Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Development of materials, surfaces and manufacturing methods for microfluidic applications CARL FREDRIK CARLBORG Doctoral Thesis in Microsystem Technology Stockholm, Sweden 2011 Development of materials, surfaces and manufacturing methods for microfluidic applications CARL FREDRIK CARLBORG Doctoral Thesis Stockholm 2011 The front cover photo shows a prototype of a miniaturised point-of-care test. A cartridge with integrated microfluidics and label-free slot waveguide ring resonator sensors. Superimposed on the image is an exploded view of the optical chip and the microfluidic distribution layers. Top right: An electron micrograph showing a small section of a super-lubricating microchannel. Water, supported by surface tension, is flown in the 30 µm wide channel between the two rows of 2 µm thick pillars. Bottom right: A biosensor microarray for detecting lactose intolerance, encapsulated with a microfluidic sticker. Each of the spots are 250 µm in diameter and consist of β-lactoglobulin (milk protein from cow). TRITA-EE 2011:058 ISSN 1653-4146 ISBN 978-91-7501-086-1 Microsystem Technology, KTH Akademisk avhandling som med tillstånd av Kungl Tekniska högskolan framlägges till offentlig granskning för avläggande av teknologie doktorsexamen den 23:e september 2011 klockan 10:00 i sal F3, Lindstedtsvägen 26, Stockholm. Thesis for the degree of Doctor of Philosophy at the Royal Institute of Technology (KTH), Stockholm, Sweden © Carl Fredrik Carlborg, September 2011 E-mail: [email protected] Tryck: Universitetsservice US AB, Stockholm 2011 iii Abstract This thesis presents technological advancements in microfluidics. The overall goals of the work are to develop new miniaturized tests for point-of-care diagnostics and robust super-lubricating surfaces for friction reduction. To achieve these goals, novel materials, surfaces and manufacturing methods in microfluidics have been developed. Point-of-care diagnostic tests are portable miniaturized instruments that downscale and automate medical tests previously performed in the central laboratories of hospitals. The instruments are used in the doctor’s office, in the emergency room or at home as self-tests. By bringing the analysis closer to the patient, the likelihood of an accurate diagnosis, or a quick therapy adjustment is increased. Already today, there are point-of-care tests available on the market, for example blood glucose tests, rapid streptococcus tests and pregnancy tests. However, for more advanced diagnostic tests, such as DNA-tests or antibody analysis, integration of microfluidic functions for mass transport and sample preparation is required. The problem is that the polymer materials used in academic development are not always suited for prototyping microfluidic components for sensitive biosensors. Despite the enormous work that has gone into the field, very few technical solutions have been implemented commercially. The first part of the work deals with the development of prototype pointof-care tests. The research has focused on two major areas: developing new manufacturing methods to leverage the performance of existing materials and developing a novel polymer material platform, adapted for the extreme demands on surfaces and materials in miniaturized laboratories. The novel manufacturing methods allow complex 3D channel networks and the integration of materials with different surface properties. The novel material platform is based on a novel off-stoichiometry formulation of thiol-enes (OSTE) and has very attractive material and manufacturing properties from a lab-on-chip perspective, such as, chemically stable surfaces, low absorption of small molecules, facile and inexpensive manufacturing process and a biocompatible bonding method. As the OSTE-platform can mirror many of the properties of commercially used polymers, while at the same time having an inexpensive and facile manufacturing method, it has potential to bridge the gap between research and commercial production. Friction in liquid flows is a critical limiting factor in microfluidics, where friction is the dominant force, but also in marine applications where frictional losses are responsible for a large part of the total energy consumption of sea vessels. Microstructured surfaces can drastically reduce the frictional losses by trapping a layer of air bubbles on the surface that can act as an air bearing for the liquid flow. The problem is that these trapped air bubbles collapse at the liquid pressures encountered in practical applications. The last part of the thesis is devoted to the development of novel low fluidfriction surfaces with increased robustness but also with active control of the surface friction. The results show that the novel surfaces can resist up to three times higher liquid pressure than previous designs, while keeping the same friction reducing capacity. The novel designs represent the first step towards practical implementation of micro-structured surfaces for friction reduction. iv Sammanfattning Den här avhandlingen presenterar tekniska framsteg inom forskningsfältet mikrofluidik. De övergripande målen med arbetet är att utveckla nya miniatyriserade analysinstrument för patientnära medicinsk diagnostik och stabila mikrostrukturerade ytor för friktionsreduktion i vätskeflöden över fasta ytor. För att uppnå dessa mål, presenterar avhandlingen nya material, ytor och tillverkningsmetoder för mikrofluidik. I analysinstrument för patientnära medicinsk diagnostik förminskas och automatiseras medicinska tester, som tidigare gjorts på sjukhusens centrallaboratorier, till små, portabla enheter. Användningsområdet för dessa analysinistrument återfinns på husläkarmottagningar, i akutrum på sjukhus eller i hemmet som självtester. Genom att förflytta analysen närmare patienten, kan diagnoser och dosjusteringar av medicineringar göras snabbare och mer exakta. Redan idag finns enklare produkter tillgängliga på marknaden, exempelvis blodsockertester, tester för vissa bakterieinfektioner och graviditetstester. För att kunna utveckla nästa generations instrument och för att kunna diagnostisera mer avancerade sjukdomstillstånd, är integration av avancerade mikrofluidiska komponenter, som sköter masstransport och provpreparering en förutsättning. Trots att mycket stora forskningsinsatser har lagts ned på området, har mycket få tekniska lösningar implementerats kommersiellt. Problemet idag är att de material och metoder, som används inom akademisk forskning, inte är anpassade för ändamålet och behöver förbättras eller bytas ut för att kunna tillverka relevanta kommersiella prototyper. Den första delen av arbetet ägnas åt att utveckla nya tillverkningsmetoder för att adressera ovanstående problem samt att utveckla av en ny polymerbaserad materialplattform, speciellt avpassad för de extrema krav som ställs på ytor och material i dessa miniatyriserade laboratoriemiljöer. De nya tillverkningsmetoderna möjliggör komplexa 3D strukturer och integration av material med olika ytegenskaper. Den nya materialplattformen, baserad på nya icke-stökiometriska tiolen-formuleringar (OSTE), har attraktiva material och processegenskaper för prototyptillverkning av miniatyriserade medicinska test, t.ex. kemiskt stabila ytor, låg absorption av biologiskt material, enkel och billig tillverkningsmetod samt en biokompatibel sammanfogningsprocess. Eftersom OSTE-platformen kan spegla egenskaperna på kommersiellt använda polymerer och ha en tillverkningsprocess som är tillräckligt enkel och billig för de flesta laboratorier, har den även potential att överbrygga gapet mellan forskning och kommersiell produktion. Friktionsreduktion i vätskeföden över ytor är ett viktigt område inom mikrofludik, där friktion är den dominerande kraften, men även för marina tillämpningar, där friktionsförluster svarar för en stor del av den totala energiåtgången, till exempel för oljetankers. Mikrostrukturerade ytor kan drastiskt minska friktionsförluster genom att fånga ett lager av gasbubblor på ytan som fungerar som smörjmedel för vätsketransporten. Problemet är att dessa ytor kollapsar vid de vätsketryck som på träffas i praktiska tillämpningar. Den andra delen av arbetet ägnas åt att utveckla nya lågfriktionsytor för att öka stabiliteten men även för att aktivt manipulera friktionen på ytan. De nya ytorna klarar upp till tre gånger högre vätsketryck med bibehållen friktionsreduktion än vad som kunnat visas tidigare och representerar ett första steg mot praktiska tillämpningar. v Till Pappa, som hade sett fram emot den här dagen Contents Contents vi List of publications ix Objectives and Overview xiii Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii 1 Introduction to lab-on-chip devices 1.1 Towards improved healthcare: point-of-care tests 1.1.1 What is a point-of-care test? . . . . . . . 1.1.2 Market and opportunities . . . . . . . . . 1.2 Advantages of microfluidics for medical testing . 1.3 Lab-on-chip development . . . . . . . . . . . . . 1.3.1 Background on polymer technology . . . . 1.3.2 Polymer materials . . . . . . . . . . . . . 1.3.3 Rapid prototyping . . . . . . . . . . . . . 1.3.4 Polymer microstructuring methods . . . . 1.3.5 Back-end processes . . . . . . . . . . . . . 1.4 The ideal prototyping system for labs-on-chip . . 1.5 Summary and conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Novel manufacturing methods for labs-on-chip 2.1 Dual surface-energy adhesive for the integration and packaging of an optical label-free sensor . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.1 Dual surface-energy adhesives . . . . . . . . . . . . . . . . . . 2.1.2 Background on optical biosensing . . . . . . . . . . . . . . . . 2.1.3 Background on mass transport . . . . . . . . . . . . . . . . . 2.1.4 Microfluidics design and manufacturing . . . . . . . . . . . . 2.1.5 Integration and bonding of the sensor cartridge . . . . . . . . 2.1.6 Biosensing results . . . . . . . . . . . . . . . . . . . . . . . . 2.1.7 Discussion and outlook . . . . . . . . . . . . . . . . . . . . . 2.2 High yield process of vertical interconnects in PDMS for batch manufacturing 3D microfluidics devices . . . . . . . . . . . . . . . . . . . 2.2.1 PDMS polymerisation process . . . . . . . . . . . . . . . . . . vi 1 2 2 3 4 4 4 6 8 8 10 13 15 17 17 17 19 21 23 25 27 27 28 30 CONTENTS 2.3 vii 2.2.2 Residual-free interconnects by local inhibition 2.2.3 Direct bonding using the inhibited surface . . 2.2.4 3D microfluidic networks for labs-on-chip . . 2.2.5 Discussion and outlook . . . . . . . . . . . . Summary and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 30 30 31 33 3 OSTE: a novel material toolbox for labs-on-chip 3.1 Thiol-ene "click" chemistry . . . . . . . . . . . . . . . . . . . . . . . . 3.2 OSTE: Off-stochiometry thiol-enes . . . . . . . . . . . . . . . . . . . 3.2.1 Residual activity through off-stoichiometry . . . . . . . . . . 3.2.2 Tuneable mechanical properties . . . . . . . . . . . . . . . . . 3.2.3 Direct patternable surface modification . . . . . . . . . . . . 3.2.4 Biocompatible low-temperature bonding . . . . . . . . . . . . 3.2.5 Low absorption of molecules . . . . . . . . . . . . . . . . . . . 3.2.6 Solvent resistant channels . . . . . . . . . . . . . . . . . . . . 3.2.7 A rapid and scalable manufacturing process . . . . . . . . . . 3.3 Facile integration of microfluidics with microarrays: the Biosticker . 3.3.1 Introduction to microarrays . . . . . . . . . . . . . . . . . . . 3.3.2 Mass-transport limitation in microarrays . . . . . . . . . . . . 3.3.3 The Biosticker: a micropatterned OSTE-sticker for microarrays 3.3.4 Preliminary results . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Summary and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 35 36 36 39 40 41 42 42 43 43 43 44 45 46 47 48 4 Introduction to low fluid-friction surfaces 4.1 Motivation . . . . . . . . . . . . . . . . . 4.2 Surface friction in liquid flows . . . . . . . 4.3 Superhydrophobic surfaces . . . . . . . . . 4.3.1 Mechanism of operation . . . . . . 4.3.2 Stability limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 49 50 51 51 53 5 Novel robust super-lubricating surfaces 5.1 A model for friction reduction in a microchannel . . 5.2 Fractal surfaces: temporary life support . . . . . . . 5.3 Active switching: wet or dry . . . . . . . . . . . . . . 5.4 Regulating the air pocket pressure to avoid collapse 5.4.1 Active regulation . . . . . . . . . . . . . . . . 5.4.2 Self-regulating air pockets . . . . . . . . . . . 5.4.3 Performance . . . . . . . . . . . . . . . . . . 5.5 Summary and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 55 57 57 58 58 59 60 61 . . . . . . . . . . . . . . . . . . . . . . . . . 6 Conclusions 63 Appendix: Tables 65 viii CONTENTS Summary of appended papers 69 Acknowledgement 73 References 75 Paper reprints 87 List of publications The thesis is based on the following papers in international peer reviewed journals: 1. "A packaged optical slot-waveguide ring resonator sensor array for multiplex label free assays in labs-on-chip", C. F. Carlborg, K. B. Gylfasson, A. Każmierczak, F. Dortu, M. J. Bañuls Polo, A. Maquieira Catala, G. M. Kresbach, H. Sohlström, T. Moh, L. Vivien, J. Popplewell, G. Ronan, C. A. Barrios, G. Stemme, and W. van der Wijngaart. Lab on a Chip, vol. 10, no. 3, pp. 281-290, February 2010 2. "On-chip temperature compensation in an integrated slot-waveguide ring resonator refractive index sensor", K. B. Gylfasson, C. F. Carlborg, A. Kaźmierrczak, F. Dortu, H. Sohlström, L. Vivien, C. A. Barrios, W. van der Wijngaart, and G. Stemme Optics Express, vol. 18, no. 4, pp. 3226-3237, February 2010 3. "Large scale integrated 3D microfluidic networks through high yield fabrication of vertical vias in PDMS", C. F. Carlborg, K. T. Haraldsson, M. Cornaglia, G. Stemme, and W. van der Wijngaart IEEE Journal of Microelectromechanical Systems, vol. 19, no. 5, pp. 1050-1057, October 2010 4. "Beyond PDMS: off-stoichiometry thiol-ene (OSTE) based soft lithography for rapid prototyping of microfluidic devices" C. F. Carlborg, K. T. Haraldsson, K. Öberg, M. Malkoch, and W. van der Wijngaart. Lab on a Chip, vol. 11, no. 18, pp, 3136-3147, July 2011 5. "Click Wafer Bonding for Microfluidic Devices" F. Saharil, C. F. Carlborg, K. T. Haraldsson, and W. van der Wijngaart. Lab on a Chip, submitted September 2011 6. "Sustained superhydrophobic friction reduction at high pressures and large flows" C. F. Carlborg, and W. van der Wijngaart. Langmuir, vol. 27, no. 1, pp. 487-493, December 2010 ix x LIST OF PUBLICATIONS The thesis is also based on the following international peer reviewed conference proceedings: 7. "Continuos flow switching by pneumatic actuation of the air lubrication layer on superhydrophobic microchannel walls C. F. Carlborg, M. Do-Quang, G. Stemme, G. Amberg, and W. van der Wijngaart. in IEEE Proceedings of the 21th Int. Conf. on Micro Electro Mechanical Systems, Tuscon, USA, January 2008, pp. 599-602 8. "Biosticker: Patterned microfluidic stickers for rapid integration with microarrays" C. F. Carlborg, M. Cretich, K. T. Haraldsson, L. Sola, M. Bagnati, M. Chiari and W. van der Wijngaart. in Proceedings of the 11th Int. Conf. on Miniaturized Systems for Chemistry and Life Sciences (µTAS), Seattle, USA, October 2011, accepted The contribution of Carl Fredrik Carlborg to the different publications: 1. part of design, all packaging and microfluidic fabrication, major part of experiments and major part of writing 2. part of design, all packaging and microfluidic fabrication, part of experiments, and writing 3. major part of design, fabrication, all experiments and major part of writing 4. major part of design, all of fabrication, major part of experiments and writing 5. major part of design, part of fabrication, experiments and writing 6. major part of design, all fabrication, experiments and writing 7. major part of design, all fabrication, major part of experiments and writing 8. all design, fabrication, major part of experiments and all writing xi The work presented in the thesis has also been presented at the following international peer reviewed conferences: 1. "Reliable batch manufacturing of miniaturized vertical vias in soft polymer replica molding" C. F. Carlborg, K. T. Haraldsson, G. Stemme, and W. van der Wijngaart. in Proceedings of the 11th Int. Conf. on Miniaturized Systems for Chemistry and Life Sciences (µTAS), Paris, France, October 2007, pp. 257-259 2. "Microchannels with Substantial Friction Reduction at Large Pressure and Large Flow" C. F. Carlborg, G. Stemme, and W. van der Wijngaart. in Proceedings of the 22rd Int. Conf. on Micro Electro Mechanical Systems, Sorrento, Italy, January 2009, pp. 39-42, oral presentation 3. "A packaged optical slot-waveguide ring resonator sensor array for multiplex label free assays in labs-on-chip", K. B. Gylfasson, C. F. Carlborg, A. Kaźmierczak, F. Dortu, M. J. Bañuls Polo, A. Maquieira Catala, G. M. Kresbach, H. Sohlström, T. Moh, L. Vivien, J. Popplewell, G. Ronan, C. A. Barrios, G. Stemme, and W. van der Wijngaart. in Proceedings of the 13th Int. Conf. on Miniaturized Systems for Chemistry and Life Sciences (µTAS), Jeju Island, South Korea, November 2009, pp. 2004-2006, oral presentation 4. "Large scale integrated 3D microfluidic networks through high yield fabrication of vertical vias in PDMS", C. F. Carlborg, K. T. Haraldsson, M. Cornaglia, G. Stemme, and W. van der Wijngaart in IEEE Proceedings of the 23rd Int. Conf. on Micro Electro Mechanical Systems, Hong Kong, P.R. China, January 2010, pp. 240-243, oral presentation 5. "Beyond PDMS: off-stoichiometry thiol-ene (OSTE) based soft lithography for rapid prototyping of microfluidic devices" C. F. Carlborg, K. T. Haraldsson, K. Öberg, M. Malkoch, and W. van der Wijngaart. in Proceedings of the 14th Int. Conf. on Miniaturized Systems for Chemistry and Life Sciences (µTAS), Groningen, Netherlands, October 2010, pp. 70-72, oral presentation 6. "Low temperature "Click" wafer bonding of off-stoichiometry thiol-ene (OSTE) polymers to silicon" C. F. Carlborg, F. Saharil, K. T. Haraldsson, and W. van der Wijngaart. in Proceedings of the 15th Int. Conf. on Miniaturized Systems for Chemistry and Life Sciences (µTAS), Seattle, USA, October 2011, accepted xii LIST OF PUBLICATIONS Other international peer reviewed journal papers, by Carl Fredrik Carlborg, not included in the thesis: 1. "Thermal boundary resistance between single-walled carbon nanotubes and surrounding matrices", C. F. Carlborg, J. Shiomi, and S Maruyama. Physical Review B, vol. 78, no. 20, pp. 205406, 2008 2. "Poly(vinyl alcohol) as a temporary carrier for fabrication of fragile membranes and 3D fluidic networks" J.M. Karlsson, T. Haraldsson, C.F. Carlborg and W. van der Wijngaart. Journal of Micromechanics and Microengineering, manuscript 3. "Low-stress transfer bonding and assembly of multiple wafer-sized polymer layers using floatation" J.M. Karlsson, T. Haraldsson, C.F. Carlborg and W. van der Wijngaart. Sensors and Actuators B, submitted August 2011 Objectives and Overview This thesis presents technological advancements in microfluidics. The overall goals of the work are to develop new miniaturized tests for point-of-care diagnostics and robust super-lubricating surfaces for friction reduction. To achieve these goals, novel materials, surfaces and manufacturing methods in microfluidics have been developed Structure Chapter 1 gives a general introduction to the development of laboratories-on-chip with specific focus on polymer materials. Chapter 2 introduces novel manufacturing methods for the development and integration of microfluidics with labs-on-chip, and their applications. This chapter also describes some of the design issues encountered in the design of microfluidic components for medical sensors. Chapter 3 introduces a novel, highly versatile prototyping material aiming to bridge the gap between academic proof of concept devices and commercial products. Chapter 4 gives an introduction to super-lubricating surfaces, their uses and their limitations. Chapter 5 presents two approaches to break the robustness limitation of current superlubricating surfaces towards implementation in liquid flow conditions encountered in realistic applications. The concluding chapter summarizes the work presented in this thesis. xiii Chapter 1 Introduction to lab-on-chip devices Accidentally your leg is scratched while out jogging, there is some blood but it does not looks so serious so you forget about it. In the evening your leg hurts and you notice that from the scar, dark lines are travelling up along your leg. You get a little worried and call your local clinic. The nurse urges you to immediately go to the emergency room for a check-up. At the hospital they suspect blood poisoning and take blood samples to determine the type of the bacteria and thus what antibiotics they can use. Normally this test is done in the central hospital laboratory and takes days. You do not have that much time, because by now you have a fever and are shaking uncontrollably. The first thing the doctors do is to give you a large dose of a broad spectrum antibiotics and hope that the bacteria will be responsive to at least one of them. Unfortunately, repeated use of broad spectrum antibiotics is one of the major causes of antibiotic resistance. Luckily, the hospital was recently equipped with the latest point-of-care system for rapid analysis of antibiotic resistance bacteria, and with only a small amount of blood the device quickly determines exactly what antibiotics should be used. Twenty minutes after you arrive, you are intravenously administered the correct antibiotics. You are able to leave the hospital after two days of observation. In another scenario, you suspect your three years old son is allergic, but you do not know to what. He recently got red rashes over his whole body when he ate and you are worried what other symptoms he will show. At the doctors office, they suggest to do an allergy test by applying small droplets of different allergy causing substances over the whole back of your son and use a small needle to make it penetrate into the skin. You know that this is painful and could induce a serious allergic shock if the allergy is severe. Another clinic uses point-of-care devices for allergy testing that need only a droplet of blood from the index finger, to exactly determine what your son is allergic to. These are two examples of scenarios for which labs-on-chip are developed at the Microsystem Technology Laboratory at KTH. This introduction gives a brief overview of the intended usage of labs-on-chip and the most common materials and manufacturing processes. Also, the reasons for the lack of success of academic research in the consumer market are discussed briefly. 