Integrating Biosensors for Air Monitoring and Breath-Based ...808742/...EHD ElectroHydroDynamic ESP...

Integrating Biosensors for Air Monitoring andBreath-Based Diagnostics

NIKLAS SANDSTRÖM

Doctoral ThesisStockholm, Sweden 2015

Front cover picture:Top: A droplet of water on a silicon diaphragm with 25 µm perforations.Middle: A 0.9 mm thick microfluidic OSTE+ cartridge fabricated by reactioninjection molding and direct covalent bonding.Bottom: Two different biosensor cartridges integrated with quartz crystal resonatorsusing mechanical clamping (left) and covalent bonding (right).

TRITA-EE 2015:020ISSN 1653-5146ISBN 978-91-7595-560-5

KTH Royal Institute of TechnologySchool of Electrical Engineering

Department of Micro and Nanosystems

Akademisk avhandling sommed tillstånd av Kungliga Tekniska högskolan framläggestill offentlig granskning för avläggande av teknologie doktorsexamen i elektriskmätteknik fredagen den 22 maj 2015 klockan 10.00 i Q2, Osquldas väg 10,Stockholm.

Thesis for the degree of Doctor of Philosophy at KTH Royal Institute of Technology,Stockholm, Sweden.

© Niklas Sandström, May 2015

Tryck: Universitetsservice US AB, Stockholm, 2015

iii

Abstract

The air we breathe is the concern of all of us but nevertheless we onlyknow very little about airborne particles, and especially which biologicalmicroorganisms they contain. Today, we live in densely populated societieswith a growing number of people, making us particularly vulnerable to airtransmission of pathogens. With the recent appearance of highly pathogenictypes of avian influenza in southeast Asia and the seasonal outbreaks ofgastroenteritis caused by the extremely contagious norovirus, the need forportable, sensitive and rapid instruments for on-site detection and monitoringof airborne pathogens is apparent.

Unfortunately, the integration incompatibility between state-of-the-artair sampling techniques and laboratory based analysis methods makesinstruments for in-the-field rapid detection of airborne particles an unresolvedchallenge.

This thesis aims at addressing this challenge by the development of novelmanufacturing, integration and sampling techniques to enable the use of label-free biosensors for rapid and sensitive analysis of airborne particles at thepoint-of-care or in the field.

The first part of the thesis introduces a novel reaction injection moldingtechnique for the fabrication of high quality microfluidic cartridges. Inaddition, electrically controlled liquid aspiration and dispensing is presented,based on the use of a thermally actuated polymer composite integrated withmicrofluidic cartridges.

The second part of the thesis demonstrates three different approaches ofbiosensor integration with microfluidic cartridges, with a focus on simplifyingthe design and integration to enable disposable use of the cartridges.

The third part to the thesis presents a novel air sampling techniquebased on electrophoretic transport of airborne particles directly to microfluidiccartridges. This technique is enabled by the development of a novelmicrostructured component for integrated air-liquid interfacing. In addition,a method for liquid sample mixing with magnetic microbeads prior todownstream biosensing is demonstrated.

In the fourth part of the thesis, three different applications for airborneparticle biosensing are introduced and preliminary experimental results arepresented.

Niklas Sandström, niklas. sandstrom@ ee. kth. seDepartment of Micro and Nanosystems, School of Electrical EngineeringKTH Royal Institute of Technology, SE-100 44 Stockholm, Sweden

iv

Sammanfattning

Vi är alla måna om en bra luftkvalité men trots det vet vi förvånansvärtlite om vilka partiklar som luften innehåller och speciellt vilka biologiskamikroorganismer som finns på dessa partiklar. Vi lever idag i tätt befolkadesamhällen med en stadigt växande befolkning, vilket gör oss extra sårbaraför luftburen smitta. Med de senaste årens uppkomst av högpatogena typerav fågelinfluensa i sydostasien och de årligen kommande utbrotten av vinter-kräksjuka så har behovet av snabba, känsliga och portabla analysinstrumentför detektion och övervakning av luftburna patogener blivit uppenbart.

Tyvärr så är nuvarande tekniker för insamling av luftprover och laborato-riebaserad analys inte kompatibla för att kunna integreras till ett instrumentsom möjliggör fältbaserad och snabb detektion av luftburna partiklar. Denutmaningen återstår att lösas.

Denna avhandling presenterar nya tekniska framsteg som avser attmöjliggöra användningen av biosensorer för snabb och känslig analys avluftburna partiklar vid ”point-of-care” eller i fält.

Den första delen av avhandlingen introducerar en ny formsprutningsteknikför tillverkning av högkvalitativa mikrofluidikchip. Dessutom presenteras enmetod för elektriskt styrd vätskehantering i mikrofluidikchip som är baseradpå en termisk expanderbar polymerkomposit.

Den andra delen av avhandlingen demonstrerar tre olika metoder förintegration av biosensorer med mikrofluidikchip, med fokus på att förenkladesignen och möjliggöra engångsanvändning.

Den tredje delen av avhandlingen presenterar en ny elektrostatisk in-samlingsteknik för luftburna partiklar direkt till mikrofluidikchip. Dennateknik möjliggörs genom utvecklandet av en ny integrerad mikrokomponentsom exponerar västkan i ett mikrofluidikchip för omgivande luft. Dessutomdemonstreras en metod för snabb omskakning och blandning av vätskeprovermed magnetiska mikropartiklar i direct anslutning till ett mikrofluidikchip.

I den fjärde delen av avhandlingen presenteras tre olika applikationer förmätning med biosensorer av luftburna partiklar tillsammans med preliminäraexperimentella resultat.

Niklas Sandström, niklas. sandstrom@ ee. kth. seAvdelningen för Mikro- och nanosystem, Skolan för elektro- och systemteknikKungliga Tekniska Högskolan, 100 44 Stockholm

v

Till min älskade familj

Contents

Contents vi

List of Publications ix

Abbreviations xv

Introduction to the Thesis xviiScope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviiObjective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviiContributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviiStructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviii

1 Introduction to Biosensing of Airborne Particles 11.1 What’s in the air? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Airborne pathogens . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3 State-of-the-art techniques for bioaerosol sampling . . . . . . . . . . 21.4 Labs-on-chip and biosensors - but only for liquids... . . . . . . . . . . 31.5 The interdisciplinary incompatibility problem . . . . . . . . . . . . . 4

2 Novel Materials and Manufacturing Methods for Labs-on-Chip 72.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.1.1 Prototyping of microfluidic polymer cartridges . . . . . . . . 82.1.2 Industrial manufacturing of microfluidic polymer cartridges . 82.1.3 OSTE and OSTE+ thermosetting polymers for labs-on-chip . 82.1.4 Methods for on-chip liquid actuation in disposable cartridges 10

2.2 Prototyping of high-quality microfluidic OSTE+ cartridges . . . . . 102.2.1 Reaction injection molding of microstructured OSTE+ parts 102.2.2 OSTE+ parts assembly by direct covalent bonding . . . . . . 112.2.3 Demonstration of OSTE+ microfluidic cartridges . . . . . . . 12

2.3 Electrically controlled single-use liquid aspiration and dispensing . . 142.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3 Advances in Biosensor Integration with Microfluidic Cartridges 17

vi

CONTENTS vii

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173.1.1 Label-free biotransducer technologies . . . . . . . . . . . . . . 173.1.2 Quartz crystal microbalance . . . . . . . . . . . . . . . . . . . 183.1.3 Rupture event scanning . . . . . . . . . . . . . . . . . . . . . 223.1.4 Anharmonic detection technique . . . . . . . . . . . . . . . . 223.1.5 Silicon photonic ring resonators . . . . . . . . . . . . . . . . . 243.1.6 State-of-the-art biosensor integration approaches . . . . . . . 25

3.2 Design considerations for microfluidic biosensor cartridges . . . . . . 263.2.1 Cartridges for quartz crystal resonators . . . . . . . . . . . . 263.2.2 Cartridges for photonic ring resonators . . . . . . . . . . . . . 28

3.3 Novel approaches to biosensor integration . . . . . . . . . . . . . . . 293.3.1 Mechanical clamping in thermoplastic cartridges . . . . . . . 313.3.2 Adhesive bonding using foils and tapes . . . . . . . . . . . . . 333.3.3 Direct covalent bonding with OSTE cartridges . . . . . . . . 363.3.4 Direct covalent bonding with OSTE+ cartridges . . . . . . . 38

3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4 Direct Airborne Particle Sampling to Microfluidic Cartridges 414.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

4.1.1 Electrostatic precipitation . . . . . . . . . . . . . . . . . . . . 424.1.2 Corona discharge . . . . . . . . . . . . . . . . . . . . . . . . . 424.1.3 Air-liquid interfacing . . . . . . . . . . . . . . . . . . . . . . . 43

4.2 Perforated diaphragms for air-liquid interfacing . . . . . . . . . . . . 434.3 Air sampling to microfluidic cartridges . . . . . . . . . . . . . . . . . 45

4.3.1 Sampling of airborne particles in ambient air . . . . . . . . . 454.3.2 Sampling of aerosols in breath . . . . . . . . . . . . . . . . . 474.3.3 Impact of ESP sampling on QCM frequency stability . . . . . 50

4.4 Rapid sample mixing in an on-chip sample tube . . . . . . . . . . . . 514.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

5 Applications for Biosensing of Airborne Particles 555.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555.2 Detection of narcotics substances in filters . . . . . . . . . . . . . . . 555.3 Monitoring and detection of norovirus in ambient air . . . . . . . . . 585.4 Breath-based diagnostics of influenza . . . . . . . . . . . . . . . . . . 64

6 Discussion and Outlook 67

Summary of Appended Papers 69

Acknowledgments 73

References 75

Paper Reprints 85

List of Publications

The thesis is based on the following peer-reviewed international journalpapers and conference proceedings:

1. ”Reaction injection molding and direct covalent bonding of OSTE+ polymermicrofluidic devices”, N. Sandström, R. Z. Shafagh, A. Vastesson, C. F.Carlborg, W. van der Wijngaart and T. Haraldsson, accepted for publicationin Journal of Micromechanics and Microengineering

2. ”Liquid Aspiration and Dispensing Based on an Expanding PDMS Com-posite”, B. Samel, N. Sandström, P. Griss and G. Stemme Journal ofMicroelectromechanical Systems, vol. 17, no. 5, pp. 1254-1262, 2008.

3. ”An integrated QCM-based narcotics sensing microsystem”, T. Frisk, N.Sandström, L. Eng, W. van der Wijngaart, P. Månsson, and G. Stemme,Lab on a Chip, vol. 8, no. 10, pp. 1648-1657, 2008.

4. ”One step integration of gold coated sensors with OSTE polymer cartridgesby low temperature dry bonding”, N. Sandström, R. Z. Shafagh, C.F. Carlborg, T. Haraldsson, G. Stemme, and W. van der Wijngaart,Proceedings of the 16th IEEE International Solid-State Sensors, Actuatorsand Microsystems Conference (TRANSDUCERS), pp. 2778-2781, 2011.

5. ”Integration of microfluidics with grating coupled silicon photonic sensorsby one-step combined photopatterning and molding of OSTE”, C. Errando-Herranz, F. Saharil, A. M. Romero, N. Sandström, R. Z. Shafagh, W. vander Wijngaart, T. Haraldsson, and K. B. Gylfason, Optical Express, vol. 21,no. 18, pp. 21293-21298, 2013.

6. ”Electrohydrodynamic enhanced transport and trapping of airborne particlesto a microfluidic air-liquid interface”, N. Sandstrom, T. Frisk, G. Stemme,and W. van der Wijngaart, Proceedings of the 21st IEEE InternationalConference on Micro Electro Mechanical Systems (MEMS), pp. 595-598,2008.

7. ”Aerosol sampling using an electrostatic precipitator integrated with amicrofluidic interface”, G. Pardon* and L. Ladhani*, N. Sandström, M.

ix

x LIST OF PUBLICATIONS

Ettori, G. Lobov and W. van der Wijngaart, Sensors and Actuators:B, vol.212, pp. 344-352, 2015

xi

The contribution of Niklas Sandström to the different publications:

1. all design, major part of fabrication, experiments and writing

2. all design, fabrication and experiments of the lateral aspiration, part of writing

3. all simulations, major part of writing

4. all design, major part of fabrication, experiments and writing

5. part of design, fabrication and writing

6. all design, fabrication and experiments, major part of writing

7. all initial design, simulation and experimental validation of the concept ofelectrostatic precipitation of airborne particles to liquid collectors, part ofdesign, fabrication, experiments and writing

xii LIST OF PUBLICATIONS

In addition, the author has contributed to the following peer-reviewedinternational journal papers:

8. ”Wafer-Scale Manufacturing of Bulk Shape-Memory-Alloy MicroactuatorsBased on Adhesive Bonding of Titanium-Nickel Sheets to Structured SiliconWafers”, S. Braun, N. Sandström, G. Stemme, and W. van der WijngaartJournal of Microelectromechanical Systems, vol. 18, no. 6, pp. 1309-1317,2009.

9. ”Increasing the performance per cost of microsystems by transfer bondingmanufacturing techniques”, W. van der Wijngaart, R.M. Despont, D.Bhattacharyya, P.B. Kirby, S.A. Wilson, R.P.R. Jourdain, S. Braun, N.Sandström, J. Barth, T. Grund, M. Kohl, F. Niklaus, M. Lapisa, and F.Zimmer MST News, vol. 2, pp. 27-28, 2008.

The work has also been presented at the following peer-reviewedinternational conferences:

10. ”Integration of polymer microfluidic channels, vias, and connectors withsilicon photonic sensors by one-step combined photopatterning and moldingof OSTE”, C. Errando-Herranz, F. Saharil, A. M. Romero, N. Sandström,R. Z. Shafagh, W. van der Wijngaart, T. Haraldsson, and K. B. Gylfason Pro-ceedings of the 17th IEEE International Conference on Solid-State Sensors,Actuators and Microsystems (TRANSDUCERS), pp. 1613-1616, 2013.

11. ”Microfluidic and transducer technologies for lab on a chip applications”,D. Hill, N. Sandström, K. Gylfason, F. Carlborg, M. Karlsson, T.Haraldsson, H. Sohlström, A. Russom, G. Stemme, T. Claes, P. Bienstman,A. Ka?mierczak, F. Dortu, M. J. Banuls Polo, A. Maquieira, G. M. Kresbach,L. Vivien, J. Popplewell, G. Ronan, C. A. Barrios, and W. van der WijngaartProceedings of the Annual International Conference of the IEEE Engineeringin Medicine and Biology Society, vol. 2010, pp. 305-307, 2010.

12. ”On-chip liquid degassing with low water loss”, J. M. Karlsson, T. Haraldsson,N. Sandström, G. Stemme, A. Russom, and W. van der WijngaartProceedings of the Micro Total Analysis Systems (?TAS) Conference onMiniaturized Systems for Chemistry and Life Sciences, pp. 1790-1792, 2010.

13. ”Screening of antibodies for the development of a fast and sensitivie influenza:A nucleoprotein detection on a nonlinear acoustic sensor”, K. Leirs, N.Sandström, L. Ladhani, D. Spasic, V. Ostanin, D. Klenerman, W. vander Wijngaart, S. Ghosh, and J. Lammertyn Proceedings of the EuropeanConference on Novel Technologies for In Vitro Diagnostics, 2014.

14. ”Rapid ultra-sensitive detection of influenza A nucleoproteins using a mi-crofluidic nonlinear acoustic sensor”, K. Leirs, N. Sandström, L. Ladhani,

xiii

D. Spasic, V. Ostanin, D. Klenerman, W. van der Wijngaart, J. Lammertyn,and S. Ghosh Proceedings of the Biosensors Conference, edition 24, 2014.

15. ”A laterally operating liquid aspiration and dispensing unit based on anexpanding PDMS composite”, B. Samel, N. Sandström, P. Griss, and G.Stemme Proceedings of the Micro Total Analysis Systems (?TAS) Conferenceon Miniaturized Systems for Chemistry and Life Sciences, vol. 1, pp. 530-532,2007.

16. ”Wafer-Scale Manufacturing of Robust Trimorph Bulk SMA Microactuators”,N. Sandström, S. Braun, T. Grund, G. Stemme, M. Kohl, and W. van derWijngaart Proceedings of the Actuators 08 Conference, pp. 382-385, 2008.

17. ”Full wafer integration of shape memory alloy microactuators using adhesivebonding”, N. Sandström, S. Braun, G. Stemme, and W. van der Wijn-gaart Proceedings of the 15th IEEE International Conference on Solid-StateSensors, Actuators and Microsystems (TRANSDUCERS), pp. 845-848, 2009.

18. ”Microsystems for Airborne Sample Detection”, N. Sandström, T. Frisk, G.Stemme, and W. van der Wijngaart Proceedings of the Mini-symposium onOpto-Microfluidics for Biological Large Scale Integration, 2008.

19. ”Lab-on-a-chip microsystems for point-of-care diagnostics”, N. Sandström,K. Gylfason, F. Carlborg, M. Karlsson, T. Haraldsson, H. Sohlström, A.Russom, D. Hill, G. Stemme, and W. van der Wijngaart Proceedings of theMicrofluidics in Bioanalytical Research and Diagnostics Conference, pp. 34-35, 2010.

20. ”Robust microdevice manufacturing by direct lithography and adhesive-free bonding of off-stoichiometry thiol-ene-epoxy (OSTE+) polymer”, A.Vastesson, X. Zhou, N. Sandström, F. Saharil, O. Supekar, G. Stemme,W. van der Wijngaart, and T. Haraldsson Proceedings of the 17th IEEEInternational Conference on Solid-State Sensors, Actuators and Microsystems(TRANSDUCERS), c, pp. 408-411, 2013.

21. ”Robust microdevice manufacturing by direct lithography and adhesive-free bonding of off-stoichiometry thiol-ene-epoxy (OSTE+) polymer”, A.Vastesson, X. Zhou, N. Sandström, F. Saharil, O. Supekar, G. Stemme,W. van der Wijngaart, and T. Haraldsson Proceedings of the 17th IEEEInternational Conference on Solid-State Sensors, Actuators and Microsystems(TRANSDUCERS), c, pp. 408-411, 2013.

22. ”Rapid Fabrication Of OSTE+ Microfluidic Devices With LithographicallyDefined Hydrophobic/ Hydrophilic Patterns And Biocompatible Chip Seal-ing”, X. Zhou, C.F. Carlborg, N. Sandström, A. Haleem, A. Vastesson,F. Saharil, W. van der Wijngaart, and T. Haraldsson Proceedings of the

xiv LIST OF PUBLICATIONS

Micro Total Analysis Systems (?TAS) Conference on Miniaturized Systemsfor Chemistry and Life Sciences, pp. 134-136, 2013.

23. ”OSTE+ microfluidic devices with lithographically defined hydrophobic/hydrophilic patterns and biocompatible chip sealing : OSTEmer AllowsEasy Fabrication of Microfluidic Chips”, X. Zhou, C.F. Carlborg, N.Sandström, A. Vastesson, F. Saharil, W. van der Wijngaart, and T.Haraldsson Proceedings of the European Conference on Novel Technologiesfor In Vitro Diagnostics, pp. 26-27, 2014.

Abbreviations

Abbreviation Explanation

Ab AntibodyADT Anharmonic Detection TechniqueBSA Bovine Serum AlbuminDMA Differential Mobility AnalysisEHD ElectroHydroDynamicESP ElectroStatic PrecipitationLoC Lab-on-ChipLoD Limit-of-DetectionNP NucleoProteinNV NoroVirus

OSTE Off-Stoichiometric Thiol-EneOSTE+ Off-Stoichiometric Thiol-Ene Epoxy

OSTE+RIM Off-Stoichiometric Thiol-Ene Epoxy Reaction Injection MoldingPDMS PolyDiMethylSiloxanePBS Phosphate Buffered SalinePCB Printed Circuit BoardPoC Point-of-CareQCM Quartz Crystal Microbalance

QCM-D Quartz Crystal Micorbalance with Dissipation monitoringREVS Rupture Event ScanningRIM Reaction Injection MoldingSEM Scanning Electron MicroscopyTEM Transmission Electron MicroscopyTSM Thickness Shear ModeVLP Virus-Like-ParticeXB Expancel Beads (microspheres)

xv

Introduction to the Thesis

ScopeThe work summarized in this thesis has been conducted within the framework ofthree different research projects, RPA, Rapp-Id and Norosensor, which are describedin chapter 5. The thesis is a summary of seven peer-reviewed international journalpapers and conference proceedings. It also includes preliminary results that are yetunpublished but that have made significant contributions to the projects mentionedabove.

ObjectiveThe objective of this thesis is to bring technological advancements that enable theuse of highly sensitive and disposable microfluidic biosensors for the detection ofairborne particles.

The targeted applications include monitoring and detection of airborne pathogensand narcotic substances in the field, and breath-based diagnostics at point-of-care.

ContributionsIn pursuit of the objective of this thesis, the following technical advancements weremade (figure 1):

• Prototyping of high-quality microfluidic cartridges using an industrial-likereaction injection molding technique (paper 1)

• Electrically controlled single-use liquid aspiration and dispensing in microflu-idic cartridges (paper2 ).

• Biosensor integration with microfluidic cartridges featuring a significantreduction of design and integration complexity compared to state-of-the-art(paper 3, 4 and 5).

• Air-liquid interfacing integrated in close proximity to the sensor in mi-crofluidic biosensor cartridges, enabling rapid detection of captured airborneparticles (paper 3 and 6).

xvii

xviii ABOUT THE THESIS

• Efficient airborne particle sampling, in ambient air and synthetic breath,directly to air-liquid interfaces in microfluidic cartridges (paper 6 and 7).

• Rapid off-chip sample mixing and seamless sample transfer to microfluidiccartridges.

StructureThe thesis is structured as follows:

• Chapter 1 gives an introduction to airborne particles and state-of-the-art airsampling techniques and highlights why rapid and sensitive detection usingbiosensors is not applicable to air airborne particles with current technology.

• In chapter 2, the fabrication of microfluidic polymer cartridges is demon-strated by the use of a novel prototyping method and a thermally actuatedcomposite material for on-chip liquid handling.

• In chapter 3, the development of biosensor integration with microfluidiccartridges for point-of-care applications is presented using one conventionaland two novel approaches.

• In chapter 4, a novel technique for sampling of airborne particles directlyto microfluidic devices is demonstrated as well as an efficient technique forsample mixing with magnetic microbeads.

• In chapter 5, applications for the sampling and biosensing of airborne particlesare presented, in which technologies developed in this thesis are used.

• Chapter 6 discusses the outlook of biosensing of airborne particles with respectto the contributions made in this thesis.

xix

Chapter 3 - Advances in Biosensor Integration with Micro�uidic Cartridges

Direct covalent bonding(paper 4, 5)

Adhesive bonding(paper 3)

Mechanical clamping

Reaction injection molding of miro�uidic cartridges(paper 1)

Chapter 2 - Novel Materials and Manufacturing Methods for Labs-on-Chip

On-chip liquid aspiration and dispensing(paper 2)

Chapter 4 - Direct Airborne Particle Sampling to Micro�uidic Cartridges

Sampling of airborne particles to micro�uidic cartridges (paper 6, 7)

Air-liquid interfacing in micro�uidiccartridges (paper 3, 6, 7)

Chapter 5 - Applications for Biosensing of Airborne Particles

Detection of norovirusin breath (paper 7)

Detection of in�uensa virus, norovirusand narcotics in ambient air (paper 3) biosensor cartridge

Rapid o�-chip liquid sample mixing inconnection to micro�uidic cartridges

Figure 1: Overview of the main contributions of the thesis.

