optofluidic photonic crystal fibres for biomedical research in fibra

Dissertation im Fachbereich Physikder Friedrich-Alexander-Universität

Erlangen-Nürnberg

Optofluidische Photonische Kristallfasern zurBiomedizinischen Forschung in fibra

Optofluidic Photonic Crystal Fibres forBiomedical Research in fibra

physics of light

Optofluidic Photonic Crystal Fibresfor Biomedical Research in fibraOptofluidische Photonische Kristallfasern zur

Biomedizinischen Forschung in fibra

Der Naturwissenschaftlichen Fakultätder Friedrich-Alexander-Universität Erlangen-Nürnbergzur Erlangung des Doktorgrades Dr. rer. nat.

vorgelegt von

DIPL.-PHYS. SARAH UNTERKOFLERgeboren in Schwäbisch Gmünd

Als Dissertation genehmigt von der Naturwissenschaftlichen Fakultät derFriedrich-Alexander-Universität Erlangen-Nürnberg

"So eine Arbeit wird eigentlich nie fertig, man muß sie für fertig erklären,wenn man nach Zeit und Umständen das Möglichste getan hat."

Johann Wolfgang VON GOETHE

Abgabe bei den Berichterstattern 28. Januar 2013

Einreichung an der Universität (Umlauf) 2. April 2013

Mündliche Prüfung 28. Mai 2013

Promotionskommissionsvorsitzender Prof. Dr. Johannes BARTH

Erstberichterstatter Prof. Philip St.J. RUSSELL, D.Phil.

Zweitberichterstatterin Prof. Monika A. M. RITSCH-MARTE, Ph.D.

ZusammenfassungPhotonische Hohlkernfasern (engl.: hollow-core photonic crystal fibres – HC-PCFs) bieten dieeinzigartige Möglichkeit, eine wohldefinierte und lichtintensive optische Mode in einem Mediummit kleinem Brechungsindex zu leiten. Dies ist in einer konventionellen optischen Glasfaser einDing der Unmöglichkeit. Füllt man zudem die löchrige Struktur der Faser mit einer Flüssigkeit,so beträgt der Überlapp einer darin gelösten oder suspendierten Probe mit dem Lichtfeld fast100 %. Gleichzeitig ist die Strahlungsintensität im etwa 20µm kleinen Faserkern im Vergleichzu konventionellen opto-analytischen Methoden um mehrere Größenordnungen erhöht. Mitanderen Worten: Die Hohlkernfaser verwandelt sich in einen exzellenten optofluidischen Kanal,der eine noch nie da gewesene Licht-Materie-Wechselwirkung entlang seiner Länge erlaubt.Die Grundlagen für die vorliegende Arbeit finden sich im gesamten Gebiet der Naturwissenschaftund Technik. Ich werde daher zunächst eine einführende Übersicht der für meine Studien wichti-gen Themengebiete geben. Im Rahmen der aufstrebenden Forschungsgebiete Biophotonik undOptofluidik werde ich anschließend zwei ganz unterschiedliche Anwendungsmöglickeiten vonHC-PCFs demonstrieren, die dazu dienen sollen die Forschung in den Lebenswissenschaften vo-ranzutreiben.Zum einen werden die einzigartigen Eigenschaften optofluidischer PCFs ausgenutzt um mittelsoptischer Kräfte einzelne Zellen über bislang unerreichte Strecken zu transportieren. Durchdie Enge im Faserkern wirken auf die Zelloberfläche erhöhte Scherkräfte, welche die De-formation der Zelle hervorrufen. Dies wiederum zieht eine Veränderung der Propagations-geschwindigkeit nach sich, was wir auf praktische Weise mit einer bildgebungsfreien DOPPLER-Geschwindigkeitsmessung aufzeichnen. Aus diesem Grunde stellen optofluidische PCFs einvielversprechendes neues Werkzeug für die Untersuchung der mechanischen Eigenschaften einerZelle dar, die ein wichtiger Indikator für ihren Gesundheitszustand sind.Zum anderen werden optofluidische PCFs als photochemische Mikrodurchflussreaktoren verwen-det. Dabei zeigt sich, dass sich durch die mikrofluidische Integration der Faser die Effektivitäteiner massenspektrometrischen Analyse drastisch steigern lässt. Dies ist von besonderer Bedeu-tung für das Screening von potentiellen Medikamenten für die lichtaktivierbare Chemotherapie.Wenn man die Grenzen der Methode weiter ausreizt, wird es letztlich vielleicht sogar möglichsein detailliertere Informationen über deren Wirkmechanismus in situ zu erhalten.Die vorgestellten neuen Methoden für die Zellbiologie und Photochemie demonstrieren sowohldas vielversprechende Potential als auch die vielseitige Anwendbarkeit von optofluidischen PCFs.Dies eröffnet eine fundamental neue Möglichkeit: Biomedizinische Forschung in fibra – in derFaser.

iii

AbstractHollow-core photonic crystal fibre (HC-PCF) features the unique possibility to guide a definedand bright optical mode in a low refractive index material. This is totally impossible in con-ventional optical fibres. Moreover, when filling PCF’s holey structure with liquid medium, theoverlap of a dissolved or suspended sample with the light field is close to unity. At the sametime the irradiance in the about 20µm small fibre core is increased by several orders of magni-tude compared to conventional opto-analytical techniques. In other words: HC-PCF turns into anoutstanding optofluidic channel to allow unprecedented light-matter interaction along its length.The foundations of this thesis can be found throughout all of science and technology. An intro-ductory review of the relevant fields of study will therefore be given. Within the framework ofthe emerging fields of biophotonics and optofluidics, I will then demonstrate two rather differentapplications of optofluidic PCFs for the advancement of the life sciences.Firstly, the unique properties of optofluidic PCFs are exploited for unchallenged long-range trans-portation of individual cells by radiation forces. Moreover, due to the confined space inside thecore, shear forces on the cell surface are greatly enhanced and provoke deformation. This causeschanges in propagation speed that are conveniently monitored using a non-imaging Doppler-velocimetric technique. Therefore, optofluidic PCFs represent a budding new tool for the inves-tigation of cellular mechanics, which is an important indicator of a cell’s state of health.Secondly, optofluidic PCFs are used as microflow photochemical reactors. It becomes clear thatupon microfluidic integration of the fibre, the efficiency of a mass-spectrometry-based analysis isdrastically enhanced. This is particularly important for the screening and study of potential drugsfor photoactivated chemotherapy. Pushing the limits further, it might eventually be possible toretrieve detailed information on their mechanism of action in situ.The presented new methods for cell biology and photochemistry demonstrate both the impressivepotential and the versatile applicability of optofluidic PCFs. This creates a fundamentally newapproach: biomedical research in fibra – in the fibre.

Contents

1 Introduction to the Matter – and how to bring it to Light 1

2 Optofluidic Photonic Crystal Fibres 52.1 Optofluidics and Optical Micromanipulation in the Life Sciences . 6

2.1.1 Basic Principles of Microfluidics . . . . . . . . . . . . . . . . . 6

2.1.2 Optofluidic Sensors and Reactors for Biochemistry . . . . . 11

2.1.3 Force Transducers and Sensors for Cell Mechanics . . . . . . 16

2.1.4 Optical Micromanipulation of Particles and Cells . . . . . . 18

2.2 Hollow-Core Photonic Crystal Fibres . . . . . . . . . . . . . . . . . . 26

2.2.1 Historical Accounts . . . . . . . . . . . . . . . . . . . . . . . . 26

2.2.2 Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

2.2.3 Light Guidance Mechanisms . . . . . . . . . . . . . . . . . . . 32

2.2.4 Filling Hollow-Core PCFs with Liquid → * . . . . . . . . . . 39

2.2.5 * → yields Unbeaten Optofluidic Channels . . . . . . . . . . 44

3 Optical Guidance and Deformation Monitoring of Single Cells in fibra 513.1 Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

3.2 Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

3.2.1 Materials and Sample Preparation . . . . . . . . . . . . . . . 53

3.2.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

3.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 62

3.3.1 Long-Distance Optical Cell Guidance . . . . . . . . . . . . . . 62

3.3.2 Temperature Calibration . . . . . . . . . . . . . . . . . . . . . 63

3.3.3 Imaging-Free Deformation Monitoring . . . . . . . . . . . . 66

3.3.4 Theoretical Analysis . . . . . . . . . . . . . . . . . . . . . . . . 68

3.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

4 Optofluidic Integration of PCF Photochemical Microflow Reactors 734.1 Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

4.2 Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

4.2.1 Materials and Sample Preparation . . . . . . . . . . . . . . . 76

4.2.2 Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

v

4.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 89

4.3.1 Photoaquation of Vitamin B12 . . . . . . . . . . . . . . . . . . 89

4.3.2 Photoactivation of Potential Anticancer Compounds . . . . 90

4.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

5 Outlook 1015.1 PCF Optofluidic Microparticle and Cell Guides . . . . . . . . . . . . 101

5.1.1 Guidance of Single Eukaryotic Cells . . . . . . . . . . . . . . 101

5.1.2 Biomechanical Probing of Single Cells . . . . . . . . . . . . . 103

5.1.3 Investigation of Optically-Bound Cell Chains . . . . . . . . . 104

5.1.4 Improvement of the Setup and On-Chip Integration . . . . . 106

5.1.5 Cell Mechanics Modelling . . . . . . . . . . . . . . . . . . . . 107

5.1.6 In-Fibre Microenvironment Monitoring with Sensor Beads . 108

5.2 PCF Photochemical Flow Reactors . . . . . . . . . . . . . . . . . . . . 109

5.2.1 Direct Injection of Online Photoactivated Agents into Tis-sue Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

5.2.2 Microflow Photochemical Synthesis . . . . . . . . . . . . . . 110

Bibliography 111

Publications 137

Acknowledgements 141

11Introduction to the Matter

– and how to bring it to Light

"Wenn du eine weise Antwort verlangst, mußt du vernünftig fragen."

Johann Wolfgang VON GOETHE (1749–1832)

"And when the answer that you want is in the question that you state,come what may, come what may."

’Blood Red Summer’, COHEED & CAMBRIA (2004)

We live in an age in which mankind strives towards enhancing efficiency in allaspects of life. Clearly, this concerns foremost science and technology which havealways had a pioneering role in sociocultural evolution. The quest is thereforeto increase the handling and cost efficiency in both production (synthesis) anddetection (analysis), while at the same time decreasing safety and health risks.This requires more rapid material or information processing schemes combinedwith minimised material consumption. As a first step, miniaturisation of conven-tional approaches is the way to go, but once achieved there is much more to that.New physical effects appear and might lead to novel functionality. Moreover, thepossibility emerges to integrate several techniques into one, on a single chip.

The term chip was coined within the framework of the oldest of all microtech-nologies: microelectronics. It was already in the 1960s, when miniaturised elec-tronic components were first used in order to increase the density of computa-tional units. In this way not only the space but also the time and labour requiredto perform computational calculations could be drastically reduced. Microelec-tronics has forced the "Digital Revolution" from the Industrial Age to the Infor-mation Age.

1

Chapter 1 INTRODUCTION TO THE MATTER – AND HOW TO BRING IT TO LIGHT

To replace electronic circuits in the further development of information tech-nology, integrated photonics [POLLOCK & LIPSON, 2003; LIFANTE, 2005; HUNS-BERGER, 2009] seeks to build integrated optical circuits for the generation, ma-nipulation and detection of photons. After all, optical fibre technology has beensetting a good example with respect to information transport – replacement ofelectronic data cables had already started in the 1980s. Huge efforts are cur-rently undertaken to exploit the intriguing quantum-physical properties of pho-tons for quantum information technology which encompasses quantum comput-ing and quantum data transfer.

Far away, on the other side of the wide field of science, another microtech-nology emerged in the late 1990s which — at first sight — seems to be ratherdifferent. Microfluidics deals with fluidic circuits composed of micrometre-sizedchannels and seeks the automation of common laboratory techniques in bothchemistry and biology [SQUIRES & QUAKE, 2005; WHITESIDES, 2006; DEMELLO,2006].

In recent years these two — seemingly incompatible — technologies have beenmerged into the field of optofluidics [FAINMAN et al., 2010; HAWKINS & SCHMIDT,2010; MONAT et al., 2007; HUNT & WILKINSON, 2008; FAN & WHITE, 2011;SCHMIDT & HAWKINS, 2011; PANG et al., 2012] via the integration of opticalcomponents such as lasers, adaptive lenses, planar waveguides and detectorsinto microfluidic chips. Herein, one can distinguish two different branches. Onebranch seeks to use the properties of certain liquids to control the propertiesof guided light1 [PSALTIS et al., 2006], whereas the other branch deals withinvestigations of liquids or liquid-based samples by means of optical techniques.

In this thesis, I will deal exclusively with the latter of these two branches andmore specifically with biomedical applications thereof. In this sense, optoflu-idics can be classified within the scientific area of biophotonics, see Figure 1.1 A.From the bioscientific viewpoint the integrated functional combination of flu-idics and optics was far from unnatural, since optical techniques have a verylong tradition and unbroken popularity in chemistry and biology. The oldestoptical technique in biology is microscopy, which was introduced over four cen-turies ago. It provided a natural extension to human’s most important sense,sight, into the microcosmos. With the advent of spectroscopy in the 19th century,the distinct interaction of light with biological and/or chemical matter becameof increasing interest as a way to detect, discriminate or structurally determinecertain (bio)chemical species on a molecular level. Over the years, (bio)chemicalsensing schemes have become more versatile, more accurate and more sensitive.Fittingly, these advances in ultra-sensitive detection techniques are exactly whatis needed for investigations in small-dimension test tubes. Electrochemical meth-

1which, in fact, I would prefer to call hydrophotonics or fluido-optics

2

1.0

Figure 1.1: A) Classifying optofluidic technology. B) Google Scholar search for articles andpatents on "optofluidics" (18th December 2012, "incl. citations" disabled).

ods, mass spectrometry and – of course – optical techniques provide high spatialand temporal resolution as well as non-invasiveness.

The synergistic effects in this novel interdisciplinary research community arejust picking up pace, as is illustrated by the number of articles and patents pub-lished in the field (Figure 1.1 B), as well as the number of special issues in jour-nals2 and topical conference meetings3.

With this thesis, I would like to introduce liquid-filled hollow-core photonic crys-tal fibre (HC-PCF) as a new optofluidic tool for efficient light-matter interac-tion in the life sciences. HC-PCFs are outstanding hollow optical waveguides[RUSSELL, 2006] which when filled with aqueous medium serve as marvellousoptofluidic channels. This means light is guided efficiently along a microfluidicchannel comprised by the fibre core, which is typically around 20µm in diame-ter. Hence, low waveguide losses of a few dB m-1 allow the use of long opticalpath lengths, greatly enhancing the effective light-matter interaction, while thesample volume amounts only to some nL per cm of interaction length. In addi-tion, HC-PCFs are easily fabricated from chemically inert, high-quality silica glasswith negligible scattering, absorbance and fluorescence. Finally, the propagatinglight is confined to the hollow core, allowing it to interact strongly with the sam-ple. At comparable input powers, the resulting irradiance in the hollow core is 5orders of magnitude higher compared to conventional bioanalytical techniques,facilitating ultra-sensitive spectroscopy, highly-efficient photoactivation and op-tical guidance of biologically-relevant objects.

2e. g. in the journals: Microfluidics & Nanofluidics 4 (2008), Biomicrofluidics 4(4) (2010), Nature Photonics5(10) (2011), Lab on a Chip 12(24) (2012).

3e. g. IEEE/LEOS Topical Meetings on Optofluidics 2006 & 2008 (North America), EOS Conferences onOptofluidics (EOSOF) 2011 & 2013 (Germany), International Conferences on Optofluidics 2011–2013(China & Hong Kong).

3

Chapter 1 INTRODUCTION TO THE MATTER – AND HOW TO BRING IT TO LIGHT

In the following chapter 2 I will equip the reader with the necessary scientificbackground by introducing the topics of microhydrodynamics, photochemistry,cell mechanics, optical forces and — of course — photonic crystal fibres. A re-view of the current state-of-the-art in the respective fields is given.

Chapter 3 demonstrates that the unique properties of optofluidic PCFs canbe used for unchallenged long-range optical transportation of individual cells,see Figure 1.2 A. Moreover, shear forces on the cell surface provoke its deforma-tion. This causes changes in speed that are conveniently monitored using a non-imaging DOPPLER-velocimetric technique. We therefore propose that optofluidicPCFs will find use as a novel tool for the investigation of cellular mechanics.

In chapter 4, as illustrated in Figure 1.2 B, optofluidic PCFs are used as a novelintegrated analytical technology for photochemistry by microfluidic coupling ofan HC-PCF microflow reactor to a high-resolution mass spectrometer. Specifi-cally, potential drugs for photoactivated chemotherapy are studied, the screeningof which requires high handling and cost efficiency. Pushing the limits further, itmight eventually be possible to retrieve detailed information on their mechanismof action in situ.

Chapter 5 serves to give an outlook on what else optofluidic PCFs can do inorder to establish our fundamentally new approach towards biomedical researchin fibra4.

4I would like to suggest this latinisation as a laconic expression for our new kind of biophysical researchin the fibre, quite in analogy to: in vivo (in the living organism) – in vitro (in the test glass) – in silico (inthe silicon chip/computer).

Figure 1.2: Optofluidic photonic crystal fibres used as A) optofluidic cell guide (chapter 3)and B) photochemical microflow reactor (chapter 4).

4

22Optofluidic

Photonic Crystal Fibres

"... daß die wichtigsten Dinge durch Röhren gethan werden.Beweise erstlich die Zeugungsglieder, die Schreibfeder und unser Schießgewehr."

’Sudelbuch E’, Georg Christoph LICHTENBERG (1775)

This chapter serves to give a comprehensive introduction to the scientific back-ground of my work. It is arranged in two main parts.

The first part, section 2.1, deals with optofluidic technology and with opticalmicromanipulation techniques. It starts with the physical description of fluidicmicrosystems. After that, I will review the literature to determine the currentstate-of-the-art in optofluidic sensors for (bio)chemistry and provide the most im-portant physical parameters characterising photochemical reactions – and reac-tors. Turning from chemistry to biology, the intriguing field of single-cell biome-chanics is presented and its importance for the development of novel prophylac-tics and therapeutics of diseases is underlined. The contents of the subsequentsection are closely tied to this. After explanation of the physical foundations ofoptical forces, a review over the most recent approaches to merge optofluidicswith optical micromanipulation for cell investigation is provided.

As a promising new tool to advance the aforementioned scientific areas, inthe second part of this chapter, that is, section 2.2, I will introduce hollow-corephotonic crystal fibres. After a brief historical account, the fabrication process ofthese special kinds of microstructured glass fibres is explained. Then, I will elab-orate on their light guidance mechanisms so as to finally discuss their propertieswhen they are filled with aqueous medium – which render them outstandingoptofluidic channels for applications in both photochemistry and cell manipula-tion!

5

Chapter 2 OPTOFLUIDIC PHOTONIC CRYSTAL FIBRES

2.1 Optofluidics and Optical Micromanipulation in the LifeSciences

2.1.1 Basic Principles of Microfluidics

The field of microfluidics concerns hydrodynamic flow within small-scale bound-aries, with characteristic dimensions typically below 100µm. The fundamentalequation of hydrodynamics is the NAVIER-STOKES equation, which applies forNEWTONian, that is, perfectly inelastic fluids. It can be considered as the contin-uum version of NEWTON’s 2nd law of mechanics, m a = F , on a per-unit-volumebasis:

ρ ·(∂v

∂t+ (v ·∇)v

)︸︷︷︸

acceleration

=−∇p +η∇2v + fext︸︷︷︸forces

, (2.1)

where v denotes the velocity field, fext the external force density (in N m-3), p

is the pressure and ρ and η are the mass density and the dynamic viscosity ofthe fluid, respectively. The latter is a measure of flow resistance, therefore η∇2vdenotes the frictional force. The nonlinear term (v ·∇)v refers to convection. Inaddition, mass conservation is expressed via the continuity equation

∂ρ

∂t+∇· (ρ v ) = 0. (2.2)

From a more generalised perspective, microfluidic systems are conveniently char-acterised via a couple of dimensionless numbers, each expressing the relativeimportance of different physical phenomena [SQUIRES & QUAKE, 2005]. Threeof these are of importance for my work and therefore discussed in the following.

