onto flat glass substrate coated

with a thin (~150nm) conductive film (ITO, Indium Oxide Tin Oxide). The thicknesses

of electro spun fibers varied between 30 μm and 130 μm.

The mechanical and electrical properties of PCL such as dielectric values and

mechanical viscosity, affect the electrospinning process so the ES fiber diameter and

fiber density cab vary. The density of the resulting mesh and other properties of the

polymers are summarized in Table 1.

The sizes of deposited fibers are randomly distributed between 200nm and 2μm.

However, fibers which diameters are bigger than 1μm tend to have irregular diameters which are caused by instability of stream of liquid-state fiber jet. The volume ratio which is defined as the ratio of fibers to a unit volume is estimated to be 20% and the distribution of fiber diameters are counted and the results are shown in Fig. 5.3.

185 Figure5.3 Electro spun fibers (left) and fiber diameter distribution (right)

ES Melting temp. Glass trans. Thermal Heat Bulk mesh Solid density T (C) Temp. conductivity Capacity density m density g/cm4 T (C) W/(m K) kJ/(kg K) g/cm3 g g/cm3

58 ~ 63 −65 ~ −60 0.07* 1.3 1.14 0.24 1.2

Table 5.1 Material properties of PCL

Femtosecond laser (Clark-MXR, CPA-2110) was used as a collimated light source which was used to fabricate a tissue scaffold structuring. The laser was amplified

by a Ti:Al2O3 regenerative amplifier operating at a central wavelength of 775 nm, pulse

duration of 150 fs, and a repetition frequency of 2 kHz / 3kHz / 6kHz. An average power

of 1.6watt (2kHz) and 2.6watt (6kHz) was attenuated to several mW with a series of thin-

film polarizing beam splitters (PBS) and wave plates. The beam had a diameter of 5 mm,

an M2 quality parameter of 1.3 and was directly applied on the sample without going

through focusing optics. The linear polarization axis was normal to the scan direction and a half-wave plate optic was used to change the polarization.

The measurements of the transmittance and reflectance were done with an

integrating sphere technique. The integrating sphere is a general purpose sphere for many

applications like laser power, transmittance, and reflectance measurements. The

transmitted or reflected lights were collected by an integrating sphere (IS236A, Thorlabs, 186 Newton, NJ). It has a diameter of 2” and has an inner surface covered by highly reflecting

Poly- Tetrafluoro Ethylene (PTFE) based high reflective bulk material, collecting the transmitted or reflected light flux from a thin tissue sample. The scattering by the integrating sphere transfers the directed transmitted and reflected light to diffuse light fluxes, being independent of the direction, and thus allows it to be probed by a small photo diode at one single position. The port for the direct connection of photodiodes

(SM05PD , Thorlabs Inc.) that can measure a wide wavelength range (300nm ~ 1100nm) and both anode and cathode grounded polarities, has been specially arranged in order to prevent direct irradiation of the detector even under large divergent input beams [31]. The photodiode converts the light intensity to current and trans-impedance converts current signal to voltage signal. Oscilloscope (TDS3012B, Tektronix) is used to record the signal at 1.25GHz sampling rate and provides the relative intensity of transmitted or reflected lights. The detail system setup is depicted in Fig. 5.4.

187

Figure 5.4 System setup for transmittance and reflectance measurements

As shown in Fig.5.5, samples were mounted in the inlet or the outlet port of the

integrating sphere for transmittance and reflectance measurement of samples,

respectively. The sensor was calibrated to convert the transmittance. To calibrate the

sensor, the incoming laser beam was measured with a calibrated power meter (PM120,

Thorlabs, Newton, NJ) and calibrated with the output voltage from integrating sphere. To

see the saturation of signal, laser beam was increased up to 20mW and input power and

output sensor signal was linearly correlated and no saturation was observed. To measure

the maximum transmittance (100%), laser beam was applied without any samples and the output signal was measured. With a sample, the output signal decreases as the thickness

188 of sample increases. The ambient light was also turned off for zero calibration at 0%

transmittance.

For reflectance measurement purpose, the laser beam needed to be re-directed to

the sample, Fig.5.5 (b), with a reflecting mirror. To measure the 0% of reflectance, the output port was open so that all laser light exits to ambient which replicates 100% of absorption or 0% reflectance. To have 100% of reflectance calibration, the output port was blocked with a cap with same material as bulk material coating. The difference of two signals, with cap and without cap, provided us the scale of reflectance.

Figure 5.5 System setup for transmittance and reflectance measurements

189 5.3Results

As shown in Fig.5.6, the input power and output power of integrating sphere was compared with different thickness of PCL samples. The incident power was varied from

0.7mW to 8mW by using a half wave plate rotation and PBS. The slope was very linear

over the entire range of incident power and the slope represents the transmittance of

sample at the give thickness. The polarization was rotated by 90 degrees, but the

measurement was not dependent on laser beam polarization.

9 40um 8 50um 7 130um 30um 6 65um 90um 5 "no sample" 4 3

Output Power (mW) Power Output 2 1 0 0246810 Input Power (mW)

Figure 5.6 Output power vs. input power of samples

190 Since the thickness varies within the same sample, the experiments were done

multiple times at different locations in the same sample. The thickness measurement of

samples was done with micrometer, and variation was observed to be +/-10μm which

includes measuring force variation and sample thickness variation within the sample. The

transmittance was plotted as shown in Fig.5.7. The transmittance decreases exponentially and the result was compared to that of solid PCL’s. The solid PCL is prepared by melting the fiber meshed PCL.

By curve fitting, the slope of transmittance decay, the extinction coefficients (μt =

-1 -1 μa + μs) were estimated as 0.0013 μm and 0.0085 μm for solid PCL and fiber PCL,

respectively. In our analysis, it is assumed that the transmittance is 100% at zero

thickness. However, the zero thickness assumption is still a pending question.

Figure 5.7 Transmittance measurement for fiber PCL (circles) and solid PCL (squares) 191

As the same manner, the reflectance of PCL fiber mesh was measured as shown in Fig.5.8. The reflectance is asymptotically converging to 65% as thickness of sample

increases up to 300μm. This value can be considered as R∞ , and the ratio of Kubelka-

Munk absorption and scattering coefficients can be calculated from Eqs. (5.11) and (5.13).

The other relationship can be estimated from the experimental measurement of transmittance of samples.

μ μ += μast (5.15)

where, μa is absorption coefficient, μs is scattering coefficient

100%

80%

60%

40% Reflectance (%) 20%

0% 0 100 200 300 Thickness (μm)

Figure 5.8 Reflectance measurement

192

By using above described approaches, we analyzed the absorption and scattering

coefficients. The total extinction coefficient which combines absorption and scattering

coefficients can be estimated by curve fitting of transmittance using exponential function

fitting. By using the total extinction coefficient from the experiment, Kubelka-Munk

theory, and three flux approaximation, we can also separate the absorption and scattering

coefficients from the total extinction coefficient [22, 25, 29]. With 22 flux theory which

assumes a diffused light illumination, we have about 98% of scattering from the absorbed

light inside the material. With 3 flux theory approximation which assumes a collimated

light illumination, we found that 97% of scattering from the absorbed light inside the

material.