Chapter 1

Introduction to Biosensing ofAirborne Particles

1.1 What’s in the air?

To me, air is intriguing. We seldom give it much thought, but when we sense thatgood food is cooking, that spring is approaching or that someone just broke wind,we know that there is more in air than just oxygen. It is an invisible and, to someextent, unexplored world that is full of airborne particles, which are too small tobe seen but nevertheless affecting our daily lives.

So, what are these airborne particles? Well, there are both non-biological andbiological particulate matter in air, commonly categorized by size: particles with ahydrodynamic diameter smaller than 10 microns and a diameter smaller than 2.5microns. Specifically, the particles are considered potentially dangerous due to thatthey are able to travel deep into our lungs, causing more severe effects comparedto larger particles that get caught in the upper respiratory tract.

The non-biological particulate matter, such as dust and pollutants, consistsof solid particles and aerosol droplets, generated either by naturally occurringprocesses or by human activities. The biological particulate matter, such as bacteriaand virus, is commonly referred to as the bioaerosol and consist of both viableand nonviable particles, which might just be fragments of a complete bioparticle.Whereas both types of particulate matter can cause human diseases, particles inbioaerosols can cause acute and severe infections that can be further transmittedform person-to-person and are therefore of particular interest in the work of thisthesis since.

1.2 Airborne pathogens

Much effort is put into understanding the effects of the particulate matter in ouratmosphere on human health, particularly of the non-biological species [1]. Less

1

2 CHAPTER 1. INTRODUCTION

is known about bioaerosols and specifically about the transmission of airbornepathogens.

There are several pathogens that are known to be transmitted via air fromperson-to-person, including virus such as influenza (flu), varicella (chicken pox) andrubeola (measles), and bacteria such as mycobacterium tuberculosis (tuberculosis),bordetella pertussis (whooping cough) and methicillin-resistant staphylococcusaureus (e.g. pneumonia). These airborne pathogens can travel on solid particles,such as dust and skin flakes, or in aerosol droplets generated by breathing, talking,coughing, sneezing, vomiting, etc.

A recent investigation revealed a tight correspondence with the concentrationof bioaerosols to the activities and number of people in lecture halls [2]. Anotherinvestigation showed a high presence of certain pathogens in the ambient air inhospitals [3]. Larger aerosol droplets rapidly settle but smaller droplets quicklyevaporate to form droplet nuclei that are suspended in air for long periods of time[4]. Consequently, the pathogens spread between people in the same room but alsoover long distances, from one ward to another, potentially causing major outbreakswith significant impact on both healthcare and economy.

There are also airborne pathogens that are transmitted from other sources thaninfected persons, such as bacterial and fungal spores, bacteria in bird droppings orin liquid-based systems such as in air-conditioning. Also flushing toilets have beenfound to generate pathogen-containing aerosols [5]. When humans are exposedto the environment in which such pathogens exist, there is a risk of airbornetransmitted infection.

1.3 State-of-the-art techniques for bioaerosol sampling

There are several available techniques for sampling and detecting airborne particles.Microscale airborne particles, such as dust, can be detected and quantified but notidentified using optical particle counters. Finer airborne particles in the nanometerrange, such as gas compounds, can be analyzed using mass spectrometry. However,to analyze bioaerosols, other methods are required.

A passive method for bioaerosol sampling is to use a settle plate, to whichthe airborne particles sediment. A petri dish settle plate enables subsequentculturing, counting and identification of species. However, passive methods areonly recommended in settings where active methods are not applicable [6].

Three of the most common active sampling methods that are used for bioaerosolsare filtering, impactors and impingers. In filtering, the air is pumped through adevice, such as a cyclone (SKC Inc., USA), in which the particles are trappedin a mechanical filter with a precisely defined pore size. Airborne pathogens riskdehydration after capture but subsequent analysis, for example by microscopy, ispossible by transferring the captured species to culture media or water dissolvablegelatin.

1.4. LABS-ON-CHIP AND BIOSENSORS - BUT ONLY FOR LIQUIDS... 3

Impactors are based on devices in which high air flows make airborne particlesimpact on a solid collector surface or alternatively an adhesive or a petri dish. Multi-stage impactors, such as the Andersen cascade impactor (Thermo Fisher ScientificInc, USA), consist of several sections designed for the capture of increasingly smallerparticles. Long sampling times introduce the risk of dehydration and the particlescan rupture due to mechanical stress upon impact.

In impingers, such as the commonly used Biosampler (SKC Inc., USA), thesampled particles of a bioaerosol collide with, and become trapped in, a liquidinstead of a solid surface as in impactors. This makes the particles available forimmediate liquid-based analysis, such as polymerase chain reaction (PCR) [7].Whereas there is no risk of dehydration of the particles, mechanical stress atimpingement can cause rupture of the particles. Impingers require relatively largeliquid volumes and may therefore require subsequent up-concentration to enabledetection of low concentration samples.

The commonly used sampling methods for bioaerosols that are presentedabove rely on laboratory-based analysis of the capture particles. No establishedmethods offer a sensitive detection, identification and quantification at the point ofsampling. However, a sensitive and direct detection of airborne pathogens at point-of-care (PoC) close to the patient, or for air monitoring in healthcare and otherfacilities, would potentially enable rapid diagnostics, disease spreading and outbreakprevention, and immediate bioaerosol controls after cleansing of contaminatedfacilities.

The recent outbreak of Ebola in western Africa posed a great challenge for theworld community, which lacked sufficient resources, organization and tools to stopthe outbreak at an early stage (ref). When a similar outbreak of an airborne diseasehappens, such as an avian flu (H5N1) pandemic [8], we better be well preparedto handle the situation immediately and efficiently. Despite the lack of rapid andsensitive bioaerosol detection by currently available methods, there are emergingtechniques with great potential.

1.4 Labs-on-chip and biosensors - but only for liquids...

The last couple of decades, there has been a tremendous technology developmentthe last couple of decades for the analysis of pathogens and other bioparticles inliquid samples, such as blood, urine and nasal swabs. This is particularly true forlabs-on-chip (LoC) devices, which feature microfluidic channels that enable rapidanalysis due to short diffusion times, well controlled sample handling due to laminarflow and low consumption of liquids and reagents due to miniaturization.

The development LoC devices has been accompanied by a strong developmentof biosensors, which ideally offer rapid and sensitive detection with high specificityof particles in liquid. Integrating biosensors with LoC devices potentially makes upfor very powerful analysis systems, which we are likely to encounter more frequentlyin future healthcare.

4 CHAPTER 1. INTRODUCTION

Before that happens, there are issues that need to be addressed. Less technicallyadvanced LoC systems are already established and used extensively worldwide atPoC, such as laminar flow pregnancy tests, but devices with more sophisticatedfeatures, such as integrated biosensors, are sparse. Advanced desktop biosensorinstruments have been successfully commercialized and used for bioparticle analysisin laboratory settings. The reasons for why such instruments have not yetbeen successfully applied to PoC applications in healthcare are complex, butmight depend on factors such as technical hurdles, low user friendliness, marketunacceptance, high RnD and prototyping costs and high cost-per-test.

Considering our growing and aging population, there is an urgent need to bringin the latest technological advancements into the healthcare. Therefore, furtherresearch is required to address some of the above mentioned and remaining issues.

The potential market is largely fragmentized, with applications that differsignificantly in needs and requirements. Therefore, access to straightforwardmethods for prototyping is crucial during the development of new LoC andbiosensor devices. As of today, the de facto standard material and fabricationmethod of microfluidic polymer devices in academic research is not applicable toindustrial manufacturing, thus requiring complete redesigning of the devices toenable commercialization [9].

One focus of this thesis, is to develop new manufacturing methods that bridge thefabrication technology gap between research prototyping and industrial manufactur-ing of microfluidic devices.

When diagnostic tests are performed at the PoC or when measurements areconducted in fieldwork, the test cartridges are typically disposed after use. This isdue to the risk of pathogen transmission and contamination. Therefore, the costper test must be low, in the range of a few euros. This poses a great challenge formicrofluidic biosensor cartridges, partly due to the cost of all subcomponents, butprimarily due to the fabrication cost in the back-end processing when the biosensoris integrated with the microfluidic cartridge [10].

Another focus of this thesis, is to reduce the design and integration complexity ofbiosensor cartridges by the development of novel integration approaches

1.5 The interdisciplinary incompatibility problem

Unfortunately, the multifaceted tool box that now exists for LoCs and biosensorsis exclusive for liquid based samples. To open up the same possibilities for thedetection and analysis of airborne particles, such as bioaerosols, compatible airsampling and air-to-liquid interfacing techniques must be developed. Additionally,concatenation of airborne sampling and on-chip detection can be used in purificationsteps for other biosensing applications, as exemplified by narcotic sensing in 5.

1.5. THE INTERDISCIPLINARY INCOMPATIBILITY PROBLEM 5

To solve the incompatibility problem of the air sampling and liquid-based sensingtechnologies, an interdisciplinary approach is required.

A third focus of this thesis, is to develop technology that enables us to use existinghigh performance LoC and biosensor devices for the direct detection of airborneparticles, and in particular airborne pathogens.

Enjoy!

Chapter 2

Novel Materials andManufacturing Methods forLabs-on-Chip

This chapter presents a new prototyping method for manufacturing of microfluidicdevices, aiming to address the fabrication technology gap between academicresearch and industrial manufacturing as discussed in chapter 1.4.

First, a background to existing fabrication technologies of microfluidic polymercartridges is given in section 2.1. Thereafter, a novel high-fidelity replicationtechnique based on reaction injection molding is demonstrated in section 2.2 forthe fabrication of microfluidic polymer cartridges with exceptionally low residualstress. In section 2.3, a thermally responsive composite is demonstrated for single-use liquid aspiration and pumping in disposable microfluidic devices.

2.1 Introduction

Microfluidic cartridges are predominantly made in glass, silicon and polymers usingvarious fabrication techniques, including photolithography, etching, machiningand replication. Micro structuring of glass and silicon is typically based onsemiconductor processing methods and is expensive for relatively large chips, suchas centimeter-scale microfluidic devices. In contrast, polymer microstructuringfeatures relatively low material and fabrication costs and is compatible with a widerange of polymers with different material characteristics. Therefore, microfluidicpolymer cartridges are favorable for use in PoC and fieldwork applications thatrequire disposable cartridges.

7

8 CHAPTER 2. FABRICATION

2.1.1 Prototyping of microfluidic polymer cartridges

Since Whitesides et al. introduced the thermosetting polymer polydimethylsiloxane(PDMS) for the fabrication of microfluidic cartridges [11], it has become the workhorse in academic research and is by far the most commonly used material formicrofluidic cartridges. Microstructured parts are replicated through casting usingmicrostructured molds, such as photolithographically defined SU-8 structures onsilicon wafers. The parts are typically assembled by oxygen plasma bonding togenerate complete microfluidic cartridges [12] but other techniques also exist [13].

Recently, new promising thermosetting polymers were introduced, includingpolyurethane [14] and thiol-ene based resins [15], featuring significantly shorterprocessing times compared to PDMS.

2.1.2 Industrial manufacturing of microfluidic polymercartridges

For industrial manufacturing of microfluidic polymer cartridges, thermoplasticpolymers are exclusively used, including cyclic olefin polymer (COP), cyclic olefincopolymer (COC), polymethylmethacrylate (PMMA), polycarbonate (PC) andpolystyrene (PS) [16]. In contrast to PDMS, thermoplastic parts can be rapidlymanufactured in large quantities using industrial replication techniques, of whichmicro injection molding is the most established method [10].

However, the thermoplastic polymers and fabrication techniques suffer froman inherent and significant drawback: the fabricated parts have surfaces thatare chemically non-reactive to themselves or to common biosensor materials, thusprecluding seamless and facile surface modifications and heterogenous integration.

To build a complete and functional microfluidic device from two or morethermoplastic parts, laborious and/or difficult to control back-end processes arerequired that depend on either modification of the polymer surface by techniquessuch as plasma bonding, thermal bonding or ultrasonic bonding, or on the additionof new material by chemical bonding or adhesive bonding [17]. Using such methodsfor the heterogenous integration of biosensors with polymer cartridges severelylimits the ability to produce low cost and disposable products, thus eliminatingthe potential use in PoC and fieldwork applications.

2.1.3 OSTE and OSTE+ thermosetting polymers forlabs-on-chip

Off-stoichiometric thiol-ene (OSTE) polymers are a new type of thermosettingpolymers for the fabrication of micro structured devices, offering the unique abilityof direct covalent bonding of molded and microstructured parts. An overview of itscapabilities is given by Haraldsson et al. [18].

2.1. INTRODUCTION 9

Single vs. dual polymerization thiolene polymers

The thermoset currently exists in two different versions, OSTE [19] and OSTE+[20]. OSTE is a UV curable material and is based on thiol and allyl functionalizedmonomers, respectively, whereas OSTE+ is a UV/UV or UV/thermally curablematerial formed by dual polymerization reactions and is based on thiol, allyl andepoxy functionalized monomers, respectively.

The first polymerization reaction of OSTE+ is the same as in OSTE and isbased on the UV polymerization of the monomers with thiol and allyl groups.In OSTE, the resulting fully polymerized and off-stoichiometric material has anexcess of either thiol or allyl and has either soft or stiff mechanical propertiesdepending on the selected formulation. In OSTE+, the partially polymerized andoff-stoichiometric material has an excess of thiol and has soft mechanical properties.

In the second reaction of OSTE+, the remaining monomers with unreactedthiol are polymerized with the monomers that have epoxide groups in a second UVor thermally-initiated curing step. This makes the material fully polymerized sothat it achieves its end chemical and mechanical properties, which is either soft orstiff depending on the selected formulation. OSTE+ is then stoichiometric and hashydroxyl groups on its surface and, in contrast to most other common polymers formicrofabrication, it is natively hydrophilic [20].

In back-end processing, the remaining functional surface groups of polymerizedOSTE (thiol or allyl) and partially polymerized OSTE+ (thiol and epoxide) areavailable for surface reactions, including surface modifications [21] and/or covalentbonding [22]. During the back-end processing, the second reaction of OSTE+ isinitiated and the material becomes fully polymerized.

Casting of OSTE and OSTE+

OSTE and OSTE+ can be be molded by casting to form microfluidic parts, seefigure 2.1. Firstly, the polymer precursor is poured into a microstructured moldthat is either left open or closed with a transparent lid. Secondly, the thiol-allylpolymerization reaction is initiated and completed by exposure to UV-light. Lastly,the mold is opened and the microstructured and polymerized OSTE or partiallypolymerized OSTE+ part is demolded. The molded part is then subjected toback-end processing or, in the case of OSTE+, directly finished by the secondthiol-epoxide polymerization initiated by a last UV or thermal curing step.

The casting of OSTE+, particularly in PDMS molds, suffers from three signif-icant drawbacks. The first thiol-allyl polymerization is exothermic and generatesexcessive heat. This is not an issue in OSTE, however, in OSTE+ the generatedheat during molding can trigger the second thiol-epoxide polymerization reaction,causing a premature complete polymerization of the material. Consequently, themolded part risks irreversible bonding with the mold and the unique back-endprocessing capabilities based on the direct covalent reactions are lost. Secondly,casting does not allow for the molding of vertical interconnects without squeeze-


film formation. Thirdly, the material shrinkage during polymerization limits thereplication fidelity and introduces residual stress in the molded part.

2.1.4 Methods for on-chip liquid actuation in disposablecartridges

In fieldwork applications or in PoC applications where contagious patient samplesare handled, it is preferred to have liquids self-contained in the microfluidic cartridgefor safe and easy disposal and for limiting the risk for contamination. This requireson-chip liquid handling. Several on-chip liquid dispensing methods in microfluidicpolymer cartridges have been proposed [23–25], but liquid aspiration, which in manycases is needed to introduce samples and reagents prior to subsequent downstreamdispensing, has often been neglected needs to be addressed.

2.2 Prototyping of high-quality microfluidic OSTE+cartridges

2.2.1 Reaction injection molding of microstructured OSTE+parts

Reaction injection molding of OSTE+, termed OSTE+RIM, is a novel fabricationmethod developed during the work of this thesis and is presented in paper 1 andin figure 2.1. OSTE+RIM combines the merits of OSTE+ polymers, by efficientlyseparating the dual polymerization reactions, with the molding of microstructuredparts with high replication fidelity. This is enabled by two important features ofOSTE+RIM, temperature stabilization and shrinkage compensation.

Temperature stabilization for separated dual polymerizations

Temperature stabilization of the mold maintains a low polymerization temperatureduring molding, despite the exothermic reaction, and thereby separates the dualpolymerization reactions in the first curing step. This is achieved by using a moldmaterial with high heat conductance, e.g. Al, optionally assisted by active coolingthrough e.g. a Peltier element.

Shrinkage compensation for high-fidelity replication

Shrinkage compensation during molding is achieved by using the special shrinkagebehavior of OSTE+, which solidifies at a late stage of the first thiol-allylpolymerization. During shrinkage, the liquid OSTE+ precursor refills from inletand outlet reservoirs into the mold cavity. Consequently, the partially polymerizedmolded part features high replication fidelity with low residual stress. The secondthiol-epoxide polymerization, used in the back-end processing, introduces very littleshrinkage and residual stress (ref).

2.2. OSTE+ PROTOTYPING 11

Overview of the OSTE+RIM process

The OSTE+RIM process begins with the assembly of the mold and is followed bythe injection of the OSTE+ precursor into the mold cavity via an injection portwhile air escapes through a ventilation port. To initiate the first polymerizationstep, the OSTE+ precursor in the mold is exposed to UV-light. While the allyl-thiol reaction lasts, the generated heat is dissipated via the mold and the liquidOSTE+ precursor backfills into the mold from the injection and the ventilationports to compensate for shrinkage. The mold is then opened and the replicatedOSTE+ part is demolded. Unwanted polymer structures from the mold injectionand ventilation channels are cut off and removed.

OSTE+RIMUV polymerization of allyl-thiol

Casting/Injection

Casting

Molding Fabricated part Parts assembly Bonding Fabricated device

a)

b)

open through-holes

Al mold

PDMS mold

shrinkagecompensation

precursor back�lls themold cavity from

inlet/outlet reservoirs

temperaturestabilization

heat dissipation throughthe mold to prevent the

initiation of thethiol-epoxide reaction

UV polymerization of allyl-thiol

OSTE+: UV/heat polymerization of thiol-epoxide

OSTE: heat-accelerated reaction of allyl-thiol

polymerizationshrinkage

the shrinkage leads tolow replication �delity and

high residual stress

exothermicpolymerization

the generated heat risksinitiating the second

thiol-epoxide reactionof OSTE+

OSTE+

OSTE or OSTE+

OSTE+: UV/heat polymerization of thiol-epoxide

low residual stress

high residual stress

low shrinkage

high shrinkage

partial polymerization

premature dual polymerization closed through-holes

squeeze-�lm formation

warpageOSTE+

OH OH OH OH

less reactive surfaceafter casting

high replication �delitygives sharp corners

no warpageOSTE+

OSH O SH

highly reactive surfaceafter OSTE+RIM

inaccurate replicationgives rounded corners

MAGNIFICATIONS

MAGNIFICATIONS

Figure 2.1: Processing schemes for a) casting of OSTE and OSTE+ polymers and b)OSTE+RIM for OSTE+ polymers.

2.2.2 OSTE+ parts assembly by direct covalent bonding

After OSTE+RIM, the replicated and intermediately polymerized OSTE+ part isavailable for back-end processing. In a typical process, the OSTE+ part is alignedto, and brought in contact with another part for direct covalent bonding. Thestack is exposed to UV-light and/or heated in an oven to initiate and/or accelerate,depending on which OSTE+ formulation is being used, and complete the secondthiol-epoxide polymerization to achieve a covalent bonding between the parts andto obtain the final material properties of the polymer.


2.2.3 Demonstration of OSTE+ microfluidic cartridgesMicrofluidic OSTE+ cartridges made by combining OSTE+RIM and direct cova-lent bonding, demonstrated in figure 2.2 and 2.3, are characterized by strong leak-tight bonding, native hydrophilic surfaces, high optical transparency, accuratelyreplicated microstructures, no detectable warpage, homogenous thickness and littleresidual stress.

Figure 2.2: Photographs of a microstructured aluminum mold for OSTE+RIM and the resultingOSTE+ cartridge.

2.2. OSTE+ PROTOTYPING 13

10 mm

d)

side view

10 mm

OSTEMER 322 micro�uidic device

a) top view

0.3 mm

b) meander channel

magni�ed top view

0.9 mm

e) tubetube connector

magni�ed side view

2 mm

0.4 mm 0.2 mm

c) chamber straight channel

magni�ed top view

Figure 2.3: Photographs of an OSTE+ microfluidic device fabricated by OSTE+RIM


2.3 Electrically controlled single-use liquid aspiration anddispensing

In the work of this thesis, an on-chip liquid aspiration and dispensing method basedon a thermally responsive polymer composite was developed, see paper 2. Thecomposite consists of Expancel microspheres, also referred to as expandable beads(XB), which have liquid hydrocarbon enclosed in a hard thermoplastic polymershell. The shell softens upon heating, allowing the spheres to irreversibly expand.

The expandable microspheres were embedded in a soft polymer matrix con-sisting of PDMS. The thermally responsive PDMS-XB composite is actuated bylocalized heating, which triggers a local volume expansion. Using selective bonding,the composite was integrated with a microfluidic system on a PCB with resistiveheaters, see figure 2.4 a.

a) Illustration of a micro�uidic pump based on a PDMS-XB composite

b) The inlet channel is �lled with liquid c) The liquid �lls the formed micro�uidic chamber

d) The inlet/outlet valves are closed/opened e) The liquid is pumped by closing the chamber

1 mm

PCBPDMS-XB

PDMSplastic sheet

inlet channel �rst chamber actuation

actuated inlet valve actuated outlet valve last chamber actuation

channel heaters selective bonding

Figure 2.4: a) Illustration and b-e) photographs showing the sequence of liquid aspiration anddispensing using the thermally actuated PDMS-XB composite.

Upon actuation, the volume change of the composite allow for the formationand closing of microfluidic cavities and channels. Thereby, a liquid can first beaspirated into a formed microfluidic cavity, see figure 2.4 b-c, whereafter it can bedispensed into a downstream microchannel by closing and opening inlet and outletvalves, respectively, see figure 2.4 d-e. The resistive heater that was used to closethe formed fluidic cavity was designed such that the heat propagation is alignedwith the direction of the liquid flow to avoid trapping of liquid during dispensing.

The on-chip liquid handling system was electronically controlled and wasbased on heterogenous integration with a PCB, which could also be used for theintegration of quartz crystal biosensors as shown below in section 3.3.4. Thispotentially enables the fabrication of disposable biosensor cartridges with self-contained on-chip liquid handling for fieldwork and PoC applications.

2.4. DISCUSSION 15

2.4 Discussion

OSTE+RIM brings a strong advancement in the field of prototyping of microstruc-tured parts, enabled by heat stabilization and shrinkage compensation. It isbased on an industrial-like molding technique for thermosetting polymers withlow requirements on technical setup and offers an easy-to-use and straight-forwardprocess suitable for research and development.

Combined with the facile back-end processing of OSTE+ polymers, OSTE+RIMprovides high-fidelity manufacturing of microfluidic cartridges that feature lowresidual stress, high optical transparency and high mechanical stiffness. Theresulting cartridges are very similar to industrially manufactured thermoplasticcartridges. Thus, OSTE+RIM can potentially bridge the fabrication technologygap between academic research and industrial manufacturing of microfluidiccartridges.

Given the additional benefits of OSTE+ processing, such as facile surface mod-ifications, strong heterogeneous integration capabilities and hydrophilic surfaces,OSTE+RIM could potentially also challenge established manufacturing techniquesin industry for certain commercial applications.