The REYNOLDS number: a measure for turbulenceFor any microfluidic system, the key parameter is the REYNOLDS number Re,which is defined as the ratio between inertial and viscous forces:

Re = inertial forces

viscous forces= ρ

ηv dH, (2.3)

where v is the mean fluid velocity and dH the hydraulic diameter of the chan-nel cross section. In case of a circular channel this is identical to the channeldiameter: dH = dc. For arbitrary cross sections it is given by

dH = 4Ac

Pwet, (2.4)

where Ac is the cross-sectional area and Pwet is the (wetted) perimeter of thechannel. In essence, an arbitrary channel can be treated as equivalent to a circu-lar channel of diameter dH [BEEBE et al., 2002].

6

Optofluidics and Optical Micromanipulation in the Life Sciences 2.1

Figure 2.1: Schematic drawing of parabolic Poiseuille flow in a circular tube.

For Re . 2100, viscous interaction between the wall and the fluid is strong.This means that the flow is laminar: fluid particles move along smooth paths inlaminas or layers; no turbulent flow occurs. Hence, the NAVIER-STOKES Equa-tion (2.1) reduces to the STOKES equation by neglecting the nonlinear term:

ρ∂v

∂t=−∇p +η∇2v + fext. (2.5)

Let us further assume a steady flow (v = 0), no specific external forces ( fext = 0)and a no-slip boundary condition (vboundary = 0). Then, from Equation (2.5), theflow in a circular channel with radius rc and length Lc can be derived to have aparabolic flow profile, also known as POISEUILLE flow (Figure 2.1 ):

v(r ) = 1

4η

|∆p|Lc

(r 2

c − r 2) , (2.6)

with a mean fluid velocity of

v = r 2c

8η

|∆p|Lc

. (2.7)

With this, HAGEN-POISEUILLE’s equation for the volume flow φ (in m3 s-1) reads[HAGEN, 1839; POISEUILLE, 1840]:

φ= v Ac =πr 4

c

8η

|∆p|Lc

. (2.8)

Substituting Equation (2.7) into Equation (2.3) with dH = 2rc yields

Re = 1

4

ρ

η2

r 3c

Lc|∆p|. (2.9)

For water ρ25° CH2O =103 kg m-3 and η25° C

H2O =10-3 N s m-2. For typical dimensions of amicrofluidic channel (e. g. rc =100µm, Lc =1 cm) and flow rates <1 cm s-1, Re

is <1 and therefore well in the laminar flow regime.

7


The PÉCLET number: a measure for mixing

The absence of turbulence in low-REYNOLDS-number systems has one challeng-ing drawback: mixing, which on macroscopic scales is typically enforced by tur-bulences, needs to be established by diffusion alone. Diffusion however is typi-cally a rather slow transport phenomenon due to the underlying randomness ofBROWNian motion.

It is well known from FICK’s 1st law [FICK, 1855] that the diffusion flux isproportional to the concentration gradient ∇C of a dissolved specimen:

j =−D ∇C . (2.10)

Taking into account mass conservation, the diffusion equation (also known asFICK’s 2nd law), which describes how diffusion changes a concentration distri-bution C over time, can be derived [FICK, 1855]:

∂C

∂t=−∇ j = D ∇2C , (2.11)

where the diffusion coefficient D of a molecular species is [EINSTEIN, 1905]:

D = kBT

6πη(T )rh. (2.12)

Herein, rh denotes the hydrodynamic radius of the dissolved molecular species,kB BOLTZMANN’s constant and T the temperature. Furthermore, EINSTEIN [1905]derived the characteristic diffusion time for a molecule to cover a distance ∆x:

τD = 1

2

(∆x)2

D. (2.13)

This means in order to travel across a microfluidic channel with dc = 100µm, asmall molecule of rh ≈ 10Å would need about 25 s.

Fortunately, at this point the story is not over yet. In microfluidics another trans-port mechanism is active: advection serves to transport a dissolved substance1

due to fluid bulk motion. In mathematical terms, a conserved scalar field, whichin our case is the concentration of the dissolved specimen C =C (r , t ), is advectedby a velocity field vector v and therefore follows a continuity equation. For in-compressible fluids (∇v = 0), the advection-diffusion equation reads

∂C

∂t=− v ·∇C︸︷︷︸

advection

+D∇2C . (2.14)

1or other conserved quantities, such as e. g. energy or enthalpy

8


Figure 2.2: A) Illustration of advection dominating the diffusion in a Poiseuille flow system.An initial distribution C (r,t) emerges into one in which the equal-concentrationlines, C (r,t+Δt), are curved. As a consequence, overall diffusion is enhanceddue to strong gradients at the edges. B) Diffusion enhancement factor Deff/D formolecules with D ≈ 10-10m2 s-1 as a function of Péclet number.

We find that the advection term establishes a coupling between the velocity andconcentration fields in case the concentration gradient is not totally perpendicu-lar to the streamlines, that is, they have some aligned components. Besides, theadvection-diffusion equation might be applied to find out about the evolution ofthe concentration gradient ∇C , yielding

∂

∂t∇C =−v ·∇(∇C )+D∇2(∇C ). (2.15)

Herein, the vector gradient term v · ∇(∇C ) describes the change of the concen-tration gradient in the direction of the velocity v . In other words, with thisterm being non-zero, the velocity field helps to increase concentration gradients– which is the main criterium to increase the diffusion flux (Equation (2.10)).

In conclusion, diffusive mixing might be enhanced if the velocity flow field isnot uniform but distorted such that at the edges sharp concentration gradientsoccur which locally enhance the diffusion. This phenomenon is called TAYLOR-ARIS dispersion [TAYLOR, 1953; ARIS, 1956]. Luckily, in microfluidic systems theparabolic velocity profile of POISEUILLE flow does exactly the job, as is illustratedin Figure 2.2 A.

A measure of the effectiveness of this phenomenon can be expressed in termsof the PÉCLET number P e, which weighs the relative importance of advection

9


with respect to diffusion in a fluidic system:

P e = advection

diffusion= v dH

D= |∆p|

4ηD

r 3c

Lc. (2.16)

Based on this, an effective diffusion coefficient Deff can be attributed to the system[TAYLOR, 1953]:

Deff = D · [1+β ·P e2] , (2.17)

where β is a geometry constant and equals 1/48 for circular channels. Thisrelationship is illustrated in Figure 2.2 B.

DEMELLO [2006] has summarised alternative solutions which enforce mixingin microfluidic systems with low P e numbers.

The capillary number: a measure for "fillability"The capillary number C a weighs the viscous forces with respect to the interfacialforces and might therefore be used to determine how easy it is to initially fill ahollow microfluidic channel:

C a = viscous forces

interfacial forces= η

σv = |∆p|

8σ

r 2c

Lc, (2.18)

where σ is the surface tension. For water/air this is 72 · 10-3 N m-1 at 25° C.

The filling process of a horizontally-kept small circular tube with a viscousincompressible liquid can be described by balancing the forces working on aliquid column. This can be derived from STOKES’ Equation (2.5) by multiplicationwith the volume V = Ac ·L =πr 2

c ·L:

d

dt

(πr 2

cρLdL

dt

)=−∆pπr 2

c︸︷︷︸pressure

+2πrcσ cosθcon︸︷︷︸capillary

−8πηLdL

dt︸︷︷︸friction

, (2.19)

where L is the infiltrated fibre length and θcon = 0° the contact angle for waterand silica, which is chosen as the tube material. Equation (2.19) can be furthersimplified:

d2

dt 2L2 +Y

d

dtL2 = X, (2.20)

where the constants have been summarised into X = 4σcosθ−2∆p rcρ rc

and Y = 8ηρ r 2

c.

This nonlinear differential equation can be solved analytically to yield [NIELSEN

et al., 2005]

L(t ) =√

X

Y2 exp(−Yt )+ X

Yt − X

Y2 . (2.21)

Figure 2.3 shows the infiltrated length of the holes in a hollow-core photonicbandgap fibre (see Figure 2.16 on page 42) as a function of time at an overhead

10


pressure of 5 bar. Clearly, the different holes fill at quite different rates, whichis reflected by their respective capillary numbers. The absolute figures are quitetiny due to the fact that fibres — per definition — come at extremely large as-pect ratios Ac/Lc. Given a fibre piece of 1 m (dashed line in Figure 2.3 ), we getC a(1µm) ≈2 ·10-8, C a(5µm) ≈5 ·10-7 and C a(19µm) ≈8 ·10-6. Complete fillingof this 1 m piece would take over one hour. This needs to be taken into consid-eration as a potentially limiting factor in the practical handling of liquid-filledfibres.

Figure 2.3: Infiltrated hole length with filling time for the different hole sizes (see labelling)found in a hollow-core photonic bandgap fibre (Figure 2.16 ).

2.1.2 Optofluidic Sensors and Reactors for Biochemistry

Sensors for BiochemistryBiological and biochemical samples are necessarily aqueous-medium based. Sens-ing of chemical reactions or certain species might be based on several opticalphenomena. These include simple absorption, luminescence (both fluorescenceand phosphorescence), the RAMAN effect and surface plasmon resonance, amongothers.

With special focus on the field of optofluidics, evanescent wave guiding schemeshave found widespread use in (bio)chemical sensing. This relates to the fact thatthey mark the historical transition point from all-solid-state integrated opticaltechnologies to optofluidics. Since the optical field is established in a solid-statewaveguide and penetrates only a few 100 nm into a neighbouring fluidic chan-nel, evanescent sensing methodologies are especially useful to probe interfa-cial effects. They are therefore also a suitable framework for surface plasmon

11


resonance schemes [HOMOLA et al., 1999]. In order to receive sufficient sig-nal strength despite the small interaction, long interaction path lengths needto be exploited. Therefore, either planar waveguides [LUKOSZ, 1991] or modi-fied conventional fibres [HENRY, 1994] were regularly used. Solid-core photoniccrystal fibres and microstructured fibres with a suspended-(solid-)core advancedthe field, because they naturally come with infiltratable holes in direct neigh-bourhood of the core [AFSHAR V. et al., 2007; RUAN et al., 2007; EUSER et al.,2008a; FRAZÃO et al., 2008]. In this context, a completely liquid-filled hollow-core photonic crystal fibre (HC-PCF) has even been used for multiplexed evanes-cent wave sensing by coupling light into the glass structure instead of the core2

[JENSEN et al., 2004; RINDORF et al., 2006]. In this way, the authors exploitedthe large surface-to-volume ratio of the cladding channel holes for DNA-bindingassay studies. For further information on and applications of evanescent wave(bio)chemical sensing I would like to refer to [CHEN, 2010; PINTO & LOPEZ-AMO,2012] and the citations found therein.

Clearly, for reactions not taking place on surfaces, a larger, that is, a directlight-sample overlap is highly advantageous. In the framework of biochemicalsensors, hollow-core waveguides have for instance been used for simultaneousfluorescence excitation and detection. Sample-filled HC-PCFs have been usedfor fluorescence sensing both in core-filled HC-PCF [SMOLKA et al., 2007] andin completely filled HC-PCF [WILLIAMS et al., 2013]. While in the former, thefibre works by total internal reflection guidance and is highly multimode, thelatter preserves the PC-cladding-based fundamental-mode guidance. With anabsolute detection efficiency on the order of only an attomole of fluorophore,this recent approach outperforms the former by several orders of magnitude. Inaddition, WILLIAMS et al. [2013] demonstrated the possibility to perform intra-fibre fluorescence lifetime measurements.

Reactors for PhotochemistryBesides photophysical effects, such as fluorescence, a complete overlap of lightand sample is even more important if certain molecular species are not only tobe detected but if the sample container is used to initiate photochemical reac-tions. The field of photochemistry comprises several kinds of chemical reactionswhich take place upon photonic excitation of higher energetic states in certainspecies. Currently, the following are intensively being investigated: photodecom-position and activation of drugs for cancer therapy [ROBERTSON et al., 2009;FARRER et al., 2009] (see chapter 4), radical-induced photopolymerisation fororganic synthesis [HOFFMANN, 2008] and photoelectric effects and charge trans-port mechanisms for organic photovoltaics [SPANGGAARD & KREBS, 2004].

2that is, against its original designation

12


Figure 2.4: Important parameters in a photochemical reactor (here: standard cuvette).

What are the requirements for a perfect reactor in the study of solution-basedphotochemical reactions? Figure 2.4 shows a schematic of such a reaction vessel,a conventionally used cuvette.

For initiation of any photochemical reaction the sample needs to absorb light.The extent of photon absorption depends both on externally applied as well as onmolecule-intrinsic parameters. External factors are the power P0 and providedwavelength λ0 of a given light source (blue), as well as the reaction volume(yellow) which might (partly) be determined by the geometrical extents of thereaction vessel, most importantly its length L. The irradiance or intensity ofthe light field is then given via I0 = P0/A, where A is the effective illuminatedarea, typically corresponding to the beam cross section. Besides, every molecularspecies has an intrinsic ability to collect impinging photons of certain frequencyν0, called absorption cross section.

All these parameters are included in the well-known LAMBERT-BEER law, whichgives the absorbance of light by a bulk amount of dissolved sample (pink) in areaction vessel. It might be written down in terms of e- or 10-base:

Abs′ =− ln

(Iin

Iout

)=σN L =α′ L, (2.22a)

Abs =− log

(Iin

Iout

)= εC L =αL, (2.22b)

13


where σ is the absorption cross section per absorbing particle (in cm2) and N

is the corresponding number of absorbing particles per unit volume (in cm-3).The absorption cross section’s decadic counterpart on the basis of molar concen-tration C is the decadic extinction coefficient ε (in L mol-1 cm-1). α and α′ are theabsorption coefficients in 10- and e-base, respectively. With C = 1000 cm3

LNNA

, whereNA is AVOGADRO’s constant (in mol-1), we obtain the mutual relationships

α′ = ln 10 ·α ⇐⇒ α= log e ·α′, (2.23a)

σ= ln 10 · εC

N= ln 10 ·1000

cm3

L

ε

NA. (2.23b)

Please note that Equations (2.22) do not explicitely include the intensity decayintrinsic to the light source (e. g. by diffraction).

The consideration of absorption alone, however, is not enough. The absorp-tion of a photon leads to an excited state in the molecule, which can then takeseveral different decay routes leading to a range of different photochemical orphotophysical processes. Each of these processes has a certain probability tooccur, which can be expressed via its respective quantum yield Φ. It states thefraction of the absorbed photons which eventually lead to the reaction of in-terest. The two important intrinsic photochemical parameters can be mergedinto a single parameter Ψ, which I call the molecule’s photochemical conversionefficiency3:

Ψ= ε ·Φ. (2.24)

The irradiance denotes the light energy per time and area in units of W = J s-1 m-2.The energy of a single photon with frequency ν0 is given by multiplication withPLANCK’s constant h: E1 phot = hν0. Hence, the number of photons correspondingto an irradiance I within a time period dt and in an area A can be obtained from

Nphot =A

hν0I dt . (2.25)

The number of photons absorbed by a single molecule can be simply retrievedby substituting A in Equation (2.25) with the absorption cross section σ:

Nabs =σ

hν0I dt . (2.26)

The decrease in the number of analyte molecules due to photochemical conver-sion into the product is therefore:

dN =−ΦNabs N (t ), (2.27)

3For a fluorescent molecule this is referred to as brightness.

14


where N (t ) denotes the current number of analyte molecules.

Using Equation (2.23b) this is transferred into an expression in concentration:

dC =−Φ ln 10 ·1000cm3

L

ε I

NA hν0dt︸︷︷︸

Nabs

·C (t ). (2.28)

In a 1st order chemical reaction the kinetics follow the relation(dC (t )

dt

)=−k ·C (t ) ⇐⇒ C (t ) =C0 ·e−k t , (2.29)

where the reaction rate k is a constant. By comparison of Equation (2.28) withEquation (2.29) and using Equation (2.24), we find that the rate constant can beexpressed as:

k = ln(10) ·1000cm3

L

1

NA·Ψ · I

hν0∝Ψ · (I ·λ0). (2.30)

In addition, LAMBERT-BEER’s law in Equations (2.22) reminds us that, regardingthe sensitivity of analysis and total turnover of a reaction, the effective lengthover which the reaction is allowed to occur is also important.

Abstracting our view from the specific application of photochemistry to generalkinds of light-matter interaction, we find that the overall strength of a light-matter interaction process is given by [OKAMOTO, 2006, Chapter 5.1]∫

I (z)dz. (2.31)

With this the figure of merit for light-matter interaction can be defined as [BEN-ABID et al., 2002a]:

Ξ := λ0

P0·∫

I (z)dz, (2.32)

which also takes into account the parameters specific to the light source.

Clearly in a cuvette, the overlap of the light field with the sample is 1, whilein evanescent schemes it is only on the order of 0.1. However, this number doesnot take into account a situation in which the irradiated volume (yellow outlinein Figure 2.4 ) is much smaller than the actual sample volume (pink). This meansthat the absorbing particles randomly diffuse in and out of the illuminated spot,which is extremely disadvantageous for reactions with a small quantum yield orfast reversible reactions. Increasing the beam size is however not a good idea,because this would decrease the irradiance.

Hollow-core optofluidic waveguides provide an efficient way around this prob-lem. They will be discussed in detail in section 2.2.4, once a more profoundunderstanding of hollow-core photonic crystal fibres has been obtained.

15


Figure 2.5: A) Schematic representation of the three-dimensionally linked cytoskeleton of aeukaryotic cell (mesh much denser in reality, organelles omitted for simplicity).B) Close-up of the main three cytoskeletal filaments.

2.1.3 Force Transducers and Sensors for Cell Mechanics

The field of cell mechanics studies how cells4 move and interact, but also howthey sense, generate and respond to mechanical forces. Intriguingly, a cell’s me-chanical behaviour is inherently coupled to its physiology [GILLESPIE & WALKER,2001; BAO & SURESH, 2003; JANMEY & WEITZ, 2004]. Therefore, the cell’selasticity reflects its state of health, as is well documented for diseases like blooddisorders [MOHANDAS & GALLAGHER, 2008], asthma [FABRY & FREDBERG, 2007]and cancer [SURESH, 2007]. Deeper insights into the phenomenon might hencelead to the development of drugs which could restore healthy functioning in dis-eased cells [VANAPALLI et al., 2009]. This bears an enormous potential for a newera of prophylactic and therapeutic agents.

The reason for this mechano-chemical coupling is that on the microscopic levelthe boarders between physics, chemistry and biology vanish. The cell’s internalscaffolding structure is called cytoskeleton [FLETCHER & MULLINS, 2010], illus-trated in Figure 2.5 A. It consists of a dense, meshlike superstructure of differentkinds of protein filaments (Figure 2.5 B). These macromolecular strands, as wellas their interconnections, are continuously assembled and disassembled basedon the mechanical and chemical cues the cell experiences. As a key feature ofliving organisms, the dynamics obey the laws of non-equilibrium thermodynam-ics [POLLARD & BORISY, 2003a,b].

Materials with such a complex microstructure typically show a viscoelastic me-chanical behaviour. This means that depending on the time scale (or frequency)of an external perturbation, the material might either act more like a liquidor more like a solid. Their theoretical description is hence typically performedbased on mechanical equivalent circuits composed of NEWTONian dashpots, re-

4I only refer to cells of higher organisms, not to microbial cells such as bacteria, fungi or yeasts.

16


flecting perfect liquid behaviour, and HOOKEan springs, corresponding to perfectsolid behaviour [KOLLMANNSBERGER & FABRY, 2011]. This results in a time-/frequency-dependent stress-strain relation, based on which interesting phenom-ena appear: creep is the increase in strain under constant stress, whereas stressrelaxation is the decrease in stress under constant strain.

Taking all of the above into consideration it does not astonish that, to date,it is not possible to fully describe and understand the biomechanics of a cell[HOFFMAN & CROCKER, 2009]. Its behaviour is governed by the accumulatedcomplexity of non-equilibrium thermodynamics [BURSAC et al., 2005], viscoelas-ticity [KASZA et al., 2007] and of mesoscopic scales [PRAPROTNIK et al., 2008],which are, each on its own, challenging to theoretically simulate or model.