The ablation threshold for fiber meshed PCL was measured. The fundamentals of

femtosecond laser ablation have been summarized in previous sections and there are

many references elsewhere [32]. For our purposes we can consider single pulse removal

of a material with threshold ablation fluence Fth by a beam having a Gaussian radial fluence profile,

2 2 0 (−= 2exp)( wrFrF 0 ) (5.16) where w0 is Gaussian beam radius and F0 is maximum (or peak) fluence, related to pulse energy Ep as

2 0 = p /2 πwEF 0 (5.17)

It is noted that 0 = 2FF av . If 0 > FF th , the ablation threshold fluence, the diameter D of the area near the center of the beam from which material is removed is then

193 2 2 = 0 ()0 /ln2 FFwD th . (5.18)

2 Fth can be found by curve fitting a semi-logarithmic plot of D measured for various F0 and calculating the intercept at zero fluence.

We measured the width of ablation and fluence, and they are plotted as Fig.5.9.

2 Based on Eqs. (5.16) – (5.18), we calculated the ablation threshold as Fth=0.06 J/cm .

2 This number is lower when it is compared to bulk PCL ablation threshold, Fth=1 J/cm

[33].

1000 900 800 ) 2 700 m

μ 600 ( 2 500 400

Diameter 300 200 100 0 0.00 0.05 0.10 0.15 0.20 0.25 Fluence(J/cm2)

Figure 5.9 Ablation diameters vs. Fluence

One of reasons of lower ablation threshold is due to multiple scattering. It is believed that the multiple scattering including backscattering inside ES PCL fiber widens 194 the affected area of laser beam irradiation which in turn decreases laser fluence or ablation threshold calculation.

As we can see in Fig.5.10, initial focused spot diameter (d0=1.6μm) was applied on the surface of PCL fiber mesh, but multiple scattering occurs which redirect incident laser beams and scattered laser beams. The experimental results support this assumpition and we observed that the width of ablation was as big as 8 μm to 15 μm even lowest laser energy we applied (Fig.5.11)

Figure 5.10 Increased ablation diameter by multiple scattering

195

Figure 5.11 Ablated fiber with femtosecond laser (E=500nJ)

It is also observed that there are two distinct slopes in the graph which does not follow logarithmic curve fitting in Fig.5.9. One of reason is that a strong plasma radiation may affect the widening the width of ablation at higher fluence region which are visually noticeable. The plasma expands rapidly shortly after the femtosecond laser pulse and is observed to persist for many 10’s of nanoseconds [34, 35].

5.4Conclusions and future works

In this section, the femtosecond laser scattering inside ES fiber is presented.

Integrating sphere was used to collect the transmitted and reflected light from samples.

Based on multiscattering theories, about 97% and 98% of light absorbed inside material

196 scatters with 22 flux and 3 flux assumptions, respectively. The ablation threshold is measured to be 0.094 J/cm2 which is lower than that of bulk material 1J/cm2. The measured minimum ablation width is 8μm which is wider than theoretically calculated diffraction limited focus spot size, 1.6μm. It is believed that the multiscattering and backscattering of light contributes to lowering ablation threshold and widening ablation width. A computer simulation based on Maxwell equations can help us understand the multiple scattering phenomena inside ES PCL fiber and that remains as future work.

197 5.5References

1. D. Reneker, and I. Chun, Nanometre diameter fibres of polymer, produced by electrospinning. Nanotechnology, 1996. 7: p. 216-223.

2. P. Baumgarten, Electrostatic spinning of acrylic microfibres. J. Colloid Interface Sci., 1971. 36: p. 71-79.

3. J. Lannutti, D.R., T. Ma, D. Tomasko, and D. Farson, Electrospinning for tissue engineering scaffolds. Material science and engineering C, 2007. 27: p. 504-509.

4. K. Lee, H. Kim, M. Khil, Y. Ra, and D. Lee, Characterization of nano-structured poly( e-caprolactone) nonwoven mats via electrospinning. Polymer, 2003. 44: p. 1287-1294.

6. W. Li, C. Laurencin,E. Caterson, R. Tuan, and F. Ko, Electrospun nanofibrous structure: A novel scaffold for tissue engineering. Biomedical Mat. Res., 2002. 60: p. 613-621.

7. H. Yoshimoto, Y. Shin, H. Terai, and J. Vacanti, A biodegradable nanofiber scaffold by electrospinning and its potential for bone tissue engineering. Biomaterials, 2003. 24: p. 2077-2082.

8. C. Xu, R. Inai, M. Kotaki, and S. Ramakrishna, Aligned biodegradable nanotibrous structure: a potential scaffold for blood vessel engineering. Biomaterials, 2004. 25: p. 877-886.

10. H. Choi, J. Johnson, J. Nam, D. Farson, and J. Lannutti, Structuring electrospun polycaprolactone nanofiber tissue scaffolds by femtosecond laser ablation. Journal of Laser Applications, 2007.

11. P. Yang, H. Wei, H. Huang, B. Baum, Y. Hu, G. Kattawar, M. Mishchenko, and Q. Fu, Scattering and absorption property database for nonspherical ice particles in the near- through far-infrared spectral region. Applied Optics, 2005. 44(26): p. 5512-5523.

198 12. J. Mourant, T. Fuselier, J. Boyer, T. Johnson, and I. Bigio, Preditions and measurements of scattering and absorption over broad wavelength ranges in tissue phantoms. Applied Optics, 1997. 36(4): p. 949-957.

13. T. Tesfamichael, A. Hoel, G. Niklasson, E. Wackelgard, M. Gunde, and Z. Orel, Optical characterization method for black pigments applied to solar-selective absorbing paints. Applied Optics, 2001. 40(10): p. 1672-1681.

14. S. Mourad, P. Emmel, K. Simon, and R. Hersch, Extending Kubelka-Munk's theory with lateral light scattering. IS&T's NIP 17: Int. Conf. on digital printing technologies, 2001: p. 469-473.

15. N. Marley, J. Gaffney, J. Baird, C. Blazer, P. Drayton, and J. Frederick, An empirical method for the determination of the complex refractive index of size- fractionated atmospheric aerosols for radiative transfer calculations. Aerosol Science and Technology, 2001. 34: p. 535-549.

16. R. Molenaar, J. Bosch, and J. Zijp, Determination of Kubelka-Munk scattering and absorption coefficients by diffuse illumination. Applied Optics, 1999. 38(10): p. 2068-2077.

17. J. Lai, Z. Li, C. Wang, and A. He, Experimental measurement of the refractive index of biological tissues by total internal reflection. Applied Optics, 2005. 44(10): p. 1845-1849.