In preliminary tests, we have successfully explored the possibilities to extendthe capabilities of OSTE+RIM even further. Due to that the OSTE+ precursorhas relatively low viscosity and only solidifies upon active curing, it allows for astraightforward injection into the mold cavity. This feature enables the possibility ofbatch-fabrication by having multiple mold cavities in a single mold. In preliminarytests, we have successfully molded four parts in a single batch. These partsconstituted the top and bottom pairs of two microfluidic cartridges. After molding,the parts were handled and aligned, with the use of a carrier/release liner, in pairsof bottom parts and top parts. In the back-end processing, the bottom and topparts were bonded in a single batch, generating two complete microfluidic cartridge.Future research will include more parts in a single batch for demonstrating thisprocess.

Another intriguing possibility of OSTE+RIM processing is the in-mold het-erogeneous integration of various materials and components with the OSTE+polymer during polymerization. In preliminary tests, we patterned a mold with anelectrically conductive spray prior to molding. During OSTE+RIM , the OSTE+precursor reacted with the electrically conductive and patterned material. Upondemolding, the material was transferred from the mold to the replicated OSTE+part. Although the electrical conductivity of the transferred patterns was lowon the molded part and the process needs further optimization, it conceptuallydemonstrates the possibilities of heterogenous integration in the molding step ofOSTE+RIM.

A similar approach is to integrate a functional material in the bulk of thereplicated part by first mixing it with the polymer precursor. This was tested forthe integration of Expancel beads with molded OSTE+ parts. To retain the samefunctionality of the OSTE+ composite as PDMS-XB, an OSTE+ formulation with


soft mechanical end properties was used. The results demonstrated a successfulexpansion of the composite upon thermal actuation. With further optimization ofthe composition of the composite, liquid aspiration and dispensing can potentiallybe integrated in an OSTE+ microfluidic cartridge.

One possibility with great potential is two-component OSTE+RIM, in whichthe precursors of two different polymer formulations are subsequently injected intothe mold, generating a replicated part that consists of multiple materials. Thistechnique already exists in industrial injection molding but in contrast, OSTE+RIMfor example enables the fabrication of replicated parts that consists of two differentmaterials with very similar surface chemical properties but different mechanicalproperties. This idea is yet to be tested but is potentially attractive for a numberof applications, including the fabrication of integrated gaskets in OSTE+ biosensorcartridges.

Chapter 3

Advances in Biosensor Integrationwith Microfluidic Cartridges

This chapter introduces new methods for the manufacturing of high performancemicrofluidic biosensor cartridges for disposable use, aiming for reducing thedesign and integration complexity of current biosensor cartridges, as discussed inchapter 1.4. The work in this thesis is focused on two different types of label-free biosensors, namely acoustic quartz crystal resonators and optical silicon ringresonators.

First, an introduction to label-free biotransducer techniques and conventionalbiosensor integration approaches is given in section 3.1. Thereafter, importantdesign considerations are explained for the integration of quartz crystals andsilicon photonic ring resonators in section 3.2. Lastly, new advances for biosensorintegration with microfluidic cartridges based on three different approaches aredemonstrated in section 3.3.

3.1 Introduction

3.1.1 Label-free biotransducer technologies

Biosensing is commonly categorized into label-based and label-free biosensing.The former is based on the indirect measurement of an analyte particle thathas been labeled with an entity that is easily detected, such as a fluorophore.A well established label-based method is enzyme-linked immunosorbent assay(ELISA) [26]. Label-based biosensing can achieve detections of single molecules [27]but typically only offers end-point readouts.

In contrast, label-free biosensing refers to the direct measurement of an analyteparticle, for example by the binding of the particle to an immobilized receptor onthe surface of a transducer. Together, the receptor and the transducer form what iscommonly referred to as a biotransducer, which converts biological signals or events

17

18 CHAPTER 3. ADVANCES IN BIOSENSOR INTEGRATION

QCM and QCM-D

Quartz crystal resonator Photonic ring resonator

REVS ADT

Time

Freq

uenc

y sh

ift

Drive voltage

Sign

al o

f AU

Drive voltage

3f c

urre

nt

Wavelength

Pow

er tr

ansm

issi

on

Dis

sipa

tion

shift

TSM oscillation amplitude Increasing TSM oscillation amplitude Increasing TSM oscillation amplitude

Bond breakage

Resonance shift

No bond breakage

AT-cut quartz crystal Si3N4 waveguides on SOIimmobilized receptor

captured antigen

Optical power density distribution

Figure 3.1: Illustration of biosensor techniques based on quartz crystal resonators, includingQCM and QCM-D, REVS and ADT, and photonic ring resonators.

into a detectable readout signal.A recent review of recent developments in the field of label-free electrical,

mechanical and optical biosensors is given in [28]. Label-free biosensing typicallyfeatures real-time monitoring of the binding events and therefore allows forinstantaneous detection in the presence of analyte particles. This is particularlyfavorable for the applications that are targeted in the work of this thesis, whereparticle detection both at PoC and in fieldwork require rapid analysis.

This thesis is focused on two types of resonating label-free biosensors, which arebased on acoustic and optical transducers. One type is piezoelectric quartz crystals,which are well established biosensors that are commonly referred to as quartzcrystal microbalances (QCMs) [29]. The other type is silicon photonic sensors,which have a had strong development during the last decade particularly for LoCapplications where miniaturization is critical [30]. Whereas silicon photonic sensorsallow for significant miniaturization and multiplexing, QCM-based biosensors onlyrequire small-scale electronics to operate and are thus suitable for the integration incompact and portable instruments. These two types of biotransducer technologiesare introduced in more detail below and illustrated figure 3.1.

3.1.2 Quartz crystal microbalanceQCMs are acoustic mass sensing transducers based on piezoelectric thickness shearmode (TSM) quartz crystal resonators and were first introduced by Sauerbrey in1959 [31]. They were later demonstrated for mass sensing in liquid conditions in1980 [32, 33], which paved the way for liquid-based biosensing.

3.1. QUARTZ CRYSTAL MICROBALANCE 19

Attana Cell 200 Q-sense Omega Autoa) b)

Figure 3.2: Photographs of commercially available QCM instruments. a) Q-sense Omega Auto(QCM-D), reproduced with permission by Biolin Scientific. b) Attana Cell 200 (QCM), reproducedwith permission by Attana.

Since then, QCM has been applied to a range of different bioapplications,including the analysis of cells and of receptor-ligand interactions, and the detectionof pathogens and of biomarkers. QCM-based biosensing systems have beensuccessfully commercialized through desktop laboratory machines (Attana, Q-sense,QCMLab, Stanford Research Systems and International Crystal Manufacturing),designed primarily for industrial and academic research and development, seefigure 3.2.

AT-cut quartz crystal characteristics

In most QCM biosensing applications, AT-cut quartz crystals are used, seefigure 3.3. The quartz sensors are typically shaped into thin circular disks. Acircular electrode is located on each side of the disk. The two electrodes are ofdifferent size with the largest electrode facing the liquid, which is referred to as thetop side. When an electrical high-frequency signal is applied to the electrodes, aTSM oscillating motion is induced in the crystal at a resonance frequency, which istypically 5-30 MHz and depends on the thickness of the disk.

The oscillation amplitude is distributed with a Gaussian profile over thecrystal [34], see figure 3.4, and has a maximum in the central region, which by thatfeatures the region of highest sensitivity, whereas the oscillation is negligible towardsthe perimeter. The gradient of the shear-wave oscillation amplitude also induces anout-of-plane motion on AT-cut crystals [35–39] which must be considered for theintegration of quartz crystal resonators with microfluidic cartridges, as discussed insection 3.2.


Figure 3.3: Commercially available 5 MHz quartz crystal sensors for QCM-D applications.Reproduced with permission by Biolin Scientific.

Biosensing principle

When a particle binds to the surface of the QCM sensor and interacts with thesurface motion, the resonance frequency of the transduced signal is altered. Forthe binding of small nanoscale particles, such as nucleic acids, polypeptides orproteins, the addition of mass approximately corresponds to an increase of thethickness of the crystal and the resonance frequency decreases. The typical lowerlimit-of-detection (LoD) for QCMs in biosensing applications is approximately 5ng/ml [40]. The high sensitivity of resonating quartz crystals comes from theenormous acceleration, which is approximately 107g [41], that analytes that arerigidly bound to the transducer surface experience. If this thesis were to besubjected to this acceleration, you would have to lift approximately three hundredtons the next time you turn the page.

For the binding of large microscale particles, such as bacteria, the frequencyresponse from the sensor can be more ambiguous. For example, upon binding ofmicrobeads to a QCM sensor, the transduced signal can show an increase of theresonance frequency instead of a decrease [42]. This is due to viscoelastic energylosses of the bound receptor-bead complex on the surface of the sensor, which is

3.1. QUARTZ CRYSTAL MICROBALANCE 21

Osc

illat

ion

ampl

itude

mm0 11 22 33 44 55

max

min

Shear-wave oscillation amplitude over a 10 MHz AT-cut quartz crystal

x

y

Cross section

Particle displacement

Non-oscillating region

Figure 3.4: . The plot shows a measured oscillation amplitude profile over a 10 MHz AT-cut quartz crystal (data reproduced by permission from QCM Labs, Sweden). The illustrationindicates the oscillating and non-oscillating regions of the transducer surface. The cross sectionillustrates the shear wave lateral displacement (in the x-direction) of a rigidly bound particle uponan applied sinusoidal electric field over the thickness of the crystal (in the y-direction). Note, theshape of the amplitude profile depends on the design of the quartz crystal disk and the electrodes.

particularly prominent in liquid conditions. Therefore, QCM biosensing based onresonance frequency shift measurements is primarily suitable for the detection ofparticles below the micrometer scale.

The viscoelastic losses that can yield inaccurate results in traditional QCMsensing, can be utilized by measuring the change in quality factor (Q) of the sensor.QCM with dissipation monitoring (QCM-D) is a technique that simultaneouslyanalyzes the resonance frequency shifts and energy dissipation (D, inverse of Q)of the transduced signal as the oscillation decays upon termination of the appliedAC signal [43–45]. The technique enables biosensing of larger particles and softmaterials and has been commercialized (Q-sense, Biolin Scientific Holding AB,Sweden) and used extensively in biotechnology research [46].

Summary

QCM is a well proven and established technique successfully demonstrated bothin research and in commercial applications. The technique is highly sensitiveand in combination with, e.g. dissipation monitoring, is suitable for the real-time detection of a range of different analytes. However, most of the instrumentsthat have been developed as of today have been designed for laboratory work


and/or have been relatively large and complex. Much of the bulk and complexityin existing commercial instruments comes from advanced and automated liquidhandling systems. The QCM technique itself only requires small-scale electronicsand has great potential for integration in easy-to-use and compact instruments forPoC applications [29].

3.1.3 Rupture event scanningBiosensing principle

Rupture event scanning (REVS) is a quartz crystal resonator technique thatexclusively uses the viscoelastic energy losses on the transducer surface forsensing [47], which for example has been demonstrated for the detection of singlevirus particles [48]. In REVS, an AT-cut quartz crystal is driven close to itsresonance frequency by a sinusoidal oscillation at an increasing amplitude to causeunbinding by rupture of the bound particles from the sensor surface. When thedissociation happens, abrupt changes in the transduced electrical signal is measuredat the third harmonic (3f), i.e. at three times the resonance frequency.

Summary

It is argued that weaker interactions dissociate at lower oscillation amplitudesand hence, specifically bound particles can be distinguished from those that arespecifically bound. However, recent investigations imply that the signal from thebond rupture could originate from a cooperative process of unbinding, rather thana rupture, and experiments showed that the selectivity between specifically andunspecifically bound particles could be questioned [45, 49]. Additionally, sincethe onset of any unbinding event is mediated by thermal fluctuations, making thedissociation random in time, reproducibility is likely challenging with the REVStechnology. The method is also destructive, making repeated measurements of theanalyte impossible.

3.1.4 Anharmonic detection techniqueAnharmonic detection technique (ADT) is a newly developed measurement methodbased on quartz crystal resonators [49, 50]. It is similar to REVS in that it usesthe viscoelastic energy losses at increasing oscillation amplitudes for sensing. Incontrast to REVS, ADT relies on limited maximum oscillation amplitudes to avoidthe event of rupture of the surface bound particles, making ADT a non-destructivetechnique and enabling multiple consecutive amplitude scans.


When a measurement is performed, the viscoelastic energy losses of the boundreceptor-particle complex increase non-linearly as the oscillation amplitude is

3.1. ANHARMONIC DETECTION TECHNIQUE 23

increased. This non-linear response is transduced into an anharmonic signal bythe quartz crystal, resulting in a measurable electrical current at the third (orhigher) odd harmonic.

Large particles that are directly captured by receptors on the sensor surfacetypically provide stronger anharmonic signals compared to small particles due tothat the viscoelastic energy losses increase with the size and mass of the boundparticle [51]. The anharmonic signal can be further amplified by introducing a linkerfor the attachment of the receptor to the sensor surface. Upon binding of a particleto a receptor, the linker acts as a spring that is stretched during the oscillatingmotion of the crystal, thus increasing the viscoelastic losses on the sensor surface.Thus, for the direct binding and detection of particles, ADT is most suitable formicrometer scale particles rather than small molecules in the nanometer range andin contrast to QCM, the use of an elastic linker-receptor complex is advantageous.

ADT has been experimentally used for the direct detection of 6400 bacterialspores, which corresponds to a mass of 9.6 ng, visually confirmed using opticalmicroscopy [51]. Based on the measurements of the spores and assuming a detectionlimit corresponds to a signal-over-noise ration of 2, the reported lower limit-of-detection was 430 spores.

Discrimination of surface interactions

Interestingly, the profile of the 3f electrical current response upon amplitudescanning is different in ADT if the particle is strongly bound (chemisorbed), looselybound (physisorbed) or if the sensor surface is clean [51]. Thus, it is possible todistinguish between specific and unspecific surface interactions. In addition, theADT has the unique ability among biosensors to physically remove unwanted andloosely bound particles from the sensor surface. When the amplitude scanning isperformed, the maximum amplitude can be adjusted to provide enough energy tobreak the weak (van der Waals) bonds to the physisorbed particles while preservingthe bonds to the more strongly bound chemisorbed particles. The unspecificallybound particles can thereby be rinsed from the sensor surface, increasing theselectivity of the biotransducer in addition to the specificity provided by thereceptors.

Summary

ADT is a promising biotransducer technique but compared to the well proven andestablished QCM technique it is still at its infancy and needs further researchin terms of repeatability, quantifiability, etc. For example, it has still not beeninvestigated how the out-of-plane motion in AT-cut quartz crystals impacts on theADT response. In addition, as for all quartz crystal resonators, the signal responseis likely different depending on where on the crystal surface a particle binds due tothe difference in oscillation amplitude. This likely impacts on the reproducibilityof measurements of low concentration samples, where there is less averaging of


the surface density of bound particles. Nevertheless, in combination with tailor-made bioassays, it could potentially allow for the detection of particles in very lowconcentration samples.

The ADT technology has a promising potential as a biotransducer techniquefor PoC applications and field work due to four important factors: it features highsensitivity; it enables the discrimination between unspecific and specific surfaceinteractions; it allows for the direct detection of large particles, potentially wholevirus and bacteria, and; it potentially allows for miniaturization and integrationof the ADT instrument due to that the technique is based only on small-scaleelectronics.

3.1.5 Silicon photonic ring resonatorsResonating biotranducers can also be based on optical sensing. One example ofsuch technique is silicon photonic ring resonators [30, 52], consisting of a ringwaveguide in close proximity to a neighboring straight waveguide, which is coupledto an external light emitter and detector. Incoming light in the straight waveguidecouples to the ring resonator, in which the light resonates at effective wavelengthsthat fit an integer number of times in the ring.


In biosensing applications, the surface of the ring waveguide is coated withreceptors. When a particle binds, the refractive index changes locally in the range ofthe evanescent field of the light in the ring waveguide. The change of refractive indexintroduces measurable shifts of the resonance wavelengths. During the last decade,several photonic ring resonators have been successfully demonstrated in a range ofbiosensing applications, including the detection of proteins [53], cancer biomarkers[54] and virus [55]. The technique has recently also been commercialized (GenalyteInc., USA). The lower limit-of-detection is commonly reported as the smallestdetectable surface concentration of the analyte on the sensor surface, with state-of-the-art values of approximately 3 pg/mm2 [56].

Summary

Silicon photonic ring resonators are fabricated by standardized manufacturingprocesses available in the semiconductor industry. Their small size allows forthree important advantages compared to quartz crystal resonators: wafer-scalebatch fabrication, potentially lowering the cost per device; multiplexed biosensing,enabling the analysis of several different particles in a single measurement, and;higher absolute mass sensitivity, enabling the detection of low concentrationsamples.

The advantage of small chip size and wafer-scale batch fabrication is howeverhampered by the fact that the footprint of the microfluidic system required to

3.1. STATE-OF-THE-ART APPROACHES 25

provide the sample to the sensor is often much larger in size compare to the sensor[57]. In addition, photonic waveguides typically require off-chip light emittorsand detectors, increasing the overall size and making the integration and chip-to-instrument tolerances critical. The high sensitivity of Si photonic ring resonatorscompared to acoustic transducer techniques is particularly advantageous whendetecting small analytes, such as proteins, in low volumes, but less critical in thedetection of large analytes, such as virus and bacteria. Moreover, the LoD is limitedin practical biosensing applications due to unspecific binding, which can only bepartially circumvented by relying on surface chemistry and advanced bioassays.

3.1.6 State-of-the-art biosensor integration approachesQuartz crystal packaging

In industrial quartz crystal resonators, such as those used for timing applications,HC-49 crystal holders or similar models are used. The crystal is suspended bytwo metal wires that clamp to electrical contact pads on the edges of each sideof the crystal, where electrically conductive adhesive glue provides mechanicallystability. The crystal is protected from surrounding electrical disturbances by agrounded metal cap. However, such cartridges are not compatible with biosensingapplications.

Existing QCM biosensor cartridges, such as those from Attana, Q-sense,QCMLab, Stanford Research Systems and International Crystal Manufacturing,share a de facto standard design approach with three common denominators: firstly,the crystal is mounted in the cartridge by mechanical clamping; secondly, thecartridge consists of multiple parts, typically including O-rings for liquid sealingand screws for mounting, and; thirdly, the integration relies on manual assembly.

This type of integration and cartridge design allows for disassembly of thecartridge after measurements for thorough cleaning of the cartridge parts and forreplacing the crystal prior to new measurements. Therefore, such cartridges areespecially suitable for laboratory applications and are extensively used in academicresearch.

Electrical contacting is commonly made by: clamping a wire between the bottomo-ring and the contact pads of the quartz crystal; by clamping the bottom side of thequartz crystal and its contact pads towards contact pads an a bottom substrate,such as a PCB, or; by direct contacting to the quartz crystal via spring-loadedcontact pins in the cartridge holder. Specifically, all of these methods apply a forceto the quartz crystal, which risks affecting the performance of the crystal.

Packaging of silicon photonics

Photonic ring resonators developed in academic research are predominantly inte-grated with microfluidic PDMS cartridges by plasma bonding [57, 58] or clamping[59], whereas commercially available photonic ring resonator biosensors share a


similar clamping-based integration approach using gaskets as described above(Genalyte Inc., USA).

Packaging of disposable biosensor cartridges - an unresolved challenge

Whereas clamping of the sensor with a microfluidic cartridge that accommodatesgaskets, such as O-rings, is suitable for laboratory applications, it is less attractivefor disposable cartridges in fieldwork and PoC applications due to high componentand integration costs. The inability of all commonly used polymers to directly bondto other components for heterogeneous integration hinders the development andcommercialization of technically advanced polymeric microdevices [10]. Whereassensors, such as quartz crystals and silicon photonic ring resonators, are industriallymanufactured in large quantities at low cost and well-characterized surface biofunc-tionalization protocols are readily available, new ways to heterogeneously integratesensors with polymer cartridges are needed to realize high-performance biosensorinstruments with lower cost disposable cartridges for PoC and fieldwork.

In this thesis, the development of microfluidic biosensor cartridges has beenfocused on two fields of applications: laboratory use for the development ofbiosensing platforms, and; disposable use in end applications at PoC or fieldwork.To be able to integrate a biosensor with a microfluidic cartridge, it is first importantto understand existing design limitations and considerations.

3.2 Design considerations for microfluidic biosensorcartridges

3.2.1 Cartridges for quartz crystal resonators

Physical aspects

Quartz crystals are acoustic resonators with a physical motion, and therefore,special consideration is required for their physical integration with microfluidiccartridges. To combine quartz crystal biosensors with liquid samples, a liquid flowcell above the crystal is required. An open flow cell is not advised due to theundefined height of the liquid layer, which is discussed below. Hence, a closed flowcell with a defined height and a fluidic inlet/outlet is preferred.

As shown in figure 3.4, AT-cut quartz crystals have no, or very little, oscillationtowards the perimeter of the crystal. The non-oscillating area of the crystalis therefore suitable for mounting the crystal into a cartridge, for example byclamping. In this way, dampening of the crystal oscillation as a result of themounting is limited or even absent.

However, when strong clamping forces are applied to the crystal, even atits perimeter, the crystal risks being mechanically stressed which risks affectingthe oscillation performance. Therefore, the clamping force should not be more

3.2. DESIGN CONSIDERATIONS 27

than what is required to ensure leak tightness during normal fluidic operation.Alternatively, other mounting and liquid sealing approaches can be explored.

AT-cut quartz crystals are susceptible to interference from compressionalacoustic waves in liquid that are generated out-of-plane by the shear wave oscillationof the crystal [35–39]. To circumvent the interference of the crystal oscillation fromthe standing waves with wavelength λ generated in the liquid flow cell, optimal flowcell heights, h, must be chosen and are calculated by:

h = (2n+ 1)λ4 [m]

λ = cf [m],

where c is the speed of sound in liquid (1500 m/s at RT ) and f is the resonancefrequency of the quartz crystal. For 10 MHz crystals, which are used in the QCMapplications, the optimal flow cell heights are h = 37.5, 112.5, 187.5, . . . µm, whereasfor 14.3 MHz crystals, which are used in the ADT applications, the optimal flowcell heights are h = 26.2, 78.7, 131.1, . . . µm.

Alternatively, the shape of the flow cell ceiling can be non-parallel to the crystalsurface, for example dome-shaped, to distribute the reflected compressional acousticwaves towards the side of the flow cell rather then straight back towards theoscillating crystal surface.

All quartz crystal resonators are susceptible to temperature fluctuations butAT-cut crystals have, in contrast to many other crystal cuts, a cubic temperaturedependence, making them relatively frequency stable over a certain temperaturerange. However, even minor changes in temperature can have impact on highlysensitive measurements. To ensure that the crystal, reagents and buffers all havethe same temperature during measurements, a temperature controller integrated inthe cartridge holder of the instrument is advised, particularly in field applications.In laboratory settings, the temperature of all parts are more easily controlled andequalized. Therefore, we have not employed an active temperature controller formeasurements presented herein.

Electrical aspects

The electrical contacting of the quartz crystal is important. Both the contactresistance and the contacting force must be simultaneously low for optimal electricaland mechanical performance of the crystal, respectively.

Conventional electrical contacting methods in biosensor cartridges rely onapplying a mechanical force to the contact pads of the quartz crystal, with therisk of affecting the mechanical performance of the crystal. Other methods thatavoid direct contacting have been demonstrated, such as wireless and electrodelesscontacting in QCM applications [60], however, no direct comparison withconventional methods have been made.