Connected to this intrinsic intricacy, in order to obtain deeper cellbiologicalinsights it is necessary to retrieve data on the single-cell level [DI CARLO & LEE,2006]. In this way the loss of information resulting from ensemble averagingin bulk approaches is circumvented. This is crucial because cells — even fromthe same strain — are highly individual with respect to their morphology andphysiology. However, the detailed information obtained from single-cell sensi-tive methods needs to be associated with high-throughput analysis to achievestatistical relevance [ANDERSSON SVAHN & VAN DEN BERG, 2007], which is oftena challenging task.

A variety of different measurement techniques to probe and manipulate sin-gle cells at forces < 1 pN and displacements < 1 nm is available. The numberof review articles on this topic reflects the enormous research efforts currentlyundertaken, see for example [BAO & SURESH, 2003; MOFRAD, 2009; HOFFMAN

& CROCKER, 2009; KOLLMANNSBERGER & FABRY, 2011]. Basic means of probingcellular mechanics comprise tension (uniaxial or biaxial), compression, shear,hydrostatic pressure, bending and twisting, as well as combinations thereof.These techniques either examine adherent cells or cells in suspension. As op-posed to blood cells, which are naturally in a suspended state, tissue cells dependmuch more on their cytoskeleton, since it is their natural designation to undergostrong cell-cell as well as cell-extracellular matrix5 interactions. However, alsonaturally adherent cells might be transferred into suspended state for investi-gation. The mentioned techniques have in common that it remains difficult toquantify accurately the forces applied to or generated by cells, and that it is notclear how forces are distributed between various subcellular structures. As aconsequence, different experimental techniques yield very different viscoelasticparameters, even for the same cell type and under similar conditions.

Several of these techniques make use of optical forces which provide a con-tactless way of investigation [ZHANG & LIU, 2008; RAMSER & HANSTORP, 2010].

5The extracellular matrix fills the interspaces between united cell structures and is mainly composed offibrous structures (collagens), as well as densely packed (glyco)polysaccharides and glycoproteins.

17


Being a vast field of research on its own, the foundations of optical forces shallbe discussed in detail in the upcoming section.

2.1.4 Optical Micromanipulation of Particles and Cells

A photon has no mass but it carries both energy and momentum:

Ephot = hν (2.33a)

pphot =h

λ= Ephot

c0. (2.33b)

While Equation (2.33b) refers to the photon momentum in vacuo, it had been along-standing question if a photon when entering a dielectric medium of refrac-tive index n would lose or gain momentum – the first proposed by ABRAHAM

[1909], the latter by MINKOWSKI [1908]:

pABRAHAM = h

nλ, (2.34a)

pMINKOWSKI = n h

λ. (2.34b)

Even more confusingly, over the years several different experiments had beenperformed and would support either one or the other expression of photon mo-mentum [LEONHARDT, 2006]. Only recently, BARNETT [2010] could resolve thecentury-long "ABRAHAM-MINKOWSKI dilemma". He found that the two versionsdo not contradict each other but are rather due to different underlying mathe-matical approaches. ABRAHAM’s version refers to the kinetic momentum, that is,the mechanical punch of the photon regarded as a particle. MINKOWSKI’s versioninstead refers to the canonical momentum, which is somewhat more subtle, buttied to the wave nature of the light. PFEIFER et al. [2011] have rediscussed BAR-NETT’s findings with special respect to what is presented in the following sectionsand explain why for (most of) the applications discussed below MINKOWSKI’sviewpoint is of more practical use.

Optical ForcesWhere there is a change in momentum, there is a force:

F = dp

dt, (2.35)

but the overall momentum of a single photon is small and so is the associatedforce. Nevertheless, if there are many photons, optical forces might indeed haveobservable effects onto the matter they are interacting with. In fact, in 1871MAXWELL assumed that a "flat body exposed to sunlight would experience [...]pressure on its illuminated side, and would therefore be repelled from the side on

18


which the light falls." (recited from [MILONNI & BOYD, 2010]). The first un-ambiguous corroborating experiments were performed by LEBEDEW [1901] andNICHOLS & HULL [1903a,b] – a noteworthy achievement given that forces on theorder of nN were acting on a macroscopic piece of mirror 13 mm in diameter.

Despite for decades being a well-known phenomenon, useful applications ofoptical forces had to await the advent of the laser as a bright and directed lightsource. In 1970, Arthur ASHKIN demonstrated that in a weakly-focused laserbeam the momentum transfer of the photons when being reflected and refractedat the boundaries of a transparent dielectric particle (Figure 2.6 A) is large enoughto induce a directed movement. The total optical force F net can be separated intoa lateral gradient force F grad ∝∇I that tends to pull the particle into the centreof highest intensity, and into an axial scattering force F scat ∝ I that pushes it inpropagation direction of the beam, see Figure 2.6 B [ASHKIN, 1970].

This is both true for very small dielectric particles, such as atoms and molecules,but also for beads several micrometres in size. Theoretical description, however,requires different proceedings [NEUMAN & BLOCK, 2004]. In the general (ormesoscopic) case, full LORENTZ-MIE scattering theory has to be applied. Yet, inthe case of very small particles dp ¿ λ RAYLEIGH scattering (dipole) theory suf-fices and in the case of rather large particles dp Àλ a ray optics model is typicallyapplied as in [ASHKIN, 1992]. The applications discussed in this thesis are onlyconcerned with the latter case.

Optical TrappingIn his original paper ASHKIN [1970] also demonstrated that stable optical trap-ping of microparticles in all three dimensions of space is possible by balanc-ing the axial scattering forces of two weakly-focused counter-propagating beams[ASHKIN, 1970], see Figure 2.6 C.

Later, he and his co-workers found that particle trapping is even more easilyestablished by using a single tightly-focused laser beam [ASHKIN et al., 1986],Figure 2.6D. This is because with increasing numerical aperture of the focus-ing lens, the gradient force develops a component in anti-beam direction whicheventually, that is at NA≈1, counterbalances the scattering force. This facili-tates a more versatile trapping scheme for contact-free and handy particle mi-cromanipulation. It soon became evident that even single viruses and bacteria[ASHKIN & DZIEDZIC, 1987], as well as mammalian cells [ASHKIN et al., 1987]can be trapped and micromanipulated with these optical tweezers, despite theirsmall refractive index difference to the surrounding aqueous medium. For air-way smooth muscle (tissue) cells the refractive index has been determined to benASM =1.36 [CURL et al., 2005], while it is slightly higher for red blood cells,which are densely packed with haemoglobin, nRBC =1.40 [GHOSH et al., 2006a].

19


Figure 2.6: Optical forces facilitate the micromanipulation of small particles. A) When aray of light hits a translucent microparticle it is partially reflected and refractedat the interfaces to the surrounding medium. The corresponding changes inoptical momentum are transferred to the particle. B) Optical propulsion in aweakly-focused Gaussian beam [Ashkin, 1970]. C) 3D trapping in a counter-propagating dual-beam trap [Ashkin, 1970]. D) 3D trapping in a single, tightly-focused Gaussian beam [Ashkin et al., 1986], known as optical tweezers.

Unfortunately, ASHKIN & DZIEDZIC [1987] noticed that bacterial cells wouldquickly die in the trap due to radiation damage of the 100 mW argon-ion laserbeam at 514 nm wavelength. Therefore, optical traps are nowadays typically op-

20


erated in the NIR region of the spectrum, where biological tissue absorbs leastand water has a local absorption minimum. However, care has to be taken at thelower edge of the NIR range, that is, around 800 nm. At locally high irradiancesin the focus of ∼MW cm-2 two-photon processes take place even in a continuouswave (cw) beam, leading to UVA-like radiation damage. This has been demon-strated for Chinese hamster ovary (CHO) cells, for which cell viability decreasedwhen trapped for a few minutes at 70 mW power [KÖNIG et al., 1995]. In con-trast, in a 1064 nm cw-beam, CHO and sperm cells survived trapping powers upto 400 mW for comparable periods of time [LIU et al., 1996].

Apart from radiation damage, the absorption of IR radiation leads to heating.According to PETERMAN et al. [2003] and LIU et al. [1996], the temperaturein the vicinity of a trapped particle or cell increases by about 10 K per Watt ofoptical power at 1064 nm wavelength (cw). For conventional single-beam trap-ping heating is therefore not supposed to diminish cell viability. Still, tempera-ture effects have to be taken into account in trapping studies which investigatetemperature-dependent parameters. Moreover, the amount of heating needs tobe evaluated for novel trapping schemes.

Besides the investigation of fundamental interactions of (multiple) micropar-ticles with complex optical force fields [GRIER, 2003; DHOLAKIA & ZEMÁNEK,2010; PADGETT & DI LEONARDO, 2011; DHOLAKIA & CIŽMÁR, 2011; PADGETT

& BOWMAN, 2011], optical tweezers have become an indispensable tool in bi-ological research [SVOBODA & BLOCK, 1994; STEVENSON et al., 2010; FAZAL &BLOCK, 2011]. Within this framework, it is of increasing interest to embed op-tical traps on microfluidic chips. For instance, cells might be optically sorted byinstalling optical traps [BUICAN et al., 1987; CRAN-MCGREEHIN et al., 2006], aholographic optical lattice [MACDONALD et al., 2003] or an active optical switch[WANG et al., 2005] perpendicular to a fluidic channel. As mentioned earlier,the investigation of cellular mechanical properties is of particular importance[VANAPALLI et al., 2009; KIM et al., 2009]. NÈVE et al. [2008, 2010] for instanceheld single bone, cartilage and muscle cells in place with optical tweezers in arectangular microfluidic channel while applying different flow/shear rates andmeasuring their deformational response. For comparison, they also stretchedthe cells by moving apart two separate optical tweezers beams which had beentightly focused at opposite ends of the cell body. This is a simplified version ofan original approach by HÉNON et al. [1999], who attached two microbeads tothe cell surface which acted as optical handles for the two tweezers beams. Inthis indirect way, mechanical stresses act primarily locally, but can be transferredto the cell body via membrane proteins that link to the cytoskeleton. Comparedto NÈVE et al. [2010], KANETA et al. [2001] demonstrated a more homogenousway to apply shear stresses around single cells. They directed a weakly-focused

21


laser beam along the axis of a 200µm-bore capillary6 and exploited the gradientforce to hold red blood cells in the centre while simultaneously applying a weakcounterflow. In this way, shear stresses are more symmetric and elasticity mea-surements not disturbed by the strong local forces of the optical tweezers. Theshear-induced elongation of the cells was measured by microscope imaging.

Optical StretchingWithin the cell mechanics framework, GUCK et al. [2000, 2001] demonstratedthat dual-beam optical traps provide a useful tool for a more direct and homo-geneous non-contact all-optical probing. They created a trap from the divergingfields of two standard optical fibres facing each other (Figure 2.7 A), as originallydemonstrated by CONSTABLE et al. [1993]. Later they integrated the scheme intoa microfluidic chip, perpendicular to a fluid channel [LINCOLN et al., 2004]. Ifa suspended cell is trapped and the power is ramped up to ∼1 W per beam, thecell stretches as a result of optical surface forces. The fact that soft dielectric parti-cles are stretched rather than squeezed (which one might expect intuitively) canbe rationalised by the application of MINKOWSKI’s momentum (Equation (2.34b)).This circumstance is illustrated for a single paraxial ray in Figure 2.7 B. When hit-ting the front surface of the cell, an incident ray (red, "incident") is refracted7 intothe cell’s interior (red, "refracted 1"), whereupon it gains momentum according toMINKOWSKI, because p ∝ n and ncell > nmed. In effect, the surface gains momen-tum in opposite direction, that is, outwards (blue). When this ray encounters therear cell surface, it is refracted back into the surrounding medium (red, "refracted2") and loses momentum so that here, as well, the optical force acts outwardly(orange). As a result, the cell stretches. The total scattering force acting on thecentre of mass is given by integration of the surface forces over the whole solidangle. Total scattering forces from a single beam onto mouse embryonic fibrob-last cells (BALB 3T3) have been determined to be ∼2 pN per 100 mW of opticalpower [GUCK et al., 2001].

In subsequent studies, GUCK and co-workers demonstrated the application oftheir optical cell stretcher as a novel, deformability-based screening method formalignant transformation and metastatic competence of abnormal cells [LIN-COLN et al., 2004; GUCK et al., 2005; REMMERBACH et al., 2009].

While optical cell damage is not supposed to occur in this diverging beamgeometry, heating needs to be taken into account. In the centre of the opti-cal cell stretcher it has been measured to be 13 K W-1 at 1064 nm wavelength[EBERT et al., 2007], which — given the higher optical powers — leads to alarger increase in temperature as compared to single-beam trapping. Trappingtime periods of only a few seconds seem short enough to avoid damage of the6Please note that here, the capillary was not intended to serve as a waveguide but only as a circular

microfluidic channel into which a (quasi) free-space beam is focused.7A minor portion of the ray is reflected at each surface, which is neglected in this discussion.

22


Figure 2.7: Optical cell stretching. A) Schematic of the optical cell stretcher, an on-chip fibre-based dual-beam trap with diverging beams [Guck et al., 2000, 2001; Lincolnet al., 2004]. B) An incident ray is refracted at either surface of the cell. Accordingto Minkowski’s momentum the local momentum transfers (blue and orange)act to stretch the cell.

cell, however, elastic parameters are expected to depend strongly on tempera-ture. This fact needs to be taken into consideration if the obtained parametersare to be compared with other techniques.

Optical GuidanceIn a weakly-focused beam as in ASHKIN’s first experiment [ASHKIN, 1970] (Fig-ure 2.6 B), optical particle guidance is strongly limited, because optical forces areonly strong enough if the beam spot is not larger than a few times the parti-cle’s own size. In this case the propagation distance can be approximated bytwo times the RAYLEIGH length which does not exceed µm length scales. Hence,it would be highly advantageous if one could find a way to further extend thepropulsion distance and thereby to "magically" propel particles by light. On-chiptransport through stationary fluids or against weak flows would then be possible.This might serve as a tool for directed writing of single cell layers to establishwell-defined cell-cell and cell-surface interaction patterns [ODDE & RENN, 2000].Moreover, suspended particles can be separated and sorted by means of their dif-ferent optical and fluidic forces when pushed through a liquid by a light beam– a method termed optical chromatography [IMASAKA et al., 1995]. For cells,sufficient optical separation in an on-chip free-space beam has so far only beenachieved by tagging a portion of the cell population with internalised beads toimprove the refractive index contrast [ASHOK et al., 2010].

Patently, waveguides are superior to free-space beams with respect to achiev-able propagation distance. Optical propulsion through water was first demon-

23


strated on evanescent-wave planar waveguides by KAWATA & TANI [1996], whodemonstrated the guidance of 1-5µm polystyrene beads at up to 14µm s-1 ve-locity and over maximum distances of ∼1 cm (P =80 mW). Optimisation of thetechnique yielded only slight improvements: maximum velocities of 28µm s-1

for particles of 3µm in size (P =53.5 mW) were reached [SCHMIDT et al., 2007].GRUJIC & HELLESØ [2007] have built a counter-propagating trap based on evanes-cent waveguides, which they could also actuate as a long-range trap by variationof the relative beam intensity. Interestingly in this configuration, the axial trap-ping efficiency depends on the decay of the beam intensity along the axis, whichis simply given by the waveguide loss. The lower the loss, the more shallow thetrapping potential along the beam axis.

Clearly, in evanescent waveguides the particle propulsion takes place on a sur-face, because the region of highest intensity lies within the solid core and there-fore the gradient force acts to pull the particle towards the surface. In mostcases, specifically by usage of biological saline buffers, surface forces will thenoverweigh the optical scattering forces, impeding any forward propagation ofthe particles. This circumstance becomes even worse for cells, because they arenaturally equipped with surface receptors to strongly adhere to surfaces. There-fore, proof-of-principle demonstrations constrict themselves to red blood cells,which as suspended cells come with less (sticky) surface receptors [GAUGIRAN

et al., 2005]. A maximum red blood cell speed of 6µm s-1 at 60 mW incou-pled power and a maximum propagation distance of 0.1 mm were demonstrated[AHLUWALIA et al., 2010].

In hollow waveguides cells can be guided at the centre of the waveguide mode,away from the surface8. The first microparticle experiments in a hollow waveg-uide were performed by RENN et al. [1999]. The authors have for instance pro-pelled a 7µm large polystyrene sphere through a water-filled 20µm-bore capil-lary at a maximum speed of 300µm s-1 (P =220 mW). Only one year later, themethod was applied to cells. ODDE & RENN [2000] propelled chicken embryonicspinal cord cells, about 9µm in diameter, through a liquid-filled 100µm-borecapillary over a maximum distance of 7 mm, which was limited by waveguideloss. Velocities were on the order of 50µm s-1 at a laser power of 450 mW. Themean optical force was estimated to be around 5 pN. This has been the mostpromising cell guidance experiment in the literature so far.

Guidance experiments using other types of optofluidic waveguides have onlydemonstrated the guidance of beads. MANDAL & ERICKSON [2007] have guidedmultiple 3µm polystyrene beads through the selectively filled core of a hollow-core photonic crystal fibre, which essentially results in a highly multimode total

8Please note that the different kinds of hollow optofluidic waveguides will be discussed with respect totheir physical aspects in full detail later in section 2.2.4. Here, I will therefore only quickly refer to theapproaches taken and achievements reached.

24


internal reflection waveguide. At a power of 210 mW, they reached a maximumdistance of 2 cm. Another popular on-chip optofluidic waveguide is a so-calledARROW, with which also cm-distance propagation of beads could be obtained[MEASOR et al., 2008]. Based on this principle KÜHN et al. [2009a] have also in-troduced a dual-beam particle trap. In [KÜHN et al., 2009b], the authors appliedtheir method to accumulate multiple fluorescent particles in the central trappingposition so as to increase the overall detection signal.

A big step forward was taken by introducing hollow-core photonic crystal fi-bres to the field, because they provide diffractionless low-loss guidance of a sin-gle waveguide mode. This provides long propulsion distances and a stable lateraltrapping in the centre of the hollow core. In the initial demonstration by BEN-ABID et al. [2002b], an air-filled PCF9 served to propel a 5µm polystyrene beadthrough a 20µm-core over distances of 15 cm at speeds of around v =1 cm s-1

(P =80 mW). Using PCFs also allowed particle guidance around bends for thevery first time. After his move to the MPL in Erlangen, Philip RUSSELL resumedthis interesting field of research. Under his supervision, my colleagues TijmenEUSER and Martin GARBOS have introduced the outstanding particle guidancecapabilities of hollow-core PCFs into biologically-oriented optofluidics by fillingthem with water [EUSER et al., 2009, 2010]. In liquid, propagation distanceson the order of 0.5 m are achieved on a regular basis, using optical powers of afew 100 mW only. Depending on the actual size of the particle, velocities reachinto the 1 mm s-1 regime. Due to the well-determined optical forces of the singlefibre mode, they were also able to accurately measure the hydrodynamic forcesby counterbalancing the optical propulsion with a pressure-driven counterflow[GARBOS et al., 2011a].

With the study presented in chapter 3 we took the step to real biological ap-plications. We achieved long-range laser propulsion of single red blood cells inoptofluidic photonic crystal fibre over a length of up to 24 cm. Moreover, wefound that this might be used as a new tool for biomechanical studies of individ-ual cells.

At this point I would like to ask the reader for some more patience about theachievements in optofluidic light-matter interaction which I had the chance totake part in. In order to fully appreciate our novel techniques for biomedicalresearch, it is inevitable to gain a deeper insight into the physical foundations ofhollow-core photonic crystal fibre – a truly outstanding waveguide.

9Recent advances in air-filled HC-PCF particle guidance can be found within the Ph.D.-project of my col-league Oliver SCHMIDT, see for example [SCHMIDT et al., 2012a,b].

25


2.2 Hollow-Core Photonic Crystal Fibres

Photonic crystal fibres (PCFs) integrate the concept of photonic crystals (PCs)into the area of fibre optics to create a fundamentally new kind of optical waveg-uide. In marked contrast to conventional fibres, PCFs facilitate the engineeringof certain waveguiding properties by tuning of the microstructure’s geometri-cal parameters. Most important for the work presented in this thesis are thehollow-core versions of PCF. Hollow-core photonic bandgap fibres (HC-PBGFs)and kagomé-structured fibres (KFs) provide diffractionless guidance of brightoptical modes along their hollow core, that is, in air. With their introductionabout two decades ago, completely new opportunities for the investigation andexploitation of light-matter interactions were established.