18. G. Mie, Beitrage zur Optik truber Medien, speziell kolloidaler Metallosungen. Ann. Phys, 1908. 330: p. 377-445.

19. H.C. van de Hulst, Light scattering by small particles, ed. 2nd. 1957, New York: Dover Publications. 297-328.

20. A. Cox, A. DeWeerd, and J. Linden, An experiment to measure Mie and Rayleigh total scattering cross sections. Am. J. Phys., 2002. 70(6): p. 620-625.

21. S. Chandrasekhar, Radiative Transfer. 1960, New York: Dover Publications.

22. P. Kubelka, and F. Munk, Ein Beitrag zur Optik der Farbanstriche. Zeits f. techn. Physik, 1931. 12: p. 593-601.

23. F. Fabbri, M. Franceschini, and S. Fantini, Characterization of spatial and temporal variations in the optical properties of tissuelike media with diffuse reflectance imaging. Applied Optics, 2003. 42(16): p. 3063-3072.

24. L. Yang, B. Kruse, and S. Miklavcic, Revised Kubelka-Munk theory II, Unified framework for homogeneous and inhomogeneous optical media. J. Opt. Soc. Am. A, 2004. 21(10): p. 1942-1952.

199 25. P. Mudgett, and L. Richards, Multiple scattering calculations for technology. Applied Optics, 1971. 10(7): p. 1485-1502.

26. D. Sardar, and L. Levy, Caparative evaluation of absorption coefficients of KCL:Eu2+ and CaF2:EU2+ using a spectrophotometer and an integrating sphere. J. Appl. Phys., 1996. 79(3): p. 1759-1762.

27. W. Cheong, S. Prahl, and A. Welch, A review of the optical properties of biological tissues. IEEE journal of quantum electronic, 1990. 26(12): p. 2166- 2185.

28. E. Patterson, C. Shelden, and B. Stockton, Kubelka-Munk optical properties of a barium sulfate white reflectance standard. Applied Optics, 1977. 16(3): p. 729- 732.

29. T. Burger, H. Ploss, J. Kuhn, S. Ebel, and J. Fricke, Diffuse reflectance and transmittance spectroscopy for the quantitative determination of scattering and absorption coefficients in quantitative power analysis. Appl. Spectrosco., 1997. 51: p. 1323-1329.

30. T. Burger, J. Khun, R. Caps, and J. Fricke, Quantitative determination of the scattering and absorption coefficients from diffuse reflectance and transmittance measurements applied to pharmaceutical powders. Applied Spectrsc., 1997. 51: p. 309-317.

31. Thorlabs, Inc., www.thorlabs.com, accessed 7/11/07

32. J. Kruger, and W. Kautek, Ultrashort pulse laser interaction with dielectrics and polymers. Advances in polymer science, 2004. 168: p. 247-289.

33. T. Lippert, Interaction of photons with polymers: From surface modification to ablation. Plasma process polym., 2005. 2: p. 525-546.

34. B. Salle, O. Gobert, P. Meynadier, M. Perdrix, G. Petite, and A. Semerok, Femtosecond and picosecond laser microablation: ablation efficiency and laser microplasma expansion. Appl. Phys. A - Mater. Sci. Processing, 1999. 69: p. S381-S383.

35. S. Kilmentov, T. Kononenko, P. Pivovarov, S. Garnov, V. Konov, A. Prokhorov, D. Breitling, and F. Dausinger, The role of plasma in ablation of materials by ultrashort laser pulses. Quant. Electronics, 2001. 31: p. 378-382.

200

References

1. Z. Hu, S. Zhang, and X. Zheng, Three dimensional micromachining based on AFM. Key engineering materials, 2006. 315: p. 800-804.

2. Z. Chen, Y. Gao, J. Lin, R. Su, and Y.Xie, Vacuum assisted thermal bonding of plastic capillary electrophoresis microchip imprinted with stainless steel template. Journal of Chromatography, 2004. A(1038): p. 239-245.

3. Y.H. Tak, C.N. Kim, M.S. Kim, K.B. Kim, M.H. Lee and S.T. Kim, “Novel patterning method using Nd : YAG and Nd : YVO4 lasers for organic light emitting diodes,” Synthetic Metals 138, 497-500 (2003).

4. Y. Xia, and G. Whitesides, Soft lithography. Annu. Rev. Mater. Sci., 1998. 28: p. 153-184.

5. Y. Kim, and C. Lieber, Machining Oxide Thine Films with an AFM: Pattern and Object Formation on the Nanometer Scale. Science, 1992. 257(5068): p. 375-377.

7. X. Liu, D. Du, and G. Mourou, Laser ablation and micromachining with ultrashort laser pulses. IEEE journal of quantum electronics, 1997. 33(10): p. 1706-1716.

8. X. Liu, D. Du, and G. Mourou, Laser ablation and micromachining with ultrashort laser pulses. IEEE journal of quantum electronics, 1997. 33(10): p. 1706-1716.

9. X. Liu, D. Du, and G. Mourou, Laser ablation and micromachining with ultrashort laser pulses. IEEE journal of quantum electronics, 1997. 33(10): p. 1706-1716.

10. W. Li, C. Laurencin,E. Caterson, R. Tuan, and F. Ko, Electrospun nanofibrous structure: A novel scaffold for tissue engineering. Biomedical Mat. Res., 2002. 60: p. 613-621.

11. W. Li, C. Laurencin,E. Caterson, R. Tuan, and F. Ko, Electrospun nanofibrous structure: A novel scaffold for tissue engineering. Biomedical Mat. Res., 2002. 60: p. 613-621.

201 12. W. Cheong, S. Prahl, and A. Welch, A review of the optical properties of biological tissues. IEEE journal of quantum electronic, 1990. 26(12): p. 2166- 2185.

13. V. Bhardwaj, P. Rajeev, P. Corkum, and D. Rayner, Strong field ionization inside transparent solids. J. Phys. B: At. Mol. Opt. Phys., 2006. 39: p. S397-S407.

14. Thorlabs, Inc., www.thorlabs.com, accessed 7/11/07

15. Thermofisher scientific, Microslide specification. 2007.

16. Technical Glass Products, Properties of fused-quartz. 2007.

17. T. Tesfamichael, A. Hoel, G. Niklasson, E. Wackelgard, M. Gunde, and Z. Orel, Optical characterization method for black pigments applied to solar-selective absorbing paints. Applied Optics, 2001. 40(10): p. 1672-1681.

18. T. Lippert, Interaction of photons with polymers: From surface modification to ablation. Plasma process polym., 2005. 2: p. 525-546.

19. T. Lippert, Interaction of photons with polymers: From surface modification to ablation. Plasma process polym., 2005. 2: p. 525-546.

20. T. Lippert, Interaction of photons with polymers: From surface modification to ablation. Plasma process polym., 2005. 2: p. 525-546.