Quartz crystal resonators are susceptible to electrical disturbances from nearbyobjects, which is described as the proximity effect [61]. Therefore, electricalshielding of the sensor cartridge is critical to avoid random changes of the resonancefrequency in real applications. Polymer sensor cartridges can not provide theelectrical shielding by them selves. Instead, the electrical shielding should beprovided by the cartridge holder in the QCM instrument.

Performance aspects

The quartz crystal cartridges that are presented in this thesis, have been fabricated,throughout the years, to provide proof-of-concept for a range of applications, asdetailed in 5. The performance quality requirements for these cartridges have inall cases been low drift and low noise. Thus, prior to any biosensing in-house or bypartners, the cartridges have been verified for noise level below 1 Hz and no driftwithin 10 minutes in liquid conditions. Other performance parameters have notbeen investigated because the scope of the work has, at any moment in time, beenon delivering functional components in the frame of the respective projects.

3.2.2 Cartridges for photonic ring resonatorsAspects of miniaturization

Silicon photonic ring resonators also require special design considerations fortheir integration with microfluidic cartridges, however, not so much in terms ofmechanical clamping.

In the silicon wafer-scale processing of miniaturized photonic ring resonators,a large number of chips per wafer can be fabricated. When such sensors arebatch-fabricated with integrated microfluidics, the significantly larger microfluidiccartridge determines the chip size of each photonic sensor on the wafer. To lowercost and utilize the large number of chips that are possible to produce per wafer,it is important that also the microfluidics, including fluidic connectors, can beminiaturized to a footprint that matches the photonics chips.

Due to the small size of the ring resonators, the microfluidic channels overthe waveguide rings need to be scaled down accordingly to be able to focus thesample to the sensor surface and not let it pass by on the sides of the ring.Consequently, the miniaturization leads to strict alignment tolerances of the sensorand the microfluidics, which need to be considered in the design and integration.

Optical aspects

The coupling of light to the sensor from external light emitters and detectorsrequires through-hole openings in the microfluidic cartridge. Grating couplerson the Si surface are typically in the micron size range whereas through-holesin the cartridge are significantly larger due to limitations in the used fabricationtechniques, such as punching holes in PDMS [57]. If the fabrication method would

3.3. NOVEL APPROACHES TO BIOSENSOR INTEGRATION 29

allow for a decrease in the hole size, less cartridge area but more accurate alignmentwould be required.

3.3 Novel approaches to biosensor integration

This section will present advances in biosensor integration with microfluidiccartridges using quartz crystal and silicon photonic biosensors. The main focusis on quartz crystal biosensor cartridges, which are overviewed in figure 3.5.

The biosensors have been integrated with microfluidic cartridges using threedifferent approaches: mechanical clamping, which is the conventional approach;adhesive bonding, and; direct covalent bonding. The cartridges are presented inchronological order within each integration approach.

Although these cartridges have been develop at different moments in time,within different research projects and for different applications, the overall aimhas been to reduce the design and integration complexity to address the objectiveof the thesis and enable the manufacturing of disposable biosensor cartridges.

All cartridges were connected to their corresponding instruments in customdesigned cartridge holders prior to use.

30 CHAPTER 3. ADVANCES IN BIOSENSOR INTEGRATIONcartridge

Mechanical clampingAdhesive bondingCovalent bondingReducued design and integration complexity

assembly

methods

fasten with

screws

fasten with

spring-loadedclips

press-�tfastening

by friction

verticallyconductive

foil

medical-grade

biologically inerttape

bonding ofthiol-gold

bonding ofepoxide-quartz

thiol-gold(epoxide-PCB)

quartzcrystal

10 MH

zQ

CM

10 MH

zQ

CM

10 MH

zQ

CM

10 MH

zQ

CM

10 MH

zQ

CM

14.3 MH

zA

DT

14.3 MH

zA

DT

�ow cell

height

~ 112 µm

~ 79 µm

~ 112 µm

~ 112 µm

~ 50 µm(not optim

al)

~ 80 µm(not optim

al)

~ 79 µm

fabricationm

ethods

milling

injectionm

olding,m

illing,cutting

milling

milling,

DRIE,

cutting

injection m

olding,cutting

reactioninjectionm

olding

casting

electricalcontacting

HC-49

industrialholder

PCB

externalspring-loaded

contact

verticallyconductive

foil


contact

HC-49

industrialholder


contact(alt. PCB)

#parts

121144433(alt. 4)

frequencynoise/drift

low

low

low

low

low

medium

high/low

intendeduse

lab work

lab work

lab work

�eldwork

lab work

�eldwork

(prototype)

lab work

(prototype)

�eldwork

key features and im

provements

conventional designreusable

easy manual handling

microscopy com

patiblereusable

rapid integrationfew

partsonly polym

ersm

icroscopy compatible

disposable/reusable

air-liquid interfacingno clam

ping forcesfew

partspartly reusable

injection molded parts

only UPS class VI m

aterialsno clam

ping forcesfew

partsdisposable

direct bondingno clam

ping forcesfew

parts(-) not enclcosed

disposable

strong direct bondinghydrophilic surfaces

bonds to other substratesbatch-m

anufacturing potentialno clam

ping forcesfew

partsdisposable

Figure 3.5: Overview of the QCM cartridges presented in this thesis.

3.3. NOVEL INTEGRATION APPROACHES 31

3.3.1 Mechanical clamping in thermoplastic cartridgesThree different QCM and ADT cartridges with sensor integration based onmechanical clamping were designed and fabricated for use in the development ofnovel biosensing platforms. Whereas the first two cartridges are aimed solely forlaboratory work, the third type of cartridge is designed also with consideration todisposable use, hence it features a less complicated design and integration approachcompared to the first two cartridges.

First, the design and integration method of each cartridge are introduced andthereafter a comparison between them is made.

Cartridge assembly with screws

The first type of cartridge with sensor integration based on mechanical clamping wasused for QCM sensing of norovirus in application ??, see chapter 5. The cartridgewas clamped by screws and consisted of twelve parts: a 10 MHz quartz crystaland a HC-49 crystal holder (International Crystal Manufacturing Inc., USA), twoUPS class VI EPDM O-rings (Trelleborg, Sweden), two machined microfluidicthermoplastic parts and six screws, see figure 3.6. The electrical connection tothe quartz crystal was made through the HC-49 crystal holder. The height of theliquid flow cell was approximately 112 microns.

quartz crystal

bottom o-ring

top o-ring

electrical connector

screw

top plastic part

bottom plastic part

a) b) �uidic port

threaded screw hole

�ow cell

1 cm

Figure 3.6: A QCM biosensor cartridge based on mechanical clamping using screws.

Cartridge assembly with clips

The second cartridge with sensor integration based on mechanical clamping wasused for ADT sensing of influenza and norovirus in application 5.4 and 5.3, seechapter 5. The cartridge was clamped by spring-loaded clips and consisted ofseven parts: a 14.3 MHz quartz crystal (LapTech Inc., Canada), an EPDMO-ring (Trelleborg, Sweden), two commercial microfluidic thermoplastic parts(Microfluidic ChipShop Gmbh, Germany), a double-sided medical-grade adhesivetape (Adhesive Research, Ireland), a thermoplastic spacer and a gold-coated and


patterned printed circuit board (PCB), see figure 3.6. The electrical connection wasmade by direct contacting of the gold electrodes on the quartz crystal to contactpads on the PCB. The height of the liquid flow cell was approximately 79 microns.

quartz crystal

o-ring

PCB

spacer

top plastic part

a) b)

adhesive

bottom plastic part

�uidic port �ow cell

1 cm

Figure 3.7: An ADT biosensor cartridge based on mechanical clamping using clips (not shownin the figure).

Press-fit cartridge assembly

The third cartridge with sensor integration based on mechanical clamping was usedfor ADT sensing of influenza and norovirus in application 5.4 and 5.3, see chapter 5.The cartridge was clamped with a press-fit assembly provided by two specific pillar-hole structures and consisted of four parts: a 14.3 MHz quartz crystal (LapTechInc., Canada), an EPDM O-ring (Trelleborg, Sweden), two machined microfluidicthermoplastic parts, see figure 3.6. The electrical connection was made via spring-loaded contact pins in the cartridge holder to the gold electrodes on the quartzcrystal. The height of the liquid flow cell was approximately 79 microns.

a) b)

o-ring

quartz crystal

bottom plastic part

top plastic part

channel

electrical port

press-�t pillar

press-�t hole

�ow cell

1 cm

Figure 3.8: Photograph of an ADT biosensor cartridge based on mechanical clamping using apress-fit assembly.


Comparison of mechanically clamped cartridges

The first cartridge, which is assembled with screws, is based on the conventionalintegration approach used in many of the commercial devices. In addition, it usesan industrial HC-49 crystal holder for electrical contacting of the crystal. Thus,its design and integration approach relies on well-proven methods and enablesdisassembly and reuse of all parts, making it suitable for laboratory work.

The second cartridge is also aimed for laboratory work but enables easier manualhandling compared to the first cartridge due to that is clamped by spring-loadedclips. The main technical advancement that it brings is its compatibility withmicroscopy analysis of the full sensor surface, which is enabled by moving the fluidconnections to the sides of the cartridge.

The third cartridge is aimed for laboratory work but with consideration to therequirements on disposable cartridges, making it conceptually different from thefirst two cartridges. It features a novel and rapid press-fit type of assembly, whichcan be adapted by its tightness to allow for single use or reuse, and consists of aminimal amount of parts required for mechanical clamping.

In the first cartridge, the quartz crystal is clamped between two O-rings, whereasin the second and the third cartridge the quartz crystal is clamped by an O-ring ontop but against a stiff flat surface at the bottom. Potentially, using O-rings both atthe top and bottom provides a softer contact with less risk of inducing mechanicalstress in the crystal but having a stiff bottom surface below the crystal helps tomore accurately define the flow cell height in the clamped state.

The electrical contacting was made using three different approaches. However,in the third cartridge the electrical contacting to the quartz crystal was movedoff-chip to the crystal holder. This allowed the third cartridge to only consist ofpolymer parts in addition to the quartz crystal. Moreover, the third cartridgewas assembled with a press-fit assembly, further reducing the number of parts andsimplifying the integration.

Thus, for mechanical clamping of the quartz crystal in a microfluidic cartridge,the press-fitted cartridge provided the most promising approach for a disposablebiosensor cartridge. However, biosensor cartridges based on mechanical clampingstill require multiple parts that are integrated by a serial pick-and-place assembly,inevitably limiting the reduction of manufacturing costs.

3.3.2 Adhesive bonding using foils and tapesThe number of parts in a biosensor cartridge can be reduced and the integrationsimplified by changing from double-sided to single-sided mounting of the sensor.This can be realized by using adhesive bonding, where the top side of the sensoris directly attached to the cartridge via the adhesive. The adhesive provides boththe attachment of the sensor and sealing of the liquid flow cell. Specifically, thebottom side of the quarts crystal resonator is not in contact with other parts andcan therefore oscillate without mechanical interference. No forces are applied to


the crystal in such configuration so the risk of introducing mechanical stress issignificantly reduced compared to mechanical clamping.

Below, single-sided bonding of quartz crystals with microfluidic cartridges usingtwo different types of adhesives is described. One cartridge is for liquid-basedsamples and the other is for airborne samples and includes an air-liquid interface atthe top of the flow cell. The cartridges are first presented and thereafter compared.

Assembly with a vertically conductive adhesive foil

Single-sided bonding of quartz crystals with microfluidic cartridges was demon-strated for the QCM detection of narcotic substances, as described in applica-tion 5.2, see chapter 5 and paper 3. The cartridge consisted of four parts: a 10MHz quartz crystal (Biosensor Applications AB, Sweden); a vertically conductivedouble-sided adhesive foil (VCAF) (3M, USA); a machined thermoplastic part atthe bottom and; a microfluidic silicon chip at the top, see figure 3.9. The parts werebonded by the adhesive foil, which also provided liquid confinement and electricalcontact between the electrodes on the quartz crystal and gold contact pads on thesilicon chip. Spring-loaded contact pins provided the contact between silicon chipand the QCM instrument. The silicon chip included a perforated diaphragm forair-liquid interfacing and a microfluidic system, as described in section 4.2. Theliquid flow cell height was defined by the adhesive foil which had a thickness of 50microns.

a) b)

conductive adhesive

silicon chip

quartz crystal

bottom plastic part

�uidic port electrical port

channel diaphragm�ow cell

1 cm

Figure 3.9: Photograph of QCM biosensor cartridges assembled with vertically conductiveadhesive tape.

Assembly with a medical-grade adhesive tape

Prototypes of another approach using single-sided bonding of quartz crystals withmicrofluidic cartridges were demonstrated within the frame of the Rapp-Id project.The prototypes consisted of four parts: a 10 MHz quartz crystal (QCMLab,Sweden); two commercial microfluidic thermoplastic parts (Microfluidic ChipShopGmbH., Germany) and; a medical-grade double-sided adhesive tape (Adhesive


Research Inc., Ireland), see figure 3.10. The cartridge was bonded by the adhesivetape, which has been developed for diagnostic applications and is claimed by themanufacturer to be inert to biological samples. The electrical contacting was donedirectly to the electrodes on the quartz crystal via spring-loaded contact pins inthe cartridge holder, similarly as in the press-fitted cartridge above. In the currentdesign, the cartridge prototypes did not include an air-liquid interface but wereintended for liquid samples. The liquid flow cell height was defined by the adhesivefoil which had a thickness of 80 microns.

a) b)

adhesive

quartz crystal

bottom plastic part

�uidic port electrical port

channel

top plastic part

�ow cell

1 cm

Figure 3.10: Photograph of QCM biosensor cartridges assembled with medical-grade adhesivetape.

Comparison of adhesive bonded cartridges

The first cartridge based on adhesive bonding brings two significant technicaladvancements. The first advancement is the use of the multifunctional verticallyconductive adhesive foil, which enables sensor integration, liquid confinement andelectrical contacting. The second advancement is the integration of a perforateddiaphragm, which enables air-liquid interfacing in close proximity to the biosensorin the microfluidic cartridge. The cartridge is primarily intended for single-use, butin laboratory work the cartridge parts can be reused by submerging the cartridgein isopropanol, which makes the adhesive foil swell and the parts disassembled.

The second cartridge using adhesive bonding is a prototype based on theintegration of a quartz crystal resonator with only commercial products: injectionmolded thermoplastic parts, and a double-sided adhesive tape, both of which aremade from medically certified materials. The cartridge has not been used forany specific biosensing measurement but demonstrates the potential of integratingquartz crystal resonators in commercial microfluidic products.

Using adhesives in microfluidic biosensing applications introduces several poten-tial risks. Firstly, the sample can become absorbed by the adhesive and the adhesivecan cause contamination due to leaching into the sample liquid. Secondly, the long-term mechanical stability of adhesive tapes in microfluidic biosensing applicationsis unknown. Thirdly, using adhesive tape makes it difficult to consider the optimal


flow cell height for AT-cut quartz crystals in the design due to that only certaintape thicknesses are available. To further reduce the number of cartridge parts,to simplify the sensor integration and to avoid the above mention issues, directbonding of the sensor to the cartridge was investigated.

3.3.3 Direct covalent bonding with OSTE cartridges

Thiol-functionalized molecules, such as polyethyleneglycols (PEGs), are commonlyused to form monolayers on gold surfaces, e.g. as linkers in immunoassays [62].The thiol groups spontaneously form covalent-like bonds to the gold atoms, whichare stronger than gold-gold bonds [63]. The thiol-gold complexes allow for surfacerearrangements of the bound molecules [64], thus enabling efficient formation ofdense monolayers.

Our hypothesis was that quartz crystal sensors could be directly integrated withOSTE cartridges by covalent-like bond formation between the top gold electrodeson quartz crystals sensors surfaces and the thiol functional surface groups on OSTEcartridges. In addition, reorientation of the surface-mobile thiol-gold complex onthe gold surface potentially enables the relaxation of bound polymer chains into anon-strained position during bonding, generating minimal amounts of stress at thebond interface. This hypothesis was tested by the fabrication of novel OSTE QCMcartridges.

OSTE thiol-gold bonding to quartz crystals

An OSTE QCM cartridge based on direct covalent bonding of the quartz crystal forthe integration with a microfluidic cartridge was demonstrated for QCM sensing inpaper 4. The cartridge consisted of three parts: a 10 MHz quartz crystal and a HC-49 crystal holder (International Crystal Manufacturing Inc., USA) and; an OSTEmicrofluidic part, see figure 3.11. The OSTE part was casted and directly bondedto the quartz crystal via the formation of thiol-gold bonds on the outer perimeterof the top electrode of the crystal. The electrical connection to the quartz crystalwas made through the HC-49 crystal holder. The height of the liquid flow cell wasapproximately 112 microns.

The OSTE QCM cartridge provided a void-free and leak-free bond interfacethat could withstand differential pressures of up to 2 bars. A performance analysis,reported in paper 4, showed a resonance frequency signal with no detectabledrift while monitoring the signal for ten minutes. This result indicates that thehypothesis of that the single-sided bonding featured low stress at the bond interfaceis plausible, since stress in the crystal originating from the mounting typicallygenerates drift in the resonance frequency signal. The noise level was approximately+- 4 Hz, which was undesirably high but expected from the non-encapsulatedand exposed crystal making it susceptible to disturbances from the surroundingenvironment.


quartz crystal

electrical connector

OSTE part

�uidic port �ow cell

1 cm

Figure 3.11: Photograph of a QCM OSTE cartridge assembled by direct covalent bonding.

To form a complete OSTE cartridge that encapsulates the crystal, an additionalOSTE part with an excess of allyl functional groups is needed. The additionalOSTE part would constitute the bottom part of the complete cartridge, whereasthe existing OSTE part, which accommodates the crystal, would constitute thetop part. The allyl groups of the additional OSTE part would react with the thiolgroups of the existing OSTE cartridge to form a bond, thereby enclosing the crystal.

The thiol-gold bonding between the OSTE cartridge and the crystal setslimitations to the geometrical design of the gold electrodes on the quartz crystal, i.e.the gold electrode must extend towards the non-oscillating perimeter of the crystalto be available for bonding. However, the further the electrode is extended towardsthe perimeter, the broader the oscillating region of the crystal will be. Thus, thereis a conflict in the design requirements, making the integration of quartz crystalsto OSTE cartridges less favorable.

OSTE thiol-isocyanate bonding to silicon photonic chips

Thiols have the ability to bond to other materials than gold. One exampleis isocyanate, which is commonly used as a linker for surface attachment ofproteins [65]. Previous work has demonstrated ”click” wafer bonding of OSTE tosilicon wafers by first coating the silicon surface with isocyanate [66]. By combiningthe fabrication technology for making microfluidic OSTE cartridges using castingwith the click bonding approach, we explored the possibility to directly integratesilicon photonic ring resonators with microfluidic OSTE cartridges, described inpaper 5.

Casting of OSTE is limited to single-sided replication of structures in the moldand does not allow for the fabrication of open through-holes in the OSTE partneeded for fluidic and optical interconnects. In order to circumvent these issues weadapted the casting approach to include double-sided replication of structures inthe mold and to be combined with the photolithographic capabilities of OSTEpolymers [67, 68]. This allowed the replication of relatively large fluidic tubeconnectors in the bottom side of the mold and the replication of microfluidic


channels in the top side of the mold. The open through holes were createdby photolithography that was done in the same step at the UV-light initiatedpolymerization during molding. Thus, the novel combination of double-sidedreplication during casting and photolithography promoted a straightforward single-step fabrication of the microfluidic OSTE part.

The molded OSTE cartridge was subsequently bonded with the silicon photonicchip at an elevated temperature for 10 min, at which the silicon-isocyanate surfacehad time to covalently bond with the thiol surface of the OSTE polymer. Theresulting silicon photonic sensor cartridge is shown in figure 3.12.

b)

silicon photonic chip

OSTE part

�uidic port channel

a)

optical port

ring wavegudieoptical grating

1 mm

Figure 3.12: Photograph of a silicon photonic OSTE cartridge assembled by direct covalentbonding.

The cartridge included a 100 micron channel over the ring resonator, separatedby only 75 micron wide bond areas from the through-holes used for optical probing.The tight spacing allowed for a grating coupler spacing of only 500 micron, thusreducing the wafer footprint of the silicon photonics compared to what is requiredfor PDMS cartridges [69]. Measurements of ethanol and methanol mixed inDI water yielded a volume refractive index sensitivity of 50.5 nm/RIU, in goodagreement with previously reported similar devices [70].

3.3.4 Direct covalent bonding with OSTE+ cartridgesPartially polymerized OSTE+ cartridges fabricated by OSTE+RIM provide epox-ide functional groups on its surface, in addition to thiol groups. The additionalsurface functionality gives OSTE+ bonding properties similar to epoxy glue.OSTE+ has been demonstrated to bond to a range of materials [71], including itselfand directly to silicon dioxide, i.e. quartz. Hence, we investigated the possibility touse OSTE+ to form a complete cartridge that integrates and encapsulates a quartzcrystal sensor by direct dry-bonding, independent of the gold electrode design. Thepreliminary results are presented below.

The cartridge consisted of three parts: a 10 MHz quartz crystal (QCMLab,Sweden); an OSTE+ microfluidic part and; a bottom substrate, see figure 3.13. The


bottom substrate was either another OSTE+ part or a PCB (QCMLab, Sweden),as seen in the figure.

quartz crystal

OSTE+ part

�uidic port

PCBconductive adhesive

a) b)

�ow cell

1 cm

Figure 3.13: Photograph of a QCM OSTE+ cartridge assembled by direct covalent bonding.

The OSTE+ part was fabricated using OSTE+RIM followed by direct bondingwith the quartz crystal and the bottom substrate in a single step. The quartzcrystal was only bonded to the top OSTE+ part and was separated from the bottomsubstrate. Prior to bonding, the PCB was prepared by depositing droplets of EPO-TEK conductive epoxy glue (Epoxy Technology Inc., USA) on its contact pads.Upon assembly, the conductive glue came in contact with the electrodes of thequartz crystal, thereby providing electrical contact. The bottom OSTE+ partaccommodated through-holes for direct electrical contacting of the quartz crystalby spring-loaded contact pins in the cartridge holder. The top OSTE+ part, whichwas bonded to the top surface of the quartz crystal, included flow cell with anapproximate height of 112 microns and fluidic inlet/outlet.

A preliminary analysis of the cartridge demonstrated leak-tight liquid operationand no frequency drift during a 10 minutes test period. The noise level wasapproximately 0.2 Hz, significantly lower compared to the OSTE QCM cartridge.By the use of the epoxy functional groups in the OSTE+ material, the OSTE+part can bond to both the gold and the quartz surfaces of the crystal, thus, noconsideration is needed with regard to the electrode design. The OSTE+ cartridgealso gives freedom in selecting an appropriate bottom substrate, such as anotherOSTE+ or a PCB.


3.4 Discussion

In this chapter, novel materials and integration approaches have been demonstratedfor the integration of quartz crystal and silicon photonic resonators, which arecommonly used transducers for biosensing applications.

Double-sided mechanical clamping of the biosensor is the conventional integra-tion method, used both in industry and academic research. It is not dependent onthe chemical properties of the cartridge materials, provides mechanical stability andallows for disassembly of the cartridge. However, mechanically clamped cartridgesare best suited for laboratory applications. For disposable use in PoC and fieldworkapplications, such cartridges introduces unnecessary high costs due to a relativelyhigh number of cartridge materials, often including metals, and a serial pick-and-place type of assembly. However, it is potentially possible to adapt mechanicalclamping for disposable use by using a press-fit assembly, which simplifies theintegration and allows for a reduction of the number of cartridge parts.