2.2.1 Historical Accounts

Optical fibres...Conventional optical fibres are flexible waveguides made of glass or polymer witha thickness on the order of a human hair10, that is, a few to several tens of mi-crometres. Light is guided by means of total internal reflection (TIR). The possi-bility to guide light by TIR has already been known in the early 19th century andwas first demonstrated in a jet of water by COLLADON [1842], see Figure 2.8 A.

The first practical application of optical fibres was in form of fibre bundles formedical imaging or endoscopy, which kept advancing the field over the decades[LAMM, 1930; HOPKINS & KAPANY, 1954; HIRSCHOWITZ et al., 1957].

Even after the advent of lasers in the early 1960s, the optical losses of glasseswere still too high for telecommunication purposes. For this exact reason, peopleproposed using liquid-filled waveguides for telecommunication instead of all-

10Speaking of hair, WELLS [1989] reported in Nature that grey hair would act as a — rather poor — opticalfibre and proposed to investigate ”[w]hether light transmission down hairs affects skin and hair”...

Figure 2.8: A) A beam of light is focused by a water-filled round bottom flask and onto anopening on the other side. The light is guided along the water jet by total internalreflection. B) Schematic drawing of a conventional glass fibre.

26

Hollow-Core Photonic Crystal Fibres 2.2

solid fibres [PAYNE & GAMBLING, 1972; STONE, 1972]. It were KAO & HOCKHAM

[1966] who theoretically analysed which structural and material requirementsneeded to be fulfilled to reduce losses in silica to levels below 20 dB km-1. Intheir paper they stressed the need for purified silica glass, based on which thefirst low-loss optical fibre was eventually realised by Corning researchers KECK

& SCHULTZ [1973]. To circumvent frustrated TIR and scattering from scratches,fibres composed of a guiding core and a cladding of lower refractive index aretypically used. For illustration see Figure 2.8 B

Even though still being regularly used for medical imaging purposes and hav-ing found application in laser technology and for chemical and environmentalsensors, the most important application for optical fibres to date is telecommu-nication. Fibre-based networks started to replace copper-wire-based networks inthe 1980s. This was the logical consequence of the several orders of magnitudelower losses and higher bandwidths (i. e. data rates), connected to the high op-tical frequencies and electromagnetic interference resistance of optical signals.State-of-the-art silica fibres have losses below 0.2 dB km-1 at the minimum ab-sorption wavelength of 1550 nm.

Despite the unbroken need to expand fibre-based telecommunication networks,the underlying technology has basically come to a point of stagnation — al-though at a high performance level — at which only minor quantitative im-provements in the existing and even less novel applications can be expected[RUSSELL, 2003]. To expand the modalities of fibre operation with special fo-cus on the emerging field of integrated photonic circuits, a fundamentally newmechanism of light guidance was inevitable.

...meet photonic crystals...Photonic crystals (PCs) are periodically structured media with the lattice constantbeing on the order of the effective wavelength of light11 [YABLONOVITCH, 1987;JOHN, 1987] (see Figure 2.9 A). When talking about PCs, people are typicallyreferring to man-made structures for photonic applications, but they have alsobeen found in nature12 [VUKUSIC & SAMBLES, 2003, 2004].

Photonic crystals’ analogy to crystalline solids is not restricted to their name.In a simple picture, the function of PCs can be descriptively explained basedon BRAGG’s law, that is, the reflection of X-rays in ion crystals. These representperfect 3D periodic structures with lattice constants dq corresponding to wave-

11The term light typically refers to electromagnetic waves from the ultraviolet to the infrared.12Biological examples include the iridescent wings of several butterfly species (such as the ones of

genus Morpho), the feathers of the blue peafowl (Pavo cristatus), the armors of certain bugs (such asChrysochroa vittata or Chrysina resplendens), the colourful spines of the sea mouse (genus Aphrodita) orthe scales of the (living) fossil fish coelacanth (genus Latimeria).

27


Figure 2.9: A) Schematic illustration of a "woodpile" 3D photonic crystal, similar to the onereported by Lin et al. [1998]. B) Photonic-crystal slab waveguide with a linedefect (missing row of holes) along which light propagation takes place. Themembrane is surrounded by air from both sides to confine the light by TIR invertical direction [Sugimoto et al., 2004].

lengths in the X-ray radiation regime λBRAGG,q:

mλBRAGG,q = 2dq sin θBRAGG,q , (2.36)

where m ∈ N, q denotes a certain set of lattice planes and θBRAGG,q is the anglebetween the incident ray and the scattering planes. Hence, for BRAGG reflectionto occur d needs to be on the order of λBRAGG. This is a quasi discrete condition.The reason is that for X-rays refractive indices are close to 1 for almost all ma-terials, so that the reflections at each crystal plane are very weak and thereforeconstructive interference requires an accurately determinate couple (λBRAGG,d)q .

In contrast to X-rays, electromagnetic radiation in the optical regime exhibitsa larger variance in its interaction with matter and hence larger refractive in-dex contrasts within a crystal structure. This would correspond to the BRAGG-condition being valid for a larger range of angles, that is, including different setsq of lattice planes simultaneously. If the index contrast in a PC is large enough tomake these stop bands overlap in all directions, a full 3D photonic bandgap (PBG)is created and light propagation is prohibited in every direction inside the PC.

Clearly, this cannot be mathematically substantiated by the simple pictureof BRAGG-reflection, from which the descriptive explanation was started. Theanalogy to electronic DEBROGLIE-wave propagation in semiconductors is indeedmore appropriate. After all, in crystalline solids the electron waves experienceperiodic variations in the atomic potential, while in PCs the electromagneticwaves experience periodic variations in refractive index. Hence, the eigenmodesof a PC are as well described by means of BLOCH modes [BLOCH, 1929], the

28


wavevector K of which tells about allowed (K real) and forbidden (K imagi-nary) states. Bandgaps appear due to the anticrossing of the dispersion relationas an effect of the periodic potential of the crystal on the electrons and photons,respectively [YABLONOVITCH, 1993; RUSSELL et al., 1993].

Photonic crystals provide excellent means to control and manipulate the flowof light. Hence, they find use as special filters, cavities (to build tiny, spontaneous-emission-suppressed lasers, for instance) and waveguides [JOANNOPOULOS et al.,2008]. If light is incident in the plane of 2D PCs, (2D) PBGs can exist [VIL-LENEUVE & PICHÉ, 1992] if the refractive index difference between two consti-tuting materials is at least on the order of 2.2:1 [RUSSELL, 2006]. Light propa-gation is then only prohibited in the plane of the crystal, but might be allowedalong an in-plane defect, see Figure 2.9 B.

...to form PCFs!But is it possible to have light guidance in a defect perpendicular to the photoniccrystal plane? This (or the like) must have been the question that came to PhilipRUSSELL’s mind in 1991 [RUSSELL, 2003]. Behind this was the idea of creatingoptical mode guidance in a several micron-sized hollow core by means of a fibrecladding made of a 2D photonic crystal – a photonic crystal fibre. The answer tothe question was found four years later in calculations by BIRKS et al. [1995]:HC-PCFs are indeed possible at a refractive index ratio of 1.45:1. This is valid fora fibre consisting of air holes in glass, the choice of which is important becauselithography techniques used for high refractive index materials cannot establishthe production of fibres up to kilometres of length.

However, the question of how to fabricate PCFs was not clear from the starteither. Luckily, after only a few years of research, it turned out that it is indeedmanageable and — in retrospect — even rather simple. The first successful PCFdraw took place in 1995 [KNIGHT et al., 1996, 1997]. However, at that timefabrication had not advanced to a point where elevated air-filling fractions forbandgap guidance in a hollow core would have been possible [RUSSELL, 2003].Instead, a fibre having a solid core and a surrounding hexagonal lattice of smallair holes was initially fabricated. As a virtue born out of necessity the researcherscreated a fibre with extraordinary guidance mechanism: an endlessly single-mode(ESM) fibre, this means a fibre being single mode for a broad range of wave-lengths [KNIGHT et al., 1996, 1997; BIRKS et al., 1997] (Figure 2.10A). The firstsolid-core photonic bandgap fibre (SC-PBGF) followed in 1998 [KNIGHT et al.,1998] and finally the first HC-PBGF could be experimentally realised in 1999[CREGAN et al., 1999] (state-of-the-art example: Figure 2.10 C). Three years lateranother HC-PCF with a kagomé13-lattice structure was reported, the kagomé-fibre (KF) [BENABID et al., 2002a] (state-of-the-art example: Figure 2.10D). It

13japanese: , meaning "eye of the basket" – referring to the pattern of traditional japanese baskets.

29


was actually developed by chance during an attempt to fabricate an HC-PBGF.Luckily, the fabricators were curious enough to test if the structure would guidelight. Intriguingly, it showed to have a fairly low-loss guidance over a broadtransmission bandwidth of more than 1µm in wavelengths!

Photonic crystal fibres overcome the constraints of conventional fibres, whichare based on total internal reflection, in several different respects. First of all,the majority of PCFs consists of only one type of glass, rendering it mechan-ically and thermally more stable. Even more, in TIR guidance light can exclu-sively be guided in the higher-refractive-index core, surrounded by a lower-indexcladding. In contrast, with the advent of hollow-core photonic bandgap fibres,low-loss light guidance in a hollow core became possible14. Indeed, HC-PBGFprovides a marvellous optical mirror with a nominal reflectivity of 0.99999992for every single photon bounce inside the hollow core. As a result, kilometredistances of light propagation can be achieved [RUSSELL, 2007].

To date, a variety of different PC- and other microstructured fibres for all sortsof photonic applications exists. These include nonlinear optics in solid mate-rials and gases, optomechanical interactions, optical excitation of plasmons innanowires, light-tailoring components for integrated photonics, quantum opti-cal studies and fibre lasers. The focus of this thesis are biophotonic applications,and the photonic crystal fibres used as (bio)chemical sensors, photochemical re-actors and optofluidic waveguides will be discussed in detail in the subsequentsections. Reviewing all the other fibres and their applications is far beyond thescope of this thesis. For introduction, I refer to the comprehensive review byRUSSELL [2006] and for the most recent advances in the RUSSELL group I directthe interested reader to www.pcfibre.com.

Still, I would like to make one exception and mention the most successful ap-plication of PCFs to date: Supercontinuum (SC) generation. In so-called highlynonlinear (HNL) fibres (Figure 2.10B) the mode area can be reduced to an extentwhere nonlinear effects become highly efficient15. Moreover, it is designed tohave zero dispersion at the pump wavelength to avoid pulse spreading [KNIGHT

et al., 2000]. Hence, when high-power pulses travel through this kind of solid-core fibre, the frequency spectrum is drastically broadened via an interplay ofseveral nonlinear effects (above all four-wave mixing) [RANKA et al., 2000].With an output spanning the whole visible and a broad range of the NIR spec-trum and being spatially coherent, this can basically be called a white-light laser.While in initial demonstrations mode-locked pump lasers had been used, it could

14In fairness it should be mentioned that BRAGG fibres composed of a hollow core surrounded by alternatinglayers of different refractive index materials were already proposed in 1978 by YEH et al.. However,difficulties in manufacturing — especially for the VIS-to-NIR range of the spectrum — prevent theirfrequent use until today.

15HNL fibres are basically "blown up" ESM fibres, this means they exhibit an increased air-filling fraction,compare Figure 2.10A&B.

30

http://www.pcfibre.com


be shown that a compact Q-switched microchip laser suffices to produce an SCspanning from approximately the green into the centre of the NIR region in anESM fibre [WADSWORTH et al., 2004]. Such a compact and bright white-lightsource is frequently used in our laboratory to characterise the transmission andloss spectra of on-site fabricated fibres.

Figure 2.10: Photonic crystal fibres fabricated at MPL (all to scale). Reproduced from scan-ning electron micrographs: white: glass, black: air. A) Endlessly single modefibre, B) highly-nonlinear fibre, C) 19-cell-void hollow-core photonic bandgapfibre, D) kagomé-structure 1-cell-void hollow-core fibre.

2.2.2 Fabrication

The easiest (and hence up to date most common) way to fabricate PCFs fromglass proved to be the stack-and-draw technique [KAISER & ASTLE, 1974; TONUCCI

et al., 1992; KNIGHT et al., 1996, 1997], which is illustrated in Figure 2.11 . Cap-illaries/rods are first drawn from commercially available bulky glass tubes/rods(diameter: several centimetres, length: ∼1 m) on a conventional fibre drawingtower (1). On the order of about 1 mm in diameter and cut into pieces of 1 min length, these capillaries/rods are then carefully stacked into a macroscopiccounterpart of the designated fibre structure (2). The stack is subsequently in-troduced into and held in place by another glass tube, the inner diameter ofwhich just matches the outer diameter of the stack. This tube is called the jacket(3). This preform is then drawn down to a diameter of some millimetres, result-ing in the cane (4). For HC-PBGFs this step also serves to close off the interstitialholes between adjacent capillaries by application of vacuum. After being fusedto another jacket tube (5), from the cane the actual fibre is received in anotherdrawing step (6). This also includes the coating with a UV-curable polymer for

31


Figure 2.11: Stack-and-draw technique for HC-PCF fabrication (not to scale).

improved mechanical stability. Its final total diameter typically varies between100 to 300µm while lengths up to several hundred metres can be obtained. Thisdown-scaling procedure is possible since surface tension forces tend to balanceout during the draw.

The ratio between inner to outer diameter of the original tube to be drawnto capillary (Step 1) largely determines the air-filling fraction of the fibre (fordetails, see below). Yet, it is possible to blow up the core and/or the PC structureduring the drawing by selectively applying pressure. In combination with theappropriate setting of the drawing parameters, such as the feed rate, the pullingspeed and the tension (furnace temperature), the overall geometric parametersof the fibre are adjustable at high variability.

The fibres used in my projects are all made from high-quality fused-silica glass(Suprasil® 300, Heraeus) which has transmission losses below 5·10-3 dB m-1 in awavelength range between ∼200 nm to ∼3400 nm [HERAEUS QUARZGLAS GMBH& CO. KG, 2010]. It has a softening temperature of Tg =1600° C, so that tem-peratures in the graphite furnace are typically set to between 1800 and 2000° C.

2.2.3 Light Guidance Mechanisms

General AspectsIn order to obtain a deeper understanding of the guidance mechanisms of HC-PCFs, let us first consider the laws of reflection and refraction of light at anair-silica interface, see Figure 2.12A. In the wave picture, the spatial propagation

32


of an electromagnetic wave might be described via its wavevector k

E = E0 e−i kr . (2.37)

During transit from air into glass, the absolute value of the horizontal part of thewavevector kH (green) is preserved both for the reflected and for the refractedbeam. In the fibre (Figure 2.12B), the absolute value of the equivalent wavevec-tor component is called propagation constant β, since it is a constant for a givenmode of propagation. For propagation to be possible, β needs to be smaller thannmed k0, that is the corresponding free-space wavevector in the medium with re-fractive index nmed. In turn, an effective index can be attributed to every mode βi

as neff,i = βik0

. The core modes of a cylindrical hollow-core waveguide are typicallydescribed by means of BESSEL functions of first kind [SNITZER, 1961; MARCATILI

& SCHMELTZER, 1964]. The fundamental fibre mode is linearly polarised andcalled either LP01 or EH11, the latter corresponding to the nomenclature for hy-brid modes16. It is the mode with the largest possible wavevector value, typicallyvery close to the free-space wavevector.

At this point I would like to pick up the original question in section 2.2.1 ofwhy it is possible to guide light in a defect perpendicular to the 2D PC. To this end,it is helpful to follow the upcoming explanations in the wavevector diagram inFigure 2.12 C. Propagation is not restricted to core modes per se, but there mightalso be β’s which correspond to modes propagating in the cladding17. Now, if forexample the fundamental fibre mode would like to couple to a cladding-guidedmode, then their β values (green) need to be equal as explained above. In ad-dition, the absolute value of k is ncladd-times larger (grey) than the one in thecore (black), where nmed = nair = 1. This implies that the transverse fraction of

16Recently, ZAMANI AGHAIE et al. [2009] proposed a nomenclature specific to the modes in HC-PBGFs.17either along the cladding holes, the struts/webs or the apices/strands [COUNY et al., 2007a,b]

Figure 2.12: A) Reflection and refraction at an air-silica interface. B) Mode propagation inan HC-PBGF. C) Wavevector diagram to demonstrate the prohibited couplingbetween core and cladding modes.

33


the wavevector is larger in the cladding (light blue) than in the core (blue). Inother words: there are photonic states for which the transverse resonance wave-length of the core mode λT is too large to match one of the cladding resonances.Hence, it is not the ratio of absolute refractive indices but the ratio of transverseeffective indices that matters for propagation perpendicular to a photonic crystal,neff,T,cladd/neff,T,core = kT,cladd/kT,core, which might be much larger than ncladd/ncore

[BIRKS et al., 1995, 2004]. This in turn has the effect that the PC lattice constant(called pitch Λ) necessary to create a PBG is a few times larger than the actualwavelength, on the order of several µm, which is hence feasible to fabricate.

For a core-guided mode, the incoupling has to be such that the incoming lightfield matches a value of β which cannot be guided in the cladding. Even thoughhigher-order modes typically have higher losses than the fundamental mode[EUSER et al., 2008b] they might be well excitable and a precise alignment (in-cluding a nice and flat cleave) is inevitable to achieve efficient coupling into thefundamental mode in an HC-PCF18. To this end, the numerical apertures (NAs)of the fibre and the incoupling lens need to be matched. Clearly, the numeri-cal aperture as defined for standard fibres NA = ncore sin θcrit,TIR =

√n2

core −n2cladd

cannot apply for hollow-core fibres with an inverted refractive index difference.The NA of a hollow-core fibre might therefore be approximated19 by the one fora GAUSSian beam with a radius w0 roughly being half the fibre’s core size:

NAHC-PCF ≈ nmedλ

πw0= nmed

2λ

πrcore, (2.38)

where nmed corresponds to the medium the fibre is immersed in (and filled with).For the experiments presented later on, typical values are rcore =10µm, nmed =nH2O =1.33 and λ =400–1100 nm. Hence: NAeff ≈0.03–0.09. With 0.10 NAmicroscope objectives good incoupling efficiencies into the fundamental mode≥80 % are typically obtained20.

Theoretical ModellingFor the analysis of photonic states in cylindrically symmetric PCFs, it is typicallymore convenient to arrange MAXWELL’s equations with β2 as eigenvalue, i. e. ina set of equal frequency modes rather than in a set of k2 [RUSSELL, 2006]:(

∇2 +k20 ε(r T)+ [∇ ln ε(r T)]×∇×

)H T =β2H T, (2.39)

where ε(r T) is the dielectric-function map of the structure, the index T stands forthe vector components in the transverse plane and the field vectors are given by

18PETROVICH et al. [2008] have demonstrated that an HC-PBGF with a comparably small 3-cell-void coreshows robust single-mode guidance. However, smaller cores typically come along with higher losses.

19For HC-PBGFs a theoretically exact derivation can be found in [DIGONNET et al., 2005].20The nominal NA of an objective lens refers to a beam width just matching its aperture. The NA can

therefore be decreased by decreasing the beam width with an iris aperture in front of the lens.

34


Q = QT(r T)e−iβz . The equation is formulated based on the H-field, because thisobeys continuity across dielectric interfaces, resulting in HERMITEian operatorsand hence an eigenvalue problem of N less dimensions as compared to using theE -field.

Still, the theoretical treatment of photonic crystal fibres is difficult for basicallytwo reasons: their complex two-dimensional structure, and the large refractiveindex difference between glass and air. Under these circumstances, electromag-netic treatment, this means the solution of MAXWELL’s equations, is in most casesnot possible analytically. Equation (2.39) might therefore be solved by differ-ent numerical computational techniques; for a summary see [RUSSELL, 2006,ch. IV–C]. For instance, a fixed-frequency plane-wave (FFPW) method servesto calculate an approximate density-of-states (DOS) map of a certain claddingstructure (relative to vacuum). This method is based on FOURIER expansion ofthe field [POTTAGE et al., 2003; PEARCE et al., 2005] and a corresponding com-putational code was implemented in our group by Greg PEARCE [2006].