21. T. Kim, K.C., A. Groisman, D. Kleinfeld, and C. Schaffer, Femtosecond laser- drilled capillary integrated into a microfluidic device. Appl. Phys. Lett., 2005. 86(20): p. 201106.

22. T. Itina, J. Hermann, Ph. Delaporte, and M. Sentis, Modeling of metal ablation induced by ultrashort laser pulses. Proceedings of symposium H on photonic processing of surfaces - Thin films and devices, 2004. E-MRS: p. 453 - 454.

23. T. Itina, J. Hermann, Ph. Delaporte, and M. Sentis, Modeling of metal ablation induced by ultrashort laser pulses. Proceedings of symposium H on photonic processing of surfaces - Thin films and devices, 2004. E-MRS: p. 453 - 454.

24. T. Fang, C. Weng, and J. Chang, Machining characterization of the nano- lithography process using AFM. Nanotechnology, 2000. 11: p. 181-187.

25. T. Burger, J. Khun, R. Caps, and J. Fricke, Quantitative determination of the scattering and absorption coefficients from diffuse reflectance and transmittance measurements applied to pharmaceutical powders. Applied Spectrsc., 1997. 51: p. 309-317.

202 26. T. Burger, H. Ploss, J. Kuhn, S. Ebel, and J. Fricke, Diffuse reflectance and transmittance spectroscopy for the quantitative determination of scattering and absorption coefficients in quantitative power analysis. Appl. Spectrosco., 1997. 51: p. 1323-1329.

27. Saleh, and Teich, Fundamentals of photonics. 1999: Wiley.

28. S.M. Klimentov, T.V. Kononenko, P.A. Pivovarov, S.V. Garnov, V.I. Konov, A.M. Prokhorov, D. Breitling and F. Dausinger, “The role of plasma in ablation of materials by ultrashort laser pulses,” Quant. Elec. 31, 378-382 (2001).

29. S. Valette, R. Harzic, N. Huot, E. Audouard, and R. Fortunier, 2D Calculations of the thermal effects due to femtosecond laser-metal interaction. Applied surface science, 2005. 247: p. 238-242.

30. S. Nolte, C. Momma, H. Jacobs, A. Tunnermann, B. Chichkov, B. Wellegehausen, and H. Welling, Ablation of metals by ultrashort laser pulses. J. Opt. Soc. Am. B, 1997. 14(10): p. 2716-2722.

31. S. Nolte, C. Momma, H. Jacobs, A. Tunnermann, B. Chichkov, B. Wellegehausen, and H. Welling, Ablation of metals by ultrashort laser pulses. J. Opt. Soc. Am. B, 1997. 14(10): p. 2716-2722.

32. S. Nolte, C. Momma, G. Kamlage, A. Ostendort, C. Fallnich, F. Alvenslebe, and H. Welling, Polarization effects in ultrashort-pulse laser drilling. Applied physics A - materials science & processing, 1999. 68(6): p. 563-567.

33. S. Mourad, P. Emmel, K. Simon, and R. Hersch, Extending Kubelka-Munk's theory with lateral light scattering. IS&T's NIP 17: Int. Conf. on digital printing technologies, 2001: p. 469-473.

34. S. Lai, X. Cao, and L.J. Lee, A Packaging techinque for polymer microfluidic platforms. Anal. Chem., 2004. 76: p. 1175-1183.

35. S. Kuper, and M. Stuke, Femtosecond UV excimer laser ablation. Appl. Phys. B: Photophysics Laser Chemistry, 1987. 44: p. 199-204.

36. S. Kuper, and M. Stuke, Ablation of Polytetrafluoroethylene (Teflon) With Femtosecond UV Excimer Laser Pulses. Appl. Phys. Lett., 1989. 54: p. 4-6.

37. S. Klimentov, T. Kononenko, P. Pivovarov, S. Garnov, V. Konov, A. Prokhorov, D. Breitling, and F. Dausinger, The role of plasma in ablation of materials by ultrashort laser pulses. Quantum Electronics, 2001. 31(5): p. 378-382.

203 38. S. Kilmentov, T. Kononenko, P. Pivovarov, S. Garnov, V. Konov, A. Prokhorov, D. Breitling, and F. Dausinger, The role of plasma in ablation of materials by ultrashort laser pulses. Quant. Electronics, 2001. 31: p. 378-382.

39. S. Kilmentov, T. Kononenko, P. Pivovarov, S. Garnov, V. Konov, A. Prokhorov, D. Breitling, and F. Dausinger, The role of plasma in ablation of materials by ultrashort laser pulses. Quant. Electronics, 2001. 31: p. 378-382.

40. S. Kidoaki, I. Kwon, and T. Matsuda, Mesoscopic spatial designs of nano- and microfiber meshes for tissue-engineering matrix and scaffold based on newly devised multilayering and mixing electrospinning techniques. Biomaterials, 2005. 26(37-46).

41. S. Kidoaki, I. Kwon, and T. Matsuda, Mesoscopic spatial designs of nano- and microfiber meshes for tissue-engineering matrix and scaffold based on newly devised multilayering and mixing electrospinning techniques. Biomaterials, 2005. 26(37-46).

42. S. Chandrasekhar, Radiative Transfer. 1960, New York: Dover Publications.

43. S. Baudach, J. Bonse, J. Kruger, and W. Kautek, Ultrashort pulse laser ablation of polycarbonate and polymethylmethacrylate. Appl. Surface Sci., 2000. 154: p. 555-560.

44. R. Stoian, D. Ashkenasi, A. Rosenfeld and EEB Campbell, “Coulomb explosion in ultrashort pulsed laser ablation of Al2O3,” Phys. Rev. B 62, 13167- 13173 (2000).A12

45. R. Molenaar, J. Bosch, and J. Zijp, Determination of Kubelka-Munk scattering and absorption coefficients by diffuse illumination. Applied Optics, 1999. 38(10): p. 2068-2077.

46. R. Kane, S. Takayama, E. Ostuni, D. Ingber, and G. Whitesides, Patterning proteins and cells using soft lithography. Biomaterials, 1999. 20: p. 2363-2376.

47. R. Bel Hadj Tahar, T. Ban, Y. Ohya and Yasutaka Takahashi, “Tin doped indium oxide thin films: Electrical properties,” J. App. Phys. 83, 2631-2645 (1998).

48. R. An, Y. Li, Y. Dou, H. Yang, and Q. Gong, Simultaneous multi-microhole drilling of soda-lime glass by water-assisted ablation with femtosecond laser pulses. Optics express, 2005. 13(6): p. 1855-1859.

49. Q.Chen, J. Longtin, S. Tankiewicz, S. Sampath, and R. Gambino, Ultrafast laser micromachining and patterning of thermal spray multilayers for thermopile fabrication. J. Micromech. Microeng, 2004. 14: p. 506-513.