1 2 1.1 1.1.1 CHAPTER 1. INTRODUCTION TO LAB-ON-CHIP DEVICES Towards improved healthcare: point-of-care tests What is a point-of-care test? A point-of-care test (POCT) is defined as any medical test close to the patient: in the doctor’s office, by the hospital-bed, or at home. This closeness to the patient is believed to increase the likelihood of a quick and accurate diagnosis and therapy adjustment, or give the patients the convenience of avoiding hospital appointments. In contrast to centralised laboratory machines, which are usually based on large and bulky pipetting robots, a POCT should ideally be portable, easy to use and present a first level of interpretation of the result to the user. To qualify as a successful POCT [1], it is usually required that the sensitivity and specificity are equivalent or better than centralised laboratory tests and that the cartridges are self-contained, disposable and low-cost. To realise these goals many POCT are built around a technology called labs-on-chip; miniaturised automated laboratories that use the advantages of microfluidics to compete with classical culture bottles, petri dishes and microtitre plates. The anatomy of a visionary lab-on-chip for point-of-care is displayed in Figure 1.1. It consists of a disposable part, with a loading port for the liquid sample, a sample preparation and metering unit (preconcentration, amplification, cell lysis), a microfluidic network (splitting, moving and mixing sample and reagents) and a sensor and signal transduction part with receptors for labelled or label-free detection. The disposable part is inserted into a reader, containing all the electronic part (signal processing) and the user interface (display and buttons). Figure 1.1. An example POCT as envisioned by IBM Corporation [1]. 1.1. TOWARDS IMPROVED HEALTHCARE: POINT-OF-CARE TESTS 1.1.2 3 Market and opportunities During 2009 the market for POCTs still remained only a fraction (14% or 6.9$ billion) of the total in-vitro diagnostic (medical tests) market [2]. The POCT market is dominated by a small number of established point-of-care products, most notably glucose tests, rapid streptococcal tests and pregnancy tests. However, these POCTs are technologically relatively simple and cannot handle more complex diagnostic tasks, such as nucleic acid or antibody analysis, required, in for example, antibiotic resistance screening [1]. To realise complex diagnostic devices, microfluidic components are required to handle sample transport and preparation. Miniaturisation of fluidic components is currently a major focus of many academic groups and diagnostic companies. However, there are not only technological hurdles for the wider adaptation of more advanced POCTs. The complex structure of the diagnostic market as well as stringent regulation from public health agencies contribute to the slow growth experienced the last years. Moreover, the reimbursement model for diagnostic devices, the acceptance by medical personnel, complex business models and large geographical differences may contribute to the difficulties in replacing centralised testing with POCTs [1]. Nevertheless, there are a few pioneering POCTs on the market that uses microfluidics. Among these are Abbott’s i-Stat (blood gas analysis) and Biosite’s Triage System (immunoassays), both shown in Figure 1.2. As POCTs become accepted as trusted alternatives to centralised medical testing, they have potential to revolutionise the healthcare market, in particular home-tests which represents a large untapped segment. An important user group will also be developing countries where inexpensive POCTs will significantly improve the accessibility and quality of healthcare [3]. Figure 1.2. Two commercial POCT labs-on-chip based on microfluidics currently on the market. Left: Abbott’s i-Stat for electrolytes and blood gases tests. Right: Biosite Triage System for immunoassays. 4 CHAPTER 1. INTRODUCTION TO LAB-ON-CHIP DEVICES 1.2 Advantages of microfluidics for medical testing Labs-on-chip are built in dimensions measured in micrometers and handle liquids measured in nanoliters. There are several reason why labs-on-chip are built that small. In the microscale world, things are quite different from how we are used to perceive them in macroscale world. With distances shrunk a million times, from meters to micrometers, the surface-to-volume ratio increases linearly and surface dependent forces, such as friction, dominate completely over volume related forces, such as inertia. In microfluidics, the immediate effect is that fluid flows are completely laminar, heat transfer is very fast, the diffusion times are short, and the likelihood of a molecule interacting with a channel surface is high. In addition to smaller instruments and less consumption of expensive reagents, the scaling effects brings additional advantages: • Well defined laminar flow, the sample is easy to control in the chip as the flow is completely laminar • Faster reaction times, reactions happen faster when diffusion lengths are shorter • High degree of parallelization, many tests can be done in parallel on the same droplet of sample The benefits offered by downscaling for medical diagnostics and chemical analyses are undisputed and today almost all new diagnostic kits or analytical systems use miniaturised components or effects associated with the microscale. 1.3 Lab-on-chip development The development of a lab-on-chip is a multidisciplinary challenge, involving many different disciplines. This first part of the thesis deals mainly with the engineering and integration aspects of developing prototypes for labs-on-chip, and not with the equally important biochemical and medical aspects. Successful development relies heavily on proper choice of materials and manufacturing methods. The following section aims to give a short, non-exhaustive, review on the development process of lab-on-chip devices and to identify important limitations related to materials and manufacturing methods. 1.3.1 Background on polymer technology A short introduction to polymer science In this section follows an introduction to some of the definitions in polymer science encountered later in the thesis. This is far from an exhaustive presentation of polymer science and the brave and curious is recommended to open Principles of Polymerization by George Odian [4]. 1.3. LAB-ON-CHIP DEVELOPMENT 5 General To start from the very begining: A polymer is a molecule composed of several repeating structural units, monomers, connected by covalent bonds. Polymerisation is the act of combining these monomers into long chains which can be either linear or branched. Cross-linking is the formation of covalent bonds between polymer chains. Curing is cross-linking of thermoset resins. The degree of polymerisation defines to what degree the monomers have reacted. The gel point occurs when the polymer first starts to solidify and form an infinite network, i.e. a gigantic "molecule". At the gel point, viscosity changes abruptly and the mobility of the unreacted monomers decreases. If the gel point occurs at a high degree of polymerisation, the polymer network will have very little built in stress. Polymer growth Polymer growth can occur by step-growth, chain-growth or mixed step-growth/chain-growth. In step-growth, the polymer network is formed by monomers reacting in a stepwise reaction between functional groups on the monomers to first form dimers than trimers, longer oligomers and finally chains. In step-growth polymerisation, the molecular weight increases at a slow rate, delaying the gel point until very late in the polymerisation and very high degrees of polymerisation can be obtained (>95%). Step-growth polymerisation results in polymer chains with an alternating sequence of monomers (ABABAB) and a high network homogeneity. Chain-growth polymerisation involves opening up unsaturated monomers and adding them to the reactive end groups of the growing polymer chain. In chain-growth polymerisation, the individual polymer chains grow very rapidly, creating a dense, nested, yarn-like structure that leads to an early gel-point and eventually will hide the reactive end-groups from the unreacted monomers. This leads to a lower degree of polymerisation than for step-growth polymerisation. Moreover, during chain-growth, the polymer chains can react with themselves and form loops, resulting in a non-homogenous network. In mixed chain-growth/stepgrowth the two polymerisation processes compete with each other. Radical polymerisation The polymerisation reaction can start spontaneously or more commonly with the help of an initiator, that creates a reactive intermediate compound capable of successively linking monomers to polymer chains. Initiators are commonly triggered by heat or light irradiation. The most widely used initiators produce free radicals that attack and open unsaturated bonds. The free radical polymerisation can be divided into three steps: Initiation, when the first active centre is created from which a polymer chain is formed. This is usually accomplished by an initiator that splits into two radical fragments upon actuation. Propagation, when the reactive end-groups of the growing chains react with new monomers in a serial fashion, where each reaction event recreates the active radical on the last added monomer. Termination, can be accomplished in many different ways, for instance when there are no reactive groups left, when two radicals combine or by disproportionation. Chain transfer occurs when the reactive site of a growing polymer chain is transferred to another molecule. This could have the effect of increasing 6 CHAPTER 1. INTRODUCTION TO LAB-ON-CHIP DEVICES the degree of polymerisation by transporting the reactivity of "hidden" reactive sites out of the "yarn ball", to initiate a new chain of propagation steps. Bulk properties The physical properties of the polymer are strongly dependent on the length and size of the polymer chains, to what degree they have branched and cross-linked with each other, as well as on the mechanical properties of the individual monomer molecules. The glass transition temperature Tg , defines the temperature above which the polymer chains vibrational energy exceeds the van der Waal forces keeping them together. Above the glass transition temperature the chains can slide relative to each other, and the polymer can deform. The transition does not happen exactly at Tg , but in an interval around this temperature. The size of this interval actually reveals something about the nature of the polymer network. A broad transition indicates an in-homogenous network with chains of different lengths. A narrow transition indicates a homogeneous network, with very well-defined lengths and sizes of the polymer chains. Typically, chain-growth polymers have a broad glass transition and step-growth polymers have a narrow glass transition, reflecting the difference in their polymerisation mechanisms. Figure 1.3. The loss tangent as a function of temperature for photopolymerised films of an epoxy-acrylate and a thiol-ene. The peak of the loss tangent curve indicates the glass transition temperature. The narrow glass transition of the thiol-enes compared to most other polymers indicates exceptionally homogenous networks. Reproduced from Ref. [5] . 1.3.2 Polymer materials Thermoplastics Thermoplastics are polymers that are not covalently cross-linked (Figure 1.4), and melts at temperatures above Tm and freeze to a glassy state at temperatures below Tg . From a manufacturing point of view, the main advantage of thermoplastics 1.3. LAB-ON-CHIP DEVELOPMENT 7 is their ability to be melted and reshaped against a mould, enabling production of thermoplastic parts with high throughput. From a lab-on-chip perspective, the availability of many commercial, medical grade formulations is a great advantage. The stiff mechanical properties also provide structural support and protect the sensor and the microfluidic network. However, many solvents, common in chemical analysis and separation, dissolve thermoplastics. Nevertheless, most commercial labs-on-chip are made of thermoplastics. Common thermoplastics used in microfluidics are poly(metylmethacrylate) (PMMA), polycarbonate (PC) and COC (Cyclic Olefin Copolymer). Thermosets Thermosets, are polymers that are covalently cross-linked (Figure 1.4), and thus do not melt. From a manufacturing point of view, thermosets are shaped during the polymerisation and cross-linking process. Because of the covalent bond formation, thermosets exhibits higher residual stress, shrinkage and crack-formation compared to thermoplastics. From a lab-on-chip perspective, the main advantages of the thermosets are their geometrical stability and solvent resistance. Common thermosets used in microfluidics are poly(dimethylsiloxane) PDMS (an elastomer), the hard resist SU-8 (MicroChem, USA), and the optical glue NOA81 (Norland Products, Inc, USA), that has recently been used for solvent resistant microlfuidics [6]. Figure 1.4. The difference in structure between thermoplastics and thermsets. Elastomers (PDMS) An elastomer is a rubbery and elastic polymer; it can be a thermoplastic or a thermoset. It has few cross-links between the chains and thus a low E-modulus and high yield strength, compared with other materials. In microfluidics, the thermoset elastomer poly(dimethylsiloxane) PDMS is the dominating material for prototyping microfluidic devices. PDMS is easy to handle in small laboratories, flexible but sturdy enough to manipulate, biocompatible, inert and easily bonded to silicon or 8 CHAPTER 1. INTRODUCTION TO LAB-ON-CHIP DEVICES glass using oxygen plasma treatment. Because of the few cross-links and high filler content, the polymerised films experience only moderate shrinkage (1-3%) during polymerisation [7] and can replicate nanometer sized features. However, high volume fabrication schemes for patterned PDMS layers are currently not available. From a lab-on-chip perspective, the elastic mechanical property has enabled integration of pneumatically actuated valves and pumps in microfluidic chips, and direct sealing to smooth surfaces. Moreover, PDMS is resistant to high temperatures, oxidation and many chemical and biological environments. It also exhibits high gas permeability, which is important for many living cell studies. However, an important limitation with PDMS is the difficulty to permanently modify the surface, due to the high mobility of the polymer chains. Furthermore, the polymer network absorbs small molecules [8], leaches uncured monomers [9] and swells in solvents [10]. For example, it was shown that PDMS implanted into dogs absorbed 0.7 % of its own weight in small molecules, mainly lipds [11]. Several attempts have been made to improve PDMS. Notably, Rolland et al. developed a fluorinated elastomer with similar mechanical properties as PDMS, but with dramatically improved solvent resistance and Kyung et al. [12] developed a photocurable PDMS for faster curing. Others have tackled the absorption and swelling problem by coating the inside of PDMS channels with different polymers or inorganic materials to block the diffusion of small molecules [13, 14, 15]. 1.3.3 Rapid prototyping The rapid prototyping of microfluidic devices, and in particular of labs-on-chip places high demands on the resolution of the microstructuring process. In general high-resolution replica moulding must be used to reproduce smooth microand nanoscale features. In replica moulding a replication master, with the inverse geometrical features with respect to the finished device, is first produced using lithographic techniques. The replication master is used in a polymer replication process, such as hot embossing, injection moulding or casting. An important limiting factor in lab-on-chip prototyping is the time-consuming back-end processes. These processes are often performed in a serial fashion and comprise, for instance, drilling, surface modification, biofunctionalisation and bonding. The back-end processes are estimated to make up 80% of the total cost and time of lab-on-chip prototyping, and commercial manufacturing [16]. 1.3.4 Polymer microstructuring methods Casting Casting involves pouring a liquid thermoset prepolymer on the replication master and curing using heat or UV-light (Figure 1.6). Casting is an uncomplicated process well suited for small-scale rapid prototyping, as it typically requires a very small investment in equipment (UV-lamp or oven) and requires little or no process optimisation. PDMS is casted in a process called soft lithography. However, there are 1.3. LAB-ON-CHIP DEVELOPMENT 9 Figure 1.5. A schematics of the development process of labs-on-chip. no high volume commercial casting processes for the replication of micron-sized features, which limits the capability to scale up production of the prototyped device. Casting is a planar process and the realisation of 3D structures requires drilling or punching vertical interconnects between stacked channel layers. Hot embossing Hot embossing involves heating a thermoplastic substrate to just above Tg under vacuum (Figure 1.6) and press it against a replication master to imprint microstructures. Hot embossing requires investments in expensive equipment to handle pressure, heat and vacuum control. The process requires optimisation of temperature and embossing time for each new pattern and has a cycle time for a 4" wafer around 4-15 minutes [16]. Hot embossing works well with small feature sizes and has excellent dimensional control, but has problems with high aspect ratios and can only replicate planar features [17]. Typical thermoplastic materials used are PMMA, PC and COC. The cost of a hot embossing system starts around 10k$ [16]. Microinjection moulding Injection moulding typically involves melting thermoplastic pellets that are injected into a closed replication tool at a high pressure and a high temperature. As injection moulding is the dominant replication techniques for plastics in general, most commercial labs-on-chip parts are manufactured this way. However, academic access to injection moulding equipment is limited because of very high machine costs (>75k$) as well as maintenance costs [16]. Moreover, there are many process parameters to be optimised, and it usually takes considerable time to achieve a good micro-structured polymer. The moulding process has a high degree of automation and moulding times vary from 30 sec to 2 min [17]. Thermosets can also be injection molded but the process is generally more complex. 10 CHAPTER 1. INTRODUCTION TO LAB-ON-CHIP DEVICES Figure 1.6. Comparison of the casting process (left) and hot embossing process (right) 1.3.5 Back-end processes Ports and interconnects Both casting and hot embossing are planar processes producing polymer sheets with microstructures on one side. Ports for fluidic access must be opened through the polymer layers to connect to the microfluidic channels, so called chip-to-world connections. Moreover, to realise 3D structures these layers must be stacked on top of each other and vertical interconnects must be defined between them. Ports and interconnects can be drilled in thermoplastics and punched in PDMS, but the resolution and spacing between the holes is limited. Injection moulding of thermoplastics can, in contrast to casting and hot embossing, directly mould 3D structures. Surface modification and control Surface modification in labs-on-chip devices is one of the most important steps in the manufacturing process. Common tasks for surface modification in a lab-on- 1.3. LAB-ON-CHIP DEVELOPMENT 11 chip, exemplified in Figure 1.7, includes blocking of non-specific binding of proteins [18], spatial control of surface wetting [19, 20], and attachment of bioreceptors [21]. Modification of the surface properties can be achieved in various ways: grafting of polymer chains on the surface [22], chemically attaching mono-layers of molecules [23] or physically adsorbing molecules on the substrate [24, 25]. Moreover, in many applications, surface modification must be spatially controlled to specific areas on the chip. With the proper choice of chemistry both grafting and chemical linking can be patterned using UV-light and a stencil mask. Physical deposition may be patterned using a physical mask during the deposition process or by micro-contact printing [26, 27]. However, as the deposited material is not bonded to the surface, diffusion may lead to low resolution or adsorption into the substrate. Microfluidic devices made of stiffer materials, such as thermoplastics or thermosets, provide good substrates for stable and permanent surface modification. On the contrary, devices prototyped in PDMS will have problems with permanent channel surface modifications, requiring complicated workarounds, as described earlier. Figure 1.7. A fictive lab-on-chip with a hydrophobic valve for timing of sample reaction with reagent, anti-fouling coating to avoid loss of sample at the walls and a detection zone coated first with a linker layer and subsequently spotted with antibodies. Bonding and integration When selecting bonding method for labs-on-chip, biocompatibility is a major concern as bonding normally constitutes the last step in the fabrication scheme. Typically, bio-reactive molecules, e.g. antibodies or antigens, are deposited prior to bonding. Functionalisation after bonding, in a closed off device, is a complex and slow process requiring individual filling of each channel. In Table 2.1, features of different bonding techniques for polymers in microfludics are listed. CHAPTER 1. INTRODUCTION TO LAB-ON-CHIP DEVICES 12 Medium High Medium-high Medium-high High Low High Strength Process complexity Long Short Medium Medium Low Shortmedium Short Bonding time PMMA, PMMA, PMMA PMMA, PDMS PDMS PMMA, [41] COC COC Ref Bio compatible No Yes Yes Yes No Yes Yes PMMA [42] Materials Yes PDMS [43, 44] PC, COC PC, COC Yes PC, PDMS [39, 40] [28, 29, 30] [31, 32, 33, 34] [35] [36, 37] [38] Yes Comparison of existing bonding techniques for polymer microfluidic Thermal bonding Solvents Ultrasonic welding Laser welding Plasma treatment Clamping Gluing Medium Low Low Medium High Medium Low Mediumhigh Lowmedium Low Table 1.1. devices. Adhesive films Medium-high Medium Bonding method Off-stoichiometry Medium-high LowMedium Medium Surface modification 1.4. THE IDEAL PROTOTYPING SYSTEM FOR LABS-ON-CHIP 13 As seen from the table, the biocompatibility requirement directly disqualifies thermal bonding, the most common bonding technique for thermoplastics, as well as plasma bonding, the most common technique for joining two PDMS layers. Ultrasonic welding and laser welding are capable of focusing energy only to the interface and bond thermoplastics, but require substantial process optimisation, as well as advanced equipment. This leaves solvent bonding, physical clamping, gluing, adhesive films, surface modifications and off-stoichiometry bonding. Out of these both clamping and off-stoichiometry are specific to PDMS. Rubbery materials like PDMS, can be physically clamped but the adhesion is low and high clamping pressures risks deforming the channels. By mixing PDMS in offstoichiometric ratios: one layer with excess of the base (vinyl groups) and one layer with excess of curing agent (Si-H groups), they can be covalently bonded to each other without using plasma treatment. [42]. In solvent bonding, the topmost surface of a thermoplastic polymer is dissolved using a solvent to achieve chain-entanglement across the bonding interface. Approaches where the solvent is only applied to the top layer have been shown, which makes the bonding process biocompatible[34]. Glues [39, 40] can be used to join thermoplastics at room temperature and do not require high temperatures [45]. However, the application process is critical, and care must be taken not to accidentally fill channels [46]. Double-sided adhesive films [41] do not block the channels, but are sensitive to solvents and create channels with different top and bottom surfaces. A major problem is that most glues and adhesives do not adhere to the rubbery surface of PDMS, which is the material of choice in academic proof of concept devices. The surfaces of PDMS and thermoplastics may be modified with reactive molecules that can covalently link to each other under biocompatible conditions, such as polymer coatings [43] or organofunctional silanes [44]. 