Hollow-Core Photonic Bandgap Fibre

Figure 2.13: A) SEM image of a custom-made HC-PBGF (Λ=4.07 µm, δ=143 nm,ρ=375 nm, d/Λ= 0.96, AFF= 84%). B) Derived perfect honeycomb structureas basis for calculation. C) Density-of-states plot as obtained from a FFPWcalculation. White: high DOS, black: low DOS, red: full 2D PBG.

I have applied this FFPW code to an HC-PBGF with its typical hexagonal arrayof cladding air holes, as illustrated in Figure 2.13 . Prior to running the code, thestructural parameters of a newly fabricated fibre such as the pitch Λ, the webthickness δ and the apex radius ρ had to be extracted from real scanning elec-tron micrograph (SEM) images (Figure 2.13A). Subsequently these were trans-lated into a perfect, infinitely large structure with a programme implemented byMartin FISCHER (Figure 2.13B). Out of this, the density of states is then calcu-lated for a set of normalised wavevectors βΛ−nk0Λ and for a specific range ofnormalised frequency k0Λ= ωΛ

c0, the result of which is shown in Figure 2.13 C.

Full 2D photonic bandgaps exist where the density of states in the claddingis zero and therefore light penetration is forbidden (red finger-like regions in Fig-ure 2.13 C). The abscissa is also called light line (turquoise), because here β equalsthe free-space wavevector nmed k0. Since β needs to be < ncore k0 for field propa-

35


gation, the fibre operates in the desired way in the regions of the bandgap (just)below the light line.

A crucial factor for the existence and width of a bandgap is the air-filling frac-tion (AFF), which is defined via the d-over-Λ ratio, that is, the inner claddinghole diameter (d =Λ−δ) divided by the hole-to-hole distance:

AFF = Aair

Aglass, where Aair = π

2p

3

(d

Λ

)2

and Aglass = 1− Aair. (2.40)

As a rule of thumb: the higher the air-filling fraction, the wider the bandgap.From a purely theoretical point of view, isolated strands of glass would be thebest choice, but clearly this is impossible to fabricate21.

Typically, for HC-PBGF a high air-filling fraction of at least 80 % and smallpitches on the order of a few µm are necessary to allow for PBGs somewhereinside the VIS-to-NIR region of the spectrum. The structure shown in Figure 2.13has an air-filling fraction of 84 % with a rather narrow bandgap in the region(12.0±0.5) kΛ, which corresponds to (1.60±0.05)µm in wavelength. There iseven a small-strip higher-order bandgap at around 30 kΛ=850 nm.

The number of modes for a fixed frequency in the allowed region can be ap-proximated via the fibre V -parameter by [CREGAN et al., 1999]

N = V 2

2≈ 1

2k2

0r 2core

(n2

eff,high −n2eff,low︸︷︷︸

∼some %

), (2.41)

where neff,high and neff,low are the effective indices at the respective edge of thePBG. Hence, for guided modes to exist at all, the core must be sufficiently large,on the order of a few wavelengths. Furthermore, guided modes exist only ifa wavelength inside the PBG coincides with a core resonance. As a result, theeffective bandgap guidance region is even smaller than calculated by the merephotonic crystal structure.

HC-PCF seems to be predestined to serve as low-loss waveguide due to the smallabsorption and scattering losses of light in air. Nevertheless other types of lossmechanisms are also present.

In PBGF, confinement losses can be kept small if the extent of the photoniccrystal surrounding the core is large enough [BOUWMANS et al., 2003]. Thehigher the air-filling fraction, the less cladding layers are needed. Bending losses

21POLETTI [2010] has proposed that a triangular lattice of strands would lead to a largerair-filling fraction as compared to the conventional hexagonal lattice. This would even-tually make possible an "octave spanning bandgap". Silke RAMMLER had an idea of howto stack-and-draw this structure, which we proved successful in a minimal stack-to-canedrawing test (see microscope image). However, fabricating a fibre would require an ex-tremely large stack and it is not clear yet how to frame the core.

36


are also present in HC-PBGFs, but they are typically smaller than in conven-tional TIR-guiding fibres. While bending of an HC-PBGF leads to a narrowing ofthe bandgap [HANSEN et al., 2004], for wavelengths close to the centre of thebandgap losses can be fairly low even for bending radii below 1 cm. The biggestrole in PCF losses is attributed to two main effects: The fraction of light guidedin the glass and the roughness of the glass-air interface. According to ROBERTS

et al. [2005a], the latter determines the ultimate loss due to unpreventable,thermally-driven capillary waves, which get frozen in as the glass solidifies dur-ing the drawing procedure [JÄCKLE & KAWASAKI, 1995]. Both these effects couldbe decreased if the modal light field shows a very low intensity close to thecore boundary; especially (coupling to) surface modes need(s) to be prevented[SMITH et al., 2003; WEST et al., 2004; HUMBERT et al., 2004; AMEZCUA-CORREA

et al., 2008]. These are modes with high intensity at the core to core-surroundboundary. For HC-PBGF such can be achieved by several means. Firstly, the coresize should only be increased to an extent where higher-order modes are stillunfavourable, since those are prone to bending loss [ROBERTS et al., 2005a]. Asa rule of thumb, the core diameter should be less than 15 times the operationalwavelength for quasi single-mode guidance22 [ROBERTS et al., 2005b]. Secondly,the core surround can either be designed such that it is anti-resonant with thecore mode [ROBERTS et al., 2005b] or alternatively extremely thin [SAITOH et al.,2004; KIM et al., 2004].

The best (i. e. lowest) reported attenuation values in hollow-core photonicbandgap fibres were achieved with 19-cell-void cores: 1.2 dB km-1 at 1620 nm[ROBERTS et al., 2005a], 1.7 dB km-1 at 1565 nm [MANGAN et al., 2004] and(recently in our group) 1.8 dB km-1 at 1530 nm [FROSZ et al., 2012]. This isstill significantly higher than the estimated minimum value of 0.13 dB km-1 at1900 nm wavelength, which has been proposed by ROBERTS et al. [2005a]. Todate, the minimum loss value for conventional telecommunication fibres is about0.2 dB km-1 at 1550 nm. NAGAYAMA et al. [2002] have even produced a solid-core silica fibre with 0.15 dB km-1 loss at 1570 nm.

Kagomé-Lattice FibreKagomé-lattice fibres [BENABID et al., 2002a; COUNY et al., 2006] have a claddingstructure which looks like a mesh made from ”stars of David” (C), see Figure 2.14 .The pitch of this structure is typically one order of magnitude larger as comparedto HC-PBGFs. This also implies that they are operating in regions of high nor-malised wavevector kΛ, whereas photonic bandgaps typically occur in the lowerkΛ regions23. Indeed, they exhibit a fundamentally different guidance mech-anism than photonic bandgap fibres. While losses are typically higher, on the22for 1550 nm this requires: rcore < 23µm.23Consistently, HC-PCFs with a comparably large-pitch honeycomb structure also exhibit broadband guid-

ance [BEAUDOU et al., 2008].

37


Figure 2.14: A) SEM image of a custom-made KF (Λ=9.6 µm, δ=122 nm, dcore = 18.1 µm).B) Perfect cladding structure with ”star of David” or kagomé pattern. C)Bright-field microscope image with unspecific white-light illumination from be-low (length of the fibre was a few cm). The colouration of the cladding holescorresponds to their individual optical resonances.

order of ∼1 dB m-1, they have the advantage of broadband guidance (see Fig-ure 2.14 C), easily spanning the whole range from the UV into the NIR. Moreover,KFs are typically quite resilient to bending, yet still more sensitive than HC-PBGFs. Bending radii on the order of at least 10 cm are typically no problem,while capillary fibres on the contrary need to be kept entirely straight, otherwiseguidance is lost completely. It has been shown that the corresponding bendinglosses can be further reduced by adding more cladding hole layers [BHARDWAJ

et al., 2011].

However, the guidance mechanism behind these experimental findings is upto date not being fully understood. First of all, in theoretical modelling KFs donot show regions of wavevector and -length where the density of states in thecladding becomes zero. Instead, there are broad regions with a rather low DOSalong the light line [HEDLEY et al., 2003]. These regions might be interrupted bynarrow strips of high DOS, which correspond to resonances of the glass struts:λstrut res. = 2δ

m

(n2

silica −1)1/2

,m ∈N [PEARCE et al., 2007]. In addition, and in con-trast to bandgap guidance, using more than two cladding layers does not signifi-cantly decrease the minimum loss value of this fibre (disregarding bend loss), aswas shown in finite element simulations [PEARCE et al., 2007] and which is inagreement with experimental findings. Instead, PEARCE et al. [2007] proposedthat a homogeneous strut thickness would be advantageous. Interestingly, in aminimalist structure composed of only one cladding layer, the loss was still com-parably low (∼1 dB m-1) but the fibre did not show broadband guidance (in bothexperiment and simulation) [FÉVRIER et al., 2010].

According to COUNY et al. [2007b], the interaction between core and claddingmodes is weak which is why co-existence is possible even though they are of thesame symmetry class and longitudinally phase-matched. The effective decou-

38


pling stems from the fact that the modes exhibit a strong transverse field mis-match (see again Figure 2.12 C): the intensity oscillations of the cladding modeare much faster than for the slowly varying core mode [COUNY et al., 2007b,FIG. 1]. Moreover, in KFs the core mode is observed to be more tightly con-fined than in HC-PBGFs [COUNY et al., 2006; ARGYROS & PLA, 2007], therebyexhibiting less overlap with the direct core surroundings.

Amir ABDOLVAND [2011] has further developed the idea to model the guid-ance in KFs by means of isolated BRAGG layers24 [PEARCE et al., 2007], quite inanalogy to BRAGG fibres [YEH et al., 1978]. In a BRAGG fibre with its concen-tric circular rings of higher and lower refractive index material, radial stopbandsrather than full 2D photonic bandgaps are formed. The formation of core modesmight be understood in terms of interference of the individual reflections off thestrut layers, which in a KF form three multilayer planar stacks at 60° angles.A FOURIER decomposition into spectral plane waves yields the reflected phaseand amplitude distribution. It showed that the distributions of perfectly inward-going and perfectly outward-going conical waves do not perfectly match, whichleads to a certain loss for each bounce. Even though calculated losses come outslightly higher in the model as compared to experiments, qualitative agreements(also with respect to other observations) could be obtained.

Only recently, WANG et al. [2012] managed to decrease the loss in a KF to40 dB km-1 from λ=1100 nm to at least 1750 nm by means of a 7-cell-void hypo-cycloid-shaped large core (dinner = 66µm) and three cladding layers. Such can beobtained by carefully balancing the core and cladding pressures so that duringthe drawing basically no deformation at all occurs. The benefits are two-fold: thespecial core shape provides even less overlap for the guided fundamental modewith the cladding and the cladding structure remains extremely uniform.

2.2.4 Filling Hollow-Core PCFs with Liquid → ∗

The term [∇ ln ε(r T)]×∇×H T in the HELMHOLTZ Equation (2.39) denotes the cou-pling term between the field components of H . This can be neglected if the fibrematerial and the holes have comparatively small refractive index ratios such thata paraxial scalar approximation is sufficient. Small refractive index ratios mightbe established by filling the entire holey structure homogeneously25 up with aliquid medium. Hence, the wave equation reduces to

∇2H T + (k2ε(r T)−β2)H T = 0. (2.42)

24If the strut length is much larger than the wavelength, the coupling between the glass struts, as mediatedby the apices, is negligible.

25In practise care has to be taken not to introduce any bubbles.

39


Figure 2.15: A) The refractive-index scaling law is used to choose or specifically fabricate HC-PCFs for usage with aqueous medium of n≈ 1.33. The plot shows inverted Equa-tion (2.43), taking into account wavelength dispersion. Used laser wavelengthsin this thesis are denoted, accordingly (blue: chapter 4, NIR: chapter 3). B) Lossspectra of H2O (blue), phosphate-buffered saline (PBS) (red) and D2O (orange),plotted together with the NIR-laser wavelengths available for the study reportedon in chapter 3. They are provided by a tunable continuous-wave titanium-sapphire [M Squared Lasers Ltd., 2011] and an ytterbium-fibre laser [IPGPhotonics Corp., 2010]. The ions contained in PBS do not absorb in theNIR region of the spectrum, therefore the curves for H2O and PBS coincide.

Based on this scalar version of the HELMHOLTZ equation scaling laws can bederived which tell about how the transmission properties change upon filling.Most importantly, the guidance bandgap of an HC-PBGF shifts according to therefractive-index scaling law, which has been derived by BIRKS et al. [2004] andexperimentally verified by ANTONOPOULOS et al. [2006] and COX et al. [2006].It states that the liquid-filled central wavelength λfill

c can be expressed via itsair-filled counterpart λair

c as follows:

λfillc =λair

c

√√√√√√1−(

nfilling

nsilica

)2

1−(

nairnsilica

)2 (2.43a)

⇒ λH2Oc ≈ 0.556λair

c . (2.43b)

This relationship, as illustrated for filling with H2O in Figure 2.15A, is typicallyconsulted for the fabrication of HC-PCFs intended for use with liquid medium.However, it contains only limited information about a suitable core size – bytrend, core sizes for liquid-filled fibres need to be slightly smaller.

40


Besides, due to the decreased index contrast the core mode penetrates deeperinto the core surround, leading to higher waveguide losses as compared to theair-filled counterpart. In addition, absorption of the laser light in the fillingmedium has to be taken into account. While in the blue region of the spectrum,where αmin ≈10-6 dB m-1, it might be negligible compared to the fibre loss (seechapter 4), in the NIR region of the spectrum, the loss is typically determinedby absorption (see chapter 3). For NIR applications, the local minimum between800 and 820 nm is the best choice with respect to minimising liquid heating dueto laser absorption, see Figure 2.15B. To demonstrate the validity of the indexscaling law the transmission properties of an HC-PBGF when filled with air andwith H2O are compared in Figure 2.16 . The observed guidance band of the water-filled fibre (turquoise) corresponds well with the scaling-law-expected range asderived from the experimentally obtained air-filled bandgap (grey).

Since the guidance mechanism of kagomé fibre is not fully understood, arefractive-index scaling law for the transmission cannot be directly derived. Nev-ertheless due to the linearity of MAXWELL’s equations it seems reasonable thata transmission shift similar to that in HC-PBGFs occurs. Indeed, we observea corresponding wavelength shift also for liquid-filled KFs. Figure 2.17 showsthe transmission properties for a KF with air- (grey) and D2O-filling (turquoise).Clearly, a transmission shift to shorter wavelengths takes place upon filling26.However, in both the air- and D2O-filled case it appears that single-mode guid-ance is not possible below a wavelength of around 800 nm. This demonstratesthe aforementioned importance of the core size, which for KFs has been stressedby ABDOLVAND [2011]. The larger the core with respect to the wavelength, themore higher-order modes are supported. In addition, mode conversion fromthe fundamental mode into higher-order modes might take place due to slightdistortions of the kagomé fibre.

26Note that in the short wavelength range we are limited by the edge of the supercontinuum source used,which is at ∼500 nm.

41


Figure 2.16: A) SEM of a 7-cell-void hollow-core photonic bandgap fibre fabricated at MPL(dcore=19.0 µm, Λ=3.96 µm, δ=86 nm, ρ=340 nm). I have filled a 1.0m longpiece of this fibre with H2O to determine its transmission properties. B) Compar-ison between the air-filled (black) and the water-filled (turquoise) loss spectrum.Both were obtained by a cut-back measurement. The loss of the water-filledfibre stems mainly from the absorption loss (orange – please note that this dataexhibits lower wavelength resolution). C) Mode imaging, performed on a 1.0mH2O-filled and a 2.0m air-filled fibre with 10 nm broad bandpass filters at denotedwavelengths. Core guidance was observed in good agreement with the expectedbandgap regions, as indicated in B. In the filled piece, a narrow second-orderbandgap close to 600 nm seems to be also present. The mode shown at thiswavelength is highly multimode and couples to the core surround.

42


Figure 2.17: A) SEM of a 3-cladding-rows 1-cell-void core kagomé fibre fabricated at MPL(dcore=27.7 µm, Λ=17.4 µm, δ=271 nm). I have filled a 2.3m long piece of thisfibre with D2O to determine its transmission properties. B) Comparison betweenthe air-filled (black) and the D2O-filled loss spectrum (turquoise). Both wereobtained by a cut-back measurement. The absorption loss of bulk D2O is shownin dashed orange. This specific fibre shows intermittent non-guiding regions inthe NIR when air-filled (grey shading) – a behaviour which however disappearsupon filling. C) Mode imaging, performed on a 2.3m D2O-filled and a 4.5mair-filled fibre, with 10 nm broad bandpass filters at denoted wavelengths. Coreguidance was observed in good agreement with the expected transmission regionsas obtained from the loss spectra in B.

43


2.2.5 ∗ → yields Unbeaten Optofluidic Channels

The parameters important for the efficient interaction of light and matter wereidentified in section 2.1.2 on page 15. The discussion yielded the figure ofmerit for light-matter interaction as Ξ := λ0/P0 ·

∫I (z)dz (Equation (2.32)). In

section 2.1, I have already mentioned several optofluidic solutions which havefound useful applications within this framework. Now, these shall finally be dis-cussed and compared on a physical basis.

Figure 2.18: Approaches to extend the propagation length of a light field. A) Diffraction-limited free-space weakly-focused Gaussian beam, B) diffractionless free-spaceBessel beam, C) multi-mode capillary, D) diffractionless Bessel-J02 mode inan HC-PCF.

Free-Space Diffraction-Limited BeamFor a beam focused down to its RAYLEIGH-limited spot (see Figure 2.18A) theintensity in axial direction z is given by

IFB(z) = P0

Abeam(z)= P0

πw 2(z)= P0

π

(w0

√1+

(z

LR

)2)2 = P0

Amin

(1+

(z

LR

)2) , (2.44)

44


where w(z) denotes the GAUSSian beam radius, with w0 being the minimal beam

radius and LR = πw20

λ= nmed Amin

λ0being the RAYLEIGH length. Evaluating Equa-

tion (2.31) accordingly and inserting it into Equation (2.32) then gives

ΞFB = λ0

πw 20

∣∣∣∣LR arctan

(z

LR

)∣∣∣∣∞−∞ = λ0

πw 20

·πLR =πnmed ∼ 100. (2.45)

Hence, regardless how tightly you focus a beam, the figure of merit of light-matter interaction stays the same, on the order of 1.

Free-Space Diffractionless Beam

In free space the diffraction limit might be overcome by so-called diffractionlessbeams, see Figure 2.18B. The most prominent is the so-called BESSEL beam as in-troduced theoretically by DURNIN [1987] and experimentally demonstrated soonthereafter [DURNIN et al., 1987]. Experimental production is most efficiently es-tablished with an axicon (conical) lens [MCLEOD, 1954; HERMAN & WIGGINS,1991]. The transverse beam profile consists of a series of concentric rings corre-sponding to a (finite) BESSEL-J 2

0 profile. The radius of the central spot is givenby

rcentre = 2.405

k sinβ= 2.405λ

2π sinβ≈ 2.405λ

2π sin(π−2γ). (2.46)

where β is the angle of intersection of geometrical rays with the axis. This isrelated to the axicon’s opening angle γ via β = π−γ− arcsin(nmed sinγ). Sinceaxicons are typically available up to angles of 20°, that is γ ≤ π

9 , I have usedsinγ≈ γ, arcsinγ≈ γ and assumed operation in air (nmed = 1 ⇒ λ=λ0) to get theapproximation on the right-hand side.

The axial decay of the intensity is complicated with an oscillatory intensityvariation around a mean value which eventually quickly decays after a certainlength. This has been derived to be [ARLT et al., 2001]

LBB ≈ w0

(naxi −nmed)γ, (2.47)

where naxi ≈1.46 (fused silica) refers to the refractive index of the axicon lensand w0 to the radius of the GAUSSIAN input beam. w0,max is determined by thetypical axicon diameter of 2.5 cm. With Equations (2.46) and (2.47), the figureof merit for light-matter interaction in the centre spot of an axicon-generatedBESSEL beam might be estimated via

ΞBB ≈ λ0

P0Icentre LBB =λ0

LBB

Acentre= 4.723

sin2(π−2γ)

γ· w0

λ0≈ 4.723 · (4γ) · w0

λ0, (2.48)

where a TAYLOR expansion has been used to simplify the term in γ.