204 50. P. Yang, H. Wei, H. Huang, B. Baum, Y. Hu, G. Kattawar, M. Mishchenko, and Q. Fu, Scattering and absorption property database for nonspherical ice particles in the near- through far-infrared spectral region. Applied Optics, 2005. 44(26): p. 5512-5523.

51. P. Waterman, Scattering, absorption, and extinction by thin fibers. J. Opt. Soc. Am. A, 2005. 22(11): p. 2430-2441.

52. P. Mudgett, and L. Richards, Multiple scattering calculations for technology. Applied Optics, 1971. 10(7): p. 1485-1502.

53. P. Mannion, J.M., E. Coyne, G. O'Connor, and T. Glynn, The effect of damage accumulation behabiour on ablation thresholds and damage morphology in ultrafast laser micro-machining of common metals in air. Applied surface science, 2004. 223: p. 275-287.

54. P. Mannion, J. Magee, E. Coyne, G. O'Connor, and T. Glynn, The effect of damage accumulation behabiour on ablation thresholds and damage morphology in ultrafast laser micro-machining of common metals in air. Applied surface science, 2004. 223: p. 275-287.

55. P. Kubelka, and F. Munk, Ein Beitrag zur Optik der Farbanstriche. Zeits f. techn. Physik, 1931. 12: p. 593-601.

56. P. Baumgarten, Electrostatic spinning of acrylic microfibres. J. Colloid Interface Sci., 1971. 36: p. 71-79.

57. P. Baumgarten, Electrostatic spining of acrylic microfibres. J. Colloid Interface Sci., 1971. 36: p. 71-79.

. Yavas and M. Takai, “Effect of substrate absorption on the efficiency of laser patterning of indium tin oxide thin films,” J. Appl. Phys. 85, 4207- 4212 (1999)

58. N. Marley, J. Gaffney, J. Baird, C. Blazer, P. Drayton, and J. Frederick, An empirical method for the determination of the complex refractive index of size- fractionated atmospheric aerosols for radiative transfer calculations. Aerosol Science and Technology, 2001. 34: p. 535-549.

59. N. Bulgakova, Possibility of rarefaction shock wave under short pulse laser ablation of solids. Physical Review E, 1999. 60(4): p. R3498 - R3500.

60. M. Womack, M. Vendan, and P. Molian, Femtosecond pulsed laser ablation and deposition of thin films of polytetrafluoroethylene. Appl. Surface Sci., 2004(221).

205 61. M. Unger, H.C., T. Thorsen, A. Scherer, and S. Quake, Monolithic microfabricated valves and pumps by multilayer soft lithography. Science, 2000. 288: p. 113-116.

62. M. Shirk, and P. Molian, A review of ultrashort pulsed laser ablation of materials. Journal of Laser Applications, 1998. 10(1): p. 18-28.

63. M. Sasnett, Propagation of multimode laser beam-the M2 factor, ed. D. Hall, and P. Jackson. 1989, New York: Adam Hilger. 132-142.

64. M. Roberts, J. Rossier, P. Bercier, and H. Girault, UV laser machined polymer substrates for the development of microdiagnostic systems. Anal. Chem., 1997. 69: p. 2035-2042.

65. M. Richardson, A. Zoubir, C. Rivero, C. Lopez, L. Petit, and L. Richardson, Femtosecond laser micro-structuring and refractive index modification applied to laser and photonic devices. SPIE, 2003. V: p. 15347-4.

66. M. Lenzner, J.K., S. Sartania, Z. Cheng, C. Spielmann, G. Mourou, W. Kautek, and F. Krausz, Femtosecond Optical Breakdown in Dielectrics. Physical Review Letters, 1998. 80(18): p. 4076-4079.

67. M. Giridhar, K. Seong, A. Schulzgen, P. Khulbe, N. Peyghambarian and M. Mansuripur, “Femtosecond pulsed laser micromachining of glass substrates with application to microfluidic devices,” Appl. Opt. 43, 4584-4589 (2004).

68. M. Farsari, G.F., S. Zoppel, G. Reider, and Fotakis, Efficient femtosecond laser micromachining of bulk 3C-SiC. J. Micromech. Microeng, 2005. 15: p. 1786- 1789.

69. M. Farsari, G. Filippidis, S. Zoppel, G. Reider, and Fotakis, Efficient femtosecond laser micromachining of bulk 3C-SiC. J. Micromech. Microeng, 2005. 15: p. 1786-1789.

70. M. Esch, S. Kapur, G. Irizarry, and V. Genova, Influence of master fabrication techniques on the characteristics of embossed microfluidic channels. Lab on a Chip, 2003. 3(2): p. 121-127.

71. M. Elbandrawy and M. Gupta, “Optical characteristics of femtosecond laser micromachined periodic structures in Si <100>,” Appl. Opt. 45, 7137-7143 (2006).

72. L. Yang, B. Kruse, and S. Miklavcic, Revised Kubelka-Munk theory II, Unified framework for homogeneous and inhomogeneous optical media. J. Opt. Soc. Am. A, 2004. 21(10): p. 1942-1952.

206 73. L. Sudrie, A.C., M. Franco, B. Lamouroux, B. Prade, S. Tzortzakis, and A. Mysyrowicz, Femtosecond laser-induced damage and filamentary propagation in fused silica. Physical Review Letters, 2002. 89(18): p. 186601.

74. L. Shah, Z.L. Liu, I. Hartl , G. Imeshev, G.C. Cho and M.E. Fermann, “High energy femtosecond Yb cubicon fiber amplifier,” Opt. Express 13, 4717-4722 (2005).

75. L. Shah, J.T., M. Richardson, and K. Richardson, Femtosecond laser deep hole drilling of silicate glasses in air. Applied surface science, 2001. 183: p. 151-164.

76. L. Li, T.A., and M. Esashi, Smooth surface glass etching by deep reactive ion etching with SF6 and Xe gases. J. Vac. Sci. Tech (B), 2003. 21(6): p. 2545-2549.

77. L. Keldysh, Ionization in the eld of a strong electromagnetic wave. Soviet Physics JETP, 1965. 20: p. 1307-1314.

78. K. Venkatakrishnan, P.S., and L. Lim, Femtosecond laser ablation of thin films for the fabrication of binary photomasks. J. Micromech. Microeng, 2002. 12: p. 775-779.

79. K. Tiaw, S. Goh, M. Hong, Z. Wang, B. Lan, and S. Teoh, Laser surface modification of poly(epsilon-caprolactone) (PCL) membrane for tissue engineering applications. Biomaterials, 2005. 26: p. 763-769.

80. K. Naessens, H.O., P. Daele, and R. Baets, Flexible fabrication of microlenses in polymer layers with excimer laser ablation. Applied surface science, 2003. 208- 209: p. 159-164.