1.4 The ideal prototyping system for labs-on-chip The ideal prototyping method for labs-on-chip is fast, relies on inexpensive materials, allows for 3D features (through holes) and do not require access to expensive/technically complicated facilities. The ideal prototyping material is a tuneable material that can recreate pneumatic valves and pumps but also can provide structural stability and external interface, such as manifold integration and direct tubing connections. It is chemically inert, compatible with chemical and biological samples without absorbing them and allows stable and patternable surface modification for control of wetting and biological functionalisation. Finally the material allows for biocompatible bonding to a wide range of substrates. These properties are concretized in the list below: 1. Tuneable mechanical properties. The mechanical properties of an ideal prototyping material are somewhat contradicting. On the one hand, it must mirror the stiffness of commercial thermoplastics, to produce geometrically 14 CHAPTER 1. INTRODUCTION TO LAB-ON-CHIP DEVICES stable microfludic chips with robust external chip-to-world interfaces. On the other hand, it must be soft enough to allow for the pneumatically actuated valves and pumps, commonly used in PDMS. 2. Chemical inertness and low interaction with the sample. To be able to analyse low concentration samples, the ideal material must not absorb small molecules, such as proteins or DNA from the sample, react with the sample or leach uncured components that may interact with the sample or the sensor. 3. Solvent resistance. Critical for many chemical reactions and separation processes is the use of harsh solvents. The ideal material must therefore not dissolve or swell in these solvents. 4. Direct, patternable and stable surface modifications. The spatial control of surface properties is instrumental for controlling liquids and immobillizing biological receptors on the chip. The ideal material allows for spatial control of surface modifications, preferably without the need to first activate the polymer surface by plasma or solvents. 5. Fast, scalable and utilizing inexpensive materials and processes. The prototyping method has a fast curing/microstructuring step and uncomplicated back-end processes, to allow a rapid development process. To be useable in academic research, the ideal prototyping method relies on inexpensive materials and do not require access to expensive/technically complicated facilities. To allow a fast transition to commercial production, the method is possible to scale up to medium or large-scale production. 6. Three-dimensional microfluidics. Advanced labs-on-chip must be able to handle multiple liquids, something which often requires 3D microfluidic channels with under- and overpasses. The ideal prototyping method therefore allows for efficient fabrication of multiple vertical interconnects between channel layers. 7. Biocompatible bonding. Essential for labs-on-chip is an uncomplicated and biocompatible bonding method to surfaces that are functionalized with proteins and DNA. The ideal prototyping method form a strong bond to a wide number of materials under biocompatible conditions. In academia, soft lithography in PDMS is the method predominately used for rapid prototyping of microfluidic devices. For some applications, PDMS is the perfect material, in particular for cell studies, as it is easy to structure and permeable to oxygen. However, PDMS cannot accomplish some of the basic features of an ideal lab-on-chip material, such as low absorption of molecules from the sample and stable surface modifications. Moreover, the planar fabrication method complicates fabrication of 3D microfluidic networks. Biocompatible bonding is possible using off-stoichiometric mixtures, but only to other PDMS layers. Furthermore, concerns 1.5. SUMMARY AND CONCLUSIONS 15 have been raised that devices developed in PDMS will be difficult to commercialise due to the extensive redevelopment required to transfer them into commercial thermoplastic devices [16, 47]. The question naturally arises why thermoplastics are not used directly in the prototyping process? After all, thermoplastic materials address many of the properties of the ideal prototyping material, such as chemical inertness, direct patternable surface modifications, biocompatible bonding (adhesive bonding, laser or ultrasonic welding) and 3D microfluidics (injection moulding). The reasons why thermoplastics have not gained foothold in academic prototyping lie in its cost and complexity. Although the thermoplastic material in itself is cheap, the cost of equipment and facilities is high. Moreover, extensive process-optmisation is required for each new design. Surprisingly, there have been few, if any, attempts to develop a polymer system specifically suited for lab-on-chip applications that respond to the challenges listed above. 1.5 Summary and conclusions The increasing availability of diagnostic point-of-care devices has potential to increase the quality of the healthcare system by putting the power of large centralised laboratories into the hands of doctors and patients. The portability of the tests and the closeness to the patient, are believed to increase the likelihood of a quick and accurate diagnosis and therapy adjustments at the hospital bed, in the doctor’s office, or at home. Furthermore, portable point-of-care tests will improve the accessibility and quality of healthcare in the developing world, where the access to hospital is limited. The first generation of point-of-care tests is already available on the market: blood glucose tests, rapid streptococcal tests and pregnancy tests. However, to enable more complex tasks such as nucleic acid or antibody analysis, microfluidics must be integrated to handle sample preparation and transport. The development of these technologies is currently the focus of many academic groups. However, there is a material and manufacturing bottleneck limiting the transfer of technologies from academia to commercial products. Firstly, the most commonly used material for lab-on-chip proof-of-concepts PDMS, has some serious drawbacks, preventing or complicating the development of many important functions in labs-on-chip. Secondly, prototypes in PDMS must be completely redeveloped, both from a material and manufacturing perspective, to be transferred to into commercial thermoplastic device production. The following two chapters address both of these problems. In Chapter 2, two important improvements of the PDMS manufacturing process for labs-on-chip are introduced and demonstrated. In Chapter 3, a novel prototyping material free from the limitations of PDMS and with potential to bridge the development gap between academia and commercial products is presented. Chapter 2 Novel manufacturing methods for labs-on-chip This chapter introduces and demonstrates two novel manufacturing methods for proof-of-concept POCTs, using soft lithography in PDMS. Specifically, a method for biocompatible bonding of PDMS to thermoplastics and a high yield fabrication method of 3D microfluidic devices are introduced. Both of these methods aim to remove limitations related to the PMDS material discussed in Chapter 1. In the first section, a method for joining materials with different surface energies, such as PDMS to thermoplastics, is presented. The first section also presents the packaging and integration of an optical label-free sensor, in which the novel bonding method is used. Furthermore, it discusses design-issues related to the optical sensor and considerations for the mass-transport of analyte down to the sensor surface. The second section describes a high yield process for creating vertical interconnects using a novel method to inhibit the PDMS polymerisation at specific locations. Furthermore, the second chapter briefly examplifies a lab-on-chip application in which the inhibition method is used. 2.1 2.1.1 Dual surface-energy adhesive for the integration and packaging of an optical label-free sensor Dual surface-energy adhesives Due to the difference in physicochemical properties, thermoplastic materials do not generally form irreversible bonds with PDMS, even after oxygen plasma treatment or heating. However, the very different material properties of thermoplastics and elastomers, offer many opportunities for hybrid microfluidic devices using the advantages of both. Elastomers enable pneumatic valves, thermoplastics enable a variety of reliable external interface options, such as manifold integration, direct tubing connections, and gasket connections. Therefore, a general technique for integrating materials with different surface energies in a rapid and uncomplicated fashion, is 17 18 CHAPTER 2. NOVEL MANUFACTURING METHODS FOR LABS-ON-CHIP presented in Paper 1. Figure 2.1. Dual surface-energy adhesive Only a very few strategies exist to bond PDMS to thermoplastics, notably CVD processes or silane coatings [44, 48]. These methods typically require several lengthy steps: plasma treatment (∼1 min), incubation with the silane linker (∼20 min) and bonding (∼10 min) [48]. Another alternative, both faster and more flexible, is the use of patterned double-sided pressure adhesive films. However, the glue found on most adhesives is an acrylic based glue which has no or limited adhesion to the low-energy PDMS surface. In one attempt to overcome this problem, uncured PDMS was spun and subsequently thermally cured on top of the adhesive film [49]. However, this technique involved several complicated steps in preparing the adhesive film and was considerd too elaborate. Instead, Paper 1 describes a rapid and uncomplicated bonding method using a patterned dual surface-energy adhesive film (5302A, Nitto Denko, Japan). This adhesive film, designed for the mobile phone industry to attach the key pads made out of rubber, has one side coated with a silicon based pressure sensitive glue, and one side with an acrylic based glue. The silicon glue bonds efficiently with low energy surfaces like PDMS, while the acrylic glue bonds well to the PMMA . Before bonding, openings are cut out in the adhesive film for fluidic and optics ports. The bond strength of the tape to PDMS and PMMA was characterised by peeling the tape from the substrate at 180◦ (Table 2.1). Table 2.1. Peeling strength of Nitto Denko 5302A (N/20mm width) Surface PDMS PMMA Acrylic side 0.15 20 Silicon side 5 10 The advantage of the dual surface energy film is rapid bonding of PDMS without having to use liquid glue or plasma treatment. It allows for a rapid and uncomplicated bonding, under biocompatible conditions with a high yield. The drawback is solvent incompatibility. 2.1. DUAL SURFACE-ENERGY ADHESIVE FOR THE INTEGRATION AND PACKAGING OF AN OPTICAL LABEL-FREE SENSOR 2.1.2 19 Background on optical biosensing Optical sensing is a powerful technology for label-free detection for point-of-care tests. It offers the sensitivity required for detecting the low concentrations of analytes present in bodily fluids, is free from electrical interference, has a wide dynamic range and multiplexing capability. One important drawback of optical sensors is their sensitivity to temperature which, without proper temperature control, may cause unwanted drift in the output signal. Many of the analytical instruments used in universities and pharmaceutical companies are based on optical transducers. For example, both the Biacore system (GE Healthcare, Uppsala), the gold standard in protein interaction measurements, and the AnaLight system (Farfield, UK), a system for protein conformation measurements, use optical transducers. Accordingly, to bring these powerful analytical tools into the hands of a wider user-base, there is a strong interest in integrating optical sensors in labs-on-chip. Even though an abundance of optical sensor principles has been demonstrated, very few have successfully been integrated in complete labs-on-chip. Some electromagnetism All molecules interact to a varying degree with electromagnetic fields that pass through them. The electrons in the molecule experience a force when they are exposed to the oscillating electromagnetic fields of light. If a molecule has free electrons, it will be polarised by the electric field resulting in the formation of an electric dipole. The extent of polarisation is dependent on the size, shape and orientation relative to the electric field. The quantity known as electric susceptibility, χe of a molecule quantifies its polarisability. When a polarisable molecule is placed in an oscillating electromagnetic field, such as light, the electrons within the molecules will start to oscillate and produce a current. This modifies the local relative permittivity #r of the dielectric material, which in turn modifies the local refractive index √ √ n = #r = 1 + χe . It is this local change in refractive index, that results in a modified light propagation speed, that can be measured by optical sensors, not the mass directly. This is possible since an optical waveguide in an aqueous medium will couple some of its energy into the surrounding water, this field is known as the evanescent field and extends typically some 100 nm from the surface. When biological molecules, which have a higher electrical susceptibility than water, interact with this field they change the local dielectric constant. Optical waveguide biosensors can thus directly probe their surroundings by a label-free detection method. Slot-waveguide ring resonator Great effort has been put into improving the resolution, or limit of detection, of optical biosensors through the design of resonator structures with extremely high quality factor, or Q. The quality factor is defined as Q = λ0 /∆λ, where λ0 is the centre wavelength and ∆λ is the spectral with of of the resonance determined at half the peak maximum. The higher the Q, the narrower is the resonance and a lower 20 CHAPTER 2. NOVEL MANUFACTURING METHODS FOR LABS-ON-CHIP limit of detection can be achieved. High Q-resonators can be achieved by coupling light into a circular waveguide that can fit an integer number of wavelengths around the perimeter of the circle. For these modes, standing waves or resonance will be supported in the ring. To excite the rings, light from a tuneable laser is coupled into the rings through a straight waveguide passing very close to the ring. As the wavelength of the tunable laser is swept, some wavelengths will match a resonant wavelength of the ring and light of that wavelength will be coupled into the ring and disappear from the output spectra. As the analyte attaches to the receptor ligand on the ring resonator surface, the local refractive index is modified, the resonance frequency of the ring changes and the dip in the output spectrum will move. These types of sensors have been demonstrated to have high sensitivity and limit of detection [50, 51, 52, 53]. However to increase the interaction with the sample and enhance the sensitivity, the slot-waveguide strategy was introduced [54, 55]. Essentially this strategy takes advantage of a slot in the waveguide, small enough that a large portion of the electromagnetic field power propagates inside the liquid filled open slot. In this way the interaction with the sample can be increased. Moreover, as the water filled slot has a negative thermo-optic coefficient, it can balance the positive value of both silicon oxide (bottom) and silicon nitride (wave guide) and contributes to render the waveguide temperature insensitive, as discussed in Paper 2. This solves the temperature sensitivity mentioned earlier as one of the most important drawbacks of optical sensors, and is a huge advantage for the practical implementation of POCT based on optical biosensors. In Paper 1, six active and two references ring resonators were multiplexed on the chip enabling the simultaneous measurements of several analytes. Figure 2.2. The slot waveguide enables increased light/sample interaction and thus higher sensitivity, since up to 40% of the total optical power [54] propagates in the liquid filled slot as illustrated in the cross-section above. 2.1. DUAL SURFACE-ENERGY ADHESIVE FOR THE INTEGRATION AND PACKAGING OF AN OPTICAL LABEL-FREE SENSOR 21 Optical chip The optical sensor, partly designed and fully fabricated by Dr. Kristinn Gylfasson at KTH, consisted of a silicon chip with patterned waveguides and an optical distribution network in silicon nitiride on thermally grown oxide. The surface was covered with a cladding of silicon oxide, except over sensor rings, where the cladding was removed. An overview of the optical components is shown in Figure 2.3. Before the assembly, the nitride ring resonators were selectively coated with glutaraldehyde using a process developed and applied by Dr. María José Bañuls Polo, and her collegues, at the Department of Chemistry, Universidad Politéchnica de Valencia (Valencia, Spain). It allows for functionalisation of only the nitride waveguides while leaving the silicon oxide uncoated [56]. Figure 2.3. A top view of the layout of the optical chip: light is injected at the surface grating coupler (C) and split by a multi-mode interference splitter (B) to the six transducer channels M1 to M6 and the two reference channels REF1 and REF2. Inset are: an optical micrograph of the splitter (B), electron micrographs of the grating coupler (C), and a slot waveguide ring resonator (A), with an enlargement of the coupling region between the straight waveguide and the ring. The optical chip was partly designed and fully fabricated by Dr. Kristinn Gylfasson at KTH 2.1.3 Background on mass transport A sensor with a potentially low limit of detection and high sensitivity is no guarantee for a good assay performance. The surface chemistry plays an essential role to efficiently capture the analyte from the sample and the mass transport of the analyte to the sensor is also critical in governing the dynamics of the sensor response, and 22 CHAPTER 2. NOVEL MANUFACTURING METHODS FOR LABS-ON-CHIP ultimately the performance of the assay. Three physical processes control the mass transport to the sensor surface: diffusion, convection and surface reactions. In the design of the microchannels, the interplay between these three effects must be carefully considered and modelled. The full problem can be numerically solved, but an analysis with the help of dimensionless numbers, comparing the relative importance of these three effects, is often enough for the initial design of a device. The transport-processes in biosensors have been reviewed in some excellent reviews, such as those by Gervais et al. [57] and Squires et al. [58]. Below follows a short introduction to three common dimensionless number used in the design of Labs-on-chip. Figure 2.4. Model system for mass-transport analysis [58]. Solution with target concentration c0 flows with a volumeric flow rate Q through a channel of height H and width Wc over a sensor of length L and width Ws that is functionalised with θs receptors per unit area. The binding reactions have constants kon and kof f and the analyte diffusion constant D. Imagine a microchannel where the target analyte flows with the volumeric flow rate Q through a channel of height H, width Wc and where one wall contain a sensor of width Ws , and length L. The sensor is functionalised with θs receptors per unit area, and the solution contatins target molecules with concentration c0 and diffusivity D. The analyte molecules binds to the receptors at the surface with binding constants kon and kof f . In static conditions, when there is no flow, Q = 0, diffusion is the only transport process from the bulk to√the sensor, and a depletion zone will be formed above the sensor with radius δ = Dt. The