45


Assuming an NIR wavelength of λ0 ≈1µm, the experimentally achievable max-imum ΞBB, max is on the order of 105, as obtained with a large-angle axicon(γ= 20°= π

9 ). In this case the central spot in the beam is minimal, on the orderof the wavelength, and the effective path length in air is ∼16 cm. The maximumeffective path length is achieved with a small-angle axicon (γ= 0.5°= π

360) andon the order of ∼6 m in air, while at the same time increasing the central spot inthe beam to rcentre =22µm (ΞBB =103).

Despite the huge advances of BESSEL beams compared to diffraction-limitedspots, which have for instance been demonstrated in the field of particle guid-ance [ARLT et al., 2001; GARCÉS-CHÁVEZ et al., 2002; CIŽMÁR et al., 2005] andoptical cell sorting [PATERSON et al., 2005], there are still some drawbacks intheir application to light-matter interaction experiments. Firstly, the dynamicalvariation of the intensity profile in propagation direction might introduce un-favourable effects. Secondly, BESSEL beams are free-space beams, which rendersthem prone to external perturbations such as convection or other turbulences.Thirdly, the actual interaction volume might be small, but the total sample vol-ume might be contained in a comparably large reaction vessel. This relates thefact that the efficacy of BESSEL-beam generation critically depends on the initialaperture of the lens. When generated with a microlens one cannot exceed effec-tive path lengths of ∼500µm [LEE et al., 2010], so that for this specific demon-stration in the microchip regime, ΞBB,µ ≈100, which is similar to the RAYLEIGH-limited spot.

Capillary ”Waveguide”These disadvantages might be circumvented by using small hollow waveguidesinstead. In a waveguide, a propagating optical mode decays in dependence onits loss L (typically in 10-base and in units of dB m-1), according to

I (z) = I0 e− ln(10)·L10 z

=: I0 ·e−z/Leff ,(2.49)

in which as a characteristic length, the effective path length Leff is introduced:

Leff =10 · log(e)

L. (2.50)

With the mode area being conserved in propagation direction and being approx-imately equal to the core size Acore, for a waveguide we can define

ΞWG = λ0

P0·∫ ∞

0I (z)dz = λ0

P0I0Leff =λ0

Leff

Acore. (2.51)

The first approach to achieve waveguiding in a hollow structure, see Figure 2.18 C,was by using hollow glass capillaries [SAWATARI & KAPANY, 1970]. This works by

46


so-called grazing incidence guidance, which refers to internal FRESNEL reflectionsoff the air-glass interface which are strong at an almost parallel incidence angle.As a result of this, capillary guidance is easily lost due to even slight bends, sothat one might argue whether to call a capillary a proper waveguide. MARCATILI

& SCHMELTZER [1964] have performed an elaborate analysis of optical modesin capillaries, including the derivation of the attenuation coefficient for the field,which corresponds to twice the attenuation coefficient in terms of intensity (herein e-base and units of m-1):

α′cap = 2 · α′

cap = 2 ·u2

pq

4π2

(nglass

nbore

)2 +1

2

√(nglass

nbore

)2 −1

· λ2

r 3bore

=: 2 ·u2

pq

4π2·υ · λ2

r 3bore

.

(2.52)

Herein, upq refers to the zero-point of the q-th root of the BESSEL function of 1stkind and (p−1)st order, which describes the shape of a capillary mode. This alsoimplies that higher-order modes have higher losses, because upq increases withthe order of the BESSEL function. For the fundamental mode EH11: u11 =2.405.Assuming a capillary made out of fused silica, this means nglass = nfused silica =1.45to 1.47 (for 1µm and 400 nm wavelength, respectively) and a bore filling of airor (heavy) water (nair =1.00, naq =1.33), we get

υ= 1.47 ⇒ α′aircap = 0.431

λ2

r 3bore

, (2.53a)

υ= 2.44 ⇒ α′aqcap = 0.714

λ2

r 3bore

. (2.53b)

Hence, the loss value for the fundamental mode (in dB m-1) is given by

L aircap = 10log(e) ·α′air

cap ≈ 1.87λ2

r 3bore

, (2.54a)

Laqcap = 10log(e) ·

(α′aq

cap +α′aqabs(λ)

)≈ 3.10

λ2

r 3bore

+10 log(e)α′aqabs(λ). (2.54b)

Via Equation (2.50), the effective length for a capillary can be derived to be

Laireff, cap ≈ 2.32

r 3bore

λ2, (2.55a)

Laqeff, cap ≈

(0.71

λ2

r 3bore

+α′aqabs(λ)

)−1

. (2.55b)

47


Inserting this into Equation (2.51), we get the figure of merit for light-matterinteraction in a capillary as:

Ξaircap = 0.74

n2med rbore

λ0(2.56a)

Ξaqcap =

(2.23

λ0

n2med rbore

+ α′aqabs(λ)

πrbore

λ0

2)−1

. (2.56b)

At a fixed wavelength of operation, Equation (2.56a) suggests that the figure ofmerit for light-matter interaction in a hollow capillary can be increased by in-creasing the bore size (Ξ∝ rbore). However, this relationship is limited due to thedifficulty to excite a fundamental mode in large-bore capillaries where rbore Àλ.Interestingly, the expression for a filled capillary, Equation (2.56b), exhibits a de-fined maximum Ξ-value for operation at a given wavelength. This means thatthere is an associated optimal bore size.

Liquid-Filled Hollow-Core PCFFor PCF (Figure 2.18D) circumstances are completely different because the intrin-sic waveguide loss and hence the (unfilled) effective length can be engineeredfor a desired wavelength to be on the order of a few dB m-1 for KFs and of sev-eral dB km-1 for HC-PBGFs (plus the absorption loss of the liquid). Clearly, therange of possible core sizes is stronger limited in HC-PCFs, to approximately5µm < rcore <30µm. In conclusion, for HC-PCFs we get:

ΞHC-PCF =λ0Leff, HC-PCF

Acore= 10 log eλ0

LHC-PCF · Acore= 1.38

LHC-PCF

λ0

r 2core

, (2.57)

where the loss LHC-PCF corresponds to the measured loss including the filling.By trend, Equation (2.57) favours smaller core sizes.

Table 2.1: Light-matter interaction parameters of the HC-PCFs used, compared to hypothet-ical capillaries for optimised operation with the same medium and at the samewavelength.

chapter 3: D2O, λ= 1064 nm chapter 4: H2O, λ= 488 nm

LHC-PCF |α′aqabs 5.0 dB m-1 | 0.92 m-1 3.2 dB m-1 | 0.05 m-1

HC-PBGF hypoth. capillary KF hypoth. capillary

Ξ 1.0 ·103 3.1 · 101 5.4 ·102 1.1 · 102

rcore/bore 17.5µm 63µm 19.7µm 101µm

Table 2.1 shows that in both projects, capillaries would be inferior in their light-matter interaction efficiency compared to the HC-PCFs used. Their large boresizes have additional drawbacks. First of all, as alluded to earlier, it is practically

48


impossible to achieve single-mode guidance. Yet, single-mode guidance and anarrow microfluidic channel are inevitable requirements to perform the study inchapter 3. On the other hand, the study in chapter 4 would suffer from the largersample consumption due to the increased volume in the large-bore capillary.

Other Optofluidic WaveguidesBesides capillaries and HC-PCFs, there are additional kinds of direct-overlapoptofluidic waveguides [SCHMIDT & HAWKINS, 2008, 2011] which I would liketo quickly mention for the sake of completeness.

If a capillary is filled with a medium of higher refractive index, it acts as aTIR waveguide. Many people have therefore used glass or polymer capillaries tostudy liquids of high refractive index, as pioneered by PAYNE & GAMBLING [1972]and STONE [1972]. However, this approach is challenging for aqueous mediumwith a rather low refractive index of 1.33. To date only one bulk material with alower refractive index of 1.29, a special kind of fluorinated polymer, has foundapplication as a water-filled TIR waveguide material [DRESS & FRANKE, 1996].Other clever approaches comprise the use of liquid-liquid-waveguides [WOLFE

et al., 2004] and nanoporous cladding materials to lower the average refractiveindex in the cladding [RISK et al., 2004]. Similarly, selective filling of only thecore of an HC-PCF [NIELSEN et al., 2005] has been used for fluorescence sensing[SMOLKA et al., 2007], single photon generation from quantum dots in solution[JIANG et al., 2012] and microparticle propulsion [MANDAL & ERICKSON, 2007].However, this requires additional processing steps in order to selectively collapseor plug the cladding holes prior to liquid infiltration. Moreover, as elaboratedearlier, index guidance is inferior to the intrinsic PCF guidance mechanisms, be-cause it is typically multi-mode and has inherently higher losses.

Antiresonant reflecting optical waveguides (ARROWs) have a guidance mecha-nism different from TIR- and PC-based guidance [DUGUAY et al., 1986]. If thethickness of one or several high-index cladding layers is adjusted such that theroundtrip phase shift of the transverse wave component is an integer multiple ofπ, then the FRESNEL reflection back into the core is highly efficient.

Undoubtedly, liquid-core ARROWs have drastically advanced the field of opto-fluidic waveguides in ultra-sensitive biochemical sensing [YIN et al., 2006], mi-cro flow cytometry [BERNINI et al., 2007] and on-chip optical particle manipula-tion [MEASOR et al., 2008; KÜHN et al., 2009a,b]. Despite this, ARROWs sufferfrom their cumbersome and expensive silicon-based fabrication process, whichdoes neither allow to fabricate long lengths nor to produce circular cross sec-tions. Their rectangular or arch-shaped core leads to asymmetric trapping (andshear) forces in particle guidance experiments. Losses are on the order of atleast a few 100 dB m-1, leading to Leff, ARROW =3.9 cm and ΞARROW ≈6 · 10 2 in thelowest loss ARROW reported by YIN et al. [2005].

49


In conclusion, it is clear that the overall properties of optofluidic PCFs are indeedunbeaten by any other existing waveguide, which renders them an outstandingchamber for light-matter interaction. In consequence, the studies demonstratedin this thesis could exclusively be performed by application of HC-PCFs. More-over, the two rather different applications of light-matter interaction — opticalmanipulation and photochemistry — demonstrate PCF’s versatility in the field ofoptofluidics.

50

33Long-Range Optical Guidance

and Deformation Monitoring ofSingle Cells in fibra

"Man reist ja nicht um anzukommen, sondern um zu reisen."

Johann Wolfgang VON GOETHE (1749–1832)

3.1 Framework

This chapter introduces optofluidic PCF as a unique tool to laser-propel individ-ual cells over — so far unreached — distances of 10s of cm through stationaryliquid. As opposed to previous studies on solid microparticles [EUSER et al.,2009, 2010; GARBOS et al., 2011a; GARBOS, 2011], biological cells are soft andhence deformable under the action of external forces. Therefore, during prop-agation through the narrow optofluidic channel provided by the fibre core, dy-namic changes in cell shape occur, which were investigated at constant opticalpowers of up to 350 mW [Paper II: UNTERKOFLER & GARBOS et al., 2012]. Thesecan be accurately and continuously monitored via an in-fibre DOPPLER velocime-try technique [Paper I: GARBOS et al., 2011]. This provides a convenient basisfor a non-imaging bioanalytical tool to investigate single-cell mechanics in fibra.

In this context, the advantages of liquid-filled hollow-core photonic bandgapfibres are apparent: they provide low-loss light guidance in a well-defined singlemode. This means that cells are trapped laterally at the centre of the core, typi-cally several microns away from the glass interface, which eliminates adherenceeffects and external perturbations. In turn, this results in well-defined and highlyuniform optical and fluidic forces acting on the cells.

51

Chapter 3 OPTICAL GUIDANCE AND DEFORMATION MONITORING OF SINGLE CELLS IN . . .

Figure 3.1: Schematic drawing of a red blood cell. The geometric outline and numbers followthe precise measurements by Evans & Fung [1972]. Spectrin filaments (darkgreen) are tethered to transmembrane proteins (bright green) in a two-dimensionalhexagonal geometry.

Red Blood CellsFor initial demonstration red blood cells (RBCs) with their disc-like shape werethe optimal choice for several reasons. They are the smallest of all human cells,about 7.8µm in diameter with a narrow standard deviation of only ∼8 % [EVANS

& FUNG, 1972]. This means that the whole range of sizes in an RBC populationfits into the fibre core (∼17.5µm) and the optofluidic force environment duringpropagation is similar. In addition, blood cells are naturally in suspended state.On top of that, RBCs are highly elastic, that is, their original disc-like shape iseasily deformed into a variety of different shapes by external forces [MUSIELAK,2009]. For instance, as an effect of osmosis, RBCs are well known to swell whenplaced in a hypotonic medium1 and to shrink until spiky in appearance whenplaced in a hypertonic medium. They owe their deformability to their uniqueinternal structure. Unalike other cells, they have a two-dimensionally-linked cy-toskeleton which is directly attached to their membrane envelope via membraneproteins (see Figure 3.1 ) and they have neither a cell nucleus nor any other inter-nal organelles. In consequence, RBC abnormalities (harmless or severe, chronicor acute), including malaria infection, are typically easily discriminated fromhealthy conditions via cell mechanical properties [MOHANDAS & GALLAGHER,2008]. This, in fact, had already been noticed in RBCs’ very first scientific de-scription by VAN LEEUWENHOEK [1675]: "I [...] noted that those sanguineousglobuls that make the Blood red, seemed then to be firmer and harder than they arein my Blood now; at which time my Body was very indisposed, so that I fell into asickness [...] But now I find those globuls of my Blood softer [...] and my Body in agood state of health."

1An isotonic solution, corresponding to the physiological state, has an osmolarity of 300 mOsm(ole). Os-molarity is the measure for summed-up solute concentration. At an osmolarity of ∼130 mOsm RBCs aretypically blown up into round spheres of diameter (6.78±0.32)µm [EVANS & FUNG, 1972].

52

Instrumentation 3.2

3.2 Instrumentation

3.2.1 Materials and Sample Preparation

Fibre

Figure 3.2: HC-PBGF for RBC guidance. A) Scanning electron micrograph. dcore=17.5 µm,Λ=4.7 µm, δ=120 nm. B) Normalised mode intensity profile at 1064 nm wave-length, measured at the output of a ∼ 20 cm-piece of D2O-filled fibre. C) Lossspectrum of the D2O-filled fibre (black) in the bandgap-region (mint green), com-pared to the loss of bulk D2O (orange).

The fibre used in this study is an HC-PBGF, fabricated in-house from fused sil-ica. It provides low-loss single-mode guidance in a narrow wavelength rangebetween 1000–1175 nm when filled with aqueous medium of ∼1.33 refractiveindex, see Figure 3.2 . Its core diameter is dcore = 17.5µm and its loss at the laserwavelength of 1064 nm is L =5 dB m-1 of which 80 % is due to absorption in themedium. This corresponds to an effective length of Leff =87 cm.

For cell guidance experiments the medium of choice is heavy-water-basedphosphate-buffered saline (D2O-PBS) for two reasons. Firstly, the absorptionof D2O(-PBS) at 1064 nm wavelength (αD2O =4 dB m-1) is 16 times lower thanthat of H2O(-PBS) (αH2O =65 dB m-1) (see again Figure 2.15 on page 40), ensur-ing an approximately constant optical intensity along the fibre and minimisingthe effects of laser heating of the solution. Secondly, cells need to be handledin isotonic solution to prevent them from osmotic swelling or shrinking. D2O-PBS was prepared by dissolving PBS powder (Appli-Chem) in heavy water (Sig-maAldrich).

Prior to injection into the fibre, the solution was degassed under agitation invacuum and filtered with a syringe filter of 0.22µm pore size (polyethersulfone).Filling is achieved by inserting one end of the fibre into a customised pressurecell (Figure 3.3 ) and applying a pressure of 8 bar, typically for one or two hours.

53


Figure 3.3: Technical drafts of custom-made pressure cell for fibre liquid-filling. Dead volume∼ 50 µL. A) Side view. B) Front view. C) Slanted view (body only).

The design of the pressure cell was custom-made and, to circumvent con-tamination, chosen such that no glue connections would be used. Instead, thepurchased unions fabricated from polyether ether ketone (PEEK) were tappedwith external threads and screwed into the water cell body, made from polyvinylchloride. Additional teflon tape was used for sealing. The tubing is leak-tightthrough the use of 1/16" PEEK fittings typically used for high-pressure liquidchromatography. The fibre is encased by a tightly fitting 1/16" tubing sleeveand screwed into the cell from the rear (PEEK equipment from VICI AG andUpchurch Scientific). An aluminium baseplate serves to attach the water cell tothe fibre coupling flexure stage (NanoMax, Thorlabs) with the fibre position atthe right beam height such that it can be accessed through a window (borosil-icate, thickness: 0.5 mm, Plano). This window is pressed against an O-ring ofabrasion-resistant FFKM rubber (Telle) with a cap screw2.

Once the water cell is bolted down onto this stage, the loose fibre end is placedhorizontally on a glass cover slide of 170µm thickness, to which a vertical glasswindow of 100µm thickness has been glued. This glass slide construction pro-vides optical access from below and towards the fibre facet. In contact with thisthe fibre is inserted into a V-groove, which has been mounted upside-down toanother fibre-coupling stage via a custom-made holder, see Figure 3.5 B. The fi-bre front facet is then immersed in a droplet of D2O-PBS and the filling syringedisconnected from the pressure cell to stop the flow.

2After filling of the fibre it has proven useful to flush the cell interior with several mL of fresh solution,while the cap screw is only slightly untightened. However, it might also be necessary to recleave the fibreend which requires to open the cell. By loosening of the fitting the fibre can be slid through towards thefront, cleaved, withdrawn and refixed, so as to eventually refill the pressure cell.

54

Instrumentation 3.2

3.2.2 Experiment

SetupThe experimental setup is shown in Figure 3.4 /Figure 3.5 on the following doublepage. It is a multi-level setup, where two breadboards are elevated by about20 cm, so as to host the optical tweezers setup on the table level in the gapbetween them.

It provides a choice between two lasers: a diode-pumped, single-frequency(line width δ f ≈ 70 kHz) ytterbium-fibre laser (YLR-50-LP-SF, IPG Photonics) at1064 nm wavelength with a maximum output power of 50 W [IPG PHOTON-ICS CORP., 2010] and a tunable, frequency-stabilised continuous-wave titanium-sapphire laser (SolsTiS-F-SRX, M Squared) [M SQUARED LASERS LTD., 2011],which provides a tuning range of 715 nm to 895 nm (δ f < 50 kHz). Its maxi-mum output power is ∼1.9 W at 784 nm wavelength when pumped by 7 W witha frequency-doubled Nd:YVO4 laser (Verdi-V8, Coherent) – see again Figure 2.15on page 40. The study presented in this chapter is entirely performed with theytterbium fibre laser at 1064 nm wavelength, which matches the liquid-filledHC-PBGF introduced above.

The beam is split by a polarising beam-splitter cube (PBSC), with the preced-ing half-wave plate (λ/2) allowing a continuous adjustment of the intensity ratiobetween the two exiting beams. One beam serves as trapping beam in the opticaltweezers, the other beam is to be coupled into the optofluidic PCF. This guid-ance beam is split again, thereby providing a dual-beam option (yellow shading)through the possibility of coupling into the fibre from either side. Since the right-hand breadboard is movable in fibre-axial direction, the guidance beam from thisside is delivered via a polarisation-maintaining single-mode fibre with PC/APCconnectors (P3-1064PM-FC-1, Thorlabs). In the dual-beam scheme, the intensityratio between both guidance-beam arms can be quickly switched (τswitch = 10 ns)or continuously modulated with a computer-controlled (LabVIEW) electro-opticmodulator (LM 0202 IR KD*P, Qioptiq) steered via an analog amplifier (LAV 400,Qioptiq). The purpose of dual-beam guidance in fibra will be further discussedin chapter 5. In this study, however, only the guidance beam from the left wasused and the right-hand arm was reduced to minimum intensity and blocked.The guidance beam is coupled into the fibre via a 0.1 NA IR-objective lens (LM-PLN5XIR, Olympus), which is approximately matched to the NA of the fibre.Alignment is performed by beam-walking with two mirrors and a micrometredrive at the fibre coupling stage. Incoupling efficiencies typically yielded ∼75 %.