An alternative to double-sided clamping of the sensor with the cartridge is single-sided bonding, either by the use of adhesives or directly by covalent reactions.The use of adhesives does not allow for a reduction of cartridge parts comparedto the clamped and press-fitted cartridge, but by using single-sided mounting iteliminates the need for clamping forces which potentially increases the performanceof quartz crystal resonators. It does however require and additional and complicatedalignment step when applying the adhesive, which limits its potential use indisposable cartridges.

Newly developed materials and fabrication techniques, such as OSTE+RIM,open up possibilities for a completely new type of integration approach, namelydirect bonding of the sensor with the cartridge. As demonstrated in the OSTE+QCM cartridge, the number of cartridge parts was reduced to a minimum and thedirect integration of the sensor with the cartridge simplifies the integration processto a single step. The OSTE+ parts were molded using an industrial-like reactioninjection molding process, which potentially allows for batch manufacturing ofcartridge parts. Thus, OSTE+ biosensor cartridges are potentially very promisingfor disposable use in PoC or fieldwork applications but further analysis of the sensorperformance is still required.

Chapter 4

Direct Airborne Particle Samplingto Microfluidic Cartridges

This chapter introduces novel methods for sampling of airborne particles tomicrofluidic biosensor cartridges, aiming to enable the use of existing highperformance LoC and biosensor devices for the direct detection of airborne particles.

First, an introduction is given to corona discharge and electrostatic precipitationof airborne particles in section 4.1. Also, air-liquid interfacing to microfluidicdevices is introduced. Thereafter, a novel component for robust air-liquidinterfacing in biosensor cartridges is demonstrated in section 4.2. A method ofelectrostatic precipitation directly to microfluidic cartridges is demonstrated insection 4.3. Two different applications are presented, sampling of breath-basedaerosols in section 4.3.1 and sampling of aerosols and particles in ambient airin section 4.3.2 . Lastly, a novel method for up-concentrating captured airbornesample in liquid is demonstrated in section 4.4.

4.1 Introduction

Airborne particle sampling is of interest for a range of different applications, suchas air monitoring and detection of pathogenic particles or other harmful substancesin the field but also applications in healthcare, such as breath-based diagnostics atPoC.

Conventional methods for air sampling and immediate biosensing are notcompatible. This issue is addressed in the work of this thesis by the developmentof an electrostatic precipitator integrated with microfluidic cartridges. First, abackground to relevant state-of-the-art technologies is provided.

41

42 CHAPTER 4. AIRBORNE PARTICLE SAMPLING

4.1.1 Electrostatic precipitation

Electrostatic precipitation (ESP) is based on that charged airborne particles aretransported by electrohydrodynamic (EHD) flow and captured to a collectorelectrode. The charged airborne particles are transported by three mechanisms:diffusion; migration; and convection.

Whereas today, ESP is predominantly used for filtering particulate matter,e.g. in air purifiers, the technology has also been investigated for air samplingapplications. Already in 1955, Kraemer et al. [72] explained, theoretically, thecharging and capture of aerosol droplets in the presence of electrostatic fields andrecently the technology has been applied to the sampling of bioaerosols [73, 74].ESP has also been studied as a potential technology for the capture of breathaerosol [75, 76]. However, so far, no ESP device has provided a method that allowsfor direct detection of the captured particles using a downstream biosensor.

4.1.2 Corona discharge

To utilize ESP for air sampling, the airborne particles must be charged before beingsubjected to an electric field. In this thesis, the corona discharge needles were choseto both provide charging of the particles and function as one of the electrodesgenerating the the electrostatic field. The main advantages of this technique is thatit is based on only small-scale electronics and integrates well with air sampling tomicrofluidic devices, as demonstrated below.

When corona discharge is used in ESPs, a strong electric field is generatedbetween a high-curvature discharge electrode and a low-curvature collector elec-trode, typically in a point-to-plane or wire-to-plane configuration [77]. The electricfield is generated by setting the collector electrode at electrical ground and byapplying a high voltage to the discharge electrode, which induces the coronadischarge. The corona causes ionization of the surrounding air molecules closeto the discharge electrode. The ionized particles accelerate in the electric fieldtowards the collector electrode. The polarity of the applied voltage determineswether a negative or positive corona is formed, generating either negatively orpositively charged particles, respectively.

Surrounding airborne particles are charged either by field charging or diffusioncharging. In the former case, ions generated by the corona discharge activelycollide with the airborne particles when accelerating in the electric field whereas inthe latter case, ions randomly collide with the airborne particles due to Brownianmotion.

Some investigations have shown that airborne pathogens are inactivated byionization through corona discharge [78], whereas other reports show the viabilityof certain pathogens after capture [79], which allows e.g. subsequent culturing ofbacteria. The impact of the ionization on the bioaerosol particles is much affectedby the field strength, humidity, air composition, etc. How the ionization affects

4.2. PERFORATED DIAPHRAGMS FOR AIR-LIQUID INTERFACING 43

the binding of captured pathogens in subsequent biosensing using immunoassays isunknown and needs further investigation.

4.1.3 Air-liquid interfacingCapture and direct detection of airborne particles in bioaerosols have beendemonstrated with quartz crystal resonators in open air [80–82]. However, it hasalso been questioned wether or not the receptors on the biosensor surface are ableto function properly in a dehydrated state in air [83]. To address bioreceptordenaturation, it has been suggested to use hydrogels on the biosensor surface tomaintain the activity of the antibodies in air [84]. However, biosensing based onimmunoassays is typically very sensitive to contamination and unspecific bindingand none of the above methods allow for rinsing the surface prior to measurements.

In contrast to previous research on direct biosensing of airborne particles, thisthesis focuses on the capture of airborne particles directly to the liquid environmentof microfluidic cartridges to ensure the performance and longevity of the receptorson the surface of label-free biosensors, and to allow for state-of-the-art liquid basedbioassays.

To capture airborne particles directly to a microfluidic cartridge, an air-liquidinterface on the cartridge is needed. An air-liquid interface in a macroscopic well,provides large capture interface area but does not provide robustness to floodingagainst well flooding or drying caused by gravity or pressure variations, e.g., manualhandling or liquid actuation [85].

A microscale air-liquid interface provides better robustness to flooding andcollapsing by featuring a low Bond number, i.e. that the surface tension overcomesgravitational forces and pressure variations in the liquid. Air-liquid interfacesthat are fixated by surface tension in microfluidic cartridges can be provided bymicrostructured surfaces, e.g. pillar forests. Such interface has been demonstratedfor the detection of airborne explosives [86] but suffers from long fluidic transportpaths of the sample from the air-liquid interface to the downstream sensor.

4.2 Perforated diaphragms for air-liquid interfacing

In this work, a perforated silicon diaphragm was designed and fabricated for thepurpose of air-liquid interfacing, see paper 3 and figure 4.1. The diaphragm wasfabricated in a microfluidic chip by photolithography and deep reactive ion etching(DRIE). The diaphragm has a diameter of 5 mm and was designed in differentvariations, including hexagonal-shaped perforations with an apothem of 12.5,22.5 or 40 microns, respectively. A smaller apothems provide better mechanicalrobustness whereas a larger apothems provide a higher fill-factor of the air-liquid,i.e. air-liquid over air-solid areal ratio. The thickness of the diaphragm, hence alsothe depth of the perforations, was 20 microns. In addition to the diaphragm, thesilicon chip also included microfluidic channels for the transport of liquid to, andfrom, the air-liquid interface.


a) Si chip with a perforated diaphragm and micr�uidic channels b) 25 µm pores

b) 45 µm pores

b) 80 µm pores

perforated diaphragm

micro�uidic channels

Figure 4.1: SEM images showing c) an overview of a Si chip with microfluidic channels and aperforated diaphragm a pore size diameter of b) 25 mu, c) 45 mu, and d) 80 mu.

The silicon chip was integrated in a QCM cartridge, see section 3.3.2, and wasused for development of a novel electrostatic air sampling system, see section 4.3.1,and QCM sensing of airborne narcotics in application 5.2.

A perforated diaphragm for air-liquid interfacing allows for the integration witha sensor in close proximity. In the cartridge that is demonstrated in section 3.3.2,the distance between the diaphragm and the sensor surface is only 50 microns.The short distance from the air-liquid interface to the sensor surface featuresthree important advantages. Firstly, the airborne particles that are absorbed atthe air-liquid interface have a very short distance to diffuse before reaching thesensor surface, enabling rapid sensor response. Secondly, during their diffusivetransportation the particles do not encounter other solid surfaces in the fluidicsystem, other than the diaphragm, thereby limiting the risk of sample losses dueto unspecific binding. Lastly, since the particles are not convectively transportedin long microchannels and tubing, the risk of sample dispersion greatly reduces.

Evaporation at air-liquid interfaces

One major challenge of air-liquid interfaces in microfluidic cartridges is evaporation.A large air-liquid interfacial area is advantageous for improved capture of airborneparticles but suffers from liquid loss due to evaporation. Evaporation causesincreased salinity of the capture liquid and the microfluidic cartridge risks dryingout.

The effects of evaporation can be compensated for by a continuous refilling ofthe liquid to the air-liquid interface. This is possible when using a silicon perforated

4.3. AIR SAMPLING TO MICROFLUIDIC CARTRIDGES 45

diaphragm due to that it provides enough mechanical rigidity not to severely deflector break during operation and due to that its hydrophilic surfaces provide strongenough surface tension in water to avoid a collapse of the air-liquid interface.

Each pore functions as a Laplace valve, in which the surface tension keeps theliquid in a fixed position over a certain liquid pressure range. As described in paper 3, large negative or positive pressure loads can break the valve and subsequentlycause air introduction or flooding, respectively. However, the perforated silicondiaphragms presented herein, are able to withstand typical microfluidic flow ratesrequired to compensate for evaporation at the air-liquid interface.

4.3 Air sampling to microfluidic cartridges

Above, the challenge of air-liquid interfacing in microfluidic devices was addressand in this section, a solution for compatible airborne particle sampling directly tothe air-liquid interface of a microfludic cartridge is demonstrated.

We use a novel electrode configuration in an ESP sampler where the liquid ofa microfluidic cartridge is the collector electrode. The liquid collector electrodeis a key feature that enables tree important functions that occurs simultaneously.Firstly, the airborne sample is captured directly to the liquid of a microfluidiccartridge, thus eliminating the need for subsequent sample-to-liquid transfer.Secondly, the sample is subjected to a tremendous up-concentration as it is collectedfrom liters of air into a microlitre liquid environment. Lastly, the sample can beimmediately detected by adjacent or downstream liquid-based analysis, such asintegrated biosensing.

In order to use the liquid of a microfluidic cartridge as the collector electrode,the liquid must be electrically conductive. Phosphate buffered saline, which is acommonly used buffer in biological applications and has a NaCl concentration of 0.9percent, was successfully tested to provide enough electrical conductance to workefficiently as a collector electrode in the corona discharge setup.

The work focused on two different applications: the sampling of ambient airfor the detection, monitoring, and public surveillance of airborne particles, seeapplication 5.3, and; the sampling of a patient’s breath for non-invasive PoCdiagnostics and companion diagnostics for drug development, see application 5.4.

Ambient air sampling and breath sampling have different requirements in termsof particle size, humidity, volumes, etc. Therefore, two different ESP devices weredeveloped and tested but similarly, both devices were based on a negative coronadischarge system.

4.3.1 Sampling of airborne particles in ambient airA proof-of-concept ESP device was developed and tested for the sampling ofairborne particles in ambient air, see illustration in figure 4.2 and paper 6. Thedevice included a -16 kV corona discharge needle and a microfluidic cartridge as


DC -16kV GND

Ionization

liquid electrode0.9% NaCl(aq)

micro�uidic cartridge witha Si perforated diaphragm

o-

o-

Transport Capture

corona dischargeelectrode

liquidambient air

electrid �eld collector - air-liquid interface

Absorption

o-

o-

ambient air

o-o-

o-

o-

o-

o-

o-

o-

o-

o-

o-

o-

o-

o-o-o-

o- o-

o-

o-

o-

o-

o-

Figure 4.2: Schematic illustration of ESP capture of airborne particles in ambient air to amicrofluidic device.

collector, vertically aligned and horizontally separated in open air by 15 cm. Themicrofluidic cartridge consisted of a thermoplastic part and a silicon chip witha perforated diaphragm. The thermoplastic part accommodated 150 muL of 0.9percent NaCl in DI water, which was electrically grounded to generate an electricfield between the liquid and the corona discharge needle. The silicon chip wasbonded to the thermoplastic part by double-sided adhesive tape.

The ESP device was experimentally tested by the capture of airborne smokeparticles in ambient air. Two types of smoke was used: ammonium chloride smokeused for measurements and stearic acid smoke used for visualization. Both typescontained airborne smoke particles in the size range om 0.1-1 micron (PM2.5). Thecapture of stearic acid smoke is shown in figure 4.3 a and b. The photographsvisualize how the smoke is accelerated towards the air-liquid interface of themicrofluidic cartridge when the ESP device is active by the application of a highvoltage to the corona discharge needle.

pH-measurements of the liquid in the microfluidic collector cartridge aftercapture of airborne ammonium chloride smoke particles showed significant shiftsin acidity, even after only 30 s of sampling. These particles absorb into the aqueousliquid as shown in figure 4.3 d, whereas the larger and hydrophobic smoke particlesof stearic acid only adsorb to the surface of the air-liquid interface without absorbinginto the liquid, as shown in figure 4.3 e. Thus, ESP sampling of airborne particlesto microfluidic devices is limited to water dissolvable particles, which is similar toany type of water-based analysis.


d) NH4Clc) No particles e) CH3(CH2)16COOH

a) Corona discharge OFF b) Corona discharge ON

smoke

air-liquidinterface

coronaneedle EHD wind

clean Si perforated diaphragm hydrophilic particles absorb hydrophobic particles adsorb

smoke particlesaccelerate towards air-liquid interface

micro�uidiccollector

Figure 4.3: ESP capture of airborne smoke particles in ambient air to a microfluidic device.a) the smoke is moving straight up when ESP is off. b) the smoke is accelerated towards theair-liquid interface of the microfluidic cartridge when the ESP is on.

4.3.2 Sampling of aerosols in breath

Guiding airborne particle flow in breathing tubes

Another part of the ESP development was focused on the capture of exhaledair particles used for breath-based diagnostics in application 5.4. In contrast tothe sampling of airborne particles in ambient air, breath aerosols are preferablycaptured in closed systems in order to avoid dilution of the analyte with ambientair. The ESP presented above was therefore adapted for a preliminary investigationof capture of airborne particles in a breathing pipe/tube used as part of the medicalbreathing and respiratory equipment in clinical settings.

A plastic y-junction pipe substituted a medical breathing pipe/tube during theexperimental testing, see figure 4.4. A stearic acid smoke source was positioned atthe bottom opening of the pipe and the -16 kV corona discharge needle was insertedin a small hole in the side of the pipe in its bottom region. An electrically groundedmicrofluidic collector cartridge was positioned outside one of the two top openingsof the pipe. When the ESP sampler was off, the smoke stream rised through thepipe and split in two at the y-junction, generating two equally sized smoke streamsthat each exited through one of the two top openings of the pipe, see figure 4.4 a.

When the ESP sampler was turned on, the smoke particles were ionized and thestream was accelerated and directed entirely towards the top opening where thecollector was positioned, see figure 4.4 b. No traces of smoke were visible at the


a) Corona discharge OFF b) Corona discharge ON

smoke

coronaneedle

collector

smoke particlesaccelerate towards the collector

smoke

smoke particlesexit both ends of the tube

Figure 4.4: ESP capture of airborne smoke particles transported through a plastic y-shapedpipe. a) the smoke comes out from both outlets when the ESP is off. b) the smoke is acceleratedtowards the collector at one of outlets when the ESP is on.

other top opening of the pipe. In this configuration, not all smoke particles werecaptured by the collector but some were also escaping pass the collector into theambient air, as seen in the figure.

The challenge of ionizing and trying to guide the air flow in a limited space thatis enclosed by plastic walls, as in a plastic pipe, is that the resulting charge-buildupon the plastic surface can significantly affect the ESP characteristics. In the aboveinvestigation, it was however demonstrated that it is possible to efficiently chargeand steer the airborne particles in a confined plastic environment. This resultshows promising potential for the integration of ESP sampling in existing medicalbreathing and respiratory equipment.

Capture of synthetic breath aerosols

A new ESP system was specifically developed for the sampling of exhaled air inapplication 5.4 and is described in paper 7. The sampler was integrated in ahandheld thermoplastic device that accommodated an enclosed sampling volumewith inlets and outlets for the air flow.

With consideration to design guidelines that were established by finite elementmodeling, a prototype ESP sampler for breath aerosol was fabricated, see figure 4.5a. The device included a centered inlet for the breath sample and surroundinginlets for air sheath flow. The device was circular to promote a vortex-free air flowthrough the device. Three corona discharge needles were positioned concentricallybelow the inner orifices of the sheath flow inlets at optimal locations according tothe FE simulations. A liquid collector with an open air-liquid interface was centrallypositioned at an inter-electrode distance D from the corona discharge needles. Theliquid was grounded whereas the corona discharge voltages applied to the needleswere varied in between the measurements.


0

5

10

15

20

25

0 0.5 1 1.5 2 2.5 3 3.5 4

Colle

ctio

n E�

cien

cy (%

)

Corona Current (μA)

3cm 5cm7cm

0

2

4

6

8

10

0 2 4 6 8 10 12

0cm

ESPoff

Impinger19

0

0.5

1

1.5

2

2.5

3

3.5

4

3 3.5 4 4.5 5 5.5

Coro

na C

urre

nt (

µA)

Voltage (kV)

3 cm 5 cm 7 cm

Approximate corona onset for 3cm

Waste OutletHEPA filter

Compressor

Nebulizer

HEPAfilter

85%RH air

Inlet 1

Vacuum pump

Flow sensor

HEPAfilter

Pressuresensor

Inlet 2

Outlet

Breath aerosol model

85%RH air

Aerosol droplets flow

BioSampler impinger

Liquid collector

Corona needles

-1 to -5.5 kV

GND

Ion flux

Gas flow

Aerosol particle trajectories

Sheath flow

HEPA filter

HEPA filter

ø Rnø 20

ø 100

ø 15

D

10

1

120˚

ESP dimensions

80%

20%

Electrostatic precipitator(ESP)ESP sampler

Reference Biosampler impinger

Figure 4.5: a) Illustration of the ESP sampler for capturing synthetic breath aerosol to amicrofluidic device. b) Current-voltage relationship for the ESP, plotted for D = 3 (circle), 5(square), and 7 (diamond) cm. c) Capture Efficiency of the ESP as a percentage of the estimatedinput aerosol, ndye input for D = 0 (triangle, inset), 3 (circle), 5 (square), 7 (diamond) cm. Theshaded ellipses allow for ease of visualization of the data points for each inter-electrode distance.The error bars indicate measurements made in triplicates. The efficiency of the state-of-the-artreference impinger was measurement consistent and is also indicated.


The ESP sampler was connected to an upstream breath aerosol model thatgenerated a synthetic breath aerosol containing color dye particles as targetparticles. The generated aerosol particles were 1.0-2.7 micron in size (PM2.5),thus slightly larger compared to aerosol particles generated from breath, which istypically < 1 micron in size. Downstream, the sampler was connected to a referenceBiosampler impinger, which is the gold standard in state-of-the-art air samplingequipment. More details on the measurement procedures are found in paper 7.

It was seen that no corona was produced at low negative discharge voltages,whereas a stable self-sustained corona discharge with a quasi-linear relationship tothe corona current was present at higher negative discharge voltages, see figure 4.5b.

A maximum absolute collection efficiency of 21 percent was achieved for aninter-electrode distance D = 3 cm and corona discharge voltage of V = -5.5 kV(I = 4 microA), whereas lower voltages and/or different inter-electrode distancesreduced the collection efficiency. When the ESP was turned off, the collectionefficiency of aerosol droplets was consistently < 1 percent, confirming the successof electrostatic precipitation as mechanism for aerosol capture.

The ESP sampler performed well compared to the state-of-the-art Biosamplerimpinger, which had a collection efficiency of 19 percent in our tests. In addition,due to the small volume of the liquid collector of the ESP sampler, it achieveda 10 fold higher up-concentration of the captured dye particles compared to theBiosampler impinger. The volume of the liquid collector can be further decreased, asdemonstrated in paper 6, enabling even more concentrated samples and potentiallyalso highly sensitive and rapid detection of analytes when integrated with abiosensor.

4.3.3 Impact of ESP sampling on QCM frequency stability

Many QCM oscillator circuits apply electrical ground to the top gold electrode ofthe quartz crystal, which is in contact with the sample liquid. The crystal is thenless susceptible to external electrical disturbances. Interestingly, such configurationpotentially enables the integration of QCM systems with ESP sampling, in whichthe QCM system provides electrical ground to the liquid collector electrode of themicrofluidic collector cartridge.

To test the viability of direct integration of a QCM system with an ESP system,a commercial QCM oscillator (International Crystal Manufacturing Inc., USA) wasintegrated with the ESP sampler in section 4.3.1, see figure 4.6. The electricground potential of the QCM oscillator was shared, via the top electrode of thequartz crystal, with the microfluidic collector cartridge of the ESP sampler. Thecorona discharge needle was positioned 5 cm and 15 cm above the microfluidiccartridge in two subsequent tests, in which the ESP sampler was turned on for 5 sand 10 s, respectively. The microfluidic collector cartridge included a silicon chipwith a perforated diaphragm for providing the air-liquid interface. The resonance

4.4. RAPID SAMPLE MIXING IN AN ON-CHIP SAMPLE TUBE 51

-20 0

20 40 60 80

100 120 140 160 180 200 220 240 260 280 300

0 10 20 30 40 50 60 70 80 90 100 110 120

190 Hz after 10 s of ESP

300 Hz after 5 s of ESP

stabilized 80 s after ESP onset

ESP ON

Freq

uenc

y sh

ift (H

z)

Time (s)

DC -16kV5 cm 15 cm

biosensor cartridge

Figure 4.6: The graph shows the impact of the high-voltage ESP on the QCM resonancefrequency of a microfluidic biosensor cartridge with an air-liquid interface that functioned asthe collector electrode. The ESP sampler and the QCM shared a common electrical ground.

frequency of the QCM system was recorded before, during and after the ESP systemwas on.

Unsurprisingly, the results show that the high voltage and strong electrical fieldof the ESP sampler greatly affects the QCM system by triggering a rapid increaseof the resonance frequency. It is also seen that the distance between the coronadischarge needle and the microfluidic sensor cartridge affects the magnitude of thefrequency increase; a shorter distance gives a more rapid increase of the frequency.However, after the ESP sampler is turned off, the resonance frequency decreasesand stabilizes at the same baseline as the one that was established before the ESPsampling. Thus, the ESP sampler only has a temporary and transient impact on theQCM system. It is therefore not possible to perform real-time QCM measurementsduring ESP sampling, but the integration of the QCM and ESP systems potentiallyallows for measuring frequency shifts before and after ESP sampling.