collection rate at the sensor can in this case be approximated by jD ∼ Dc0 /δ. As the depletion zone grows in the channel, the diffusion flux becomes smaller and collection rate of analyte molecules at the sensor surface decreases. Since diffusion in slow, the accumulation of enough analyte on the surface may take hours or days in low concentration solutions [59]. If convective transport is added to the model, the growth of the depletion zone 2.1. DUAL SURFACE-ENERGY ADHESIVE FOR THE INTEGRATION AND PACKAGING OF AN OPTICAL LABEL-FREE SENSOR 23 is halted and gives a steady state depletion zone above the sensor. The Péclet number characterise relative strength of convection and diffusion and the nature of the mass-transport depletion zone around the sensor. The depletion zone is thin compared to the channel if P e # 1, and the steady state diffusion zone δs above the sensor will be characterised by the distance at which the time for an analyte molecule to convect past the sensor is exactly equal to diffuse down to the sensor. At higher flow rates the depletion zone will be thinner and collection rate at the surface will increase. Contrary, if Pe $ 1, the depletion zone will extend far upstream and the transport of analyte will be limited by diffusion, thus increasing the collection time scale at the sensor. At sufficiently low Péclet numbers, the diffusion zone will exactly balance convection and all the analyte molecules in the sample can be collected by the sensor, but at the expense of decreased collection rate at the sensor surface. The time scale of the surface reaction compared to the diffusion time scale give rise to another dimensionless parameter called the Damköhler number which captures how quickly the sensor equilibrates. If Da # 1, the equilibration is limited by the rate of target diffusion to the sensor, whereas the reaction itself limits the sensor kinetics if Da $ 1. Normally one strives to engineer a microfluidic system so that equilibration is reaction-limited in order to measure the reaction kinetics rather than the diffusion kinetics. In summary, if the chip is designed for short assay times, the mass transport should be optimised to be reaction-limited with a high Peclet number to minimise the size of the diffusion-limited depletion zone. This can be realised for example by using a surface chemistry with high ligand density and use a high flow rate. If the chip must economise with a small volume of analyte, and require a high capture rate, the flow rate must be kept low enough to balance diffusion. Table 2.2. Dimensionless numbers relating the three transport processes in Labson-chip. Re Pe Da 2.1.4 Reynolds Péclet Damköhler ρU0 H/η Q/DWs kon θs H/D inertial/viscous diffusive time/convective time reaction time/diffusive time Microfluidics design and manufacturing This section deals with the design and manufacturing of the microfluidic network for the optical sensor chip. Mass transport analysis In the design of the microfuidics, the dimensionless numbers were calculated to ensure the channel dimensions and flow speed would not limit the performance of 24 CHAPTER 2. NOVEL MANUFACTURING METHODS FOR LABS-ON-CHIP the sensor. As a model system we used the anti-BSA to BSA binding reaction [60] which was also tested during the experiments. With a flow of Q = 10 µL/min, a typical diffusion constant for small proteins of D = 10−10 m2 /s and with a sensor size of Ws = 200 nm, the Peclet number was calculated to, P e = 8340 # 1, ensuring a thin depletion layer compared to the channel height. Using the the surface area of a BSA molecule (56 nm2 ) and its molecular weight (66 kDa) [61], the BSA surface coverage was estimated to 1.9 ng/mm2 . The Damköhler number for the system was calculated to Da = 1 × 10−5 , indicating a reaction-limited system. However, there was a concern that the small dimension of the slot, 200 nm wide and 300 nm high, would limit the mass transport of analytes to and from the inner surfaces of the slot to such a high degree that it would impair the performance of the sensor. A numerical investigation of the transport of analyte down to the sensor also revealed that the concentration of analyte on the inner sides of the slot compared to the top surface of the waveguide was up to ten times lower, limiting the advantage of the slot for fast binding kinetic measurements. Figure 2.5. Results of the numerical investigation of mass transport. Mass transport into the slot is diffusion limited, illustrated by the stream lines, and the concentration of analyte (surface plot) is up to ten times lower in the slot than above the slot at steady state. This will require longer incubation time than expected to take full advantage of the slot waveguide’s increased sensitivity. Layout The microfluidic layer, illustrated in Fig 2.6 (C) consisted of an array of six channels (M1–M6), each addressing one slot waveguide ring resonator. The channels are 200 µm wide and 20 µm high over the sensor. The distance between the channels is set by the minimum spacing required to avoid channel-to-channel leakage, and the minimum channel width by the need for manual alignment of the microfluidic layer to the optics chip. The elastomer PDMS was chosen as material for the microfluidic network partly because of its availability and its ease to manufacture in the laboratory, but also for its adhesion to silicon oxide through plasma activation of the PDMS. The microfluidic layer was molded using soft lithography and through holes were punched as liquid input and output ports as well as for laser access. The total area of the 2.1. DUAL SURFACE-ENERGY ADHESIVE FOR THE INTEGRATION AND PACKAGING OF AN OPTICAL LABEL-FREE SENSOR 25 Figure 2.6. (A) A photograph of one of the fabricated cartridges. The steel tubes glued to the hard plastic shell provide a stable fluidic interface to the microfluidic network below. At the front long edge, the through hole for laser access down to the optical chip is visible. The edge of the optical chip itself can also be seen in the cutout region of the hard plastic shell. (B) The inset shows one of the ring resonators in its fluidic channel. (C) The fluidic layout with individual channel to each sensor (M1 to M6). silicon chip is 40 x 15 mm2 , most of which is used as support for the microfluidic channels and connectors. The array format with one channel for each sensor, enable the simultaneous measurements of different liquids, and a more reliable diagnosis. Reference channels also enables a thermal compensation technique that is instrumental for achieving low detection limits in optical sensors. 2.1.5 Integration and bonding of the sensor cartridge Microfluidics and optical multiplexing was integrated on-chip, while fluidic pumps, tuneable laser source, read-out sensor and electronics were external in an instrument developed by Dr. Andrzej Kaźmierczak and Dr. Fabian Dortu at Multitel (Mons, Belgium). An exploded schematic of the integrated cartridge is shown in Figure 2.7 and a photo of the chip, ready to be inserted into the read out system in Figure 2.6. Reversible clamping was first evaluated as a bonding method for the microfluidic layer to the optical chip, but it was difficult to create an even clamping pressure over the chip, and the 20 µm high channels were often compressed in some part of the 26 CHAPTER 2. NOVEL MANUFACTURING METHODS FOR LABS-ON-CHIP chip. Instead, the PDMS microfluidic layer was plasma activated to expose Si-OH groups, which were covalently reacted with the uncoated silicon oxide surface of the optical chip to form Si-O-Si bonds. Fig 2.6 (B) shows photo of a ring resonator sensor in one of the microfluidic channels. The hard plastic manifold in PMMA and the dual surface-energy adhesive film (Nitto Denko, 5302A) were micro-milled with openings for optics and fluidic ports. The adhesive film was subsequently aligned and applied to the PMMA manifold, which in turn was aligned and applied to the laminated optical and microfluidic chip. Figure 2.7. A schematic exploded view of the sensor cartridge above the alignment platform, exposing the 4 permanently bonded layers of the cartridge: the optical chip, the microfluidic layer, the adhesive film, and the hard plastic shell. Cutouts in the hard plastic shell free the edge of the precision cut silicon optical chip for accurate alignment against 3 pins protruding from the alignment platform of the read-out instrument. Light is coupled in from the top via a surface grating coupler, and collected at the long edge of the optical chip by imaging the output facets on a 1D InGaAs photodiode array. Fluidic ports for sample injection are formed by steel tubes glued into the hard plastic shell. 2.1. DUAL SURFACE-ENERGY ADHESIVE FOR THE INTEGRATION AND PACKAGING OF AN OPTICAL LABEL-FREE SENSOR 2.1.6 27 Biosensing results The performance of the cartridge was evaluated both on spotted and unspotted sensors. In Figure 2.8 (A), the sensor is functionalised only with glutaraldehyde and injected anti-BSA attaches non-specifically to the surface. In Figure 2.8 (B), the sensor has been spotted with BSA on top of the glutaraldehyde surface before the assembly. Anti-BSA is flushed through the channel and specifically binds to the spotted BSA molecules. The spotting of the surface was done by Dr. Gerhard Kresbach at Zeptosens - A Division of Bayer (Schweiz) AG, (Witterswil, Switzerland). Further results, available in Paper 1, shows a surface mass detection limit of 0.9 pg/mm2 , which is among the best reported for integrated ring resonator sensors. Figure 2.8. A comparison of real-time sensing results on unspotted and spotted sensors. (A) Shows the resonant wavelength shift as a function of time during injection of an increasing concentrations of anti-BSA on a sensor with only glutaraldehyde. (B) Shows the resonant wavelength shift of a sensor spotted with BSA on top of the glutaraldehyde surface. More results are available in Paper 1. 2.1.7 Discussion and outlook The design, fabrication, and characterisation of a packaged array of optical refractive index sensors, integrated with microfluidic sample handling in a compact cartridge was demonstrated. Using the individually addressable transducers available on the chip, we separated and compensated for different kinds of external disturbances, resulting in much improved noise level, compared to our previously published results. The multiplexed layout, with multiple sensor with separated flow compartment also allowed for temperature compensated measurements, demonstrated in Paper 2. The temperature compensation, combined with efficient mass transport and the high sensitivity of the fabricated slot-waveguide ring resonators, yielded one of the best mass detecion limis reported so far for integrated ring resonator systems. The result is also good compared to other label-free methods, but it must be kept in mind that for a POCT ten minutes is a relatively long time and the smallest concentration detectable during this time was 0.4 nM of anti-BSA (0.0615 µg/ml) which is higher than the concentration of many biomarkers in the blood (pM to fM) 28 CHAPTER 2. NOVEL MANUFACTURING METHODS FOR LABS-ON-CHIP [1]. Part of the problem is the diffusion limitation of analytes down into the 200 nm wide slot, which limits the performance of the slot waveguide for sensitive surface mass measurements. From a manufacturing point of view, the novel use of a dual surface-energy adhesive to bond PDMS to thermoplastics, enabled a rapid and biocompatible bonding method with tightly spaced microfluidic ports. By using the adhesive film instead of gluing or surface modification, the time spent on back-end processes could be reduced. However, from a material point of view, PDMS itself was problematic. If the PDMS adsorbed proteins, it could have led to a lower concentration of analyte over the sensor than in the initial sample. If this was the case, it would have led to an underestimation of the sensor performance. Furthermore, PDMS absorbed the methanol and ethanol used for the calibration of the sensor and slowly released them in all channels . However, a solution for the PDMS problem has to wait until Chapter 3. 2.2 High yield process of vertical interconnects in PDMS for batch manufacturing 3D microfluidics devices In the quest to integrate more fluidic functions in labs-on-chip the complexity of the channel network inevitably increases. To freely interconnect different on-chip regions requires three dimensional (3D) channel topology with under- and overpasses to allow the liquids to cross without mixing. Although, injection molding can to some extent produce 3D microfluidic chip parts, both casting and hot embossing are planar fabrication methods. Three-dimensional microfluidic fabrication using these methods, must use at least two microfluidic layers with well defined vias, i.e. out of plane interconnecting fluidic channels to connect the layers (Figure 2.9). Figure 2.9. Planar replication methods require vertical interconnects to create 3D devices and fluidic ports. These are often punched or drilled as a part of the back-end processes. In both hot embossing and casting, interconnecting vias can only be defined after the microstructuring process, by drilling [62, 63], laser ablation [64] or dry etching [65, 66] or punching (PMDS). This manual approach is acceptable for a low number of vias or fluidic ports spaced far apart, but for multiple tightly spaced interconnects the via creation must be integrated in the microstructuring process. 2.2. HIGH YIELD PROCESS OF VERTICAL INTERCONNECTS IN PDMS FOR BATCH MANUFACTURING 3D MICROFLUIDICS DEVICES 29 Many labs-on-chip make use of integrated pneumatic valves in PDMS[42]. These devices consists of two layers of PDMS, one layer for fluidics and one layer with pneumatic control channels. When the pneumatic channels are actuated the membrane, formed where fluidic and pneumatic channels cross, deflects and blocks the fluidic channel. However, in many applications it is necessary to connect these two layers. The problem is to create a large number of vias for large scale integrated (LSI) microfluidic networks. In previous attempts, the strategy have been to remove the prepolymer at the location of the via holes before polymerisation. Anderson et. al introduced the use of two-level masters with protruding posts at the via locations [67] (Figure 2.10). After pouring the liquid polymer on the master, a cover plate is pressed on top at high pressure to squeeze away the prepolymer of the top level mold features [67, 68, 69]. After curing the plate is removed and the resulting PDMS film with open vias holes is aligned and bonded to a second layer. Unfortunately it is difficult to achieve a high yield using this method, since it is not possible to squeeze away all prepolymer from the protruding vias features with only clamping and a blocking, thin residual membrane often have to be removed manually or via some dry etching process. Other techniques involve spinning [70, 71] or blowing [72] with gas to remove the prepolymer, all suffering from uneven surface at the vias locations due to surface tension effects. Figure 2.10. Previous attempts for batch manufacturing of vertical interconnects have focused on removing prepolymer before polymerisation. In this work, the goal was to develop a more reliable process for the creation of residual-free interconnects for applications in labs-on-chip. Because of the difficulties in completely squeezing away the prepolymer on top of the protruding mold features, a novel strategy was used in Paper 3, to prevent the prepolymer to polymerise at the interconnecting via locations. 30 CHAPTER 2. NOVEL MANUFACTURING METHODS FOR LABS-ON-CHIP 2.2.1 PDMS polymerisation process Most researchers working with microfluidics and labs-on-chip have at some point mixed PDMS. However, its composition and how it polymerises is less known. There are two main suppliers of PDMS: Dow Corning (Sylgard 184) and Momentive (RTV 615). They are both provided as two components: one base and one cross-linking agent. The curing proceeds through hydrosilylation involving the addition reaction between polysiloxane in the base containing vinyl groups and the cross-linking agent containing Si-H functional groups [4]. The reaction relies on a platinum catalyst. 2.2.2 Residual-free interconnects by local inhibition By removing the platinum from the prepolymer exactly at the via locations, polymerisation will be inhibited at those specific locations. In the literature, many substances have been reported to inhibit the polymerisation of PDMS, but one of the most efficient are the tertiary amines. The tertiary amines create a chelating complex with the platinum atom and prevents it from catalysing the polymerisation [73]. When a glass plate covalently coated with a layer of aminosilanes is clamped on top of the PDMS prepolymer, the amines capture the platinum close to the surface. Because the diffusion of new platinum catalyst is limited in the squeeze film on top of the protruding vias features platinum will be completely depleted only at those locations. The result is polymerisation everywhere except at the via locations. The inhibiting plate can be removed after polymerisation and with it follows the liqud prepolymer at the vias location, leaving membrane-free vias in the PDMS film. 2.2.3 Direct bonding using the inhibited surface A bonus, resulting from the above technique, is an alternative bonding method. At the surface of the inhibiting glass plate, enough new platinum can be resupplied from the bulk to polymerise the prepolymer, but to a lesser degree of cross-linking. When demolded, this surface is a little sticky and tape-like. It has good adhesion to most surfaces and if new platinum is resupplied, for example by stamping, it can bond covalently to another PDMS layer. 2.2.4 3D microfluidic networks for labs-on-chip The inhibition technology has so far been successfully implemented in lab-on-chip projects at KTH by Mikael Karlsson. In a project to detect antibiotic resistance bacteria from whole blood, the technology is used to create vias and ultra-thin membranes [74]. The project integrates on-chip PCR with optical detection of bacterial DNA. Figure 2.13 shows a dual layer microfluidic device for on-chip PCR, fabricated in PDMS using the inhibition technology. Another application is 3D distribution network for cell sorting during sample preparation [75]. 2.2. HIGH YIELD PROCESS OF VERTICAL INTERCONNECTS IN PDMS FOR BATCH MANUFACTURING 3D MICROFLUIDICS DEVICES 31 Figure 2.11. PDMS is catalysed by Pt atoms and polymerises at elevated temperatures. When a chelating agent (AEAPS) is coated on glass plate it captures the Pt atoms and prevent PDMS from polymerising. 2.2.5 Discussion and outlook Using localised inhibition of the PDMS polymerisation we have demonstrated an uncomplicated batch process with a high yield to create membrane-free vertical vias in PDMS using soft lithography. The method enables complex 3D microfluidic devices with small and tightly spaced interconnects (Figure 2.12) by eliminating the need for manual punching of holes in PDMS layers or cleaning blocked vias from residual membranes and has already been successfully implemented in a number of applications. While succesful in many applications, the procedure is limited to surface inhibition and requires two-level molds. 32 CHAPTER 2. NOVEL MANUFACTURING METHODS FOR LABS-ON-CHIP Figure 2.12. This 3D basketweave structure was created using the localised inhibition technique described in this section. Figure 2.13. An example application of the via inhibition technology. A dual layer microfluidic chip for on-chip PCR with vertical interconnects, fabricated by Mikael Karlsson (Microsystem Technology, KTH) 2.3. SUMMARY AND OUTLOOK 2.3 33 Summary and outlook This chapter introduced two novel improvements of the manufacturing of PDMS microfluidic devices and exemplified them through applications. The dual surface-energy adhesive film enabled the integration of PDMS microfluidic layers with thermoplastic cartridge shells and allowed for rapid and leak-tight assembly with multiple, tightly spaced fluidic ports without leakage. The localised inhibition technology of PDMS enabled complex 3D microfluidic devices to enable handling of multiple liquids on a single chip. However, many material issues still remain with PDMS as a lab-on-chip material and it is time to set the stage for a successor. Chapter 3 OSTE: a novel material toolbox for labs-on-chip This chapter introduces a novel prototyping polymer system aiming to bridge the gap between academic proof-of-concepts and commercial products. The novel polymer toolbox was developed with consideration to the requirements of an ideal prototyping system discussed in Chapter 1 (Section 1.4). The main objective was to develop a versatile toolbox for academic development of lab-on-chip proof-of-concept devices, free from many of the adverse properties of PDMS, while at the same time being compatible with commercial prototyping, i.e. to mirror the properties of commercial thermoplastics. In the first section of this chapter, a background of the versatile thiol-ene polymer system is given. Next, the novel off-stoichiometry thiol-ene (OSTE) polymer is presented along with its powerful features for rapid prototyping of microfluidics for labs-on-chip (Paper 4 and Paper 5 ). Finally, an example application of the OSTE-polymers for microfluidic integration of microarrays is presented. 3.1 Thiol-ene "click" chemistry The field of thiol-ene chemistry has during the last decades obtained a renewed interest, due to the unique chemistry, an the versatility of thiol-ene chemistry as a polymer forming reaction in many applications. Thiol-enes are formed by the radical initiated reaction between a multifunctional thiol and an "ene" monomer. The "ene" stands for alkene, which is a hydrocarbon with at least one carbon-to-carbon double bond. The unique feature of the thiolene systems is that they polymersise mainly through step-growth instead of to chain-growth, which is more common in radical polymerisation reactions. This is especially important during network formation of thermoset polymers where the step-growth meachnism will lead to high control of the polymerisation process, a homogenous structure, a late gel-point, minimal amount of unreacted spieces, and significantly lower stresses than in other thermosets. 