The single-beam optical tweezers are custom-built and established by a longworking distance (2.5 mm) 1.1 NA water-immersion microscope objective (CFIPlan 100×, Nikon). Imaging onto camera 1 (colour CMOS, uEye, iDS) (CAM1, Figure 3.6 B), the tweezers microscope has a circular field of view of 120µm

55


Figure 3.4: Cell guidance setup, Part I. A) Photograph of left side. C) Scheme →

56

Instrumentation 3.2

Figure 3.5: Cell guidance setup, Part II. B) Photograph of mounted fibre. C) ← Scheme

57


in diameter, as has been calibrated with a glass-slide scale (Plano). Its effectivemagnification is then 18.4×. Besides imaging, it facilitates transportation of aselected cell to the core entrance (for detailed illustration see Figure 3.7 ). Tothis end, a hot mirror serves to separate the VIS illumination light (bright-fieldmicroscopy) from the IR trapping beam (λedge ≈ 700 nm, Edmund Optics). Thetweezers microscope is movable in all three dimensions of space: the tweezersobjective resides on an x-y-z-flexure stage (MBT, Thorlabs), which is mounted ontop of a long-range x-y-travelling stage (∆x =∆y = 2 cm, Newport). To keep thisflexibility, the trapping beam is delivered to the tweezers platform by a single-mode fibre with FC/APC connectors (P3-980A-FC-1, Thorlabs).

Launching ProcedureAfter alignment of the fibre, a cell suspension is freshly prepared — in the currentstudy, approximately 10µL of fingertip blood were taken from one of two healthyvolunteers and diluted in 2.5 mL of DULBECCO’s phosphate-buffered saline (PBS)(Gibco, Invitrogen). A few µL of cell suspension are then added to the D2O-PBS-droplet on the glass slide. From this dilute suspension, a single RBC is trappedusing the optical tweezers microscope (Ptrap ≈80 mW). Its current position withrespect to the fibre is known from its scattering by simultaneously watching cam-era 4 (x-y) and camera 1 (y-z), as shown in Figure 3.6 . When the cell is just infront of the fibre core, the horizontal guidance beam is switched on, causing theRBC to be pushed out of the tweezers trap and into the core by optical scatteringforces (Figure 3.6 B(2-3)). Quickly thereafter pressure is applied to the rinsingsyringe so as to prevent other cells from accidentally entering the core as well.Once inside the fibre, the cell is propelled by the guided optical mode. The RBCswere observed to propagate near the centre of the core with their long axis par-allel to the fibre axis. This is the result of optical gradient forces that pull the cellmembrane into the region of highest intensity at the centre of the core.

DOPPLER velocimetryDuring propagation the cell’s speed can be accurately monitored by a recently de-veloped in-fibre DOPPLER-velocimetric technique [Paper I: GARBOS et al., 2011].This has several advantages as compared to a velocity measurement with exter-nal video-based imaging. Firstly, data acquisition and analysis are simplified andhence accelerated. Secondly, while in video imaging the range over which theparticle can be tracked is limited by the narrow field of view of the camera, ourtechnique showed to be applicable to particle propagation distances of at least∼50 cm. Moreover, the field of view of a camera is traded off against resolu-tion, which leads us to the third and most important point: the accuracy withwhich the particle can be tracked by DOPPLER velocimetry is extremely high, theprincipal limit being on the order of a few hundred nanometres, because it is aninterferometric technique [YEH & CUMMINS, 1964].

58

Instrumentation 3.2

Figure 3.6: Cell launching procedure. The RBC is optically trapped by the tweezers micro-scope. A) x-y-front facet imaging by CAM4, while moving (1-2) the cell towardsthe core entrance (3). The fibre structure is illuminated with a flash lamp from itsrear end, the trapped cell’s scattering is clearly detected. B) Once brought intothis position (1), the cell is subsequently pushed into the fibre core by unblockingthe guidance beam (red arrow) (2-3). Upon launching, the RBC aligns its longaxis in parallel to the fibre axis (3). The apparent stretching of the cell in (3) iscaused by an imaging artefact due to a relatively long exposure time.

The measurement principle is illustrated in Figure 3.7 A. A tiny fraction (< 1 ‰)of the light scattered by the moving cell (velocity vP) is reflected into the back-ward-propagating fibre mode and travels back to the fibre entrance (orange ar-row). The frequency of the backscattered light is DOPPLER-shifted with respect tothe original laser frequency f0 to

fD = f0

(1− 2n vP

c0

), (3.1)

where c0 is the speed of light and n the effective index of the fundamental mode3,which is almost the same as the refractive index of the medium nD2O-PBS =1.33.

This interferes with a similar fraction of unshifted light which is reflected at theentrance face of the fibre during incoupling (red arrow). These two fractions oflight are collinear in a self-aligned manner and propagate in backward direction.A beam splitter serves to direct them onto a photodiode (PD) which monitorstheir temporal interference beating in intensity I (t ) at a sampling rate of 20 kHz.The detected beat frequency ∆ f is then directly proportional to the cell velocity:

∆ f = f0 − fD = 2n vP

c0f0 = 2n

λ0vP = vP

Lbeat. (3.2)

The beat length Lbeat is the factor of proportionality and equals 400 nm in ourcase (n ≈1.33, λ0 = 1064 nm). Therefore, the velocity vP(t ) can be accurately

3For simplicity, the presence of higher order modes is neglected here. Co-existence of different modeswould lead to an oscillating intensity pattern along the fibre core as a result of their individual effectivemode indices. We could measure this by means of particle guidance [Paper I: GARBOS et al., 2011].

59


Figure 3.7: In-fibre Doppler velocimetry. A) When laser light of frequency f0 is coupled intothe fibre, a small amount of < 1‰ is reflected off the fibre’s front facet (red). Thisis overlaid with backward-propagating Doppler-shifted light scattered from themoving particle (orange). The interference signal is picked up with a photodiode(PD) after spatial filtering of the core region. B) Data processing procedureillustrated for a 7.5 µm polystyrene particle propelled at P=221mW. Inset: Theraw signal from the diode I(t) contains high-frequency fluctuations on the orderof several hundred Hz, directly corresponding to the particle speed vP.

60

Instrumentation 3.2

retrieved from the original data by performing a FOURIER analysis. Such hasbeen implemented in Matlab using a fast FOURIER transform in a moving timewindow. The step size was 50 ms and the width of the window was 100 ms,leading to an f -resolution of 10 Hz, and hence a vP-resolution of 4µm s-1.

As compared to the data shown in our introductory study on non-deformableborosilicate beads [Paper I: GARBOS et al., 2011], for cell guidance measurementsit is necessary to record the speed over extended distances. To avoid a bufferoverflow, the data was collected in blocks of ∼50 s, having gaps of 6 s in between.In this way, the diode signal was recorded over several 10s of minutes. TheMatlab programme had to be modified in order to automatically concatenate theseries of data files. Moreover, it was extended to perform temporal integration ofthe data for both intensity and speed in order to obtain these data as a functionof propagation distance z. For illustrative purposes, these data were eventuallyaveraged over intervals of 5000 data points by reduction to evenly spaced distances(Origin) of 0.5 cm. The procedure is illustrated in Figure 3.7 B.

Force AnalysisAt this point, it is necessary to state how the magnitude of the particle velocity isdetermined by the fundamental laws of physics. Clearly, for a steady propulsion,the driving optical force needs to balance the viscous drag force:

Fopt = Fdrag, (3.3)

which is almost instantaneously reached in a highly viscous medium such as anaqueous solution. The optical force acting on the particle can be written as:

Fopt = q ·P, (3.4)

where P is the optical power and q the optical force parameter (in pN W-1), whichdepends on the shape, size and refractive index of the particle, as well as on itsoverlap with the optical mode.

For spherical particles moving along the central axis of a narrow tube, theexpression for the drag force is:

Fdrag = 6πrP K1 vPη, (3.5)

where η is the viscosity of the liquid and rP the radius of the sphere. K1 is acorrection factor accounting for the influence of the nearby walls [LADENBURG,1907; AL QUDDUS et al., 2008]. It is a function of particle-to-tube size Γ = rP

rcore

and approaches unity for Γ→ 0 (free STOKES drag), while it diverges for Γ→ 1.However, for non-spherical particles, the drag force parameter ζ (in µm) used asin

Fdrag = ζvPη (3.6)

61


is a more convenient parameter since in this case a characteristic radius rP is notdefined. ζ depends on the size and shape of the particle and the tube diameter.It cannot be calculated analytically, but it can relatively easily be obtained fromfinite element hydrodynamics simulations as will be described in section 3.3.4on page 68.

The optical force parameter q and the fluidic force parameter ζ might bemerged into a single quantity, which is herein henceforth called the optofluidicforce parameter ξ (in pNµm-1 W-1):

ξ := q

ζ. (3.7)

The larger the value of ξ, the more mobile the particle is in a given optoflu-idic environment. Using Equations (3.3), (3.4) and (3.6), ξ can be retrieved fromexperiment via

ξ= vP ·ηD2O-PBS

P. (3.8)

For non-deformable particles ξ is constant. This means at constant viscositythe particle velocity is proportional to the optical power, as recently confirmedexperimentally in optofluidic HC-PBGF using glass microspheres [EUSER et al.,2009, 2010; GARBOS et al., 2011b]. For deformable particles, however, ξ servesas a measure of dynamic deformation under the action of optical and fluidicforces. By a knowledge of ζ the optical force parameter q and hence the totaloptical force on the cell can be derived (or vice versa).

3.3 Results and Discussion

3.3.1 Long-Distance Optical Cell Guidance

Figure 3.8: External imaging (CAM 2) of an RBC propagating through a 7 cm long sectionof the fibre. Video frames (1-5) are taken at 1min intervals. The position of theRBC is estimated from its side-scattered light and indicated by the pink arrows.

62

Results and Discussion 3.3

Figure 3.8 shows video-frame shots of an RBC propelled by an optical power of349 mW, as recorded with camera 2. The camera is placed in some distance tothe fibre to provide a large field of view. The position of the cell is indicatedvia its side-scattered light (see pink arrows). The average speed for this cell wasabout 300µm s-1, corresponding to 40 cell diameters per second4. The maximumpropagation distance was 24 cm, limited only by the physical length of the fibrepiece used. This corresponds to two orders of magnitude further optical celltransportation than reported in a previous attempt using a 100µm-bore capillary[ODDE & RENN, 2000].

Using DOPPLER velocimetry the cell speed was measured as a function of po-sition. The results for four different RBCs, each propelled at a different opticalpower, are plotted in Figure 3.9 A. At elevated optical powers the cells show adrastic increase in speed at ∼5 cm. This corresponds to the edge of the glassplate used to attach the fibre to the V-groove coupling mount. The velocity jumpwas therefore suspect to be caused by an abrupt change in the immediate envi-ronment surrounding the fibre.

Figure 3.9: Speed traces A) for four individual RBCs, each propelled at another optical power,and B) for a 7.5 µm polystyrene particle, propelled at four comparable powers (seecolour coding).

3.3.2 Temperature Calibration

The temperature in the fibre core depends sensitively on the surroundings of thePCF, which was investigated in a particle guidance study in air-filled HC-PBGF byOliver SCHMIDT et al. [2012b]. Since the viscosity is a temperature-dependent

4Try to imagine you — being cuddled up into spherical shape — are pushed through a narrow water-filledtube at ∼60 kilometres per hour!

63


quantity η = η(T ), this affects the drag force and hence the propagation char-acteristics of the cells according to Equations (3.6) and (3.8). As illustrated inFigure 3.10 C, in our approach, the first 5 cm of the fibre were situated on theglass slide and hence immersed in D2O-PBS which provides both convective andconductive cooling of the outer fibre surface. In the subsequent air-suspendedsection of the PCF, the cooling of the outer surface is less efficient because it isbasically established by convection alone.

To quantify the effect of laser-induced heating along the length of the fibre, areference measurement with a solid polystyrene (PS) microparticle (rP =3.75µm,K1 =4.15 ⇒ ζ=293µm, nPS = 1.57, Duke Scientific) was performed. Accordingto Equation (3.8), deviations from the linear relationship vP ∝ P can only stemfrom a change in viscosity in case a non-deformable particle is propelled. Hence,the velocity of the bead for different optical powers has been measured, see Fig-ure 3.9 B for four of these measurements.

Furthermore, the theoretical temperature-dependent viscosity of heavy-water-based phosphate-buffered saline had to be approximated by taking into accountthe literature values for D2O, H2O [KESTIN et al., 1985] and (H2O-)PBS [SEIYAMA

et al., 1993], as illustrated in Figure 3.10A&B. At a reference temperature of25° C (turquoise box), the viscosity increases by 3 % upon addition of PBS saltsto H2O, so that supposedly (turquoise index finger) the same factor applies whengoing from ηD2O to ηD2O-PBS. At ambient temperature T =20° C this leads to∼1.301 mPa s. Ambient temperature corresponds to the limit of low opticalpower (no heating) in the calibration experiments with the PS bead. Therefore,according to Equation (3.8), the ξ-value characteristic to the bead can be obtainedvia linear extrapolation of the vP

P (P )-data to zero power: ξPS =1.40 pNµm-1 W-1.The viscosity at different optical powers can then be easily retrieved via:

ηD2O-PBS(P ) = ξPSP

vP. (3.9)

A polynomial fit of 5th order applied to the inversely plotted viscosity-tempera-ture relationship T (η) (Figure 3.10B) allows these viscosity profiles η(z;P ) alongthe fibre to be converted into power-dependent temperature profiles T (z;P ), seeFigure 3.10D.

While in the cooled region the temperature never exceeds the upper bodytemperature limit of 42° C for optical powers used with RBCs (P <354 mW),in the air-suspended region (red shading) the temperature increases beyond thisbiologically critical point for powers above ∼225 mW (see Figure 3.10 E).

It is well known that beyond a (body) temperature of 42° C cell death occursdue to protein denaturation. According to various studies reviewed by MUSIELAK

[2009], the RBCs’ deformability is constant in the range between 25° C and 37° C,while beyond 37° C they seem to become more easily deformable.

64


Figure 3.10: A) The temperature dependence of D2O-PBS viscosity (red) is estimated fromliterature values for H2O, D2O [Kestin et al., 1985] and (H2O-)PBS [Seiyamaet al., 1993]. B) Temperature as a function of D2O-PBS viscosity. C) Schematicdrawing of the environment of the fibre. Over the initial 5 cm it is resting on aglass slide and immersed in liquid, which provides efficient cooling. Beyond the5 cm point it is suspended in air where cooling is much less efficient. D) Tem-perature profiles at three different optical powers, obtained from the referencemeasurements with the polystyrene bead (Figure 3.9 B). E) Temperature aver-aged over the range z=5.5 to 12 cm as a function of power in the air-suspendedregion. From the linear fit a heating of h=80.1KW-1 (T0=21.9°C) is retrievedin the uncooled region.

65


3.3.3 Imaging-Free Deformation Monitoring

These reference measurements with the PS bead were used for independent vis-cosity calibration, as needed to obtain unbiased ξ(z;P )-data for the cells. Fig-ure 3.11 shows the results for ten individual disc-shaped RBCs in isotonic D2O-PBS solution (A,B) and for four osmotically swollen (130 mOsm) RBCs (C). Theinitial values of ξ are narrowly distributed around (0.34±0.06) pNµm-1 W-1 forthe disc-shaped RBCs and (0.11±0.02) pNµm-1 W-1 for the rounded ones. Sinceξ is a direct measure of the size, shape and refractive index of the cells, this indi-cates that the initial cell morphology is homogeneous for each cell type, which isin accord with the findings of EVANS & FUNG [1972]. Moreover, the fact that theinitial values of ξ are independent of power suggests that no instantaneous celldeformation due to optical stretching forces takes place. This is in agreementwith the calculations by GUCK et al. [2001], where it is shown that for a largeratio of optical mode diameter to (spherical) cell thickness, optical stretchingforces by a single beam are negligibly small.

Surprisingly however, the optofluidic force parameter changes dynamically asthe cell progresses along the fibre, proving that extensive cellular deformationtakes place. Furthermore, the cells’ individual elastic responses differ notably.For illustrative purposes, in Figure 3.11 we have therefore grouped the ten disc-shaped RBCs into two figures. Four RBCs being propelled at higher opticalpowers P >250 mW, exhibiting temperatures >42° C in the uncooled region areshown in Figure 3.11A, and six RBCs being propelled at moderate optical powersP <250 mW, which proceed at moderate temperatures <42° C in Figure 3.11B.As shown in Figure 3.11A, ξ drastically increases when the cell moves into thehot fibre region, suggesting that excessive heating induces rapid cell deforma-tion. However, a slow increase in ξ can already be observed prior to this criticalpoint. This becomes even clearer in Figure 3.11B, where the rate of this early de-formational behaviour in the cooled region is more diverse, the RBC propelled at237 mW even showing an abrupt increase at ∼3.5 cm. This spread in deforma-tion rate reflects the cells’ individuality. While we could not detect any obviousdifferences between the cells from the two donors, a spread in cell age couldexplain the difference in elasticity [WAUGH et al., 1992; KANETA et al., 2001].

Osmotically swollen RBCs are grouped together in Figure 3.11 C. The three cellslaunched at powers of 292, again 292 and 312 mW were initially spherical inshape, while the fourth, launched at 348 mW, had an asymmetric, irregular ap-pearance. Intriguingly, the initially round cells show a steep increase in ξ alreadywithin the first 2 cm, followed by a slow decrease, before they gain mobility againtowards the hot region (∼47° C). On the other hand, the response of the fourth,irregularly-shaped cell resembles more that of the disc-shaped ones, except for alower initial value of ξ.

66


Figure 3.11: Optofluidic force parameter ξ as a function of propagation distance z for 14individual RBCs, propelled at different optical powers as indicated in the legend(error bars account for mismatch in power between RBC measurement and η(z)-measurement via PS bead). A) 4 initially disc-shaped RBCs propelled at highoptical powers (> 250mW, Tmax>42°C). B) 6 initially disc-shaped RBCs pro-pelled at moderate optical powers (< 250mW, Tmax<42°C). C) 4 osmoticallyswollen RBCs. The cell at P=348mW showed a slightly irregular shape priorto launching.

67


Figure 3.12: (top rows) Top- and front-view microscope images of RBCs taken before (A: reg-ular disc-shape, D: osmotically swollen) and after (B&C) guidance experimentsat optical powers indicated above. (bottom rows) Illustration of the correspond-ing finite element (FE) calculations of surface shear stress τ at vP ≡ 1 µms-1.Table 3.1 summarises the optofluidic parameters obtained.

3.3.4 Theoretical Analysis

The mobility of a laser-driven RBC through the static liquid column of the mi-crofluidic fibre core is determined by a delicate interplay of optical and fluidicforces. The detailed numerical modelling required to understand the dynamicsof this process is beyond the scope of this study. However, to gain more insightinto the deformational behaviour of the RBCs, we controllably rinsed some of thecells out of the fibre core and imaged them in the high-resolution tweezers mi-croscope. The top row of Figure 3.12 (labelled "microscope") shows (A) an RBCprior to launching together with (B-C) cells that have been propelled at differentoptical powers. In (D) an unperturbed osmotically swollen RBC is shown.

For the originally disc-shaped cell propelled at 155 mW (B), i. e., at a maximumtemperature of 34.5° C, the disc becomes elongated by 10 % in the direction ofthe fibre axis and was observed to slowly recover into a circular disc-shape. Strik-

68


ingly, the cell guided at 266 mW irreversibly folded up into a shape resembling acoffee bean (C). The same behaviour was observed in the experiment performedat 354 mW (bright red colour-code in graphs, image not shown).