81. K. Miura, J. Qiu, H. Inouye, T. Mitsuyu, and K. Hirao, Photowritten optical waveguides in various galsses with ultrashort pulse laser. Appl. Phys. Lett., 1997. 71(23): p. 3329-3331.

82. K. Lee, H. Kim, M. Khil, Y. Ra, and D. Lee, Characterization of nano-structured poly( e-caprolactone) nonwoven mats via electrospinning. Polymer, 2003. 44: p. 1287-1294.

83. K. Lee, H. Kim, M. Khil, Y. Ra, and D. Lee, Characterization of nano-structured poly( e-caprolactone) nonwoven mats via electrospinning. Polymer, 2003. 44: p. 1287-1294.

84. K. Ke, E. Hasselbrink, and A. Hunt, Rapidly Prototyped Three-Dimensional Nanofluidic Channel Networks in Glass Substrates. Anal. Chem., 2005. 77(16): p. 5083-5088.

207 85. K. Davis, K. Miura, N. Sugimoto, and K. Hirao, Writing waveguides in glass with a femtosecond laser. Opt. Lett, 1996. 21(21): p. 1729-1731.

86. J.R. Buckley, F.W. Wise, F.O. Ilday and T. Sosnowski, “Femtosecond fiber lasers with pulse energies above 10 nJ,” Opt. Lett. 30, 1888-1890 (2005).

87. J. Zhang, K. Tan, G. Hong, L. Yang, and H. Gong, Polymerization optimization of SU-8 photoresist and its applications in microfluidic systems and MEMS. J. Micromech. Microeng, 2001. 11: p. 20-26.

88. J. Rossier, C. Vollet, A. Carnal, G. Lagger, V. Gobry, H. Girault, P. Michel, and F. Reymond, Plasma etched polymer microelectrochemical systems. Lab on a Chip, 2002. 2(3): p. 145-150.

89. J. Park, D.L., N. Takano, and N. Morita, Diamond tip cantilever for micr/nano machining based on AFM. Materials Science Forum, 2006. 505: p. 79-84.

90. J. Narashimhan, a.I.P., Polymer embossing tools for rapid prototyping of plastic microfluidic devices. J. Micromech. Microeng, 2004. 14: p. 96-103.

91. J. Mourant, T. Fuselier, J. Boyer, T. Johnson, and I. Bigio, Preditions and measurements of scattering and absorption over broad wavelength ranges in tissue phantoms. Applied Optics, 1997. 36(4): p. 949-957.

92. J. Mecomber, D. Hurd, and P. Limbach, Enhanced machining of micron-scale features in microchip molding masters by CNC milling. International Journal of Machine Tools & Manufacture, 2005. 45: p. 1542 - 1550.

93. J. McDonald, V. Mistry, K. Ray, and S. Yalisove, Femtosecond pulsed laser direct write production of nano- and microfluidic channels. Appl. Phys. Lett., 2006. 88(183113): p. 1-3.

94. J. McDonald, D.D., J. Anderson, D. Chiu, H. Wu, O. Schueller, and G. Whitesides, Fabrication of microfluidic systems in PDMS. Electrophoresis, 2000. 21: p. 27-40.

95. J. McDonald, D.D., J. Anderson, D. Chiu, H. Wu, O. Schueller, and G. Whitesides, Fabrication of microfluidic systems in PDMS. Electrophoresis, 2000. 21: p. 27-40.

96. J. Lannutti, D.Reneker, T. Ma, D. Tomasko, and D. Farson, Electrospinning for tissue engineering scaffolds. Material science and engineering C, 2007. 27: p. 504-509.

208 97. J. Lannutti, D.R., T. Ma, D. Tomasko, and D. Farson, Electrospinning for tissue engineering scaffolds. Material science and engineering C, 2007. 27: p. 504-509.

98. J. Lai, Z. Li, C. Wang, and A. He, Experimental measurement of the refractive index of biological tissues by total internal reflection. Applied Optics, 2005. 44(10): p. 1845-1849.

99. J. Kruger, and W. Kautek, Ultrashort pulse laser interaction with dielectrics and polymers. Advances in polymer science, 2004. 168: p. 247-289.

100. J. Kruger, and W. Kautek, Ultrashort pulse laser interaction with dielectrics and polymers. Advances in polymer science, 2004. 168: p. 247-289.

101. J. Kruger, a.W.K., Femtosecond-pulse visible laser processing of transparent materials. Applied surface science, 1996. 96(98): p. 430-438.

102. J. Krüger and W. Kautek, “Ultrashort Pulse Laser Interaction with Dielectrics and Polymers,” Adv Polymer Sci 168, 247–289 (2004).

103. J. Ihlemann, a.B.W.-R., Excimer laser micro machining of inorganic dielectrics. Applied surface science, 1996. 106: p. 282-286.

104. J. Cheng, M.Y., C. Wei, and Y. Chuang, and T. Young, Crack-free direct writing on glass using a low power UV laser in the manufacture of a microfluidic chip. J. Micromech. Microeng, 2005. 15: p. 1147-1156.

105. Adesida, A.M., E. Andideh, M. Khan, D. Oslen, and J. Kuznia, Reactive ion etching of GaN in silicon tetrachloride plasmas. Appl. Phys. Lett., 1993. 63(20): p. 2777-2779.

106. H.C. van de Hulst, Light scattering by small particles, ed. 2nd. 1957, New York: Dover Publications. 297-328.

107. H. Yoshimoto, Y. Shin, H. Terai, and J. Vacanti, A biodegradable nanofiber scaffold by electrospinning and its potential for bone tissue engineering. Biomaterials, 2003. 24: p. 2077-2082.

108. H. Yoshimoto, Y. Shin, H. Terai, and J. Vacanti, A biodegradable nanofiber scaffold by electrospinning and its potential for bone tissue engineering. Biomaterials, 2003. 24: p. 2077-2082.

109. H. Klank, J.K., and O. Geschke, CO2 laser micomachining and back end processing for rapid production of PMMA based microfluidic systems. Lab on a chip, 2002. 2(4): p. 242-246.

209 110. H. Huang, H.Z., and Y. Liu, Experimental investigations machiniability of Ni50.6Ti49.4. Smart Materials and Structurs, 2005. 14: p. S297-S301.

111. H. Huang, H. Zheng, and Y. Liu, Experimental investigations machiniability of Ni50.6Ti49.4. Smart Materials and Structurs, 2005. 14: p. S297-S301.

112. H. Endert, R.P., and D. Basting, Excimer laser: A new tool for precision micromachining. Optical and Quantum electronics, 1995. 27(12): p. 1319-1335.

113. H. Endert, R. Patzel, and D. Basting, Excimer laser: A new tool for precision micromachining. Optical and Quantum electronics, 1995. 27(12): p. 1319-1335.

114. H. Dehghani, B. Brooksby, K. Vishwanath, B. Pogue, and D. Paulse, The effects of internal refractive index variation in near-infrared optical tomography: a finite element modeling approach. Phys. Med. Biol., 2003. 48: p. 2713-2727.