35 36 CHAPTER 3. OSTE: A NOVEL MATERIAL TOOLBOX FOR LABS-ON-CHIP A typical thiol-ene polymer is made up of two types of monomers, one with a thiol-functional group xR1 -(SH)m , and the other with an "ene" functional group, yR2 -(CH=CH2 )n , where x and y are the number of monomers of each type and m and n are the number of functional group on each. As with all previously demonstrated thiol-ene systems, one normally strives to maximize the mechanical strength, and the ultimate goal is to have an exact equal amount of functional groups xm = yn, with few or no reactive groups remaining. The resulting polymer has a non-reactive surface that requires surface treatment, e.g plasma exposure, to change the surface properties. Bowman and Hoyle [76] have shown that the kinetics can be varied to a large extent by appropriate choice of "ene" monomers. One particular interesting group of "enes" is the allyls. They polymerise uniquely via step-growth (no homopolymerisation via chain-growth), and have an almost complete (99.9%) degree of polymerisation, and minimal homopolymerisation (Figure 3.1 [76]. The reaction between allyls and thiols also belongs to a family of especially well-behaved lock-and-key reactions often called "click" reactions. The term click chemistry, coined by Sharpless [77] is a class of efficient and very selective chemical reactions that are used to join molecules together in a rapid manner with high yield, high purity and little or no by-products. In essence, two specific monomers are joined as easily as "clicking" together two matched buckle pieces. This is particularly important during surface modifications, when only one specific surface reaction is desired without unexpected side-reactions. To picture the network structure produced by a stoichiometric mix of multifunctional thiols and allyls, it could be useful to compare it with the crystal structures of solids. For example, the ideal polymer network produced by a tri-functional thiol and di-functional allyl, would have similarities with haematite (Fe2 O3 ) shown in Figure 3.2. Each thiol monomer has three covalent bonds and each allyl monomer has two covalent bonds. Recently, using commercial thiol-ene glues (e.g. NOA 81, Norland Products) microfluidic devices have been fabricated with multiple layers, each of which is partially cured prior to lamination creating a "sticky" interface that, when re-exposed to UV-light, complete the polymerisation. The partially cured layers are useful for bonding but the the method allows little control of the thickness of the inhibited layer. Furthermore, it provides no discrimination between thiol or "ene" groups on the surface and the sticky surface has limited shelf life. 3.2 3.2.1 OSTE: Off-stochiometry thiol-enes Residual activity through off-stoichiometry In Paper 4, off-stoichiometric formulations of thiol-enes are introduced for the first time in microfluidics. These materials have an excess of one of the functional groups, xm &= ym, to achieve a polymer with remaining unreacted functional groups both in the bulk and on the surface (Figure 3.3). The novel use of intentional large 3.2. OSTE: OFF-STOCHIOMETRY THIOL-ENES Figure 3.1. Mechanism of an ideal radical thiol-ene coupling (e.g. thiol-allyls and thiol-vinyls). Figure 3.2. Because of the pure step-growth mechanism in the thiol-ene reaction, the structure of haematite (Fe2 O3 ) can be qualitatively compared with to the networks produced by tri-functional thiol monomers and di-functional allyls. 37 38 CHAPTER 3. OSTE: A NOVEL MATERIAL TOOLBOX FOR LABS-ON-CHIP off-stoichiometry mixing ratios in thiol-enes opens up the possibility to fine-tune the mechanical properties and the glass transition temperature as well as producing robust and rapid surface modification and bonding processes. Instead of a fully cross-linked network, some of the multifunctional monomers in the OSTE-polymers are not fully reacted and are only partially attached to the network, similar to Figure 3.2 but with some links free and detached from the network. This has immediate effects on the mechanical properties of the polymer. Furthermore, the unreacted groups are available to use as anchors for further click surface modification or bonding after complete polymerisation, thus circumventing the need for breaking up the polymer surface before grafting. The unreacted groups remain stable in ambient atmosphere after months of shelf-time. Figure 3.3. Excess reactive groups using off-stoichiometric ratios The pure step-growth polymerisation of the thiol-allyl systems enables a high degree of control of the number of unreacted groups available after polymerisation. By choosing the appropriate multifunctional monomers, surprisingly large off-stoichiometric ratios, up to 100% excess of either thiol or allyl functionality can be achieved without losing mechanical stability and without excessive leakage of non-crosslinked material. Under the assumption of total mobility of all the monomers and equal reactivity of all the functional groups, the minimum number of monomers that has no bonds to the network can be approximated by multiplying the probability that each of the functional groups on the monomer in excess will not find a corresponding partner. For instance, in a polymer composed of tetrathiol and triallyl, with 50% excess of thiol groups (xm/yn = 1.5), the probability is (1/3)4 = 1.3% that a tetrathiol monomer will not attach to the network and be leachable. Lower functionality of the monomer in excess increases the leaching. Figure 3.4 shows the theoretical amount of non-crosslinked monomer for the two dual monomer systems used in Paper 4 and Paper 5 together with the experimentally determined leakage for the same systems. The leaching from both systems is slightly higher than the minimum theoretical value which is due to the fact that not only monomers but also oligomers and initiators that are not attached to the network can be leached. This increases the amount of extractable material. 3.2. OSTE: OFF-STOCHIOMETRY THIOL-ENES 39 Figure 3.4. Theoretical leaching of non-crosslinked components from the OSTEpolymers as a function of off-stoichiometry ratio. The line represents excess of monomers with n=4 functional groups and the dotted line with n=3 functional groups. The experimental results for OSTE Allyl (30) (n=3) and OSTE Thiol (90) (n=4) is also plotted. 3.2.2 Tuneable mechanical properties The advantage of the OSTE-polymers over standard thiol-ene polymers is that the mechanical properties can be fine tuned to the exact demands of the application by adjusting the off-stoichiometric ratio, without changing the type of monomers. In off-stoichiometry, the monomers have fewer cross-links to the network, which in turn affects the E-modulus and the glass transition temperature. Mechanical properties have been demonstrated ranging from harder than PMMA to soft like PDMS, using off-stoichiometry of thiol groups, and glass transition temperatures Tg , ranging from below 30 ◦ C to 84 ◦ C (Figure 3.5 left). Particularly useful is the ability to use heat to tune the stiffness of the polymers, which is explained further in Paper 8. By choosing the glass transition temperature slightly above room temperature, the mechanical properties of the OSTE-polymers can be tuned in a narrow temperature interval so that the polymer rapidly transforms from a glassy material (E-modulus in the GPa range) at room-temperature, to a rubbery material (E-modulus in the MPa range) at a temperature only a few 40 CHAPTER 3. OSTE: A NOVEL MATERIAL TOOLBOX FOR LABS-ON-CHIP tens of degrees higher. For example, the OSTE-polymer with 90% excess described in Paper 4 remains medium stiff (200 MPa) up until 30 ◦ C, when it suddenly starts to soften and is ten times softer at 50 ◦ C (∆ T = 20 ◦ C). This is also the case for OSTE Thiol (70), which is developed with biocompatible bonding in mind (Figure 3.5 right) and which softens at Tg = 37 ◦ C. Figure 3.5. Left: E-modulus and glass transition temperature of the OSTE Thiol as the off-stoichiometric mixing ratio is varied. Right and bottom photos: Temperature tuning of the OSTE Thiol (70). Heating to Tg allows for perfect sealing during bonding. 3.2.3 Direct patternable surface modification The OSTE-polymers provide anchors on their surfaces that can be used to chemically link or graft functional groups or polymer chains using UV-initiated thiol-ene click chemistry. With the OSTE-polymers, the surface coverage can be controlled by the ratio of off-stoichiometry. At 5% off-stoichiometry, macroscopic surface properties such as contact angle, can be modified when grafting hydrophilic PEG monomers. At higher off-stoichiometry ratios the PEG grafting results in an even larger modification of the contact angle up to a threshold value, when steric effects limits the surface coverage. The direct modification is an advantage compared to thermoplastics and PDMS, where the polymer surface first must be broken up to expose reactive groups, for example by plasma treatment and the density of active sites, as well as their surface homogeneity, is less controlled. An additional advan- 3.2. OSTE: OFF-STOCHIOMETRY THIOL-ENES 41 tage is that the grafting of functional groups can be UV-patterned using a stencil mask, creating areas with different surface properties on a single chip, for example hydrophobic and hydrophilic areas, as shown in Figure 3.6. Figure 3.6. Principle of direct UV-grafting on the OSTE-polymers 3.2.4 Biocompatible low-temperature bonding The OSTE-polymers can form covalent bonds directly with other materials, using the anchors of unreacted groups available on the surface. In contrast to solvent bonding and tape bonding, no leachable compounds that reduce bond integrity remain in the bond area. In one type of bond, the OSTE-polymer is directly covalently bonded to another OSTE-polymer with the opposite type of anchors exposed. For example, an OSTE-polymer with an excess of thiol groups bonds covalently with an OSTE-polymer with an excess of allyls groups after unfiltered UV-exposure. A lower wavelength (< 250 nm) [78] is required for the reaction to occur since no initiator is available at the interface. Another type of bonding situation occurs when the OSTE-polymer is bonded to a surface with a coating capable of reacting with thiols or allyls. This can for example be a surface of isocyanate (a common coating on sensors) or vinyl silane (Paper 5 ), a gold surface [79] (a common sensor surface) or activated esters (NHS) (Paper 8 ). The particular advantage with the OSTE- polymers compared to thermoplastics in bonding, is the temperature tuning 42 CHAPTER 3. OSTE: A NOVEL MATERIAL TOOLBOX FOR LABS-ON-CHIP as described above. When heated above Tg , the OSTE-polymers conform perfectly to micro-irregularities on the substrate and can react with a very high yield and form strong bonds (Figure 3.7). Figure 3.7. Principle of bonding the OSTE-polymers at their glass transition temperature. 3.2.5 Low absorption of molecules The dense network of the OSTE-polymers results in a significantly lower absorption of small molecules than in PDMS. The low absorption enables handling of analytes in very small concentrations, as very little of the sample is lost into the channel material. In Figure 3.8, the absorption of Rhodamine B in OSTE and PDMS was compared after 24 h of exposure. Only small amounts are adsorbed at the wall of the microchannel in the OSTE-polymer wheras a considerable amount of Rhodamine B has been absorbed by the PDMS. Figure 3.8. Comparison of absorption of small molecules (Rhodamine B) between OSTE and PDMS after 24 hours of exposure. No diffusion is observed in the OSTE sample. In the PDMS sample a large portion of the Rhodamine B has diffused into the walls. 3.2.6 Solvent resistant channels Solvent resistance is an important feature in particular for applications in chemical analysis and separation. The OSTE polymers based on the monomers used in Paper 4, 5 and 8 are compatible with common solvents such as toluene, iso-propanol, 3.3. FACILE INTEGRATION OF MICROFLUIDICS WITH MICROARRAYS: THE BIOSTICKER 43 ethanol, methanol and glycerol. As seen in table 3.1, an OSTE-polymer microchannel bonded to silicon showed good compatibility with alcohols and toluene, but solvents with large dipole moments, i.e. DMSO and acetone was not well tolerated. To increase the compatibility with these solvents, the monomers must be replaced. Table 3.1. Solvent compatibility of the OSTE-polymer using the monomer composition from Paper 5. Solvent Isopropanol Methanol Acetone Toluene Glycerol DMSO DI water 3.2.7 After 24 hours No visible effect No visible effect Bulk material failure No visible effect No visible effect Bulk material failure No visible effect A rapid and scalable manufacturing process The OSTE-polymers can be casted on the same type of masters as PDMS, and polymerised in seconds using light from a standard table top UV-lamp (365 nm). During demolding, the master can be heated to the glass transition temperature to facilitate the release of the polymer. Very stiff OSTE polymers, with an Emodulus close to 2 GPa, was successfully demolded from SU-8 masters using this technique. After demolding, the chip can be cut using the same trick to heat it to the glass transition temperature during processing or diced at room temperature using an automated dicing machine. Moreover, due to the low built-in stresses, the OSTE material lends itself well to machining and milling producing smooth edges without cracks. The low shrinkage of the OSTE-polymers also allows for wafer-level integration of silicon substrates (Paper 5 ), also shown in Figure 3.9, which significantly decreases the back-end process time and enables integration with CMOS electronics [80]. 3.3 3.3.1 Facile integration of microfluidics with microarrays: the Biosticker Introduction to microarrays Microarrays have become powerful tools in biochemical analysis, primarily because of their capability for highly multiplexed analysis. A microarray consists of a solid surface with a dense matrix of receptor ligands, probes, deposited in certain volumes by a liquid dispensing robot, producing "spots" 44 CHAPTER 3. OSTE: A NOVEL MATERIAL TOOLBOX FOR LABS-ON-CHIP Figure 3.9. The simultaneous manufacturing of multiple bonded chips reduces the back-end process time. on the surface. Briefly, a sample containing the analyte, target, is loaded on the array for specific capture to a matching probe. Next, a secondary affinity reagent, usually tagged with a fluorescent dye, is added to visualise which spots were reacting with the analyte. Microarrays are used in a number of growing research areas within the life sciences related to DNA and protein analysis. Everyday users of microarrays are found in universities, medical companies, hospitals and central laboratories (e.g clinical diagnostics or cancer diagnostics). Microarrays are divided into two groups: DNA-arrays and protein-arrays. DNA microarrays In DNA-arrays, DNA molecules are deposited on the surface and used for specific capture, hybridisation, of complementary strands in the sample. DNA-arrays are primarily used for monitoring gene expressions, but other applications such as singlenucleotide polymorphism (SNP) genotyping, where anomalies in the DNA strands are detected, have also been demonstrated [81]. The use of DNA-arrays has become ubiquitous in biochemical research and DNA-arrays are the largest segment in the microarray market. Protein microarrays A protein microarray comprises many different types of probes (typically antibodies or antigens) that are deposited on the surface. Each probe captures its target protein from a sample, typically a serum or cell lysate, and the captured proteins are subsequently detected and quantified using affinity reagents [82]. Protein arrays are used primarily in proteomics, the study of the function and structure of proteins, and recently also in diagnostics [83, 84] 3.3.2 Mass-transport limitation in microarrays The performance of microarrays depends on the conditions of the assay, for instance the buffer and temperature, and on the substrate surface on which the probes are 3.3. FACILE INTEGRATION OF MICROFLUIDICS WITH MICROARRAYS: THE BIOSTICKER 45 immobilised [23]. Another key factor for the performance of microarrays is the mass transport of analyte down to the spots. Normal incubation times for microarrays, which relies solely on diffusion for sample transport, is several hours up to days [85]. Because the diffusion coefficient, D, for nucleic acids in aqueous solutions is on √ the −11 2 −1 order of 10 m s [86], the typical distance traversed soley by diffusion, L = Dt, in 24 h is 1 mm. Considering that the horizontal length scale of a microarray is on the order of a few centimeters, hemispherical depletion volumes will form around each probe and the efficiency of the binding will be very low [59]. These effects are similar or worse in the case for protein microarrays since proteins vary in shape and size, resulting in different diffusion speeds for different analytes. Thus, there is a strong demand for technologies enabling more efficient mass transport and reliable mixing to avoid sample depletion over the spots and break the diffusion limitation. Microfluidic integration The integration of microfluidics with microarrays provides better mass transport by adding controlled convection over the microarray to avoid depletion of the sample and circumvent the diffusion limitation. Microfluidics also offers the advantage of multi-sample capabilities on one single chip, which may be of importance in genetic mutation analysis or clinical diagnosis, where direct comparison between different samples on the same chip would be preferable because the quality of slides with probe arrays varies from batch to batch [87]. Cross talk between different spots is also eliminated by separating the sample in compartments [88]. Microfluidics have previously been integrated with microarrays, and assay times have been significantly reduced. In some cases a more than thirty-fold reduction in assay time has been achieved [59, 89]. However, previous integration methods utilized either a clamped or plasma bonded PDMS channel layer on a spotted substrate of bare silicon/glass/PDMS [90, 91], double sided adhesive film [92, 93] or a plastic foil covering a thermoplastic substrate containing spots in custom-machined channels [94, 95]. All these methods are limited in practice, either because of poor material properties (adsorption of small molecules, channel deformation or tape dissolving monomers), complicated assembly processes, or because of limitation to custom-made plastic substrates. 3.3.3 The Biosticker: a micropatterned OSTE-sticker for microarrays The Biosticker, developed in Paper 8, is a microfluidic add-on for microarrays, that can be bonded to almost any substrate available on the market. It was developed based on the off-stoichiometry thiol-ene (OSTE) polymer platform presented earlier. By making use of the narrow glass transition temperature, the active surface, and the excellent micromolding capability of the OSTE polymer, integration, handling and bonding are greatly simplified. The probes are typically immobilised to the microarray surface via an epoxysilane, amino-silane, lysine or polyacrylamide linker. These linkers can also be 46 CHAPTER 3. OSTE: A NOVEL MATERIAL TOOLBOX FOR LABS-ON-CHIP Figure 3.10. A Biosticker flow cell attached to a protein microarray. The spots are visible though the polymer. To guarantee an even flow profile over the spots, branched inlet and outlet channels are used used for attaching the microfluidic flow cell. The OSTE-polymer has free thiol groups on its surface that can either directly covalently react with many standard microarray surfaces, or be designed to react via secondary functionalization (e.g. epoxyallyl or amineallyl monomers) with virtually any microarray surface. Heating the Biosticker to 37 ◦ C softens it enough for easy application to the microarray surface, where it seals conformally and reacts with the micoarray surface. The bonded fluidic layer protects the microarray during handling and the assay. The Biosticker can be removed after the assay is completed by heating above its glass transition temperature Tg =37 ◦ C, when the Biosticker softens and the interface to the surface is easily broken. This allows for read-out of the results in standard fluorescent scanners. 3.3.4 Preliminary results The feasibility of the Biosticker concept was tested on a high performance threedimensional microarray surface developed by Prof. Marcella Chiari and her group at Istituto di Chimica del Riconoscimento Molecolare (ICRM), C.N.R, in Milano, Italy. This particular surface consists of a copoly(DMA-NAS-MAPS) polymer receptor linker layer, recently demonstrated to improve sensitivity and limit-of-detection [96]. To guarantee an even flow, with homogenous concentration of the analyte over all the spots on the microarray surface, a branched microfluidic layout was designed for the Biosticker (Figure 3.10). With the help of Dr. Marina Cretich and Dr. Laura Sola, also at ICRM, two bioassays were tested with the Biosticker, a fluorescent protein experiment with spotted β-lactoglobulin detecting 1 ng/ml of anti β-lactoglobulin antibody, and a DNA hybridization test using spotted 23 mer 5’- amine modified oligonucleotides 3.4. SUMMARY AND OUTLOOK 47 and target complemetary oligonucleotide (1 µM) (Figure 3.11). For a first comparison, similar incubation times as for static experiments were used. The result showed uncomplicated application, no leakage and excellent signal and spot homogeneity, demonstrating the potential for using the Biostickers to optimise microarray assays while avoiding the need for special tools, complicated clamping, lamination or sub-optimal materials. Figure 3.11. The results of the scanned microarrays. The results from the βlactoglobulin protein assay and the DNA hybridization are very promising and show homogenous spots and excellent intensity. At the time of the writing of this thesis, the assay protocol and the channel geometry are optimised to achieve a reduce assay time, while keeping the same uncomplicated handling, high sensitivity and specificity as already demonstrated. 