4.4 Rapid sample mixing in an on-chip sample tube

The ESP sampling of airborne particles to liquid for subsequent biosensing does notnecessarily require direct sampling to the microfluidic biosensor cartridge but couldbe done to a separate liquid container, e.g. as the one used for breath sampling


b) Before mixing c) After 4 s mixing

d) Before trapping e) After �ushing/trapping

a) Illustration of bead mixing and trappingbeads aggregated beads dispersed

magnetunder channel

trappedbeadsmagnet under channel

vibrator

sample tube waste tube

micro�uidic device sensor

�exible �uidic joint

outlet

mixing b-c)

trapping d-e)

�owdirection

beadsin sample

tube

Figure 4.7: a) Illustration of the magnetic bead mixing and trapping in a microfluidic device.b) The beads are magnetically aggregated in a sample tube that is connected to the microfluidicdevice at the bottom and an electromechanic actuator at the side (side view). c) The beads arehomogeneously dispersed in the liquid sample in the tube after 4 seconds of active mixing (sideview). d) A magnet is positioned underneath a micro channel of the microfluidic device (topview). e) The sample liquid is flushed through the device and the magnetic beads are trapped inthe channel (top view). Subsequently, fresh liquid can be restored in the channel.

in section 4.3.2. Specifically, for the detection of influenza in application 5.4,sample pre-treatment is required to be separate from the biosensor flow cell in themicrofluidic cartridge due to the initial use of strong lysis buffers that inactivates theimmobilized receptors on the sensor surface. After exposure to the lysis buffer, thetarget particles are mixed with, and bound to, receptor-coated superparamagneticmicrobeads. Thereafter, the lysis buffer is replaced with standard buffer liquid toenable subsequent biosensing. This procedure is done off-chip in a sample tube,which potentially could be used as the collector in the ESP sampling.

A solution for seamless integration of sample pre-treatment in the sample tubewith subsequent biosensing in a microfluidic cartridge was developed, see figure 4.7.A 500 muL sample tube (Microfluidic ChipShop GmbH., Germany) was connectedto a microfluidic chip (Microfluidic ChipShop GmbH., Germany) by a flexible fluidicjoint in the form of silicone tube. The sample tube was positioned in an ellips-formed metal ring, connected to a small electromechanical solenoid actuator.

The tube was filled with phosphate buffered saline (PBS) containing 2.8 micronsuperparamagnetic beads (Life Technologies, Thermo Fisher Scientific Inc.,USA)that were temporarily aggregated at the side of the tube using a permanent magnet, see figure 4.7 b. The tube was then shaken by applying a 30 Hz sinusoidal ACsignal to the actuator, which moved the upper part of the tube in an ellips-liketrajectory as it followed the motion of the metal ring. The shaking of the tubestarted an immediate dispersion of the beads in the liquid, simulating mixing andbinding of target analytes with receptor-coated beads. After 4 s of actuation, thebeads had been dispersed into a homogenous distribution in the liquid, see figure 4.7c (side-view) and d (top view).

Thereafter, the liquid in the tube was aspirated through the microfluidiccartridge, see figure 4.7 e. The beads were separated from the sample liquid in a

4.5. DISCUSSION 53

microfluidic channel by an external permanent magnetic underneath the cartridge.While the beads were immobilized in the channel by the magnet, the sample liquidwas replaced with fresh buffer, thereby enabling downstream biosensing of thebeads.

4.5 Discussion

In this work, sampling of airborne particles directly to microfluidic cartridges wasdemonstrated. This was enabled by the development of a novel component for on-chip air-liquid interfacing and an ESP technique that integrated with microfluidiccartridges.

The air-liquid interface was provided by a perforated silicon diaphragm in whichsurface tension fixated the liquid meniscus in the pores. Thereby, the air-liquidinterface was able to withstand gravitation and pressure variations, making it robustfor practical use. In addition, it was possible to compensate for evaporation at theair-liquid interface by flowing liquid under the diaphragm without risking floodingor a collapse of the interface.

Air-liquid interfacing in microfluidic devices has previously been demonstratedby the use silicon pillar arrays but relied on downstream analysis of the capturedparticles. In contrast, the perforated diaphragm allowed to integrate a biosensor inclose proximity of only 50 microns to the air-liquid interface, enabling immediatedetection of the captured particles with no risk of sample losses to parasitic surfacein the microfluidic system.

However, for biosensing of airborne particles at PoC or in fieldwork, disposablecartridges are required. The perforated silicon diaphragms demonstrated in thiswork take up a relatively large area on a silicon wafer and is fabricated by DRIEthrough the wafer. Hence, the resulting cost per silicon diaphragm is too high tobe a viable alternative for disposable use.

Preferably, also the diaphragm should be made of polymer but unfortunately,polymers do not offer the same mechanical stiffness as silicon, which potentiallywould make the diaphragm too weak and flexible to hold the air-liquid interface ina steady position. A polymer microstructure for air-liquid interfacing would needto use another type of design. One example is micro pillars, which can be moldedsimultaneously as the rest of the cartridge.One drawback of a polymer pillar arrayis similar to the silicon pillar array in that it does not allow for placing the biosensorin close proximity to the air-liquid interface. However, that is a compromise thatmight be required in a disposable device.

Combining on-chip air-liquid interfacing with the developed ESP techniquefor capturing airborne particles directly to the liquid of a microfluidic biosensorcartridge opens up a range of new possibilities for biosensing applications. Byeliminating the need for sample handling after capture and prior to subsequentanalysis, the proposed method potentially enables automated and rapid sampling


and detection of airborne particles. This fits well with the requirements in PoC andfieldwork.

However, in certain situations sample pre-treatment and amplification is stillrequired. As discussed in application 5.4, the sample pre-treatment is notcompatible with the immunoassay on the sensor surface. Therefore, an off-chip sample pre-treatment technique was developed in which the sample first iscollected to a sample tube and thereafter seamlessly transferred to a flow cell ina microfluidic cartridge. To enable this process for use in an automated fashionin a disposable device, liquid aspiration and dispensing offered by the thermallyactuated composite, PDMS-XB, could be integrated in the microfluidic cartridge.This would allow for a completely electrically controlled device, from ESP airsampling, to PDMS-XB liquid handling and QCM/ADT biosensing.

One challenge in air sampling is the capture of unwanted dust particles. Suchparticles could clog the air-liquid interface and interfere with the biosensing byunspecific interactions with the sensor surface. However, ionized airborne particlescan be separated based on their size and number of acquired charges with anindustry standard technique called differential mobility analysis (DMA), in whichparticles of different size impact at different locations on the collector. Combiningthe ESP technique demonstrated in this work with DMA would enable a sizediscrimination of particles that are captured to the air-liquid interface, thus furtherup-concentrating and purifying the sample.

Chapter 5

Applications for Biosensing ofAirborne Particles

5.1 Introduction

In this chapter, three applications for biosensing of airborne particles are intro-duced, enabled by technological advancements demonstrated in this thesis, andinitial experimental results are presented.

First, detection of narcotic substances via airborne sample transport from asampling filter to the liquid of a biosensor cartridge is presented in section 5.2.Thereafter, monitoring and detection of norovirus in ambient air is presented insection 5.3. Lastly, breath-based diagnostics of influenza is presented in section 5.4.

These three applications have been or are currently targeted in research projectsin which framework the work of this thesis has been conducted. As of today,the primary task of most biosensor cartridges described in this thesis has been tosupport bioassay and instrument development, in which liquid-based samples arepredominantly used.

The measurement results presented below are achieved during the developmentphase of those larger biosensor projects. Whereas they do not represent the bestpossible sensitivities or LoDs of the QCM and ADT biosensors, they demonstratethe relevance, function and performance of the technical work in this thesis.

5.2 Detection of narcotics substances in filters

This work was based on a collaboration with Biosensor Applications AB, Sweden.We provided the miniaturization and integration of the biosensor cartridge andmicrofluidic air-liquid interfacing whereas they provided reagents and QCMinstrumentation. The collaboration resulted in paper 3.

This project was the start of our work on QCM technology and the integrationof quartz crystals with microfluidic cartridges. It was also in this work that we first

55

56 CHAPTER 5. BIOSENSING APPLICATIONS

developed the perforated silicon diaphragm for air-liquid interfacing.

Background

Biosensor Applications are focused on providing rapid on-site detection of drugs andexplosives, which is critical for the border control and customs, narcotics police,traffic police, prisons, and rehabilitation centers. Narcotic traces are typicallypresent on surfaces, fabrics, skins and in oral fluids. The traces are sampled by theuse of filters that are swiped on suspected surfaces and then subsequently processedand analyzed in a liquid-based system.

In this work, we investigate the possibility of transferring narcotic substancesfrom a filter, via air and into the liquid of a miniaturized microfluidic biosensorcartridge. A filter-air-liquid sample transfer potentially enables a rapid upconcentration of the narcotic substances and, in addition, also a rapid detectiondue to the close proximity of the biosensor to the air-liquid interface.

Contributions

The work of this thesis has contributed with:

• QCM biosensor cartridge, presented in section 3.3.2,

• microfluidic air-liquid interfacing, presented in section 4.2,

Measurement results

The QCM biosensing system is based on a competitive immunoassay, including anarcotics substance, Ab, and antigens. The Ab has higher affinity to the narcoticsmolecules than to the antigen. The antigens are pre-immobilized on the sensorsurface. Thereafter, the Ab are introduced to the sensor and bind to the antigen.Subsequently, narcotic molecules that are captured in a sample filter are vaporizedby a heat pulse. The airborne molecules absorb into the liquid of the microfluidiccartridge via the air-liquid interface. The molecules quickly diffuse 70 microns downto the sensor surface where the Ab release from the immobilized antigens and bindto the narcotics molecules in the liquid. This generates a frequency increase of theQCM due to the unloading of mass on the sensor surface.

Measurement results from the detection of 200 ng ecstasy and 100 ng cocaineis shown in figure 5.1 and figure 5.2, respectively. The spikes in the resonancefrequency at the beginning of each narcotics deposition originate from the thermalactuation used to vaporize the sample. However, the blank runs demonstrate thatthe resonance frequency quickly stabilizes at the original baseline after each spike.

For this project, a novel biosensor cartridge was developed that only consistedof four parts, which were integrated using adhesive bonding with a verticallyconducting adhesive foil. The tape provided the integration of the parts, liquid

5.2. DETECTION OF NARCOTICS SUBSTANCES IN FILTERS 57

Detection of ecstacy

Figure 5.1: The graph shows the binding of Ab, unbinding of Ab upon subsequent transfer andabsorption of ecstasy via the air-liquid interface and a control performed as a blank run, i.e. athermal vaporization step without narcotics molecules .

Detection of cocaine

Figure 5.2: The graph shows the binding of Ab, unbinding of Ab upon subsequent transfer andabsorption of cocaine via the air-liquid interface and two controls performed as blank runs beforeand after the addition of cocaine.

confinement in the microfluidic system and electrical contacting to the quartzcrystal sensor.

In addition, this work demonstrated the integration of an air liquid interfacewith a microfluidic biosensor cartridge that was positioned in close proximity, i.e.70 microns, to the sensor surface. This enabled the rapid transfer of airbornenarcotics molecules into the microfluidic cartridge and subsequent detection by thebiosensor.


5.3 Monitoring and detection of norovirus in ambient air

In this section I describe the work that has been conducted within the frameworkof one Swedish research project, RPA - Rapid Pathogen Analyzer (ended), and oneEuropean research project, Norosensor (ongoing).

The RPA project was an interdisciplinary collaboration among Swedish researchgroups where the KTH Micro and Nanosystems department provided the technol-ogy development. Other project partners contributed with the immunoassay de-velopment, including the production of norovirus (NV) virus-like-particles (VLPs)and antibody (Ab) production.

RPA was our starting point for the development of biosensor and samplingtechnologies for the rapid detection of airborne virus. This included the novelapproach of direct sampling of airborne particles to microfluidic cartridges and fur-ther development of perforated diaphragms for air-liquid interfacing and biosensorintegration.

Norosensor is an ongoing research project with partners from several Europeancountries that contribute by the development of assays, biosensing techniquesand instrumentation, whereas the KTH Micro and Nanosystems departmentcontributes with the development of biosensor cartridges, air sampling technologyand integration.

Much of the technology development that started in RPA has been transferredto, and is being further developed in, Norosensor. Norosensor also includesindustrial partners and is more focused on the targeted application compared toRPA, which in comparison was more research oriented. In addition to the QCMtechnique, Norosensor also explores the possibility of using ADT as a biosensorplatform for the detection of virus.

Background

Norovirus, also known as norwalk virus, is one of the most common causes ofacute gastroenteritis [87], leading to symptoms such as stomach pain, nausea andsevere diarrhea and vomiting. It is most common in the winter season but existsthroughout the year [88].

A person who is infected typically develops symptoms after 12-48 hours afterbeing exposed to the virus and the illness lasts 1-3 days. No long-lasting immunityis developed by persons who have had the illness. It is foremost among youngchildren and older adults that the illness can be serious, due to the risk of severedehydration.

Norovirus infection is associated with approximately 90 percent of epidemicnon-bacterial acute gastroenteritis worldwide [87]. In the U.S., approximatelytwenty million people get the illness each year and it causes up to seventy thousandhospitalizations and eight hundred deaths per year according to the Center forDisease Control and Prevention, USA.

5.3. AIR MONITORING OF NOROVIRUS 59

The virus transmits directly from person-to-person [89] or indirectly viacontaminated food or water [90]. Several studies have indicated that a commoncause of infection is via aerosolized virus, primarily generated from vomiting [89,91, 92].

Outbreaks commonly occur in daycare centers, nursing homes, schools andcruise ships. However, it is in hospitals where the illness is particularly serious,leading to outbreaks that have a major impact on both the healthcare and economy.

In one documented case, an outbreak caused illness in 69 percent of the staff and33 percent of the patients in the emergency room, and a total number of 635 infectedpersons in the hospital staff [92]. Such outbreaks not only lead to discomfort forthe infected persons, but also causes isolation or temporary shut-down or roomsand entire wards, postponed and/or cancelled operations and a prolonged illnessand an increased risk of fatality among immunosuppressed patients [93].

State-of-the-art methods methods for detection and diagnostics includes thecollection of whole stool or vomitus and analysis by PCR, ELISA or transmissionelectron microscopy (TEM) [94]. It has been demonstrated that PCR can detectas few as 20 virions [95]. However, such methods rely on laboratory facilities andtake several hours, or up to days, before providing an answer. A recent studyon the binding of norovirus virus-like-particles (NV-VLPs) to glycosphingolipids insupported lipid bilayers has demonstrated label-free biosensing by using QCM-D,but with the purpose to evaluate strain-specific binding characteristics [96]. A novellabel-based method demonstrated a limit-of-detection (LoD) of approximately 106

NV-VLPs/ml using a fluorescent TIRF-enabled readout [97], however, the reportedassay time was 2 hours and required an advanced imaging setup for readout.

If the presence of aerosolized norovirus can be rapidly detected on-site, largeoutbreaks can potentially be prevented. This work focuses on two primaryapplications: air monitoring for early detection and warning of the presence ofnorovirus in ambient air, and; rapid detection of norovirus in ambient air aftercleansing and disinfection of rooms in which infected patients have been present.Potentially, the proposed method can also be adapted for rapid diagnostics ofpatients that are suspected to have a norovirus infection.

Contributions


• QCM and ADT biosensor cartridges, presented in section 3.3.1,

• ESP sampling of airborne particles in ambient air directly to microfluidiccartridges, presented in section 4.3.1,


• sample-pretreatment and microbead mixing in an on-chip sample tube,presented in section 4.4.


Measurements with the QCM system have been conducted in our laboratory,as part of this thesis work, whereas the ADT-based measurements have beenconducted by our collaboration partners.

QCM measurement results

Preliminary measurement results for the detection of NV-VLPs in liquid have beenproduced in the work of this thesis and are shown in figure 5.3.

The surface of the quartz crystal was coated with either monoclonal orpolyclonal Ab that were specific to norovirus, see figure 5.3 a. The immobilizationof Ab was done either by direct unspecific binding of the Ab to the gold electrodeor by binding to pre-immobilized protein A. The surface was subsequently blockedwith bovine serum albumin (BSA) to prevent further unspecific binding. Non-pathogenic NV-VLPs were used as model particles to substitute real pathogenicnorovirus sample.

The tests were performed using a commercial QCM instrument (QCMLab,Sweden) and the cartridge presented in section 3.3.1. First, a stable baselinefrequency was obtained in PBS buffer solution followed by an injection of 5 µg/mlof NV-VLPs for 10 minutes. Thereafter, the flow cell and sensor surface were rinsedwith PBS buffer and a new baseline frequency was obtained. All injections weremade with a peristaltic pump and a constant liquid flow of 50 µl/min.

The resulting frequency shifts were 18 Hz for the sensor with polyclonal Ab and30 Hz for the sensor with monoclonal Ab. Given a total injection volume of 0.5 ml,the resulting mass sensitivities were 139 ng/Hz and 83 ng/Hz, respectively. Thesefigures are comparable to the results of a similar study where the detection of 5µg/ml of dengue virus in PBS solution resulted in frequency shifts ranging from10 to 40 Hz [98]. In that study, the binding of the virus to antibodies immobilizedvia protein A resulted in the largest frequency shifts compared to when otherimmobilization protocols were used.

A steady-state binding/unbinding condition or saturation of the sensor surface,i.e. that all binding sites were occupied, were not observed during the measurement,indicated by the continuos and steady decline in resonant frequency. A control wasconducted with the injection of BSA, resulting in no significant frequency shift. Nodrift was observed during the measurement and the noise level was equal to, orbelow, the frequency resolution of 1 Hz of the instrument (i.e. 1 sample readoutper second).

The short-term measurement in figure 5.3 a shows a linear decrease of theresonance frequency upon exposure of NV-VLP to the sensor. A long-termmeasurement was conducted to evaluate the binding kinetics and wether a steady-state condition between binding and unbinding or saturation of bound particles onthe surface is reached.

First, a stable baseline frequency was obtained in PBS buffer solution followedby a continuous injection of 5 µg/ml of NV-VLPs until the sample liquid wasconsumed and the liquid flow stopped after 2.5 hours. The resulting frequency


protein Amono-Ab

NV-VLP poly-Ab

Freq

ueny

shi

ft (H

z)

Time (s)

injection of 5µg/ml NV-VLP rinsing with bu�er

monoclonal Ab

polyclonal Ab

sample consumed,liquid �ow stopped

injection of 5µg/ml NV-VLP

monoclonal Ab

Freq

ueny

shi

ft (H

z)

Time (s)

protein A

mono-Ab

NV-VLP

a) Binding of NV-VLP to monoclonal and polyclonal Ab b) Long-term binding of NV-VLP to monoclonal Ab

Figure 5.3: QCM measurements were conducted in a) a short time period for detection of NV-VLP captured by immobilized monoclonal or polyclonal Ab, and in b) an extended period oftime for detection of NV-VLP captured by immobilized monoclonal Ab. The measurements wereconducted using the cartridge presented in section 3.3.1.

shift could not be evaluated due the sudden stop of liquid flow and due to that itwas not possible to distinguish the binding of NV-VLPs from unspecific bindingsince no rinsing with PBS buffer was performed.

The response in resonance frequency during the long-term measurement shows arelatively steady decrease, still with no significant indication of reaching a steady-state condition or saturation on the surface. This indicates that the biosensorsystem is either limited by the reaction kinetics or by the mass transport of sampleto the sensor.

Since the performance of the produced antibodies were analyzed using ELISAduring the development stage, which relied on over-night incubation times andonly generated end-point readouts, there is no available information about theassociation and dissociation rate constants of the Ab. Even though the Ab showedexcellent binding capabilities in the ELISA tests, there is no guarantee that theyare suitable for rapid detection in a label-free biosensor system.

However, the above results could also indicate a limitation in mass transportof the relatively large VLPs. Approximating the VLPs as particles with a radiusof 19 nm, their diffusion coefficient in water can be estimated to be 1.1 ∗ 10−11

m2/s [96]. In no flow conditions the binding at the sensor surface is likely diffusionlimited. A continuous liquid flow, as used in the these experiments, maintains theconcentration of VLPs in the liquid close to the sensor surface and limits the riskof sample depletion. However, based on the flow cell height, the liquid flow and thediffusion coefficient of the VLPs, it is estimated that approximately only 10 percentof the VLPs have time to diffuse down to the sensor surface before being flushedaway out from the flow cell. Thus, optimal flow conditions needs to be investigatedin future work.


Another uncertainty in the above measurements is the availability of bindingsites on the sensor surface. The results of the Ab immobilization on the sensorsurface, which originated from the protocols used for ELISA and TEM analysis ofnorovirus, was not thouroughly characterized. Other immobilization methods thatrely on chemisorption rather than physisorption, such as the use of thiolated PEGmolecules and EDC/NHS chemistry [99], have been proven successful and might beconsidered in future work.

The above measurements were conducted to evaluate if a rapid detection ofNV-VLP using a QCM biosensor was possible. Although the presented results arepreliminary, they indicate that the sensitivity is too low for the targeted applicationwith the current biosensor. This is addressed in ongoing research projects by: 1)shifting to alternative assays combined with novel signal amplification techniquesand improvement of the QCM instrumentation, and by; 2) the use of ADT, whichis described below.

Nevertheless, the QCM measurements provided evidence of the high perfor-mance and reliability of the QCM biosensor cartridge. The cartridge remainedleak-tight under a liquid flow for hours of operation. In addition, the sensor featuredno observable drift and low noise levels.

ADT measurement results

Our collaboration partners in the University of Cambridge and Loughborough Uni-versity have developed the method of ADT and corresponding ADT instruments.ADT potentially promises better sensitivity and lower limits of detection comparedto QCM and is therefore being explored for the use of rapid virus detection.

In Norosensor, a sandwich assay is used for the ADT detection of norovirus. Itincludes primary Ab or aptamers that are immobilized on the sensor surface andsecondary Ab or aptamers that are immobilized on magnetic microbeads. Both theprimary and the secondary Ab or aptamers are specific to norovirus. When thenorovirus particles bind to both the sensor surface and the microbead, they forman immobilized complex that potentially provides large viscoelastic energy lossesas the quartz crystal oscillates and consequently strong ADT signals.

For the development of the ADT and the immunoassays, ADT biosensorcartridges for liquid-based samples were designed and fabricated. The cartridgesare presented in section 3.3.1.

Preliminary measurement results provided by our collaboration partners areshowed in figure 5.4. The measurements were conducted using the cartridgepresented in section 3.3.1. The aim of the measurement was to evaluate the responseof the sensor upon binding of the bead-anti-NV complex to the sensor surface, whichis a part of the immunoassay that is being developed. A solution of 106 beads/mlwas used.

The results in figure 5.4 a show a characteristic ADT response where thespecifically bound bead-anti-NV-anti-IgE complex gives a high 3f current at highdrive amplitudes whereas the addition of IgG coated beads gives a lower and similar


Figure 5.4: An ADT measurement of beads coated with anti-norovirus IgE that are specificallybound to the sensor surface by immobilized anti-IgE. A control with unspecific binding of beadsthat were coated with IgG was performed. a) The raw 3f signal is plotted against the drive voltage.b) The modulus of the difference in complex 3f signal from the baseline signal is plotted againstthe 1f current. The measurements were conducted using the cartridge presented in section 3.3.1.

signal as pure buffer solution. In figure 5.4 b, the same measurement data isnormalized to the signal generated in pure buffer and plotted against the 1f current.This provides a clearer representation of the ADT output signals upon specific andunspecific binding.

Consider an ideal case where beads bind to the sensor surface in a sandwichassay with NV-VLPs as in the immunoassay described above. In addition, considerthat each bead binds only to one VLP, and that all bead-VLP complexes bindto the sensor surface and generate a similar response as seen in figure 5.4.Given the number of beads used in the measurement above, the generated signalwould correspond to a detection of 106 VLPs/ml. Noro-VLPs have a molecularweight of 1.17 ∗ 104 kDa [97], thus the detected NV-VLP concentration wouldbe approximately 20 pg/ml. This concentration is approximately five orders ofmagnitude lower compared to the one used in the QCM measurement aboveand comparable to [97]. Thus, the detection of 106 microbeads/ml presented inthe measurement above demonstrates the potentially high sensitivity of the ADTcombined with a microbead-based sandwich assay.