115. H. Choi, J.J., J. Nam, D. Farson, J. Lannutti, Structuring electrospun polycaprolactone nanofiber tissue scaffolds by femtosecond laser ablation. Journal of Laser Applications, 2007.

116. H. Choi, J. Johnson, J. Nam, D. Farson, and J. Lannutti, Structuring electrospun polycaprolactone nanofiber tissue scaffolds by femtosecond laser ablation. Journal of Laser Applications, 2007.

117. H. Choi, D.F., J. Bovatsek, A. Arai, and D. Ashkenasi, Direct-write patterning of ITO films by high pulse repetition frequency femtosecond laser ablation. Applied Optics, 2007.

118. H. Choi, D. Farson, J. Bovatsek, A. Arai, and D. Ashkenasi, Direct-write patterning of ITO film by high pulse repetition rate femtosecond laser ablation. Applied Optics, 2007.

119. H. Cao, Z.Yu, J. Wang, J. Tegenfeldt, R. Austin, E. Chen, W. Wu, and S. Chou, Fabrication of 10nm enclosed nanofluidic channels. Appl. Phys. Lett., 2002. 81(1): p. 174-176.

120. H. Becker, and U. Heim, Hot embossing as a method for the fabrication of polymer high aspect ratio structures. Sensors and Actuators, 2000. 83: p. 130-135.

121. G. Whitesides, Soft Lithography. 1998, WTEC hyper-librarian.

122. G. Whitesides, E.O., S. Takayama, X. Jiang, and D. Ingber, Soft lithography in biology and biochemistry. Annu. Rev. of Biomedical Engineering, 2001. 3: p. 335-373.

210 123. G. Mie, Beitrage zur Optik truber Medien, speziell kolloidaler Metallosungen. Ann. Phys, 1908. 330: p. 377-445.

124. G. Mainfray, a.C.M., Multiphoton ionization of atoms. Reports on progress in physics, 1991. 54(10): p. 1333 - 1372.

125. G. Lee, C. Hung, B. Ke, G. Huang, and B. Hwei, Micromachined pre-focused 1XN flow switches for continuous sample injection. J. Micromech. Microeng, 2001. 11: p. 567-573.

126. G. Kopitkovas, T.L., C. David, A. Wokaun, and J. Gobrecht, Surface micromachining of UV transparent materials. Thin solid films, 2004. 453: p. 31- 35.

127. G. Gibson, T.L., and C. Rhodes, Tunneling ionization in the multiphoton regime. Physical Review A, 1990. 41(9): p. 5049 - 5052.

128. F. Vidal, T.J., S. Laville, O. Barthelemy, M. Chaker, B. LeDegoff, J. Margot, and M. Sabsabi, Critical-point phase separation in laser ablation of conductors. Physical Review Letters, 2001. 86(12): p. 2573 -2576.

129. F. Tay, J.K., F. Watt, and W. Choong, A novel micro-machining method for the fabrication of thick-film SU-8 embedded micro-channels. J. Micromech. Microeng, 2001. 11: p. 27-32.

130. F. Ilkov, J.D.a.S.C., Ionization of atoms in the tunneling regime with experimental - evidence using Hg atoms. Journal of Physics B - Atomic molecular and optical physics, 1992. 25(19): p. 4005 - 4020.

131. F. Fabbri, M. Franceschini, and S. Fantini, Characterization of spatial and temporal variations in the optical properties of tissuelike media with diffuse reflectance imaging. Applied Optics, 2003. 42(16): p. 3063-3072.

132. E. Patterson, C. Shelden, and B. Stockton, Kubelka-Munk optical properties of a barium sulfate white reflectance standard. Applied Optics, 1977. 16(3): p. 729- 732.

133. E. Illy, D.B., M. Withford, and J. Piper, Enhanced polymer ablation rates using high-repetition-rate ultraviolet lasers. IEEE journal of quantum electronic, 1999. 5(6): p. 1543-1548.

134. E. Gamaly, A.R., B. Luther-Davies, and V. Tikhonchuk, Ablation of solids by femtosecond lasers: Ablation mechnasim and ablation thresholds for metals and dielectrics. Physics of plasma, 2002. 9(3): p. 949-957.

211 135. E. Chanson, S.P., J. Poate, J. Borland, M. Current, T. Rubia, D. Eaglesham, O. Holland, M. Law, C. Magee, J. Melngailis, and A. Tasch, Ion beams in silicon processing and characterization. J. Appl. Phys., 1997. 81(10): p. 6513-6561.

136. D. Yuan, and S. Das, Experimental and theoretical anlysis of direct-write laser micromachining of polymethyl methacrylate by CO2 laser ablation. Journal of Applied Physics, 2007(101): p. 024901-1:6.

137. D. Sardar, and L. Levy, Caparative evaluation of absorption coefficients of KCL:Eu2+ and CaF2:EU2+ using a spectrophotometer and an integrating sphere. J. Appl. Phys., 1996. 79(3): p. 1759-1762.

138. D. Reneker, and I. Chun, Nanometre diameter fibres of polymer, produced by electrospinning. Nanotechnology, 1996. 7: p. 216-223.

139. D. Hwang, T. Choi, and C. Grigoropoulos, Liquid-assisted femtosecond laser drilling of straight and three-dimensional microchannels in glass. Appl. Phys. A - Matl. Sci. & Processing, 2004. 79(3): p. 605-612.

140. D. Gomez, I.G., I. Lizuain, and M. Ozaita, Femtosecond laser ablation for microfluidics. Optical engineering, 2005. 44(5): p. 051105:1-8.

141. D. Farson, H. Choi, C. Lu, and L. Lee, Femtosecond laser bulk micromachining of microfluid channels in poly(methylmethacrylate). Journal of Laser Applications, 2006. 18(3): p. 210-215.

142. D. Du, X. Liu, G. Korn, J. Squier, and G. Mourou, Laser induced breakdown by impact ionization in SiO2 with pulse widths from 7ns to 150fs. Appl. Phys. Lett., 1994. 64(23): p. 3071-3073.

143. D. Ashkenasi, M.L., R. Stoian, and A. Rosenfeld, Surface damage threshold and structuring of dielectrics using femtosecond laser pulses: the role of incubation. Appl. Surface Sci., 1999. 150: p. 101-106.

144. D. Ashkenasi, G. Muller, A. Rosenfeld, R. Stoian, I.V. Hertel, N.M. Bulgakova and E.E.B. Campbell, “Fundamentals and advantages of ultrafast micro- structuring of transparent materials,” Appl. Phys. A-Mater. Sci. Processing. 77, 223-228 (2003).

145. Clark-MXR, CPA 2110 User manual. 2 ed. 2004: Clark-MXR.

146. C.H. Fan, J.P. Longtin, “Modeling optical breakdown in dielectrics during ultrafast laser processing,” Appl. Opt. 40, 3124-3131 (2001).