3.4 Summary and outlook In this chapter, a novel polymer platform, based on off-stoichiometry thiol-enes (OSTE) was introduced and characterised. A comparison with the wish-list in the end of Chapter 1, shows that the OSTE-polymer addresses most of the listed criteria for an ideal prototyping system for labs-on-chip. In Table 3.2 below, and in and Table 1 in the Appendix, the OSTE-polymer is compared with other materials currently available for microfluidic components prototyping. The OSTE-polymer platform is in many respects a better alternative than PDMS for rapid prototyping of labs-on-chip devices. The OSTE-polymers do not suffer from the adverse properties of PMDS, but are processed in an inexpensive and uncomplicated process, which is very similar to soft lithography. OSTEs significantly simplify the back-end processes by using built in anchors for surface functionalisation and biocompatible bonding. By also being able to mirror many of the properties of commercial thermoplastics the OSTE-polymers potentially minimise the redevelopment required to transfer a research proof-of-concept into a commercial thermoplastic device. 48 CHAPTER 3. OSTE: A NOVEL MATERIAL TOOLBOX FOR LABS-ON-CHIP 3.5 Thermoplastics NOA81 Glass Tuneable mechanical proeprties (stiff to elastomer) Chemical inertness, low interaction with sample Solvent resistance Direct, patternable and stable surface modifications Three-dimensional microfluidics Fast, scalable and inexpensive Biocompatible bonding PDMS 1) 2) 3) 4) 6) 7) 8) OSTE Process Material Table 3.2. Comparison of how common materials for lab-on-chip, as well as the novel OSTE polymer, prototyping compare to important material and processing properties for rapid prototyping of labs-on-chip x x x x x x x x x x x x x x x x x x Future work The development of the OSTE-polymer is hopefully the first step to an even more powerful and versatile polymer platform for labs-on-chip prototyping and production. There are still problems that need to be addressed. Most significant is maybe the leakage of non-crosslinked monomers in the materials with high off-stoichiometric ratio. Moreover, high temperature applications such as PCR requires a higher Tg than currently achievable in OSTE. However, both of these problems could be solved using ternary systems where a third monomer could react with the excess of the thiol [97], using a two step curing process [76]. First the thiol and allyl monomers are polymerised as previously, leaving a polymer with excess of thiol groups that can be used for surface functionalisation. Secondly, after lamination of the patterned fluidic layers, the thiol excess is reacted with the third monomer, using an appropriate initiator. This would minimise the leaching but still allow surface functionalisation before initiation of the second polymerisation. Although 3D microfluidic devices with interconnecting via was not shown in the OSTE-polymer in this work, it has been demonstrated in preliminary laboratory experiments and with similar photocurable stoichiometric thiol-ene polymers [98]. For a truly seamless transition from a research proof-of-concept to a commercial prototype, the possibilities of adapting the OSTE-polymer for commercial mass production must also be investigated, for instance its compatibility with injection moulding. All of the above is currently under development at KTH Microsystem Technology. Chapter 4 Introduction to low fluid-friction surfaces In this work, surface effects in liquid flows over solid surfaces is studied. It is shown, that by microstructuring the solid surface, microscopic surface effects can be manipulated and enhanced to have an impact on the macroscale flow patterns. The concept of super lubricating surfaces is introduced in this chapter, while my work on improving the stability of these surfaces for flow conditions encountered in realistic applications is presented in Chapter 5. 4.1 Motivation Drag, or fluid resistance, in liquid flows is the critical limiting factor in many microand macroscale applications, such as: lubrication, energy conversion, marine propulsion, flow switching, chemical separation and mixing on lab-on-chips. In a macroscopic flow, the drag has two components: the turbulent and the laminar losses. The laminar component is due to surface friction and is directly affected by the surface properties. In microchannels, where the flow is completely laminar, the laminar drag is the only source of energy loss. Furthermore, since the surface to volume ratio is large, surface friction becomes a major limitation for throughput in micro and nanofluidic devices. Also on the macroscale, there exist a thin layer at the solid-liquid interface, called the laminar sub-boundary layer, where the flow is laminar, as illustrated in Figure 4.1. The drag from this layer could be as high as 60-70 % of the total drag for large sea vessels [99]. The goal of this work was to develop effective means to reduce and manipulate the laminar drag using micromachined surfaces. The experiments, performed in microchannels, could also be applied to the laminar sub-boundary layer of macroscale flows. Research on super-lubricating surfaces, also known as superhydrophobic surfaces, is based on two approaches: the approach using self-assembled surface coatings and the approach using micro-structured surfaces. This work uses the latter approach. Self-assembled coatings have many advantages, such as application to a wide number 49 50 CHAPTER 4. INTRODUCTION TO LOW FLUID-FRICTION SURFACES Figure 4.1. Left: illustration of a macroscopic flow, with a thin laminar subboundary layer. Right: a microchannel with laminar flow. of materials and uncomplicated manufacturing. Micro-structuring, however, allows for greater control and, as we will see, more friction reduction. Recently, micro- and nanostructured surfaces with close to zero surface friction were fabricated [100]. Nevertheless, there are still technological hurdles remaining, before these superlubricating surfaces can be used in practical and commercial applications. First, the robustness dilemma must be solved. The lubricating state of the surfaces is fragile and collapses already at low liquid pressures. This makes the surfaces difficult to use, in particular since a collapsed superhydrophobic must be dried out before it can be reused, which is the second problem. Moroever, a collapsed superhydropohbic surface will quickly be fouled and loose its hydrophobicity. Paper 6 and Paper 7 of this thesis introduce two potential solutions to these limitations; robustness, to counter collapse at liquid pressure levels encountered in practical applications, and active control that can actively restore and collapsed surfaces to a super-lubricating state. 4.2 Surface friction in liquid flows In fluid mechanics, zero-slip is the universally accepted boundary condition, at a solid-liquid boundary. It postulates that the fluid is stationary at the solid surface, because of high friction at the interface. In reality, this condition has proven untrue, but for most practical applications the surface velocity is so small that it can safely be ignored. For the situations in which the zero slip boundary condition does not hold, the slip boundary condition model was introduced. It relates the fluid velocity at the surfaceÚ u0 , the slip velocity, to the magnitude of the shear rate experienced by the fluid at the surface: ! ! ! ∂u ! (4.1) u0 = λ !! !! , ∂y where λ is called the slip length. Its geometrical interpretation is illustrated in Figure 4.2. 4.3. SUPERHYDROPHOBIC SURFACES 51 Figure 4.2. Illustration of the slip length λ at a solid-liquid interface. The slip length is a practical measure to compare the friction reducing capacity of different surfaces. However, due to differences in measurement setups, the reported slip length on similar surfaces may vary. Smooth hydrophobic surface, for example, have a reported slip lengths of zero to a few nanometers [101, 102]. Is that enough to significantly affect the total fluid resistance in a microchannel? By using a simple model of a pressure-driven flow between two infinite parallel plates separated by a distance H, where one surface supports slip, a relation between the volume flow rate per unit depth q, and the slip length λ can be derived: " dp H3 − q= 4µ dx #$ % 1 λ + . 3 λ+H (4.2) From the above equation, it can be seen that for a given pressure gradient dp/dx, and fluid viscosity µ, the volume flow rate q can be significantly enhanced only if the slip length λ on the same order of magnitude as the height of the channel H. However, on a smooth hydrophobic surface, such as Teflon, the slip length is only a few nanometers. This seems discouraging, since typical microchannels in lab-on-chip applications are at least a few micrometers in height, and the laminar sub-boundary layer in turbulent flows can be up to several milimeters [103]. How can surfaces be made more hydrophobic than Teflon? 4.3 4.3.1 Superhydrophobic surfaces Mechanism of operation Superhydrophobic surfaces were originally inspired by the water-repelling features of plants, such as the lotus leaf. As shown in Figure 4.3, lotus leaves have a rough hydrophobic surface that is able to trap air between itself and a liquid, to create a lubricating air cushion, very much like an inverted hoover craft. As shown in Figure 4.3, the liquid on a superhydrophobic surface can be in two states. It can fully wet and penetrate the roughness (Wenzel state, no lubrication) or it can rest on top of the roughness, suspended by surface tension (Cassie or fakir 52 CHAPTER 4. INTRODUCTION TO LOW FLUID-FRICTION SURFACES Figure 4.3. A superhydrophobic surface. Left: Water droplets on a lotus leaf can not wet the rough hydrophobic surface and experience extremely low friction. The inset shows a close up of the surface of the lotus leaf. Right: Schematic figure of a superhydrophobic surface. The water is suspended on air between the solid fractions and the water can accelerate, on top of the air pockets, to a finite slip velocity before it reaches the next solid area where it is slowed down to zero. state, air lubrication). In the latter state, the liquid will experience zero friction on top of the air-pockets created in the rough surface and accelerate to a finite slip velocity. However, at the solid-liquid contact points, the liquid will be slowed down by friction. On average, this gives rise to an effective slip velocity and an effective slip length over the entire surface. The specific geometry of superhydrophobic surfaces can vary: posts, ribs, spikes or unstructured. In this work, we use ribs oriented perpendicular to the flow, as they are robust in the sense that if one airpocket collapses, it will only fill a small section of the surface. Moreover, they are uncomplicated to fabricate with high precision using standard UV-lithography. In a parallel plate flow, the effective slip length on such a surface was predicted by Lauga and Stone [104]: λ= " # 1 L ln , 2π cos(Fc π/2) (4.3) where the air fraction (shear free fraction) Fc is expressed as Fc = a/L, i.e. the gas-liquid interface length between two ribs a divided by the pitch of the ribs L. From this equation, which has only two factors to play with, we see that the air fraction Fc must be maximised to reduce drag, but beyond a certain point (Fc → 1), the slip can only be made larger by increasing the pitch L. This is why the slip length on nano-structured surfaces will be limited (Fc high but L low). Luckily, it is practically possible to engineer superhydrophobic surfaces with slip lengths reaching several micrometers [105, 106, 107, 108, 109, 110]. Although the measurement techniques and measured results can vary for similar surfaces, the slip on a microstructured surface can reach hundreds of micrometers under optimised conditions [100]. The next section describes in more detail how much the friction can be lowered in microchannels using these surfaces — and to what cost. 4.3. SUPERHYDROPHOBIC SURFACES 4.3.2 53 Stability limitations There is a fundamental limitation to all superhydrophobic surfaces. At some point, when the air pockets are made larger and larger, the surface tension will not be strong enough to support the liquid meniscus. The liquid will then penetrate into the superhydrophobic surface, and ruin the lubricating properties. This limitation is described by the Young-Laplace equation, which relates the maximum sustainable static pressure drop over the air-liquid interface, ∆Pmax , to the surface tension and geometry of the air pocket. In the case of parallel ridges, the maximum interfacial pressure is 2γ cos(θ) , (4.4) a where PL is the liquid pressure, PG is the air pocket pressure, γ is the surface tension, θ is the contact angle of the smooth hydrophobic material and a is the air pocket cavity size. Because the air pockets are isolated from the surroundings and their air pressure is fixed to ambient conditions during priming, this introduces a dilemma: large air pocket cavities a are needed to generate large slip, in order to have a large effect on the frictional losses of the flow, but large air pockets collapse easily. For instance, microfluidic channels are typically around 10 µm in height and if the air pockets size is 5 µm, the slip will be enough to affect the flow (4.2). However, it will only support a liquid pressure up to 15 kPa (4.4), which is less than encountered in most practical applications. This reduces the advantage of using a superhydrophobic surface, since the flow rate is also limited by the pressure. ∆Pmax = PL − PG = Chapter 5 Novel robust super-lubricating surfaces With the understanding of the mechanism and limitations of superhydropohobic surfaces from Chapter 4, this chapter introduces two ways to circumvent the YoungLaplace limitation and allow sustained lubrication at high pressures and large flows. The main theme is to pneumatically connect the air pockets and actively or passively counter the increased interfacial pressure drop at high liquid pressures. In the first section, a model is presented for the practically achievable friction reduction in a superhydrophobic channel limited by the Young-Laplace collapse pressure. In the second and third sections, previous attempts to increase the robustness and to manipulate the lubrication of superhydrophobic surfaces are discussed. The fourth section presents my work on a simple active external regulation of the air pocket pressure (Paper 7 ) to allow switching between superhydorphobic states, and a mechanism for passive adaptation of the air pocket pressure to the liquid pressure by a moving piston effect (Paper 6 ). Both were first-time demonstrations of superhydrophobic surfaces stable beyond the theoretical Young-Laplace pressure. 5.1 A model for friction reduction in a microchannel To explore the robustness dilemma, an expression for the total friction ν in a superhydrophobic microfluidic channel, relative to that of a smooth channel, limited by the Young-Laplace collapse pressure was derived in Paper 6 : $ ln(1/ cos(Fc π/2)) ν ≥ 12 +1 πWF Fc %−1 . (5.1) This expression is valid for parallel plate flows using a superhydrophobic surface with ribs perpendicular to the flow. It contains only two parameters, the air fraction of the surface Fc and a novel dimensionless number WF , also introduced in Paper 6, dubbed the channel’s energy carrying capacity; WF = PL Dh . γ cos(θ) 55 (5.2) 56 CHAPTER 5. NOVEL ROBUST SUPER-LUBRICATING SURFACES Here, Dh is the generalised hydraulic diameter of an arbitrary cross-section, defined as Dh = 4A/P , where A is the cross-sectional area and P the wetted perimeter. Whereas the slip length can only describe the friction reducing properties of the surface, WF can be thought of as a figure of merit for the complete superhydrophobic flow system and makes it possible to compare the flow capacity of different flow systems that have different geometries and surface energies. The higher WF , the higher is the capacity for large flows and highly pressurised liquids (large Dh and PL ). In a sense WF , is the energy carrying capacity of the flow. Figure 5.1. Graph of the relative friction factor ν as a function of the dimensionless figure of merit WF for some values of the shear free fraction Fc . Increased performance of the channel, that is, higher friction reduction and higher flows, is achieved by moving toward the bottom right of the figure. For each fixed value of Fc , there is a minimum obtainable surface friction factor ν uniquely expressed by the dimensionless parameter WF . The area to the right of each line represents the forbidden values, where friction reduction is not allowed because of collapse of the superhydrophobic surface. The minimum channel friction ν, for a few different values of air fractions Fc , is plotted as lines in Figure 5.1 using (5.1). This figure visualises the limitations of friction reduction due to air pocket collapse. The area to the left of each curve represents possible combinations of channel geometry, pressure and surface energy that are stable with regard to surface tension. Points to the right side of each Fc curve represent configurations exceeding the Young-Laplace pressure limitation where the flow system will be unstable and loose its superhydrophobic properties. 5.2. FRACTAL SURFACES: TEMPORARY LIFE SUPPORT 57 The black area in the bottom right corner, limited by Fc = 1 (only air, no solid) represents the ultimate goal — extreme lubrication with highly pressured liquids and large flows. The plot gives a good idea of what is practically possible when designing a superhydrophobic flow system. 5.2 Fractal surfaces: temporary life support Previously, the way to avoid the complete loss of super-lubricating properties above the Young-Laplace collapse pressure was to create multiple layers of air pockets, i.e. a surface with a roughness on multiple length scales. This is a well tested strategy also used by many water repelling leaves. When examining a lotus leaf (Figure 4.2) in a microscope, several levels of roughness can be observed. The largest roughness will trap the biggest air pockets, but when it collapses air will still remain in the smaller air pockets, which are more robust to higher pressures. This can go on several levels as the pressure increases. In this way the lubrication properties are never completely lost, only reduced step by step. This ideas was used by Onda et al. [111] when they presented a fractal superhydrophobic surface made in wax. Recently however, Lee et al. [100] demonstrated that fractal surfaces, contrary to popular belief, actually can reduce the lubcrication in cases when the air fraction Fc of the largest roughness scale already is large. Adding a second roughness scale will smoothen the sharp tips of the solid fractions and allow the liquid interface to recede slightly into the cavity of the largest roughness, which will decrease the lubrication. Thus, fractal surfaces can limit the effect of a collapse, but cannot prevent it and in extreme cases also reduce the lubrication. 5.3 Active switching: wet or dry Once a superhydrophobic surface is wetted, the liquid must somehow be expelled from the liquid filled pockets before it can regain its super-lubricating properties. Although there are many ways to change the wettability of a surface, such as thermosensitive polymers [112], pH switchable surfaces [113], roughness switching of a surface [114] or optical switching [115, 116], these techniques can only transform a dry super-lubricating state into a wetted, non-lubricating state. Not vice versa. The other direction, from wetted state to a dry state, has also been separately demonstrated [117, 118, 106]. Here, larger amounts of external energy must be provided to move the liquid out of the pockets and dry them. For instance, a short burst of electrical current through wetted droplets on a superhydrophobic surface will evaporate the water in the cavities [117] and push up the droplets. Acoustics has also been used to shoot wetted droplets off lotus leaves [118] and very recently also dewetting using electrolytically generated gas bubbles [106] was shoen. There have been no demonstrations combining these two effects to actively manipulate the states of a superhydrophobic surface between wet and dry in a continuous flow. 58 5.4 CHAPTER 5. NOVEL ROBUST SUPER-LUBRICATING SURFACES Regulating the air pocket pressure to avoid collapse Figure 5.2. Experimental setup for active regulation of the air pockets. Two parallel channels on top of a superhydorphobic surface with perpendicular gratings. The air pockets formed in the grooves are connected to the air channel through the grating. In a first experiment the same driving pressure was connected to both channels, thus creating a zero pressure drop over the air liquid interface. In a second experiment, the air channel pressure was manipulated enabling switching from dry to wet states in the liquid channel. 5.4.1 Active regulation The first solution for circumventing the Young-Laplace collapse limitation, presented in Paper 7, pneumatically connects the air pockets in a superhydrophobic surface to the same pressure source as the the liquid driving pressure. Two flow channels were placed perpendicular to a ribbed superhydrophobic surface, in such a way that the channels could communicate with each other, through the grooves in the surface (Figure 5.2). Water was flowed in one channel and air was flowed in the other. As both air and liquid flows were connected to the same upstream pressure source, the pressure drop over the air-liquid interface, PL − PG , in the liquid channel could be kept constant even as the liquid pressure was increased above the theoretical Young-Laplace collapse pressure for the specific surface geometry. This active regulation of the air pocket pressure, prevents collapse of the air pockets. The principle is presented in Figure 5.2. 