However, the above exercise was based on ideal conditions, which are typicallyunrealistic in practice, and does not necessarily represent the expected outcome ina real measurement. Nevertheless, since the beads are the main contributor to the3f current signal and the main function of the analyte particles, being intact viruscapsids or only protein fragments, is to attach the beads to the sensor surface, theabove measurement is a strong indication of the potential of ADT.

The ADT cartridge used in the experiment demonstrated great performance andreliability. The quality factor of the sensor was approximately 2800 in liquid andstrong 3f currents were recorded even in low concentration samples, as demonstrated


above. This type of cartridge has been used by our project partners for threeconsecutive years without failure.

5.4 Breath-based diagnostics of influenza

This work has been conducted within the framework of the European multidisci-plinary research project Rapp-Id (ongoing), which includes technical competence,industrial participation and clinical expertise. Several different technology plat-forms are being developed for rapid diagnostics of infectious diseases. We areinvolved in a platform based on ADT where the aim is to develop a point-of-carebiosensor instrument for non-invasive breath-based diagnostics of influensa.

In Rapp-Id, we provide the development of biosensor cartridges, air-liquidinterfacing, aerosol sampling technology and integration, whereas our collabora-tors provide the development of ADT instrumentation, immunoassays, sample-pretreatment protocols, etc. As in Norosensor, much of the technology that wasdeveloped in RPA has been transferred to Rapp-Id for further development.

Background

Over the past few years, the concern of an influenza pandemic has increased sincethe emergence of a highly pathogenic avian influenza A (H5N1) in southeast Asia,causing what is commonly referred to as the bird flu [8]. The strain is maintainedin bird populations but according to investigations made by the world healthorganization, WHO, a total number of 630 human cases with a confirmed infectionof H5N1 was reported. The infections resulted in the death of 375 people. Theseconcerns were further supported by the emergence of a pandemic caused by a newinfluenza virus (H1N1).

Aerosol transmission is considered to be major cause for the spreading ofinfluenza A [100]. Further studies have strengthen this hypothesis and it suggeststhat aerosol transmission, in particular, is responsible for the most severe cases ofdisease involving viral infection of the lower respiratory tract [101].

Rapid influenza diagnostics is used for screening patients with suspectedinfluenza and provides timely results to support clinical decision making [102]. Itcan help to facilitate antiviral treatment, decrease inappropriate use of antibioticsand reduce the duration of treatment at hospitals. In addition, rapid influenza testsare also used for investigating suspected influenza outbreaks, such as those causedby H5N1 and H1N1, to facilitate prompt implementation of control measures [103].

However, commercial tests for rapid diagnostics of influenza suffer from lowsensitivity [103] and require confirmatory test results provided by conventionallaboratory based methods, such as shell vial culture, tissue cell viral culture,or RT-PCR [103]. A recent review of emerging biotechnology-based detectionsystems for rapid diagnostics of influenza identifies that there has been considerableprogress in the development of new tests, but also highlights the urgent demand formore sensitive, simple and rapid detection [104]. According to the review, further

5.4. BREATH-BASED DIAGNOSTICS OF INFLUENZA 65

development of diagnostic assays should aim at achieving affordable point of caredetection ability to facilitate diagnostics in the field.

This work focuses on the development of an ADT-based biosensor system forrapid and highly sensitive PoC influenza diagnostics. In addition, we forego theconventional sampling methods, which include sputum and swab samples and thatcause discomfort for the patient, with a non-invasive sampling of breath aerosolbased on our developed ESP system.

Contributions


• ADT biosensor cartridges, presented in section 3.3.1,

• ESP sampling of breath-based aerosol directly to microfluidic cartridges,presented in section 4.3.2,


• sample-pretreatment and microbead mixing in an on-chip sample tube,presented in section 4.4.

Measurements with the ADT system has been conducted by our collaborationpartners.

Measurement results

An immunoassay for the biosensor is being developed in Rapp-Id by our projectpartners. However, the direct detection of influenza virus using a label-freebiosensor is challenging due to the lack of conservative epitopes that are easilyaccessible for surface immobilized Ab or aptamers. In Rapp-Id, this issue iscircumvented by lysing the virus and target nucleoproteins (NPs), which are lesssubjected to mutation. A single influenza virus holds about one thousand NPs.Thus, the release of the NP generates a thousand-fold sample up-concentration,which is advantageous for the downstream biosensing.

The biosensor is based on a sandwich immunoassay, including magneticmicrobeads. The released NPs are first bound in solution to microbeads that arecoated with primary anti-NV or aptamers. Thereafter, the NV-bead complex bindsto secondary anti-NV or aptamers that are immobilized on the sensor surface.

A disadvantage is that the lysis buffer inactivates the secondary Ab on thesensor surface. Therefore, the first binding step must take place outside of thesensor flow cell. The lysis require a relatively large volume of liquid. Therefore, wedeveloped a method for sample lysis and mixing with microbeads that is performedin a sample tube, connected to the microfluidic cartridge, see section 4.4. Afterlysing, mixing and binding, the lysis solution can be replaced with buffer solutionwhile the NV-bead complexes are held in position by magnetic forces. Thereafter,


~1000 in�uenza nucelproteins (NP)control

linkersecondary anti-NP

bead withprimary anti-NP

in�uenza nucelprotein (NP)

Drive amplitude (V)

3f c

urre

nt (A

U)

Figure 5.5: An ADT measurement of 1000 influenza nucleoproteins, approximatelycorresponding to 1 virus particle, which gives a significant difference in the 3f current comparedto the baseline signal taken before the capture of the nucleoproteins. The measurements wereconducted using the cartridge presented in section 3.3.1.

the NV-bead complexes introduced to the flow cell and captured onto the sensorsurface by the secondary Ab or aptamer.

A preliminary measurement provided by our collaboration partners is shownin figure 5.5. The measurement was conducted using the cartridge presented insection 3.3.1. The aim of the measurement was to evaluate the response of the sensorupon binding of the NV-bead complex. A solution of 1000 purified nucleoproteinswas used.

The measurement shows significantly stronger 3f current upon binding of theNP-bead complex compared to the baseline signal in buffer solution. The resultsindicate that potentially very low concentration samples of influenza virus can bedetected. It also demonstrates that small particles, such as proteins, can be detectedusing ADT when combined with a microbead-based sandwich assay. However, theseare preliminary results and further testing needs to be conducted to confirm thereproducibility, quantifiability, sensitivity, specificity, etc.

Similarly to the cartridges used in the previously presented measurements,the ADT cartridge used in this experiment was based on mechanical clamping.However, it is conceptually different in terms of design. It is mounted by a press-fitassembly and consists of a minimal amount of parts to allow for future disposableuse. The cartridge demonstrates leak-tight operation, rapid assembly/disassemblyand excellent performance during ADT measurements. This type of cartridge hasbeen used by our project partners for one year without failure.

Chapter 6

Discussion and Outlook

In this thesis, novel techniques were demonstrated for the integration of airborneparticle sampling to microfluidic biosensor cartridges and were designed anddeveloped for use at PoC or in-the-field applications. Specifically, the presentedtechniques for airborne particle sampling, sample pretreatment and mixing, on-chipliquid actuation and biosensing are all based on small-scale electronic instrumen-tation. The airborne sampling and subsequent biosensing were demonstrated tobe electrically compatible and the sample transport from air to liquid to sensorwas seamlessly integrated. The results potentially enable the development of anintegrated system in a portable instrument.

We also developed new methods for manufacturing and heterogeneous inte-gration with the aim to close the fabrication technology gap between academicresearch and industrial production and thus enable a quick and seamless transferof technological advancements from science to society.

In particular, a new industrial-like and easy-to-use reaction injection moldingtechnique was developed for the fabrication of thermoplastic-like microfluidiccartridges. This is a major step forward to closing the fabrication technology gapto industrial manufacturing compared the workhorse in academia today, namelycasting of PDMS.

Additionally, we demonstrated benefits of the new molding technique comparedto industrial injection molding of thermoplastic parts, e.g. by the fabricationof disposable QCM biosensor cartridges. The integration of the QCM with thecartridge was based on direct covalent bonding, which enables a much morestraightforward and uncomplicated back-end process compared to conventionalintegration approaches and reduces the amount of cartridge parts to a minimum,i.e. microfluidic polymer parts and a QCM. Furthermore, the reaction injectionmolding and direct integration techniques potentially allow for batch processing,which would make a significant difference in the reduction of fabrication costscompared to the conventional serial pick-and-place type of integration. Thus, thenovel molding and integration techniques potentially address many of the existing

67

68 CHAPTER 6. OUTLOOK

challenges of making a heterogeneously integrated biosensor cartridge for disposableuse.

However, the new reaction injection molding technique needs further opti-mization and adaption before proving itself suitable for industrial manufacturing,e.g. by further decreasing the molding cycle-times. Therefore, additionally,advancements were made in the integration of biosensors with thermoplasticmicrofluidic cartridges. By employing a press-fit type of assembly, the designand integration complexity of a new biosensor cartridge was substantially reducedcompared to commercially available cartridges. Future development of OSTE+RIMtowards two-component injection molding, i.e. simultaneous molding of bothstiff and rubbery materials, could potentially allow integrating the o-ring as astructural part of the polymer cartridge. In my view, this is how far conventionalmanufacturing technologies could be used for adapting current biosensor cartridgesdesigned for laboratory settings to the use as disposables at PoC or in the field.

To conclude, I believe that future instruments for biosensing of airborne particleswill open a new world for analysis, where we can learn how to better understand thepresence and transport of airborne particles. This will hopefully lead to improvedhealthcare, better prevention of disease outbreaks, and potentially also enable newexciting applications that are yet undiscovered.

Summary of Appended Papers

Paper 1: Reaction injection molding and direct covalent bonding of OSTE+polymer microfluidic devicesIn this paper, we present OSTE+RIM, a novel reaction injection mold-ing process that combines the merits of off-stoichiometric thiol-ene epoxythermosetting polymers with the fabrication of high quality microstructuredparts. OSTE+RIM uniquely incorporates temperature stabilization andshrinkage compensation of the OSTE+ polymerization during molding,enabling rapid processing, high replication fidelity and low residual stress.The replicated parts are directly and covalently bonded in the back-endprocessing, generating complete devices. The method is demonstrated withthe fabrication of optically transparent, warp-free and natively hydrophilicmicrofluidic cartridges.

Paper 2: Liquid Aspiration and Dispensing Based on an Expanding PDMSCompositeThis paper introduces a method for for on-chip liquid aspiration and dispens-ing in microfluidic cartridges. The actuation mechanism is based on single-use thermally expanding polydimethylsiloxane (PDMS) composite, whichallows altering its surface topography by means of individually addressableintegrated heaters. The method is demonstrated by the fabrication ofmicrofluidic devices that are designed to create an enclosed cavity in thesystem, due to locally expanding the initially unstructured composite. Thisenables negative volume displacement and leads to the event of liquidaspiration. Subsequently, the previously formed cavity is closed by asecondary actuation step, leading to downstream liquid dispensing in themicrofluidic cartridge.

Paper 3: An integrated QCM-based narcotics sensing microsystemIn this paper, we present the design, fabrication and successful testingof an QCM biosensor for airborne narcotics detection. The biosensorcartridge consists of only four parts, including a perforated diaphragm forair-liquid interfacing. The parts are integrated using a vertically conductivedouble-sided adhesive foil, which provides both electrical contacting and

69

70 SUMMARY OF APPENDED PAPERS

liquid containment. The biosensor absorbs airborne narcotics molecules andperforms a liquid immunoassay on the sensor surface. The system wasdemonstrated with a successful detection of 100 ng of cocaine and 200 ngof ecstasy.

Paper 4: One step integration of gold coated sensors with OSTE polymercartridges by low temperature dry bonding

This paper introduces a novel method for the integration of biosensorswith microfluidic OSTE cartridges by direct covalent bonding. OSTEthermosetting cartridges are casted and then bonded to gold-coated quartzcrystals by gold-thiol reactions. The resulting bond interface is shownto be completely homogenous and void free and the cartridge is testedsuccessfully to a differential pressure of up to 2 bars. This approach delivers astraightforward and rapid high-quality integration aiming for the productionof robust, low cost and disposable biosensor cartridges.

Paper 5: Integration of microfluidics with grating coupled silicon photonic sensorsby one-step combined photopatterning and molding of OSTE

In this paper, we present a novel integration method for packaging siliconphotonic sensors with polymer microfluidics cartridges, designed to be suitablefor wafer-level production methods. We demonstrate the fabrication, in asingle step, of an OSTE microstructured part and the subsequent unassisteddry bonding with a grating coupled silicon photonic ring resonator sensor chip.The microfluidic layer features photopatterned through holes (vias) for opticalfiber probing and fluid connections, as well as molded microchannels andtube connectors. The method addresses the previously unmet manufacturingchallenges of matching the microfluidic footprint area to that of the photonics,and of robust bonding of microfluidic cartridges to biofunctionalized surfaces.

Paper 6: Electrohydrodynamic enhanced transport and trapping of airborneparticles to a microfluidic air-liquid interface

This paper introduces a novel approach for efficient sampling of airborneparticles in ambient air to a microfluidic cartridge by integrating an elec-trohydrodynamic air pump with a microfluidic air-liquid interface. In oursystem, a negative corona discharge partially ionizes the air around a sharpelectrode tip while the electrostatic field accelerates airborne particles towardsan electrically grounded air-liquid interface, where they absorb. The air-liquid interface is fixated at the microscale pores of a perforated silicondiaphragm, each pore functioning as a static Laplace valve. Our system wasexperimentally tested using airborne smoke particles of ammonium chlorideand aqueous salt solution as the liquid. We measured that EHD enhancedtransport of the particles from the air into the liquid is enhanced over 130times compared to passive trapping.

71

Paper 7: Aerosol sampling using an electrostatic precipitator integrated with amicrofluidic interfaceIn this paper, we present development of a point-of-care system aiming forbreath-based sampling of airborne particles directly to a microfluidic cartridgefor subsequent analysis. The system involves an electrostatic precipitatorthat uses corona charging and electrophoretic transport to capture aerosoldroplets onto a microfluidic air-to-liquid interface for downstream analysis.A theoretical study of the governing geo- metric and operational parametersfor optimal electrostatic precipitation is presented. The fabrication of anelectrostatic precipitator prototype and its experimental validation using alaboratory-generated synthetic breath aerosol is described. Collection effi-ciencies were comparable to those of a state-of-the-art Biosampler impinger,with the significant advantage of providing samples that are at least 10 timesmore concentrated.

Acknowledgments

Now, a long but worthwhile micro journey has come to an end and without thegreat help of others I would not have gotten this far.

First of all, I would like to express my sincere gratitude to, Wouter. Thank youfor supervising me during my doctoral studies with lots of enthusiasm, inspirationand trust as well as for providing me the opportunity to try my wings on thismultidisciplinary adventure through science.

I would also like to thank and congratulate Göran for creating and preservingan exceptionally motivating, innovative and multicultural research environment aswell as for an outstanding leadership that allows an open and positive atmospherein the group.

Many thanks also to Tommy for his unwearied support in which he always bringsa down-to-earth perspective on things, independent on the subject of matter, andalso for bringing all aspects of doctoral studies to a higher level.

I would like to thank Björn for introducing me to this exciting field of researchand for supervising me during my master thesis project, which was the first leg ofthis journey into a miniaturized world.

I would especially like to thank my former master thesis student Reza, who isnow my colleague, office roommate, and close friend. Thank you for all the cheerfulmoments, hard work and contributions you have made to this thesis, despite all thechallenges that life has brought you. You are a fighter and there is nothing thatcan stop you!

Thank you for the financial support for the work of this thesis, which was partlyprovided by: Vinnova, Swedish Foundation for Strategic Research and the SwedishResearch Council in the RPA project; Innovative Medicines Initiative in the Rapp-ID project; and the 7th Framework Programme by the European Commission inthe Norosensor project.

I would also like to thank: my old office roommate Kristinn for sharing yourimmense knowledge, both while working and eating Italian food; Fredrik C foryour very contagious positive attitude; Gaspard for interesting and thoughtful long-into-the-night-at-some-hotelroom discussions; Andreas for the exciting time-lapseplanning and for diving deep into fun tech-talks; Staffan for giving me the laughsof my life with your extraordinary nice Hollywood voice; Hans for sharing yourviews and knowledge of, not only the technological but also the artistic aspect, of

73

74 ACKNOWLEDGMENTS

photography; Laila for being a great coworker but in the same breath a caring andsincere friend; Jonas for making everyone smile by bringing irony into any seriousmatter; Stefan B for a long and true friendship, despite you being a badass ”mc-knutte” ;-) Thomas for all the joyful storytelling on one of my favorite subjects,adventures in the archipelago; Alexander for letting me share train delay complaintsover a lunchbox; Mikael S for inspiring me to live life more like a scout; Mikael H forbringing some Småland atmosphere to the capital; and Umer for being the originalrole model for beard growing students.

I am also very grateful and would like to express my appreciation to ourtechnicians, the mechatronic wizards Kjell and Mikael B, and the queen of silicon,Cecilia. In addition, work would have been much more difficult without the helpfrom our truly outstanding and hardworking administrators, Erika and Ulrika.Thank you!

At the department of Micro and Nanosystems, I have met some wonderfulcolleagues that have shared their knowledge, friendship and awesomeness with meand I would therefor like to say thank you to: Frank, Joachim, Niclas, Oleksandr,Fredrik F, Mikhail, Bernhard, Simon, Valentin, Carlos, Xuge, Henrik F, HenrikG, Hithesh, Maoxiang, Miku, Gabriel, Jessica, Floria, Mina, Federico, Stephan,Fritzi, Xiaojing, Xinghai, Xiamo, Nutapong, Martin, Adit, Mikael A, Farizah andZargham.

I have also had the chance to work with some great master thesis students andI would like to thank and wish good luck to Sveinbjörg, Jutta, Gleb, and Javad.

The various projects that I have worked in have given me the opportunity tocollaborate with many different people in Europe, and I would particularly like tothank: Jeroen, Dragana, Karen, and Francesco in Leuven; David and Victor inCambridge; Sourav, Igor, Shilpa and Carlos in Loughborough; Lieven and his teamat JnJ; Johan and Jimmy at Getinge; Lars and Vasile at QCMLab; Hans, Berit,Rolf and Tony at KI; Lennart, Marie and Johan at LiU; and Mats and Annika atSU.

On a personal note, I am forever grateful for the unreserved support given bymy mum and dad and parents-in-law and would like to thank you for all yourencouragement, pep talks, advice, and practical support as well as for your loveand caring of my family.

Lastly, I want to thank my family. Tuva and Elliot, you are my greatest sourceof inspiration in life and everyday you challenge me to explain complex things moreeasily, which I consider is the most important skill of a scientist. Mia, thank youfor all the love, support, strength and understanding that you give to me. Withthese last words of the thesis, I want to thank you for always standing by my side,I love you!

Niklas Sandström,Stockholm, April 26, 2015

References

[1] R. M. Harrison and J. Yin, “Particulate matter in the atmosphere: whichparticle properties are important for its effects on health?” The Science ofthe total environment, vol. 249, no. 1-3, pp. 85–101, Apr. 2000.

[2] H. Brandl, A. von Däniken, C. Hitz, and W. Krebs, “Short-termdynamic patterns of bioaerosol generation and displacement in an indoorenvironment,” Aerobiologia, vol. 24, no. 4, pp. 203–209, 2008.

[3] I. Eames, J. W. Tang, Y. Li, and P. Wilson, “Airborne transmission of diseasein hospitals,” Journal of The Royal Society Interface, vol. 6, no. 6, pp. 407–702, Oct. 2009.

[4] X. Xie, Y. Li, A. T. Y. Chwang, P. L. Ho, and W. H. Seto, “How far dropletscan move in indoor environments – revisiting the Wells evaporation–fallingcurve,” Indoor Air, vol. 17, no. 3, pp. 211–225, Jun. 2007.

[5] D. Johnson, R. Lynch, C. Marshall, K. Mead, and D. Hirst, “AerosolGeneration by Modern Flush Toilets,” dx.doi.org, vol. 47, no. 9, pp. 1047–1057, Jul. 2013.

[6] B. M. Andon, “Active air vs. passive air (settle plate) monitoring in routineenvironmental monitoring programs.” PDA journal of pharmaceutical scienceand technology / PDA, vol. 60, no. 6, pp. 350–355, Nov. 2006.

[7] J. S. West, S. D. Atkins, J. Emberlin, and B. D. L. Fitt, “PCR to predictrisk of airborne disease,” Trends in microbiology, vol. 16, no. 8, pp. 380–387,Aug. 2008.

[8] A. Abbott and H. Pearson, “Fear of human pandemic grows as bird flu sweepsthrough Asia.” Nature, vol. 427, no. 6974, pp. 472–473, Feb. 2004.

[9] H. Becker, “Mind the gap!” Lab on a Chip, vol. 10, no. 3, pp. 271–273, 2010.

[10] U. M. Attia, S. Marson, and J. R. Alcock, “Micro-injection moulding ofpolymer microfluidic devices,” Microfluidics and Nanofluidics, vol. 7, no. 1,pp. 1–28, 2009.

75

76 REFERENCES

[11] J. C. McDonald and G. M. Whitesides, “Poly(dimethylsiloxane) as a materialfor fabricating microfluidic devices.” Accounts of chemical research, vol. 35,no. 7, pp. 491–499, Jul. 2002.

[12] D. C. Duffy, J. C. McDonald, O. Schueller, and G. M. Whitesides, “Rapidprototyping of microfluidic systems in poly(dimethylsiloxane),” AnalyticalChemistry, vol. 70, no. 23, pp. 4974–4984, 1998.

[13] M. A. Eddings, M. A. Johnson, and B. K. Gale, “Determining theoptimal PDMS–PDMS bonding technique for microfluidic devices,” Journalof Micromechanics and Microengineering, vol. 18, no. 6, p. 067001, Jun. 2008.

[14] J. S. Kuo, L. Ng, G. S. Yen, R. M. Lorenz, P. G. Schiro, J. S. Edgar, Y. Zhao,D. S. W. Lim, P. B. Allen, G. D. M. Jeffries, and D. T. Chiu, “A newUSP Class VI-compliant substrate for manufacturing disposable microfluidicdevices,” Lab on a Chip, vol. 9, no. 7, pp. 870–876, 2009.

[15] D. Bartolo, G. Degré, P. Nghe, and V. Studer, “Microfluidic stickers,” Lab ona Chip, vol. 8, no. 2, pp. 274–279, 2008.

[16] H. Becker and C. Gärtner, “Polymer microfabrication technologies formicrofluidic systems,” Analytical and Bioanalytical Chemistry, vol. 390, no. 1,pp. 89–111, 2008.

[17] C.-W. Tsao and D. L. Devoe, “Bonding of thermoplastic polymermicrofluidics,” Microfluidics and Nanofluidics, vol. 6, no. 1, pp. 1–16, 2009.

[18] T. Haraldsson, C. F. Carlborg, and W. van der Wijngaart, “ OSTE: a novelpolymer system developed for Lab-on-Chip,” SPIE MOEMS- . . . , vol. 8976,pp. 897 608–897 608–6, Mar. 2014.

[19] C. F. Carlborg, T. Haraldsson, K. Öberg, M. Malkoch, and W. van derWijngaart, “Beyond PDMS: off-stoichiometry thiol–ene (OSTE) based softlithography for rapid prototyping of microfluidic devices,” Lab on a Chip,vol. 11, no. 18, pp. 3136–3147, Sep. 2011.