212 147. C. Zhang, C. Yang, and D. Ding, Deep reactive ion etching of commercial PMMA in O2/CHF3, O2/Ar based discharges. J. Micromech. Microeng, 2004. 14: p. 663-666.

148. C. Xu, R. Inai, M. Kotaki, and S. Ramakrishna, Aligned biodegradable nanotibrous structure: a potential scaffold for blood vessel engineering. Biomaterials, 2004. 25: p. 877-886.

149. C. Schaffer, Interaction of femtosecond laser pulses with transparent materials, in Physics. 2001, Harvard University: Cambridge, MA. p. 173.

150. C. Schaffer, Interaction of femtosecond laser pulses with transparent materials, in Applied Physics. 2001, Harvard University: Cambridge.

151. C. Schaffer, A. Brodeur, J. Garcia, and E. Mazur, Micromachining bulk glass by use of femtosecond laser pulses with nanojoule energy. Opt. Lett, 2001. 26(2): p. 93-95.

152. C. Schaffer, A. Brodeur and E. Mazur, Laser induced breakdown and damage in bulk transparent materials induced by tightly focused femtosecond laser pulses. Meas. Sci. Technol., 2001. 12: p. 1784 - 1794.

153. C. Molpeceres , S. Lauzurica , J.L. Ocana, J.J. Gandia, L. Urbina and J. Carabe, “Microprocessing of ITO and a-Si thin films using ns laser sources,” J. Micromech. Microeng. 15, 1271-1278 (2005).

154. C. Li, D. Johnson, and R. Kovacevic, Modeling of waterjet guided laser grooving of silicon. International Journal of Machine Tools & Manufacture, 2003. 43: p. 925 - 936.

155. C. Geankoplis, Transport Processes and Unit Operations. 3rd ed. 1993, Englewood Clifs: Prentice Hall.

156. C. Choi, Fabrication of optical waveguides in thermosetting polymers using hot embossing. J. Micromech. Microeng, 2004. 14: p. 945-949.

157. C. Anthony, P. Docker, P. Prewett, and K. Jiang, Focused ion beam microfabrication in Foturan(TM) photosensitive glass. J. Micromech. Microeng, 2007. 17: p. 115-119.

158. C. Aguilar, Y. Lu, S. Mao, and S. Chen, Direct micro-patterning of biodegradable polymers using ultraviolet and femtosecond lasers. Biomaterials, 2005. 26: p. 7642-7649.

213 159. B.J. Luff, J.S. Wilkinson and G. Perrone, “Indium tin oxide overlayered waveguides for sensor applications,” Appl. Opt. 36, 7066-7072 (1997).

160. B. Stuart, M. Feit, and S. Herman, Nanosecond to femtosecond laser-induced breakdown in dielectrics. Physical Review B, 1996. 53(4): p. 1749-1761.

161. B. Stuart, M. Feit, A. Rubenchik, B. Shore, and M. Perry, Laser-induced damage in dielectrics with nanosecond to subpicosecond pulses. Physical Review Letters, 1995. 74(12): p. 2248-2252.

162. B. Seddon, Y. Shao, J. Fost, and H. Girault, The application of excimer-laser micromachining for the fabrication of disc microelectrodes. Electrochimica Acta, 1994. 39(6): p. 783-791.

163. B. Salle, O. Gobert, P. Meynadier, M. Perdrix, G. Petite and A. Semerok, Femtosecond and picosecond laser microablation: ablation efficiency and laser microplasma expansion,” Appl. Phys. A – Mat. Sci. Processing 69, S381-S383 (1999).

164. B. Rethfeld, A. Kaiser, M. Vicanek, and G. Simon, Ultrafast dynamics of nonequilibrium electrons in metals under femtosecond laser irradiation. Physical Review B, 2002. 65: p. 214303-1 - 214303-11.

165. B. Ratner, A. Hoffmann, F. Schoen, and J. Lemons, Biomaterials science, An Introduction to Materials in Medicine. 1996, San Diego: Academic press.

166. B. Min, G. Lee, S. Kim, Y. Nam, T. Lee, and W. Park, Electrospinning of silk fibroin nanofibers and its effect on the adhesion and spreading of normal human keratinocytes and fibroblasts in vitro. Biomaterials, 2004. 25(1289-1297).

167. B. Jo, L. Lerberghe, K. Motsegood, and D. Beebe, Three-dimensional Micro- channel fabrication in PDMS elastomer. Journal of microelectromechanical systems, 2000. 9(1): p. 76 - 81.

168. B. Fisette and M. Meunier, “Three-dimensional microfabrication inside photosensitive glasses by femtosecond laser,” JLMN-J. Laser Micro/Nanoengineering, 1, 7-11 (2006).

169. B. Chichkov, C. Momma, S. Nolte, F. Alvensleben, and A. Tunnermann, Femtosecond , picosecond and nanosecond laser ablation of solids. Applied Physics A: Materails Science & Processing, 1996. 63(2): p. 109-115.

170. B. Bhushan, and V. Koinkar, Nanoindentation hardness measurements using AFM. Appl. Phys. Lett., 1994. 64(13): p. 1653-1655.

214 171. Salleo, T. Sands, and F. Genin, Machining of transparent materials using an IR and UV nanosecond pulsed laser. Appl. Phys. A - Mater. Sci. Processing, 2000. 71(6): p. 601-608.

172. Tseng, Y. Chen, and K. Ma, Fabrication of high-aspect-ratio microstructures using excimer laser. Optics and Lasers in Engineering, 2004. 41: p. 827-847.

173. Rosenfeld, M. Lorenz, R. Stoian, and D. Ashkenasi, Ultrashort laser pulse damage threshold of transparent materials and the role of incubation. Appl. Phys. A, 1999. 69(Suppl.): p. S373-S376.

174. Holland, C. Cher, G. Rosengarten, and D. Simon, Plasma etching of thermoset polymer films for microchannels. Smart Mater. Struct., 2006. 15: p. S104-S111.

175. Glover, E. Illy, and J. Piper, High speed UV micro machining of polymers with frequency doubled copper vaport lasers. IEEE journal of quantum electronic, 1995. 1(3): p. 830-836.

176. Cox, A. DeWeerd, and J. Linden, An experiment to measure Mie and Rayleigh total scattering cross sections. Am. J. Phys., 2002. 70(6): p. 620-625.

177. Ben-Yakar, and R. Byer, Femtosecond laser ablation properties of borosilicate glass. J. Appl. Phys., 2004. 96: p. 5316 - 5323.

178. A. Ben-Yakar, A. Harkin, J. Ashmore, R. Byer, and H. Stone, Thermal and fluid processes of a thin melt zone during femtosecond laser ablation of glass : the formation of rims by single laser pulses. J. Appl. Phys. D, 2007. 40: p. 1447 - 1459.

215