5.4. REGULATING THE AIR POCKET PRESSURE TO AVOID COLLAPSE 59 In theory there is no limit to the liquid pressure such a system can handle, but at elevated upstream pressures the air-pockets may burst if the pressure in the gas filled channel exceeds the pressure in the liquid channel. The bursted air pockets introduces a small bubble in the liquid flow that disrupts the measurements. In another configuration, the air channel was connected to a computer controlled pressure source, while the upstream liquid pressure was connected to a separate pressure source. As the air pressure in the air control channel was manipulated, the air-liquid interface could be retracted into the deep grooves (wet state) or restored to full lubrication (dry state). The active switching of the surface friction enabled manipulation of the liquid flow rate through the channel while keeping a constant upstream liquid pressure. 5.4.2 Self-regulating air pockets Figure 5.3. The princple behind the self-regulating superhydrophobic design. The air pockets are pneumatically connected to the liquid through feedback channels. When liquid enters the feedback channels the liquid pressure compresses the air behind the grating and lowers the pressure drop over the interface. The second solution, presented in Paper 6, pneumatically connects the air pockets in the superhydrophobic surface to the bulk liquid pressure through feedback channels, as illustrated in Figure 5.3. When the liquid pressure increases, some water will enter feedback channels, which, like a piston, will compress the air in the air pockets, increasing the air pressure and reducing the interfacial pressure drop over the air-liquid interface. In this way, liquid pressure exceeding the Young-Laplace collapse pressure for the pattern can be tolerated without loss of super-lubrication and without the need for external pressure control. 60 CHAPTER 5. NOVEL ROBUST SUPER-LUBRICATING SURFACES Figure 5.4. The performance gain of active regulation and self-regulation of the air pocket pressure compared to previous superhydrophobic surfaces with isolated air pockets. The active regulation can in theory balance much higher pressures but at high pressures air bubbles are easily introduced into the liquid flow channel. The self-regulating design is ultimately limited by air diffusion from the air pockets into the liquid. 5.4.3 Performance Active regulation The active configuration was tested to up to three times higher liquid pressure than the collapse pressure of an identical superhydrophobic surface, with isolated air pockets. Figure 5.4 illustrates to performance gained by pneumatically connecting the air pockets to an extrenal pressure source. The novel surfaces, with regulated air pockets, are compared to reference superhydrophobic surfaces with isolated air pockets, in terms of maximum achievable energy carrying capacity WF , at fixed relative channel friction ν. The novel designs with regulated air pockets, can move into the theoretically forbidden area, with higher pressures and larger flows, than previously possible. In Figure 5.5, the feasibility of the active switching configuration is demonstrated by actively switching the superhydrophobic state of the surface. In this example, 5.5. SUMMARY AND OUTLOOK 61 the mass flow manipulation capacity was only 5%, due to the the non-optimised geometry and low air fraction. Figure 5.5. Switching of the flow rate in a superhydrophobic channel by manipulating the air pocket pressure to either retract or push back the air-liquid interface into the deep pockets. At air channel pressures under the Young-Laplace pressure, the water is sucked all the way into the air filled channel, through the connecting grooves, and introduces liquid droplets. These droplets can not be removed and terminate the function of the device. Self-regulating The self-regulating design was likewise abel to withstand roughly three times the Young-Laplace collapse pressure for an identical superhydrophobic surface without the feed back mechanism, as illustrated in Figure 5.4. However, the feedback mechanism is ultimately limited by the gas exchange over the air-liquid interface and at elevated air pressures gas starts dissolving into the liquid and slowly empties the feedback channels, as illustrated by Figure 5.6, in which the life-time of superhydrophobicity is plotted as a function of relative over-pressure, above the Young-Laplace collapse pressure. 5.5 Summary and outlook Two approaches were demonstrated to increase the robustness of superhydrophobic surfaces and, for the first time, allow sustained friction reduction at liquid pressures exceeding the Young-Laplace collapse pressure. The first principle pneumatically connected the air pockets to an external pressure source, which not only allows high liquid pressure with sustained lubricating properties, but also, for the first time, active switching between lubricating and non-lubricating states in a liquid flow. The second principle regulated the pressure of the air pockets by pneumatically connecting them to the liquid flow pressure. This self-regulating configuration required no external actuation. However, due to the diffusion of air into the liquid at elevated 62 CHAPTER 5. NOVEL ROBUST SUPER-LUBRICATING SURFACES Figure 5.6. Life time of the super-lubricating property at pressures exceeding the Young-Laplace collapse pressure. At prolonged times exceeding the Young-Laplace pressure, air diffusion over the interface empties the air pockets. pressures, super-lubrication at high liquid pressures and large flows could only be sustained for a limited time. The work represent the first steps towards constructing superhydrophobic surfaces with high lubrication, that can also be applied at liquid pressure levels encountered in real life. Later, the self-regulating principle has been further developed and refined in the work by Lee et al. [106], where air is electrolytically generated in the air pockets to stabilise the air pressure for longer time periods at high pressure. Chapter 6 Conclusions This thesis presented four novel concepts in microfluidics related to the development of new materials, surfaces and manufacturing methods. The following two novel manufacturing improvements of PDMS prototyping for lab-on-chip devices were introduced and demonstrated: 1. The first adhesive packaging method based on dual-surface adhesive films to bond substrates with different surface energies. The novel concept of dual surface-energy adhesive films to bond substrates with different surface energies enabled bonding of a microfluidic layer in PDMS, with a thermoplastic cartridge shell. The method was demonstrated in the microfluidic integration and packaging of a label-free optical ring resonator sensor. Biosensing results showed a mass detection limit of 0.9 pg/mm2 , one of the best reported values for integrated ring resontor sensors. 2. The first microfluidic manufacturing method for 3D devices based on localised inhibition of the polymerisation reaction of PDMS. The novel concept of localised inhibition of the PDMS polymerisation reaction enabled high yield fabrication of small, tightly spaced vertical interconnects using soft lithography with PDMS. This technology enabled the facile and rapid fabrication of three-dimensional microfluidic channel systems subsequently implemented in several labs-on-chip. The following new material for prototyping lab-on-chips devices was introduced and demonstrated: 3. The first polymer material platform specifically developed to meet the complete needs of lab-on-chip prototyping.. The novel concept of off-stoichiometry thiol-enes (OSTE) enabled the development of a highly versatile prototyping platform, aiming to bridge the gap 63 64 CHAPTER 6. CONCLUSIONS between academic proof of concept devices and commercial products. The prototype platform features attractive features for lab-on-chip rapid protoyping: a) Tuneable mechanical properties (PDMS-like to thermoplastics-like) b) Chemical inertness and low interaction with the sample (minimal absorption of small molecules) c) Solvent resistance (compatible with common solvents) d) Direct, patternable and stable surface modifications (UV-patternable using "click" chemistry) e) Low-temperature, biocompatible bonding (bonding at 37 ◦ C without the need of plasma or glue) f) Rapid, inexpensive and scalable (compatible with soft lithography) The novel polymer platform was demonstrated in wafer bonding for microfluidic devices and biocompatible integration with protein and DNA coated microarrays. The following novel surface design for friction reduction in liquid flows was introduced: 4. The first superhydrophobic surface that can withstand liquid pressures exceeding the Young-Laplace limitation. The novel designs of superhydrophobic surfaces pneumatically connected the trapped air pockets either to an external pressure source to demonstrate active manipulation of the surface friction, or to the bulk liquid pressure through a feedback channel to demonstrate self-regulation of the air pocket pressure. The results showed the novel surfaces resist up to three times higher liquid pressures than previous designs, while maintaining the same friction reducing capacity. The novel designs represented the first step towards practical implementations of micro-structured surfaces for friction reduction. Appendix: Tables 65 APPENDIX: TABLES 66 1 2 Polymer Previous work: Polymethylmethacrylate1 Polycarbonate1 Cyclic olefin (co)polymer1 Polydimethylsiloxane2 Norland Adhesive 81(thiol-ene based)2 Current work: OSTE (off-stoichiometry thiol-enes)2 Solvent resistance Excellent Good Excellent Poor Excellent Excellent Table 1. Comparison of some mechanical and chemical properties of common polymer materials used in microfluidic prototyping. Young’s modulus (MPa) Good Good Good Good Excellent Good Excellent Excellent UV 1800-3100 2000-2400 2000-2400 Excellent Excellent Tm 250-260 260-270 190-320 400-900 Poor Excellent 90-110 (swells) Excellent Excellent 70-80 Excellent Excellent Tg 100-122 145-148 70-155 - 1300 Excellent Excellent 70-100 Acronym PMMA PC COC -135 - 20-1800 Optical transmissivity PDMS 72 - Acid/base Water resitance contact angle NOA81 30-90 72 82 82 OSTE http://www.matbase.com/ Author’s DMA experiments Injection molding Hot embossing Casting Process Medium (hoursdays) Difficult (days) Process and equipment setup Simple (hours) High (20150 k$) High (>75k$) (2-15 Low k$) Very low (<2 k$) Tooling requirements Medium (>10k$) None Invest. cost High (secmin) Long (minhours) Medium (min) Cycle times High (3D) Medium (2D) Geom. flexibility High High Medium Product automation Low Table 2. Comparison of the microreplication processes for microlfuidic prototyping. Adapted form [16] 67 Summary of appended papers Paper 1 : A packaged optical slot-waveguide ring resonator sensor array for multiplex label-free assays in Labs-on-chip We present the design, fabrication, and characterisation of an array of optical slot-waveguide ring resonator sensors, integrated with microfluidics in a compact cartridge, for multiplexed real time label-fee biosensing. We obtain a volume refractive index detection limit of 5 × 10−6 refractive index units (RIU) and a surface mass density detection limit of 0.9 pg/mm2 . A novel bonding method based on dual surface-energy adhesive films allowed for fast and leak-tight assembly of cartridges with multiple tightly spaced fluidic interconnects, decreasing the time spent on back-end processes. Paper 2 : On-chip temperature compensation in an integrated slot-waveguide ring resonator refractive index sensor array We study the temperature dependence of an integrated slot-waveguide refractive index sensor array packaged in a microfluidic cartridge. The slotwaveguide ring resonator sensors show a low temperature dependence of −16.6 pm/K, while at the same time a large refractive index sensitivity of 240 nm per refractive index unit. Furthermore, by using one channel as on-chip temperature reference, a differential temperature sensitivity of only 0.3 pm/K is obtained. We demonstrate refractive index measurments during temperature drift and show a detection limit of 8.8 × 10− 6 refractive index units in a 7 K temperature window, without external temperature control. Paper 3 : Large scale integrated 3D microfluidic networks through high yield fabrication of vertical vias in PDMS In this article we introduce, experimentally demonstrate and characterise a novel, uncomplicated single-step method for creating membrane free vertical vias in PDMS. It enables batch manufacturing of large scale integrated 3D microfluidic networks or densely perforated membranes. The method, which has 69 70 SUMMARY OF APPENDED PAPERS a 100% yield, is based on inhibiting the polymerisation of commercial PDMS only at the via locations, circumventing the problem of residual membranes blocking the vias due to inadequate or uneven clamping. Paper 4 : Beyond PDMS: off-stoichiometry thiol-ene (OSTE) based soft lithography for rapid prototyping of microfluidic devices We introduce a novel polymer platform based on off-stoichiometry thiol-enes (OSTEs), aiming to bridge the gap between research prototyping and commercial production of microfluidic devices. We demonstrate important features for a prototyping system, such as one-step surface modifications; tuneable mechanical properties; direct leakage free sealing through direct UV-bonding; rapid prototyping; uncomplicated processing and the ability to mirror the mechanical and chemical properties of both PDMS as well as commercial grade thermoplastics. Paper 5 : Low temperature "click" wafer bonding of off-stoichiometry thiol-ene (OSTE) polymers to silicon We introduce a novel wafer bonding concept designed for permanent attachment of micromolded polymer structures to functionalized silicon substrates. The method, designed for simultaneous fabrication of many identical Labson-chip devices, utilizes a chemically reactive polymer microfluidic structure which rapidly bonds to a functionalized substrate wafer via "click" chemistry reactions. The microfluidic structure consists of an off-stoichiometry thiolene (OSTE) polymer with a very high density of surface bound thiol groups and the substrate is a silicon wafer that has been functionalized with common bio-linker molecules. The method is biocompatible and is well suited for wafer-level microfluidic packaging of pre-functionalized surfaces. In this article, we demonstrate void free and low temperature (<37 ◦ C) bonding in the fabrication of a complete batch of microfluidic devices consisting of a microfluidic OSTE polymer layers bonded to a silane functionalized silicon wafer. The diced devices showed a burst pressure exceeding 4 bars, are compatible with most organic solvents, are easily surface modified and have excellent solvent barrier properties. Paper 6 : Sustained superhydrophobic friction reduction at high pressures and large flows We first introduce a figure of merit to describe and compare the total drag reduction of different superhydrophobic flow systems. We then show it im- 71 possible to achieve high friction reduction at high liquid pressure other than in thin channels only a few micrometers in height due to the Laplace pressure limitation. Secondly, we introduce a novel self-regulating design for friction reduction in laminar flow systems with large channels and high pressure. The design is based on pneumatically connecting the air pockets to the liquid pressure through feedback channels that can regulate the air pocket pressure. We demonstrated that this self-regulating design can give up to three times better performance than standard superhydrophobic channels. Paper 7 : Continuos flow switching by pneumatic actuation of the air lubrication layer on superhydrophobic microchannel walls We introduce a method for robust, active control of friction reduction in microchannels, enabling new flow control applications and overcoming previous limitations with regard to sustainable liquid pressure. The air pockets trapped at a superhydrophobic micrograting during liquid priming are coupled to an actively controlled pressure source, allowing the pressure difference over the air/liquid interface to be dynamically adjusted. This allows for manipulating the friction reduction properties of the surface, enabling active control of liquid mass flow through the channel. It also permits for sustainable air lubrication at high liquid pressures, without loss of superhydrophobic properties. With the non-optimized grating used in the experiment, a difference in liquid mass flow of 5% is obtained by alternatively collapsing and recreating the air pockets using the coupled pressure source. The method also allows for sustainable liquid pressure 3 times higher than the Young-Laplace pressure of a passive device. Paper 8 : Continuos flow switching by pneumatic actuation of the air lubrication layer on superhydrophobic microchannel walls We present a one-step, reversible, and biocompatible bonding method of a stiff patterned microfluidic "Biosticker", based on off-stoichiometry thiolene (OSTE) polymers , to state-of-the-art spotted microarray surfaces. The method improves and simplifies the batch back-end processing of microarrays. We illustrate its ease of use in two applications: a high sensitivity flow-through protein assay; and a DNA-hybridization test. Read-out was performed in a standard high-volume array scanner, and showed excellent spot homogeneity and intensity. The Biosticker is aimed to be a plug-in for existing microarray platforms to enable faster protein assays and DNA hybridizations through mass transport optimization. Acknowledgement My doctoral studies have taken me through many different areas of research, and the present work would have been impossible without the guidance and assistance from many people. First of all I would like to thank my main supervisor Wouter van der Wijngaart, for accepting me as a student in his team, for giving me the time to learn and giving me the freedom to make mistakes and the guidance not to repeat them. I would also like to thank Tommy Haraldsson for introducing me to the world of polymer science. His guidance and support have made the last part of my time as a PhDstudent a truly ex(c)iting journey. Many thanks also to Göran Stemme for having created a truly motivating and unpretentious research environment in his group. Financial support for this work was provided partly through grants from the Swedish Research Council and the European Commission via the FP6-IST-SABIO project. During my time as PhD student at the Microsystem Technology Laboratory, I have been lucky to work together with many extraordinary individuals. First I would like to thank people who have directly helped out with the experiments: Kristinn Gylfason, for our collaboration in the bio-sensing project, for being a great source of knowledge and inspiration, and for becoming a good friend; Mikael Karlsson, for sharing long days of experimenting in the laboratory at MST and for always being so positive; Farizah Saharil, for good team work with the OSTE experiments; Matteo Cornaglia for all the PDMS you inhibited for me; Thomas Moh, for helping me with the microfluidic assmeblies; Liu Yitong, for the excellent work on improving the OSTE-polymers; Kim Öberg for help with the thiol-enes and for always having time; Andrzej Kaźmierzak and Fabian Dortu at Multitel, for long days and nights of experiments in Mons; Marcella Chiari, Marina Cretich and Laura Sola at ICRM for the microarray experiments; Minh Do-Quang at KTH Mechanics for help with finite element simulations; Junichiro Shiomi at the Tokyo University for introducing me to molecular dynamics, and academic research in general. Special thanks also to those who have contributed with valuable discussions and help: Mikael Sterner, my office-mate, for patiently sharing valuable tips and tricks on many diverse topics; Niklas Sandström, for discussions on exciting applications of our research, Aman Russom, Sergey Zelenin and Sahar Ardabili at KTH Cell Physics for all the discussions on cells, DNA and proteins and for the help with the microscopes, Gerry Ronan at Farfield and Gerhard Kreshbach at Zeptosens for 73 74 ACKNOWLEDGEMENT sharing their expertise in system integration and biosensing; Michael Malkoch and Mauro Claudino at KTH Fiber and Polymer Technology for valuable discussions on polymer science and for help with the machines. All my friends, colleagues and former colleagues at MST: Adit, Andreas, Björn, Erika, Farizah, Frank, Gaspard, Göran, Hans, Henrik, Hithesh, Joachim, Fredrik F, Fritzi, Kjell, Martin, Niclas, Niklas, Nutapong, Mikael A, Mikael K, Mikael S, Staffan, Stefan, Thomas, Tommy, Umer, Wouter, Liu Yitong and Zargam. Thank you for a supporting atmosphere and the good times. I also very much appreciate the help from Tommy, Kristinn, Ebba, Hans, Wouter and my mom for proofreading different parts of this thesis. I am forever in debt to my parents for supporting me all the way, in particular to my dad who had very much looked forward to this day but was not given the time to experience it. Last but not the least, I would like to thank my beloved wife Selina, for loving me and being all that I could ever wish for. This would never have been possible without your understanding and encouragements. Carl Fredrik Carlborg, Stockholm, August 29th, 2011 References [1] L. Gervais, N. de Rooij, and E. Delamarche, “Microfluidic Chips for Pointof-Care Immunodiagnostics,” Adv. Mater., vol. 23, no. 24, pp. H151–H176, 2011. [2] F. Breussin, “Microfluidic technologies for point of care testing,” in Yole Workshop lab-on-chip for diagnostics, Yolé Développement, Mar. 2010. [3] J. Gantelius, Novel diagnostic microarray assay formats towards comprehensive on-site analysis Novel diagnostic microarray assay formats towards comprehensive on-site analysis. PhD thesis, Royal Institute of Technology (KTH), 2009. [4] G. Odian, Principles of Polymerization. Hoboken, New Jersey: John Wiley & Sons, Inc., 2004. [5] J. A. Carioscia, H. Lu, J. W. Stanbury, and C. N. Bowman, “Thiol-ene oligomers as dental restorative materials,” Dental Materials, vol. 21, pp. 1137– 1143, Dec. 2005. [6] D. Bartolo, G. Degre, P. Nghe, and V. Studer, “Microfluidic stickers,” Lab Chip, vol. 8, no. 2, pp. 274–279, 2008. [7] S. Lee and S. Lee, “Shrinkage ratio of PDMS and its alignment method for the wafer level process,” Microsystem Technologies, vol. 14, pp. 205–208, Feb. 2008. [8] M. W. Toepke and D. J. Beebe, “PDMS absorption of small molecules and consequences in microfluidic applications,” Lab Chip, vol. 6, no. 12, pp. 1484– 1486, 2006. [9] S. Yunus, de Looringhe, C. Poleunis, and A. Delcorte, “Diffusion of oligomers from polydimethylsiloxane stamps in microcontact printing: Surface analysis and possible application,” Surf. Interface Anal., vol. 39, no. 12-13, pp. 922– 925, 2007. [10] J. N. Lee, C. Park, and G. M. Whitesides, “Solvent Compatibility of Poly(dimethylsiloxane)-Based Microfluidic Devices,” Analytical Chemistry, vol. 75, pp. 6544–6554, Dec. 2003. 