[20] C. F. Carlborg, A. Vastesson, Y. Liu, W. van der Wijngaart, M. Johansson,and T. Haraldsson, “Functional off-stoichiometry thiol-ene-epoxy thermosetsfeaturing temporally controlled curing stages via an UV/UV dual cureprocess,” Journal of Polymer Science Part A: Polymer Chemistry, vol. 52,no. 18, pp. 2604–2615, Sep. 2014.

[21] X. Zhou, C. F. Calborg, and N. Sandstrom, “Rapid Fabrication of OSTE+Microfluidic Devices with Lithographically Defined Hydrophobic/HydrophilicPatterns and Biocompatible Chip Sealing,” in Proceedings of the 17thInternational Conference on Miniaturized Systems for Chemistry and LifeSciences, 2013.

REFERENCES 77

[22] F. Saharil, F. Forsberg, Y. Liu, P. Bettotti, N. Kumar, F. Niklaus,T. Haraldsson, W. van der Wijngaart, and K. B. Gylfason, “Dryadhesive bonding of nanoporous inorganic membranes to microfluidic devicesusing the OSTE(+) dual-cure polymer,” Journal of Micromechanics andMicroengineering, vol. 23, no. 2, p. 025021, Feb. 2013.

[23] E. T. Carlen and C. H. Mastrangelo, “Electrothermally activated paraffinmicroactuators,” Microelectromechanical Systems, Journal of, vol. 11, no. 3,pp. 165–174, Jun. 2002.

[24] C. G. Cooney and B. C. Towe, “A thermopneumatic dispensing micropump,”Sensors and Actuators a-Physical, vol. 116, no. 3, pp. 519–524, Oct. 2004.

[25] P. Griss, H. Andersson, and G. Stemme, “Expandable microspheres for thehandling of liquids.” Lab on a Chip, vol. 2, no. 2, pp. 117–120, May 2002.

[26] R. M. Lequin, “Enzyme immunoassay (EIA)/enzyme-linked immunosorbentassay (ELISA).” Clinical chemistry, vol. 51, no. 12, pp. 2415–2418, Dec. 2005.

[27] S. Schultz, D. R. Smith, J. J. Mock, and D. A. Schultz, “Single-target moleculedetection with nonbleaching multicolor optical immunolabels.” Proceedings ofthe National Academy of Sciences of the United States of America, vol. 97,no. 3, pp. 996–1001, Feb. 2000.

[28] S. U. Senveli and O. Tigli, “Biosensors in the small scale: methods andtechnology trends.” IET nanobiotechnology / IET, vol. 7, no. 1, pp. 7–21,Mar. 2013.

[29] P. Prakrankamanant, “Quartz crystal microbalance biosensors: prospects forpoint-of-care diagnostics.” Journal of the Medical Association of Thailand =Chotmaihet thangphaet, vol. 97 Suppl 4, pp. S56–64, Apr. 2014.

[30] A. L. Washburn and R. C. Bailey, “Photonics-on-a-chip: recent advances inintegrated waveguides as enabling detection elements for real-world, lab-on-a-chip biosensing applications,” Analyst, vol. 136, no. 2, pp. 227–236, 2011.

[31] G. Z. Sauerbrey, Sauerbrey: Use of quartz vibration for weighing thin... -Google Scholar. J. Physik, 1959.

[32] T. Nomura and O. Hattori, “Determination of micromolar concentrations ofcyanide in solution with a piezoelectric detector,” Analytica chimica acta, vol.115, pp. 323–326, Mar. 1980.

[33] P. L. KONASH and G. J. BASTIAANS, “Piezoelectric-Crystals as Detectorsin Liquid-Chromatography,” Analytical Chemistry, vol. 52, no. 12, pp. 1929–1931, 1980.

78 REFERENCES

[34] B. A. Martin and H. E. Hager, “Velocity Profile on Quartz Crystals Oscillatingin Liquids,” Journal of Applied Physics, vol. 65, no. 7, pp. 2630–2635, 1989.

[35] B. A. Martin, “Flow Profile Above a Quartz Crystal Vibrating in Liquid,”Journal of Applied Physics, vol. 65, no. 7, pp. 2627–2629, 1989.

[36] S. M. Reddy, J. P. Jones, and T. J. Lewis, “The coexistence of pressure wavesin the operation of quartz-crystal shear-wave sensors,” Journal of AppliedPhysics, vol. 83, no. 5, pp. 2524–2532, Mar. 1998.

[37] R. Lucklum, S. Schranz, C. Behling, F. Eichelbaum, and P. Hauptmann,“Analysis of compressional-wave influence on thickness-shear-mode resonatorsin liquids,” Sensors and Actuators a-Physical, vol. 60, no. 1-3, pp. 40–48, May1997.

[38] V. M. Mecea, J. O. Carlsson, and R. V. Bucur, “Extensions of the quartz-crystal-microbalance technique,” Sensors and Actuators a-Physical, vol. 53,no. 1-3, pp. 371–378, May 1996.

[39] V. M. Mecea, “From quartz crystal microbalance to fundamental principlesof mass measurements,” Analytical Letters, vol. 38, no. 5, pp. 753–767, 2005.

[40] K. Wong-ek, O. Chailapakul, N. Nuntawong, K. Jaruwongrungsee, andA. Tuantranont, “Cardiac troponin T detection using polymers coated quartzcrystal microbalance as a cost-effective immunosensor,” BiomedizinischeTechnik/Biomedical Engineering, vol. 55, no. 5, pp. 279–284, Oct. 2010.

[41] A. Arnau Vives, Ed., Piezoelectric Transducers and Applications. Berlin,Heidelberg: Springer Berlin Heidelberg, 2004.

[42] A. Pomorska, D. Shchukin, R. Hammond, M. A. Cooper, G. Grundmeier,and D. Johannsmann, “Positive Frequency Shifts Observed Upon AdsorbingMicron-Sized Solid Objects to a Quartz Crystal Microbalance from the LiquidPhase,” Analytical Chemistry, vol. 82, no. 6, pp. 2237–2242, Feb. 2010.

[43] M. Rodahl and B. Kasemo, “A simple setup to simultaneously measure theresonant frequency and the absolute dissipation factor of a quartz crystalmicrobalance,” Review of Scientific Instruments, vol. 67, no. 9, pp. 3238–3241, Sep. 1996.

[44] F. Höök, M. Rodahl, P. Brzezinski, , and B. Kasemo, “Energy DissipationKinetics for Protein and Antibody−Antigen Adsorption under ShearOscillation on a Quartz Crystal Microbalance,” Langmuir, vol. 14, no. 4,pp. 729–734, Jan. 1998.

[45] M. Edvardsson, M. Rodahl, B. Kasemo, and F. Höök, “A dual-frequencyQCM-D setup operating at elevated oscillation amplitudes.” AnalyticalChemistry, vol. 77, no. 15, pp. 4918–4926, Aug. 2005.

REFERENCES 79

[46] M. C. Dixon, “Quartz crystal microbalance with dissipation monitoring:enabling real-time characterization of biological materials and theirinteractions.” Journal of biomolecular techniques : JBT, vol. 19, no. 3, pp.151–158, Jul. 2008.

[47] M. A. Cooper, “Biosensing using rupture event scanning (REVS)™,”Measurement Science and Technology, vol. 14, no. 11, pp. 1888–1893, Nov.2003.

[48] M. A. Cooper, F. N. Dultsev, T. Minson, V. P. Ostanin, C. Abell, andD. Klenerman, “Direct and sensitive detection of a human virus by ruptureevent scanning.” Nature biotechnology, vol. 19, no. 9, pp. 833–837, Sep. 2001.

[49] S. K. Ghosh, V. P. Ostanin, and A. A. Seshia, “Anharmonic InteractionSignals for Acoustic Detection of Analyte,” Analytical Chemistry, vol. 82,no. 9, pp. 3929–3935, Apr. 2010.

[50] ——, “Anharmonic Surface Interactions for Biomolecular Screening andCharacterization,” Analytical Chemistry, vol. 83, no. 2, pp. 549–554, Dec.2010.

[51] S. K. Ghosh, V. P. Ostanin, C. L. Johnson, C. R. Lowe, and A. A.Seshia, “Probing biomolecular interaction forces using an anharmonicacoustic technique for selective detection of bacterial spores,” Biosensors &bioelectronics, vol. 29, no. 1, pp. 145–150, Nov. 2011.

[52] W. Bogaerts, P. De Heyn, T. Van Vaerenbergh, K. De Vos, S. Kumar Sel-varaja, T. Claes, P. Dumon, P. Bienstman, D. Van Thourhout, and R. Baets,“Silicon microring resonators,” Laser & Photonics Reviews, vol. 6, no. 1, pp.47–73, Jan. 2012.

[53] A. L. Washburn, M. S. Luchansky, A. L. Bowman, and R. C.Bailey, “Quantitative, label-free detection of five protein biomarkers usingmultiplexed arrays of silicon photonic microring resonators.” AnalyticalChemistry, vol. 82, no. 1, pp. 69–72, Jan. 2010.

[54] A. L. Washburn, L. C. Gunn, and R. C. Bailey, “Label-Free Quantitationof a Cancer Biomarker in Complex Media Using Silicon Photonic MicroringResonators,” Analytical Chemistry, vol. 81, no. 22, pp. 9499–9506, 2009.

[55] M. S. McClellan, L. L. Domier, and R. C. Bailey, “Label-free virus detectionusing silicon photonic microring resonators,” Biosensors & bioelectronics,vol. 31, no. 1, pp. 388–392, 2012.

[56] K. De Vos, J. Girones, T. Claes, Y. De Koninck, S. Popelka, E. Schacht,R. Baets, and P. Bienstman, “Multiplexed Antibody Detection With an Arrayof Silicon-on-Insulator Microring Resonators,” Photonics Journal, IEEE,vol. 1, no. 4, pp. 225–235, 2009.

80 REFERENCES

[57] C. F. Carlborg, K. B. Gylfason, A. Kaźmierczak, F. Dortu, M. J. B.Polo, A. M. Catala, G. M. Kresbach, H. Sohlström, T. Moh, L. Vivien,J. Popplewell, G. Ronan, C. A. Barrios, G. Stemme, and W. van derWijngaart, “A packaged optical slot-waveguide ring resonator sensor arrayfor multiplex label-free assays in labs-on-chips,” Lab on a Chip, vol. 10, no. 3,pp. 281–290, Feb. 2010.

[58] M. S. Luchansky and R. C. Bailey, “High-Q optical sensors for chemical andbiological analysis,” Analytical Chemistry, 2011.

[59] V. Zamora, P. Lützow, M. Weiland, D. Pergande, and H. Schröder, “ Highlysensitive integrated optical biosensors,” SPIE . . . , pp. 893 307–893 307–7,Mar. 2014.

[60] H. Ogi, “Wireless-electrodeless quartz-crystal-microbalance biosensors forstudying interactions among biomolecules: a review.” Proceedings of theJapan Academy. Series B, Physical and biological sciences, vol. 89, no. 9,pp. 401–417, 2013.

[61] G. Y. Yu and J. Janata, “Proximity effect in quartz crystal microbalance.”Analytical Chemistry, vol. 80, no. 8, pp. 2751–2755, Apr. 2008.

[62] N. K. Chaki and K. Vijayamohanan, “Self-assembled monolayers as a tunableplatform for biosensor applications,” Biosensors & bioelectronics, vol. 17, no.1-2, pp. 1–12, Jan. 2002.

[63] D. Krüger, H. Fuchs, R. Rousseau, D. Marx, and M. Parrinello, “Pullingmonatomic gold wires with single molecules: an Ab initio simulation.”Physical review letters, vol. 89, no. 18, p. 186402, Oct. 2002.

[64] F. P. Cometto, P. Paredes-Olivera, V. A. Macagno, and E. M. Patrito,“Density functional theory study of the adsorption of alkanethiols on Cu(111),Ag(111), and Au(111) in the low and high coverage regimes.” The journal ofphysical chemistry. B, vol. 109, no. 46, pp. 21 737–21 748, Nov. 2005.

[65] J. Escorihuela, M. J. Bañuls, J. G. Castelló, V. Toccafondo, J. García-Rupérez, R. Puchades, and Á. Maquieira, “Chemical silicon surfacemodification and bioreceptor attachment to develop competitive integratedphotonic biosensors,” Analytical and Bioanalytical Chemistry, vol. 404, no. 10,pp. 2831–2840, 2012.

[66] F. Saharil, C. F. Carlborg, T. Haraldsson, and W. van der Wijngaart,“Biocompatible "click" wafer bonding for microfluidic devices.” Lab on a Chip,vol. 12, no. 17, pp. 3032–3035, Sep. 2012.

[67] J. F. Ashley, N. B. Cramer, R. H. Davis, and C. N. Bowman, “Soft-lithographyfabrication of microfluidic features using thiol-ene formulations.” Lab on aChip, vol. 11, no. 16, pp. 2772–2778, Aug. 2011.

REFERENCES 81

[68] J. M. Karlsson, F. Carlborg, and F. Saharil, “High-resolution micropatterningof off-stoichiometric thiol-enes (OSTE) via a novel lithography mechanism,”. . . for Chemistry and . . . , 2012.

[69] E. P. Kartalov and S. R. Quake, “Microfluidic device reads up to fourconsecutive base pairs in DNA sequencing-by-synthesis,” Nucleic AcidsResearch, vol. 32, no. 9, pp. 2873–2879, May 2004.

[70] K. De Vos, I. Bartolozzi, E. Schacht, P. Bienstman, and R. Baets, “Silicon-on-Insulator microring resonator for sensitive and label-free biosensing.” OpticsExpress, vol. 15, no. 12, pp. 7610–7615, Jun. 2007.

[71] F. Saharil, L. El Fissi, Y. Liu, F. Carlborg, D. Vandormael, L. A.Francis, W. van der Wijngaart, and T. Haraldsson, “SUPERIOR DRYBONDING OF OFF-STOICHIOMETRY THIOL-ENEEPOXY (OSTE(+))POLYMERS FOR HETEROGENEOUS MATERIALLABS-ON-CHIP,” . . .Micro Total Analysis . . . , pp. 1831–1833, 2012.

[72] H. F. Kraemer and H. F. Johnstone, “Collection of Aerosol Particles inPresence of Electrostatic Fields,” Industrial & Engineering Chemistry, vol. 47,no. 12, pp. 2426–2434, Dec. 1955.

[73] T. Han and G. Mainelis, “Design and development of an electrostatic samplerfor bioaerosols with high concentration rate,” Journal of Aerosol Science,vol. 39, no. 12, pp. 1066–1078, Dec. 2008.

[74] M. Tan, F. Shen, M. Yao, and T. Zhu, “Development of an AutomatedElectrostatic Sampler (AES) for Bioaerosol Detection,” dx.doi.org, vol. 45,no. 9, pp. 1154–1160, May 2011.

[75] A. Miller, G. Frey, G. King, and C. Sunderman, “A Handheld ElectrostaticPrecipitator for Sampling Airborne Particles and Nanoparticles,” dx.doi.org,vol. 44, no. 6, pp. 417–427, Apr. 2010.

[76] L. S. Christensen, K. E. Brehm, J. Skov, K. W. Harlow, J. Christensen, andB. Haas, “Detection of foot-and-mouth disease virus in the breath of infectedcattle using a hand-held device to collect aerosols,” Journal of VirologicalMethods, vol. 177, no. 1, pp. 44–48, Oct. 2011.

[77] J.-S. Chang, P. A. Lawless, and T. Yamamoto, “Corona discharge processes,”Plasma Science, IEEE Transactions on, vol. 19, no. 6, pp. 1152–1166, Dec.1991.

[78] D. Dobrynin, G. Friedman, A. Fridman, and A. Starikovskiy, “Inactivation ofbacteria using dc corona discharge: role of ions and humidity,” New Journalof Physics, vol. 13, no. 10, p. 103033, Oct. 2011.

82 REFERENCES

[79] J.-M. Roux, O. Kaspari, R. Heinrich, N. Hanschmann, and R. Grunow,“Investigation of a New Electrostatic Sampler for Concentrating Biologicaland Non-Biological Aerosol Particles,” dx.doi.org, vol. 47, no. 5, pp. 463–471,Jan. 2013.

[80] T. W. Owen, R. O. Al-Kaysi, C. J. Bardeen, and Q. Cheng, “Microgravimetricimmunosensor for direct detection of aerosolized influenza A virus particles,”Sensors and actuators. B, Chemical, vol. 126, no. 2, pp. 691–699, 2007.

[81] J. Lee, J. Jang, D. Akin, C. A. Savran, and R. Bashir, “Real-time detection ofairborne viruses on a mass-sensitive device,” Applied Physics Letters, vol. 93,no. 1, 2008.

[82] D. R. P. Morris, J. Fatisson, A. L. J. Olsson, N. Tufenkji, and A. R.Ferro, “Real-time monitoring of airborne cat allergen using a QCM-basedimmunosensor,” Sensors and actuators. B, Chemical, vol. 190, pp. 851–857,Jan. 2014.

[83] L. RAJAKOVIC, V. GHAEMMAGHAMI, and M. THOMPSON, “Adsorp-tion on Film-Free and Antibody-Coated Piezoelectric Sensors,” Analyticachimica acta, vol. 217, no. 1, pp. 111–121, 1989.

[84] D. D. Stubbs, W. D. Hunt, S. H. Lee, and D. F. Doyle, “Gas phase activityof anti-FITC antibodies immobilized on a surface acoustic wave resonatordevice,” Biosensors & bioelectronics, vol. 17, no. 6-7, pp. 471–477, Jun. 2002.

[85] M. Michalzik, R. Wilke, and S. Buttgenbach, “Miniaturized QCM-based flowsystem for immunosensor application in liquid,” Sensors and actuators. B,Chemical, vol. 111, pp. 410–415, 2005.

[86] T. Frisk, N. Sandström, L. Eng, W. van der Wijngaart, P. Månsson, andG. Stemme, “An integrated QCM-based narcotics sensing microsystem ,”Lab on a Chip, vol. 8, no. 10, pp. 1648–1657, 2008.

[87] L. Lindesmith, C. Moe, S. Marionneau, N. Ruvoen, X. Jiang, L. Lindblad,P. Stewart, J. LePendu, and R. Baric, “Human susceptibility and resistanceto Norwalk virus infection.” Nature medicine, vol. 9, no. 5, pp. 548–553, May2003.

[88] A. W. Mounts, T. Ando, M. Koopmans, J. S. Bresee, J. Noel, and R. I.Glass, “Cold weather seasonality of gastroenteritis associated with Norwalk-like viruses.” Journal of Infectious Diseases, vol. 181 Suppl 2, no. Supplement2, pp. S284–7, May 2000.

[89] P. J. Marks, I. B. Vipond, D. CARLISLE, D. DEAKIN, R. E. Fey, and E. O.Caul, “Evidence for airborne transmission of Norwalk-like virus (NLV) in ahotel restaurant,” Epidemiology and infection, vol. 124, no. 03, pp. 481–487,Jun. 2000.

REFERENCES 83

[90] M. Noda, “Statistical analysis of attack rate in norovirus foodborneoutbreaks,” International journal of food microbiology, vol. 122, no. 1-2, pp.216–220, Feb. 2008.

[91] P. J. Marks, I. B. Vipond, F. M. Regan, K. Wedgwood, R. E. Fey, andE. O. Caul, “A school outbreak of Norwalk-like virus: evidence for airbornetransmission.” Epidemiology and infection, vol. 131, no. 1, pp. 727–736, Aug.2003.

[92] L. A. SAWYER, J. J. MURPHY, J. E. KAPLAN, P. F. PINSKY,D. CHACON, S. WALMSLEY, L. B. SCHONBERGER, A. PHILLIPS,K. FORWARD, C. GOLDMAN, J. BRUNTON, R. A. FRALICK, A. O.CARTER, J. WILLIAM G GARY, R. I. GLASS, and D. E. LOW, “25- TO 30-NM VIRUS PARTICLE ASSOCIATED WITH A HOSPITAL OUTBREAKOF ACUTE GASTROENTERITIS WITH EVIDENCE FOR AIRBORNETRANSMISSION,” American Journal of Epidemiology, vol. 127, no. 6, pp.1261–1271, Jun. 1988.

[93] S. Schwartz, M. Vergoulidou, E. Schreier, C. Loddenkemper, M. Reinwald,M. Schmidt-Hieber, W. A. Flegel, E. Thiel, and T. Schneider, “Norovirusgastroenteritis causes severe and lethal complications after chemotherapy andhematopoietic stem cell transplantation,” Blood, vol. 117, no. 22, pp. 5850–5856, Jun. 2011.

[94] R. HF, S. M, B. S, W. A, P. W, and D. HW, “Laboratory diagnosis ofnorovirus: which method is the best?” Intervirology, vol. 46, no. 4, pp. 232–238, Dec. 2002.

[95] M. JA and B. LD, “Laboratory diagnosis of norovirus.” Clinical laboratory,vol. 52, no. 11-12, pp. 571–581, Dec. 2005.

[96] G. E. Rydell, A. B. Dahlin, F. Höök, and G. Larson, “QCM-D studiesof human norovirus VLPs binding to glycosphingolipids in supported lipidbilayers reveal strain-specific characteristics.” Glycobiology, vol. 19, no. 11,pp. 1176–1184, Nov. 2009.

[97] M. Bally, M. Graule, F. Parra, G. Larson, and F. Höök, “A virus biosensorwith single virus-particle sensitivity based on fluorescent vesicle labels andequilibrium fluctuation analysis,” Biointerphases, vol. 8, no. 1, p. 4, Feb.2013.

[98] C.-C. Su, T.-Z. Wu, L.-K. Chen, H.-H. Yang, and D.-F. Tai, “Development ofimmunochips for the detection of dengue viral antigens,” Analytica chimicaacta, vol. 479, no. 2, pp. 117–123, Mar. 2003.

[99] F. Frederix, K. Bonroy, W. Laureyn, G. Reekmans, A. Campitelli, W. Dehaen,, and G. Maes, “Enhanced Performance of an Affinity Biosensor Interface

84 REFERENCES

Based on Mixed Self-Assembled Monolayers of Thiols on Gold,” Langmuir,vol. 19, no. 10, pp. 4351–4357, Apr. 2003.

[100] R. Tellier, “Review of aerosol transmission of influenza A virus.” Emerginginfectious diseases, vol. 12, no. 11, pp. 1657–1662, Nov. 2006.

[101] Tellier, “Aerosol transmission of influenza A virus: a review of new studies,”Journal of The Royal Society Interface, vol. 6, no. 6, pp. 302–790, Sep. 2009.

[102] T. M. Uyeki, “Influenza diagnosis and treatment in children: a review ofstudies on clinically useful tests and antiviral treatment for influenza.” ThePediatric infectious disease journal, vol. 22, no. 2, pp. 164–177, Feb. 2003.

[103] T. M. Uyeki, R. Prasad, C. Vukotich, S. Stebbins, C. R. Rinaldo, Y.-h. Ferng,S. S. Morse, E. L. Larson, A. E. Aiello, B. Davis, and A. S. Monto, “Lowsensitivity of rapid diagnostic test for influenza.” Clinical Infectious Diseases,vol. 48, no. 9, pp. e89–92, May 2009.

[104] T. R. Shojaei, M. Tabatabaei, S. Shawky, M. A. M. Salleh, and D. Bald,“A review on emerging diagnostic assay for viral detection: the case of avianinfluenza virus.” Molecular biology reports, vol. 42, no. 1, pp. 187–199, Jan.2015.

Paper Reprints

85

Integrating Biosensors for Air Monitoring and Breath-Based ...808742/...EHD ElectroHydroDynamic ESP...

Documents

Transcript of Integrating Biosensors for Air Monitoring and Breath-Based ...808742/...EHD ElectroHydroDynamic ESP...