75 76 REFERENCES [11] J. W. Swanson and J. E. Lebeau, “The effect of implantation on the physical properties of silicone rubber,” J. Biomed. Mater. Res., vol. 8, no. 6, pp. 357– 367, 1974. [12] K. M. Choi and J. A. Rogers, “A Photocurable Poly(dimethylsiloxane) Chemistry Designed for Soft Lithographic Molding and Printing in the Nanometer Regime,” Journal of the American Chemical Society, vol. 125, pp. 4060–4061, Apr. 2003. [13] B.-Y. Kim, L.-Y. Hong, Y.-M. Chung, D.-P. Kim, and C.-S. Lee, “SolventResistant PDMS Microfluidic Devices with Hybrid Inorganic/Organic Polymer Coatings,” Advanced Functional Materials, vol. 19, no. 23, pp. 3796–3803, 2009. [14] A. R. Abate, D. Lee, T. Do, C. Holtze, and D. A. Weitz, “Glass coating for PDMS microfluidic channels by sol-gel methods,” Lab Chip, vol. 8, pp. 516– 518, Apr. 2008. [15] J. Lee, M. J. Kim, and H. H. Lee, “Surface Modification of Poly(dimethylsiloxane) for Retarding Swelling in Organic Solvents,” Langmuir, vol. 22, pp. 2090–2095, Feb. 2006. [16] H. Becker and C. Gärtner, “Polymer microfabrication technologies for microfluidic systems,” Analytical and Bioanalytical Chemistry, vol. 390, pp. 89– 111, Jan. 2008. [17] L. J. Lee, M. J. Madou, K. W. Koelling, S. Daunert, S. Lai, C. G. Koh, Y.-J. Juang, Y. Lu, and L. Yu, “Design and fabrication of cd-like microfluidic platforms for diagnostics: Polymer-based microfabrication,” Biomedical Microdevices, vol. 3, no. 4, pp. 339–354, 2001. [18] N. Le, R. Gubala, V andGandhiraman, S. Daniels, and D. Williams, “Evaluation of different nonspecific binding blocking agents deposited inside poly(methyl methacrylate) microfluidic flow-cells,” Langmuir, vol. 27, pp. 9043–51, June 2011. [19] R. Ledesma-Aguilar, R. Nistal, A. Hernandez-Machado, and A. Pagonabarraga, “Controlled drop emission by wetting properties in driven liquid filaments,” Nature Materials, vol. 10, pp. 367–371, 2011. [20] T. Rohr, D. Ogletree, F. Svec, and J. Fréchet, “Surface functionalization of thermoplastic polymers for the fabrication of microfluidic devices by photoinitiated grafting,” Advanced Functional Materials, vol. 13, pp. 264–270, April 2003. REFERENCES 77 [21] D. Witters, N. Vergauwe, S. Vermeir, F. Ceyssens, S. Liekens, R. Puers, and J. Lammertyn, “Biofunctionalization of electrowetting-on-dielectric digital microfluidic chips for miniaturized cell-based applications,” Lab Chip, vol. 11, pp. 2790–2794, 2011. [22] D. Wu, B. Zhao, Z. Dai, J. Qin, and B. Lin, “Grafting epoxy-modified hydrophilic polymers onto poly(dimethylsiloxane) microfluidic chip to resist nonspecific protein adsorption,” Lab Chip, vol. 6, pp. 942–947, 2006. [23] D. W. Grainger, C. H. Greef, P. Gong, and M. J. Lochhead, “Current microarray surface chemistries,” Methods in Molecular Biology, vol. 381, pp. 35–57, 2007. [24] C. F. Carlborg and W. van der Wijngaart, “Sustained Superhydrophobic Friction Reduction at High Liquid Pressures and Large Flows,” Langmuir, vol. 27, pp. 487–493, Jan. 2011. [25] E. Eteshola and D. Leckband, “Development and characterization of an elisa assay in pdms microfluidic channels,” Sensors and Actuators B: Chemical, vol. 72, no. 2, pp. 129–133, 2001. [26] S. A. Ruiz and C. S. Chen, “Microcontact printing: A tool to pattern,” Soft Matter, vol. 3, pp. 168–177, 2007. [27] M. Geissler, A. Bernard, A. Bietsch, H. Schmid, B. Michel, and E. Delamarche, “Microcontact-printing chemical patterns with flat stamps,” Journal of American Chemical Society, vol. 112, no. 26, pp. 6303–6304, 2000. [28] L. P. Hromada, B. J. Nablo, J. J. Kasianowicz, M. A. Gaitan, and D. L. DeVoe, “Single molecule measurements within individual membrane-bound ion channels using a polymer-based bilayer lipidmembrane chip,” Lab Chip, vol. 8, pp. 602–608, 2008. [29] Z. Chen, L. Zhang, and G. Chen, “A spring-driven press device for hot embossing and thermal bonding of pmma microfluidic chips,” Electrophoresis, vol. 31, pp. 2512–2519, 2010. [30] B.-Y. Pemga, C.-W. Wub, Y.-K. Shenc, and Y. Lind, “Microfluidic chip fabrication using hot embossing and thermal bonding of cop,” Polymers for Advanced Technologies, vol. 21, no. 7, pp. 457–466, 2009. [31] R. KW, L. H, and K. DR, “Plastic microchip liquid chromatographymatrix-assisted laser desorption/ionization mass spectrometry using monolithic columns,” Journal of Chromatography A, vol. 1111, no. 1, pp. 40–47, 2006. 78 REFERENCES [32] H. Klank, J. P. Kutter, and O. Geschke, “Co2-laser micromachining and backend processing for rapid production of pmma-based microfluidic systems,” Lab Chip, vol. 2, pp. 242–246, 2002. [33] L. Brown, T. Koerner, J. H. Horton, and R. D. Oleschuk, “Fabrication and characterization of poly(methylmethacrylate) microfluidic devices bonded using surface modifications and solvents,” Lab Chip, vol. 6, no. 66-73, 2006. [34] Y. Hsu and T. Chen, “Applying taguchi methods for solvent-assisted pmma bonding technique for static and dynamic micro-tas devices,” Biomedical Microdevices, vol. 9, no. 4, pp. 513–22, 2007. [35] R. Truckenmuller, R. Ahrens, Y. Cheng, G. Fischer, and V. Saile, “An ultrasonic welding based process for building up a new class of inert fluidic microsensors and -actuators from polymers,” Sensors and Actuators A, vol. 132, no. 1, pp. 385–392, 2006. [36] A. Boglea, A. Olowinsky, and A. Gillner, “Fibre laser welding for packaging of disposable polymeric microfluidic-biochips,” Applied Surface Science, vol. 254, pp. 1174–1178, Dec. 2007. [37] T. Ussing, L. V. Petersen, C. B. Nielsen, B. Helbo, and L. Højslet, “Micro laser welding of polymer microstructures using low power laser diodes,” The International Journal of Advanced Manufacturing Technology, vol. 22, no. 1-2, pp. 198–205, 2007. [38] J. C. McDonald, D. C. Duffy, J. R. Anderson, D. T. Chiu, H. Wu, O. J. A. Schueller, and G. M. Whitesides, “Fabrication of microfluidic systems in poly(dimethylsiloxane),” ELECTROPHORESIS, vol. 21, pp. 27–40, Jan. 2000. [39] J. Kentsch, S. Breisch, and M. Stelzle, “Low temperature adhesion bonding for biomems,” Journal of Micromechanics and Microengineering, vol. 16, no. 4, pp. 802–807, 2006. [40] F. Dang, S. Shinohara, O. Tabata, Y. Yamaoka, M. Kurokawa, Y. Shinohara, M. Ishikawa, and Y. Baba, “Replica multichannel polymer chips with a network of sacrificial channels sealed by adhesive printing method,” Lab Chip, vol. 5, pp. 472–478, 2005. [41] D. F. Pulak Nath, Y. A. Kunde, A. Zeytun, B. Branch, and G. Goddard, “Rapid prototyping of robust and versatile microfluidic components using adhesive transfer tapes,” Lab Chip, vol. 10, pp. 2286–2291, 2010. [42] M. A. Unger, H.-P. Chou, T. Thorsen, A. Scherer, and S. R. Quake, “Monolithic Microfabricated Valves and Pumps by Multilayer Soft Lithography,” Science, vol. 288, pp. 113–116, Apr. 2000. REFERENCES 79 [43] H.-Y. Chen, A. A. McClelland, Z. Chen, and J. Lahann, “Solventless adhesive bonding using reactive polymer coatings,” Analytical Chemistry, vol. 80, no. 11, pp. 4119–4124, 2008. [44] K. S. Lee and R. J. Ram, “Plastic-PDMS bonding for high pressure hydrolytically stable active microfluidics,” Lab Chip, vol. 9, no. 11, pp. 1618–1624, 2009. [45] A. Gerlach, H. Lambach, and D. Seidel, “Propagation of adhesives in joints during capillary adhesive bonding of microcomponents,” Microsystem Technologies, vol. 6, pp. 19–22, Nov. 1999. [46] C.-W. Tsaoa and D. L. DeVoe, Lab on a Chip Technology: Volume 1: Fabrication and Microfluidics, vol. 1. Caister Academic Press, 2009. [47] D. Mark, S. Haeberle, G. Roth, F. Stetten, and R. Zengerle, “Microfluidic labon-a-chip platforms: requirements, characteristics and applications,” Chem. Soc. Rev., vol. 39, no. 3, pp. 1153–1182, 2010. [48] V. Sunkara, D.-K. Park, H. Hwang, R. Chantiwas, S. A. Soper, and Y.-K. Cho, “Simple room temperature bonding of thermoplastics and poly(dimethylsiloxane),” Lab Chip, vol. 11, no. 5, pp. 962–965, 2011. [49] J. Kim, R. Surapaneni, and B. K. Gale, “Rapid prototyping of microfluidic systems using a PDMS/polymer tape composite,” Lab Chip, vol. 9, no. 9, pp. 1290–1293, 2009. [50] H. Sohlström and M. Öberg, “Refractive index measurement using integrated ring resonators,” in The 8th European Conference on Integrated Optics, pp. 322–325, Apr. 1997. [51] E. Krioukov, Klunder, A. Driessen, J. Greve, and C. Otto, “Sensor based on an integrated optical microcavity,” Opt. Lett., vol. 27, pp. 512–514, Apr. 2002. [52] C. Y. Chao and L. J. Guo, “Biochemical sensors based on polymer microrings with sharp asymmetrical resonance,” Applied Physics Letters, vol. 83, no. 8, pp. 1527–1529, 2003. [53] A. L. Washburn, L. C. Gunn, and R. C. Bailey, “Label-Free Quantitation of a Cancer Biomarker in Complex Media Using Silicon Photonic Microring Resonators,” Analytical Chemistry, vol. 81, pp. 9499–9506, Nov. 2009. [54] V. R. Almeida, Q. Xu, C. A. Barrios, and M. Lipson, “Guiding and confining light in void nanostructure,” Optics Letters, vol. 29, no. 11, pp. 1209–1211, 2004. [55] Q. Xu, V. R. Almeida, R. R. Panepucci, and M. Lipson, “Experimental demonstration of guiding and confining light in nanometer-size low-refractiveindex material,” Optics Letters, vol. 29, no. 14, pp. 1626–1628, 2004. 80 REFERENCES [56] C. A. Barrios, M. J. Bañuls, V. González-Pedro, K. B. Gylfason, B. Sánchez, A. Griol, A. Maquieira, H. Sohlström, M. Holgado, and R. Casquel, “Labelfree optical biosensing with slot-waveguides,” Opt. Lett., vol. 33, pp. 708–710, Apr. 2008. [57] T. Gervais and K. F. Jensen, “Mass transport and surface reactions in microfluidic systems,” Chemical Engineering Science, vol. 61, pp. 1102–1121, Feb. 2006. [58] T. M. Squires, R. J. Messinger, and S. R. Manalis, “Making it stick: convection, reaction and diffusion in surface-based biosensors,” Nature Biotechnology, vol. 26, pp. 417–426, Apr. 2008. [59] K. Pappaert, J. Vanderhoeven, P. Vanhummelen, B. Dutta, D. Clicq, G. Baron, and G. Desmet, “Enhancement of DNA micro-array analysis using a shear-driven micro-channel flow system,” Journal of Chromatography A, vol. 1014, pp. 1–9, Oct. 2003. [60] B. I. Inc, Application notes #107: Binding Kinetics Analysis with SPR: Interaction between Bovine Serum Albumin (BSA) and Anti-BSA, Mar. 2010. [61] A. K. Wright and M. R. Thompson, “Hydrodynamic structure of bovine serum albumin determined by transient electric birefringence.,” Biophysical journal, vol. 15, pp. 137–141, Feb. 1975. [62] L. Chen, F. Azizi, and C. H. Mastrangelo, “Generation of dynamic chemical signals with microfluidic C-DACs,” Lab Chip, vol. 7, no. 7, pp. 850–855, 2007. [63] M. Rhee and M. A. Burns, “Microfluidic assembly blocks,” Lab Chip, vol. 8, no. 8, pp. 1365–1373, 2008. [64] C. Neils, Z. Tyree, B. Finlayson, and A. Folch, “Combinatorial mixing of microfluidic streams,” Lab Chip, vol. 4, no. 4, pp. 342–350, 2004. [65] K. Atsuta, H. Noji, and S. Takeuchi, “Micro patterning of active proteins with perforated PDMS sheets (PDMS sieve),” Lab Chip, vol. 4, no. 4, pp. 333–336, 2004. [66] J. Cha, J. Kim, S.-K. Ryu, J. Park, Y. Jeong, S. Park, S. Park, H. C. Kim, and K. Chun, “A highly efficient 3D micromixer using soft PDMS bonding,” Journal of Micromechanics and Microengineering, vol. 16, pp. 1778–1782, Sept. 2006. [67] J. R. Anderson, D. T. Chiu, R. J. Jackman, O. Cherniavskaya, J. C. McDonald, H. Wu, S. H. Whitesides, and G. M. Whitesides, “Fabrication of Topologically Complex Three-Dimensional Microfluidic Systems in PDMS by Rapid Prototyping,” Analytical Chemistry, vol. 72, pp. 3158–3164, July 2000. REFERENCES 81 [68] B. H. Jo, L. M. Van Lerberghe, K. M. Motsegood, and D. J. Beebe, “Threedimensional micro-channel fabrication in polydimethylsiloxane(PDMS) elastomer,” Microelectromechanical Systems, Journal of, vol. 9, pp. 76–81, Mar. 2000. [69] C.-H. Hsu, C. Chen, and A. Folch, “"Microcanals" for micropipette access to single cells in microfluidic environments,” Lab Chip, vol. 4, no. 5, pp. 420–424, 2004. [70] E. P. Kartalov, C. Walker, C. R. Taylor, W. F. Anderson, and A. Scherer, “Microfluidic vias enable nested bioarrays and autoregulatory devices in Newtonian fluids,” Proceedings of the National Academy of Sciences, vol. 103, pp. 12280–12284, Aug. 2006. [71] R. J. Jackman, D. C. Duffy, O. Cherniavskaya, and G. M. Whitesides, “Using Elastomeric Membranes as Dry Resists and for Dry Lift-Off,” Langmuir, vol. 15, pp. 2973–2984, Apr. 1999. [72] J. H. Kang, E. Um, and J.-K. Park, “Fabrication of a poly(dimethylsiloxane) membrane with well-defined through-holes for three-dimensional microfluidic networks,” Journal of Micromechanics and Microengineering, vol. 19, no. 4, pp. 045027+, 2009. [73] N. Sabourault, G. Mignani, A. Wagner, and C. Mioskowski, “Platinum Oxide (PtO2): A Potent Hydrosilylation Catalyst,” Organic Letters, vol. 4, pp. 2117–2119, June 2002. [74] M. Karlsson, T. Haraldsson, N. Sandström, G. Stemme, A. Russom, and W. van der Wijngaart, “On-chip liquid degassing with low water loss,” in Proceedings Micro Total Analysis Systems (muTAS), pp. 1790–1792, 2010. [75] J. Hansson, J. M. Karlsson, T. Haraldsson, W. van der Wijngaart, and A. Russom, “Inertial particle focusing on parallel microfluidic channels for highthroughput filtration,” in The 16th International Conference on Solid-State Sensors, Actuators and Microsystems, pp. 1777–1780, IEEE, 2011. [76] C. E. Hoyle and C. N. Bowman, “Thiol-Ene Click Chemistry,” Angewandte Chemie International Edition, vol. 49, no. 9, pp. 1540–1573, 2010. [77] H. C. Kolb, M. G. Finn, and K. B. Sharpless, “Click Chemistry: Diverse Chemical Function from a Few Good Reactions,” Angewandte Chemie International Edition, vol. 40, no. 11, pp. 2004–2021, 2001. [78] N. B. Cramer, J. P. Scott, and C. N. Bowman, “Photopolymerizations of thiolene Polymers without Photoinitiators,” Macromolecules, vol. 35, pp. 5361– 5365, July 2002. 82 REFERENCES [79] N. Sandström, R. Z. Shafagh, C. F. Carlbor, T. Haraldsson, G. Stemme, and W. van der Wijngaart, “One step integration of gold coated sensors with OSTE polymer cartridges by low temperature dry bonding,” in The 16th International Conference on Solid State Sensors, Actuators and Microsystems, pp. 2778–2781, IEEE, 2011. [80] P. Man, D. Jones, and C. Mastrangelo, “Microfluidic plastic capillaries on silicon substrates: a new inexpensive technology for bioanalysis chips,” in Micro Electro Mechanical Systems, 1997. MEMS ’97, Proceedings, IEEE., Tenth Annual International Workshop on, pp. 311 – 316, 1997. [81] C. A. Afshari, “Perspective: Microarray Technology, Seeing More Than Spots,” Endocrinology, vol. 143, pp. 1983–1989, June 2002. [82] G. MacBeath, “Protein microarrays and proteomics,” Nature Genetics, vol. 32, pp. 526–532, Dec. 2002. [83] S. F. Kingsmore, “Multiplexed protein measurement: technologies and applications of protein and antibody arrays,” Nature Reviews Drug Discovery, vol. 5, pp. 310–321, Mar. 2006. [84] M. Hartmann, J. Roeraade, D. Stoll, M. Templin, and T. Joos, “Protein microarrays for diagnostic assays,” Analytical and Bioanalytical Chemistry, vol. 393, pp. 1407–1416, Mar. 2009. [85] M. Schena, Protein Microarrays. Jones and Bartlett Publishers, Inc, July 2004. [86] G. L. Lukacs, P. Haggie, O. Seksek, D. Lechardeur, N. Freedman, and A. S. Verkman, “Size-dependent DNA Mobility in Cytoplasm and Nucleus,” Journal of Biological Chemistry, vol. 275, pp. 1625–1629, Jan. 2000. [87] S. Russell, L. Meadows, and R. Russel, Microarray Technology in Practice. Elsevier, 2008. [88] C. Situma, M. Hashimoto, and S. Soper, “Merging microfluidics with microarray-based bioassays,” Biomolecular Engineering, vol. 23, pp. 213–231, Oct. 2006. [89] H. H. Lee, J. Smoot, Z. McMurray, D. A. Stahl, and P. Yager, “Recirculating flow accelerates DNA microarray hybridization in a microfluidic device,” Lab Chip, vol. 6, no. 9, pp. 1163–1170, 2006. [90] D. Erickson, X. Liu, U. Krull, and D. Li, “Electrokinetically Controlled DNA Hybridization Microfluidic Chip Enabling Rapid Target Analysis,” Analytical Chemistry, vol. 76, pp. 7269–7277, Dec. 2004. REFERENCES 83 [91] J. Liu, B. A. Williams, R. M. Gwirtz, B. J. Wold, and S. Quake, “Enhanced Signals and Fast Nucleic Acid Hybridization By Microfluidic Chaotic Mixing,” Angewandte Chemie International Edition, vol. 45, no. 22, pp. 3618–3623, 2006. [92] N. B. Adey, M. Lei, M. T. Howard, J. D. Jensen, D. A. Mayo, D. L. Butel, S. C. Coffin, T. C. Moyer, D. E. Slade, M. K. Spute, A. M. Hancock, G. T. Eisenhoffer, B. K. Dalley, and M. R. McNeely, “Gains in Sensitivity with a Device that Mixes Microarray Hybridization Solution in a 25-μm-Thick Chamber,” Analytical Chemistry, vol. 74, pp. 6413–6417, Dec. 2002. [93] Y. Liu, “DNA probe attachment on plastic surfaces and microfluidic hybridization array channel devices with sample oscillation,” Analytical Biochemistry, vol. 317, pp. 76–84, June 2003. [94] M. Noerholm, H. Bruus, M. H. Jakobsen, P. Telleman, and N. B. Ramsing, “Polymer microfluidic chip for online monitoring of microarray hybridizations,” Lab Chip, vol. 4, no. 1, pp. 28–37, 2004. [95] C. Situma, Y. Wang, M. Hupert, F. Barany, R. McCarley, and S. Soper, “Fabrication of DNA microarrays onto poly(methyl methacrylate) with ultraviolet patterning and microfluidics for the detection of low-abundant point mutations,” Analytical Biochemistry, vol. 340, pp. 123–135, May 2005. [96] M. Cretich, G. di Carlo, R. Longhi, C. Gotti, N. Spinella, S. Coffa, C. Galati, L. Renna, and M. Chiari, “High sensitivity protein assays on microarray silicon slides,” Analytical Chemistry, vol. 81, no. 13, pp. 5197–5203, 2009. [97] J. A. Carioscia, J. W. Stansbury, and C. N. Bowman, “Evaluation and control of thiol–ene/thiol–epoxy hybrid networks,” Polymer, vol. 48, pp. 1526–1532, Mar. 2007. [98] J. F. Ashley, N. B. Cramer, R. H. Davis, and C. N. Bowman, “Soft-lithography fabrication of microfluidic features using thiol-ene formulations,” Lab Chip, vol. 11, no. 16, pp. 2772–2778, 2011. [99] K. Fukuda, J. Tokunaga, T. Nobunaga, T. Nakatani, T. Iwasaki, and Y. Kunitake, “Frictional drag reduction with air lubricant over a super-water-repellent surface,” Journal of Marine Science and Technology, vol. 5, pp. 123–130, Dec. 2000. [100] C. Lee and C.-J. Kim, “Maximizing the Giant Liquid Slip on Superhydrophobic Microstructures by Nanostructuring Their Sidewalls,” Langmuir, vol. 25, pp. 12812–12818, Nov. 2009. [101] C. C. Bizonne, B. Cross, A. Steinberger, and E. Charlaix, “Boundary Slip on Smooth Hydrophobic Surfaces: Intrinsic Effects and Possible Artifacts,” Physical Review Letters, vol. 94, pp. 056102+, Feb. 2005. 84 REFERENCES [102] C. H. Choi, Johan, and K. S. Breuer, “Apparent slip flows in hydrophilic and hydrophobic microchannels,” Physics of Fluids, vol. 15, no. 10, pp. 2897–2902, 2003. [103] T. C. Island, W. D. Urban, and M. G. Mungal, “Mixing enhancement in compressible shear layers via sub-boundary layer disturbances,” Physics of Fluids, vol. 10, no. 4, pp. 1008–1020, 1998. [104] E. Lauga and H. A. Stone, “Effective slip in pressure-driven Stokes flow,” Journal of Fluid Mechanics, vol. 489, no. -1, pp. 55–77, 2003. [105] Y. C. Jung and B. Bhushan, “Biomimetic structures for fluid drag reduction in laminar and turbulent flows,” Journal of Physics: Condensed Matter, vol. 22, pp. 035104+, Jan. 2010. [106] C. Lee and C. J. Kim, “Underwater Restoration and Retention of Gases on Superhydrophobic Surfaces for Drag Reduction,” Physical Review Letters, vol. 106, pp. 014502+, Jan. 2011. [107] C. Lee and C.-J. Kim, “Influence of Surface Hierarchy of Superhydrophobic Surfaces on Liquid Slip,” Langmuir, vol. 27, pp. 4243–4248, Apr. 2011. [108] R. J. Daniello, N. E. Waterhouse, and J. P. Rothstein, “Drag reduction in turbulent flows over superhydrophobic surfaces,” Physics of Fluids, vol. 21, no. 8, pp. 085103+, 2009. [109] B. Woolford, D. Maynes, and B. Webb, “Liquid flow through microchannels with grooved walls under wetting and superhydrophobic conditions,” Microfluidics and Nanofluidics, vol. 7, pp. 121–135, July 2009. [110] P. Joseph, C. C. Bizonne, J. M. Benoit, C. Ybert, C. Journet, P. Tabeling, and L. Bocquet, “Slippage of Water Past Superhydrophobic Carbon Nanotube Forests in Microchannels,” Physical Review Letters, vol. 97, pp. 156104+, Oct. 2006. [111] T. Onda, S. Shibuichi, N. Satoh, and K. Tsujii, “Super-Water-Repellent Fractal Surfaces,” Langmuir, vol. 12, pp. 2125–2127, Jan. 1996. [112] T. Sun, G. Wang, L. Feng, B. Liu, Y. Ma, L. Jiang, and D. Zhu, “Reversible Switching between Superhydrophilicity and Superhydrophobicity,” Angewandte Chemie International Edition, vol. 43, no. 3, pp. 357–360, 2004. [113] F. Xia, L. Feng, S. Wang, T. Sun, W. Song, W. Jiang, and L. Jiang, “DualResponsive Surfaces That Switch between Superhydrophilicity and Superhydrophobicity,” Adv. Mater., vol. 18, no. 4, pp. 432–436, 2006. [114] J. Lee, B. He, and N. A. Patankar, “A roughness-based wettability switching membrane device for hydrophobic surfaces,” Journal of Micromechanics and Microengineering, vol. 15, pp. 591+, Mar. 2005. REFERENCES 85 [115] A. Athanassiou, M. I. Lygeraki, D. Pisignano, K. Lakiotaki, M. Varda, E. Mele, C. Fotakis, R. Cingolani, and S. H. Anastasiadis, “Photocontrolled variations in the wetting capability of photochromic polymers enhanced by surface nanostructuring.,” Langmuir : the ACS journal of surfaces and colloids, vol. 22, pp. 2329–2333, Feb. 2006. [116] H. S. Lim, J. T. Han, D. Kwak, M. Jin, and K. Cho, “Photoreversibly Switchable Superhydrophobic Surface with Erasable and Rewritable Pattern,” Journal of the American Chemical Society, vol. 128, pp. 14458–14459, Nov. 2006. [117] T. N. Krupenkin, J. A. Taylor, E. N. Wang, P. Kolodner, M. Hodes, and T. R. Salamon, “Reversible Wetting Dewetting Transitions on Electrically Tunable Superhydrophobic Nanostructured Surfaces,” Langmuir, vol. 23, pp. 9128– 9133, Aug. 2007. [118] J. B. Boreyko and C. H. Chen, “Restoring Superhydrophobicity of Lotus Leaves with Vibration-Induced Dewetting,” Physical Review Letters, vol. 103, pp. 174502+, Oct. 2009. Paper reprints 87 TRITA-EE 2011:058 ISSN 1653-4146 ISBN 978-91-7501-086-1 www.kth.se