The dynamical changes in ξ originate from indirectly correlated changes in q

and ζ. While precise theoretical determination of the optical forces is difficult forthe complicated shapes of RBCs [MOREIRA et al., 2012], the drag force param-eter ζ can be fairly easily retrieved for any of the observed ’snap shots’ of RBCshapes using hydrodynamical finite element modelling (FEM) (COMSOL Multi-physics engineering simulation software, version 4.2a). For simplicity, the cellshapes were modelled as solid bodies of simplified geometry under steady-stateconditions. As illustrated in the lower part of Figure 3.12A, the unperturbed cellwas approximated by a toroid with a minor radius of 1.29µm and a major ra-dius of 2.62µm, combined with a disc of 1.25µm thickness to fill up the centralvoid. Average RBC dimensions were used with a disc diameter of dRBC =7.82µmand surface area ARBC =135µm2 [EVANS & FUNG, 1972]. The same geometrybut elongated by a factor of 1.1 was used to account for the stretched cell (Fig-ure 3.12B); both volume and surface area remain approximately equal to that ofthe unstretched cell (less than 1 % change). The folded cell (Figure 3.12 C) wasapproximated by a prolate ellipsoid with a semi-minor axis of 2.15µm and asemi-major axis of 4.76µm, as obtained from the microscope image for the cellpropelled at 266 mW. In this folded case, the volume of the cell was assumed tobe equal to that of the unstretched cell.

The core was modelled as a cylindrical tube with a diameter of 17.4µm, cor-responding to the hydraulic diameter of the slightly hexagonal fibre core. Thelength of the cylindrical tube was chosen to be 20 times its radius, an aspect-ratiowhich avoids any influence of the axial boundaries [AL QUDDUS et al., 2008]. Inaccordance with experimental observations, the long axis of the cell was keptcollinear with the fibre axis, while moving at constant speed. A particle speedof 1µm s-1 was assumed in the modelling to allow for better comparison of thefluidic force parameters between the different cell shapes.

Taking account of the temperature-dependent viscosity in the experimentsηexp, the surface shear rate γ was then calculated using a standard function in thesoftware (spf.sr), which solves the 3D NAVIER-STOKES equations for an incom-pressible fluid at no-slip boundary conditions. The surface shear stress τ is theneasily found by multiplication with the given viscosity: τ = η · γ. For the unper-turbed (A) and the stretched RBC (B), the shear stress is inhomogeneous alongthe cell, attaining a maximum value of τmax (see Table 3.1) on the side closest tothe wall of the fibre core. As a result the cell stretches in axial direction. The netfluidic drag force Fdrag was obtained by integrating the shear stress over the en-tire surface of the RBC. For symmetry reasons, only the force component alongthe direction of propagation remains non-zero. From Fdrag, ζ can be obtained

69


using Equation (3.6). In addition, the actual drag force in the experiments F expdrag is

calculated by multiplying the modelled forces by the actual observed velocity vP,because in this simplified solid body model the relationship between drag forceand velocity is linear (see Equation (3.6)).Table 3.1 summarises the (experimental) optofluidic, the (calculated) fluidic

and the deduced optical force parameters, as well as the maximum surface shearstress both at a speed of 1µm s-1 and at the propagation speed for which theparticular shape was observed. The values obtained for the optical forces areon the same order of magnitude as found for other types of cells under slightlydifferent experimental circumstances [ODDE & RENN, 2000; GUCK et al., 2001].

Table 3.1: Force parameters for microscopically observed RBC shapes (Figure 3.12 ).

A B C D

parameter unit circular 10 % stretched folded osmotically

disc disc "coffee bean" swollen

P mW 78† 155 266 292‡

ηexp mPa s 1.07 (28° C) 0.93 (34° C) 0.77 (43° C) 0.93 (34° C)

ξexp pNµm-1 W-1 0.33± 0.05 0.66 0.93 0.11± 0.02

vP observed µm s-1 31† 110 333 29‡

... at position z cm 0.25† 6.25 5.25 0.25‡

ζFEM(1µm s-1) µm 135 129 99 197

τmax(1µm s-1) mPa 1.26 1.00 0.61 1.06

τmax(vP) mPa 39 110 203 29

q = ξ ·ζ pN W-1 45 85 92 22

Fopt = F expdrag pN 4 13 25 6

† Here, the initial value at P = 78 mW was used as an example.‡ Initial value at P = 292 mW.

Our results show that the optofluidic mobility of a laser-driven RBC is indeedan intricate combination of optical and viscous drag forces. For instance, al-though the drag force parameter for the stretched cell is reduced by only 5 %compared to the unperturbed cell, the large increase of ξ implies that the opticalforces must have increased by a factor of ∼2 due to stretching. However, if thestretched cell folds, q increases by less than 10 % only, whereas on the otherhand the drag force parameter decreases by ∼25 %.

The large increase in q between the disc-shaped RBC and the folded RBC isreasonable, since a larger and more curved fraction of the cell membrane islocated close to the intensity maximum at the centre of the waveguide, enhanc-

70

Conclusions 3.4

Figure 3.13: A) Backscattered intensity obtained by the photodiode (Doppler velocimetryraw signal) as a function of propagation distance of some of the RBCs. Anincrease indicates an accompanying increase in optical force due to simultaneousdeformation of the respective cell, as replotted in B).

ing the overall scattering force. This is corroborated by variations in backscat-tered intensity recorded with the photodiode in the DOPPLER velocimetry unit,which we could observe for some of the RBCs. As demonstrated in Figure 3.13Afor the cells propelled at 237, 266, 354 and again 354 mW, the (normalised)backscattered intensity increases in step with the optofluidic force parameter(Figure 3.13B) — notably, the two cells propelled at 266 mW (yellow circles) and354 mW (red squares) were the ones observed to fold into "coffee bean shape".The absolute backscattered intensity does however not allow for quantitativecomparisons of the optical forces between the individual experiments because itcritically depends on the alignment of the setup, such as for instance the exactaperture of the iris diaphragm used for spatial filtering.

3.4 Conclusions

Liquid-filled hollow-core photonic crystal fibre allows single cells to be laser pro-pelled over distances of 10s of cm under the action of optical forces of ∼50 pNper Watt (refer to q in Table 3.1). This unique feature may find applications inon-chip cell transport, for instance in cases where fluidic transportation is notpossible. Moreover, with HC-PBGFs it is even possible to guide cells aroundtight bends, as suggested by experiments with solid beads [GARBOS, 2011, ch.5.4]. The velocity of the cell can be conveniently and accurately measured witha non-imaging DOPPLER-velocimetric technique which circumvents cumbersomehigh-resolution imaging data acquisition and elaborate image analysis. More-over, omittance of a large microscope facilitates a fully integrated, miniaturisedbioanalytical method. The DOPPLER technique allows continuous monitoring ofthe optofluidic force parameter, which is a measure of the shape of the laser-

71


driven cell. This allows to detect major dynamic deformations in the movingcells, which in the current setup are caused by mainly two different stress fac-tors.

We suspect laser-induced heating to strongly contribute to cellular deformationin a non-physiological way in the uncooled region of the fibre (P >225 mW andz >5 cm), because it has been found before that dead cells exhibit less resistanceto deformational forces [NÈVE et al., 2010]. Switching to a titanium-sapphire-laser wavelength of around 810 nm and equipping the fibre with a cooling sheathwould avoid excessive heating over longer propagation distances. Nevertheless,even at moderate temperature notable deformations could be detected, espe-cially for osmotically swollen RBCs. It is therefore reasonable to assume thatdeformations are induced by the shear forces acting on the surface of the mov-ing cells; these would increase at larger cell-to-core-diameter ratios.

It is interesting to note that the deformations observed in our experimentsoccur over time scales of minutes. This is rather slow compared to other cell rhe-ological techniques, which typically work at higher shear forces but with shorterimpact [MUSIELAK, 2009]. However, exploring the longer time-scale regimemight be important with respect to the fact that RBCs in vivo are constantly flow-ing through the blood vessels. Only occasionally they need to squeeze throughnarrow blood capillaries. In the literature I could only find one study whichperformed biomechanical experiments on single RBCs over similarly extendedtimes. MARKLE et al. [1983] stretched RBCs between two micropipettes andfound that this would also produce slow distortions over time scales of a minute,leading to a permanent re-arrangement of the cytoskeleton.

Besides the time scale, there is yet another thing that distinguishes our novelapproach from the established. RBCs are known to be able to take all kindsof interesting shapes when pushed through microchannels by a pressure-drivenPOISEUILLE flow [SECOMB et al., 1986; NOGUCHI & GOMPPER, 2005; POZRIKIDIS,2005; KAOUI et al., 2009]. In our experiments, in contrast, RBCs are opticallypropelled through a static column of liquid at a good distance from the channelwalls. Intriguingly so, we could observe complete interfolding of RBCs (’coffeebean’ shape), which had previously only been observed in systems where thecapillary is much smaller than the cell itself [SKALAK & BRÅNEMARK, 1969]. Ourresults therefore complement previous studies on pressure-driven transport anddeformation of RBCs by the parabolic profile of a POISEUILLE flow.

In order to achieve full compatibility with optofluidic technology, it is necessaryto incorporate HC-PCFs into existing microfluidic circuitries. We have recentlybeen able to achieve this by constructing a suitable optofluidic interface, with themain scope being to use optofluidic PCFs as photochemical flow reactors [PaperIII: UNTERKOFLER et al., 2012]. This will be the main topic of the next chapter.

72

44Optofluidic Integration of

PCF Photochemical MicroflowReactors for the Analysis of

Light-Activated Drugs

"A little fire is quickly trodden out;Which, being suffered, rivers cannot quench."

’Henry VI’, William SHAKESPEARE (1564–1616)

4.1 Framework

The study presented in this chapter demonstrates the usefulness of optofluidicPCFs as photochemical microflow reactors for analytical chemistry. In the past,several experimental studies have shown that optofluidic PCFs are potent pho-tochemical reactors by means of in-fibre absorption spectroscopy on a restingsample. Efficient photoconversion of low-quantum-yield reactions [CHEN et al.,2010] as well as in situ kinetics monitoring of photolytic [KHETANI et al., 2008],photocatalytic [CUBILLAS et al., 2012] and of photoswitching reactions at sub-picomole sensitivity [WILLIAMS et al., 2012] were successfully demonstrated.Yet, intrinsic to absorption spectroscopy is its limited information content con-cerning the structural determination of the reaction products. Such is typicallyachieved by nuclear magnetic resonance (NMR) spectroscopy, electron param-agnetic (EPR) spectroscopy or mass spectrometry (MS). Hence, to achieve betterknowledge of the products of a photo-induced reaction, PCF photochemical re-actors need to be combined with one of these more sophisticated, non-opticalanalytical schemes in a continuous-flow setting. In addition, the online inte-

73

Chapter 4 OPTOFLUIDIC INTEGRATION OF PCF PHOTOCHEMICAL MICROFLOW REACTORS

gration of excitation and detection inherently allows for a more rapid analysiscompared to the conventional batch approach which comprises the irradiationof a sample solution in a cuvette over extended periods of time (up to hours).

The immediate analysis is especially advantageous in the study of the reactionpathways of photoactivatable drugs, the development of which is the field of re-search of our collaborating group headed by Peter SADLER at the University ofWarwick (UK). To enhance analytical, handling and cost efficiency in the screen-ing of these drugs, it is inevitable to achieve a more rapid analysis combinedwith minimised sample consumption – a perfect job for optofluidic PCFs!

Implementation of our technique via the construction of an appropriate optoflu-idic interface has proven successful for a model photoreaction and was the sub-ject of Paper III: UNTERKOFLER et al., 2012. In combination with a high-resolutionmass spectrometer, we could both accelerate the analysis and reduce the sam-ple consumption by a factor of 50 as compared to the conventional cuvette-irradiation approach. In a follow-up study [Manuscript IV: MCQUITTY et al., 2013,in preparation] we have further developed our technique to study the mechanismof action of potential photoactivatable drugs. These intriguing compounds shallbe briefly introduced in the following paragraphs.

Prodrugs for Photoactivated ChemotherapyMetal-based drugs have found wide-spread use in medicine, for instance in thetreatment of arthritis (Au-based), ulcers (Bi-based), or a sudden rise in bloodpressure (FeII-based: sodium nitroprusside) [GUO & SADLER, 1999]. Yet, themost important application of these drugs is anticancer therapy where cisplatin1

and its analogues are to date the most widely used of all chemotherapeuticals[SIDDIK, 2003]. These platinum-based drugs are known to bind to guanine (G)bases in the DNA so as to induce certain intra- and interstrand crosslinks whicheventually lead to cell death. However, several tumours exhibit resistance, bothacquired and intrinsic, to cisplatin treatment. One mechanism of resistance in-volves inactivation of the drug caused by binding to glutathione (GSH) and otherintracellular thiols (–SH groups). Moreover, cisplatin has severe side-effects in-cluding contingent hearing loss and critical toxicity to the kidneys.

A way around the intoxication of the whole body is tumour targeting [TOR-CHILIN, 2000; VANNEMAN & DRANOFF, 2012], the idea of which had alreadybeen proposed 100 years ago by Paul EHRLICH, the pioneer of chemotherapy2

[EHRLICH, 1913]. A particularly useful approach to not only spatially but also

1or more exactly: cis-[PtII(Cl)2(NH3)2] – cis-diamminedichloroplatinum(II)2”[Die Leitgruppe] soll das therapeutische Agens wesentlich nur zu den Parasiten, aber nicht zu den Kör-

perzellen führen. [...] [Der Stoff] würde in diesem Sinne genau den Immunprodukten des Organismusentsprechen, die ihrerseits nach Art von Zauberkugeln ihren Feind, den Parasiten, isoliert treffen.” [EHRLICH,1913, p. 349]. Translation: ”[The head group] needs to guide the therapeutic agent exclusively to the par-asite, but not to the cells of the body. [...] In this respect it would exactly match the immune products of theorganism, which themselves – alike magic bullets – would target only their enemy, the parasite.”

74

Framework 4.2

temporally target a tumour is photodynamic therapy (PDT), which, for example,comprises the treatment of cancer with a so-called photosensitiser [SVANBERG

et al., 2010; ROBERTSON et al., 2009], see Figure 4.1 . Upon irradiation, suchan organic compound builds up highly reactive species, which then attack theirradiated cells [PHILLIPS, 1995]. However, this mechanism relies on the pres-ence of oxygen, yet a key feature of solid-cancer metabolism is hypoxia, that is, astate of oxygen depletion as compared to healthy tissue [WARBURG, 1956; HAR-RIS, 2002]. This intrinsic contrariety strongly limits the principal merits of PDT[BROWN & GIACCIA, 1998].

Figure 4.1: Principal scheme of desired spatio-temporal tumour targeting in PDT and PACT.A patient with a solid tumour (brown) (1) is injected with a photosensitiser orprodrug (pink) which, if targeted, is only accumulated in the tumour (2). Uponirradiation the agent is activated and kills the cancer cells with high specificity(3) to finally release the patient as cured (4).

Consequently, it is highly worthwhile to combine these two important con-cepts of medical treatment by developing novel metal-based photoactivatable(pro)drugs, which do not rely on the presence of oxygen. This approach isnamed photoactivated chemotherapy (PACT) in order to distinguish it from tra-ditional PDT [FARRER et al., 2009]. To achieve this goal, transition-metal-basedcomplexes are particularly promising, because they are both well-known for theiranticancer- [BRUIJNINCX & SADLER, 2008; FRICKER, 2007; DESOIZE, 2004] andfor their photoactivity [PATMORE, 2008]. In contrast to organic molecules, metalcomplexes have excited states corresponding to the UVA to VIS range of thespectrum. Upon excitation intricate decay pathways are taken. At any stageduring the decay, photochemical reactions might occur, such as ligand dissocia-tion/substitution or redox processes leading to radical formation [FARRER et al.,2009; PATMORE, 2008]. Besides synthesis and cell testing, the development ofPACT prodrugs relies on the understanding of the mechanism of action of thesenovel compounds. With our new technique we have taken a step closer to the insitu determination of their photoreaction products.

75


4.2 Instrumentation

The fabrication of the fibre, the construction of an optofluidic interface mountand the implementation of the setup, as well as running the optofluidic side ofthe experiment was my task, while my colleague Ruth MCQUITTY was in chargeof preparing the samples, operating the mass spectrometer and analysing themass spectra.

4.2.1 Materials and Sample Preparation

FibreThe fibre used for this study needed to be custom-designed and -fabricated in or-der to fulfil both certain optical and fluidic requirements. Firstly, light guidancein the blue region of the spectrum3 had to be established when filled with aque-ous medium. This requires both a small pitch Λ and a very thin web thickness δsince light of short wavelength couples easily to glass resonances [BENABID et al.,2009]. Secondly, high flow rates are facilitated at lower pressures if the overallholey structure is large, see Equation (2.8). Thirdly, the fibre had to be equippedwith a specifically thick jacket tube to avoid deformations of the fibre (cladding)when screw-tightening the fibre into the optofluidic interface mount illustratedin Figure 4.3 .

Taking all these considerations into account, a kagomé-type HC-PCF is the bestchoice since it has an overall larger structure than a bandgap fibre guiding in thesame wavelength range. The broadband-guidance feature of the kagomé fibre(KF) was not exploited in this specific study, however, it offers the additionalpossibility of combining our non-optical analysis scheme with in-fibre absorption(or other) spectroscopy in the future. In order to achieve guidance in the bluerange of the spectrum when the PCF is filled with aqueous medium, we usedthe refractive-index scaling law for photonic bandgap fibres as a guideline, asintroduced on page 40. Eventually, the obtained, water-filled fibre would showsingle-mode guidance of 405 nm and 488 nm wavelengths. When filled with air,the guidance range4 is between ∼550 and 780 nm, the upper edge of which isfurther in the blue than expected according to the scaling law (870 nm).

A scanning electron micrograph of this particular KF is shown in Figure 4.2 Atogether with a fundamental mode obtained at 488 nm wavelength when filledwith water (Figure 4.2 B). Its loss was determined by cutting back a water-filledpiece of initially ∼1.75 m to ∼0.2 m in several steps and measuring the transmis-sion at 405 nm wavelength. This was done twofold, as can be seen in Figure 4.2 C.

3As mentioned earlier, the excited states of metal-based drugs are typically in the UV. However, due to theextremely high irradiance in the PCF, we are able to excite these transitions (far) off-peak in the blue.

4The lower wavelength was limited by the supercontinuum source; it seems probable that guidancereached further into the blue even for the air-filled fibre.

76

Instrumentation 4.2

Figure 4.2: KF used as photochemical microflow reactor. A) Scanning electron micrograph.dcore=19.7 µm, Λ=10.57 µm, δcore=139 nm, δ=245 nm. B) Fundamental-modeguidance at 488 nm wavelength imaged through the surface of the microfluidicchannel used as optofluidic interface (Figure 4.3 ). C) Loss measurement by cut-back technique. The transmission was simultaneously measured with a powermeter (red) and by imaging the core mode with a CCD (blue).

A power meter served to detect the total transmission (red). However, as indi-cated before, light in the short wavelength range and through a short piece offibre might also effectively be guided in the glass. Therefore, the fibre modewas at the same time captured on a CCD camera and postprocessed in a Matlabprogramme to account for the transmission in the core region only (blue). Thisloss value of L = ( 3.2±0.6) dB m-1 at 405 nm wavelength (Leff =1.36 m) is morereliable. Note that the global absorption minimum of H2O is at 475 nm wave-length: α=1.8 ·10-4 cm-1 [TAM & PATEL, 1979]. Hence, contrary to the situationin chapter 3, in this study the loss is not dominated by the medium but by thewaveguide loss. The obtained value agrees well with the loss values measuredfor core guidance in the air-filled fibre, ranging between 1.2 and 3.7 dB m-1 (at610 and 740 nm, respectively).

Construction of the Optofluidic Interface MountThe main challenge in the implementation of our setup was to find a solutionfor the optofluidic interface: the fibre needs to be connected to a high-flow-rate(∼ µL s-1) microfluidic circuitry in a way that would withstand the correspond-ing elevated pressures (several bar), while at the same time providing opticallyflat surfaces through which efficient incoupling can be achieved. Moreover, wewanted to have the ability to remove or replace the fibre if necessary. The liquidpressure cell introduced in the previous chapter (Figure 3.3 on page 54) providesa good solution with respect to the latter points, however its large dead volumeof ∼50µL constitutes a severe drawback. It would not only lead to enormoustotal dead times on the order of hours but also, due to material expansion, the

77


system would additionally require another couple of minutes in order to achievea stable flow rate.

Therefore, the aim was to find a true microfluidic solution. A good startingpoint was to use an off-the-shelf microfluidic chip (MFC). I chose a through-hole chip with 16 separate straight channels of rectangular geometry, each ofwhich has a cross section of 200µm×100µm and a length of 18 mm, and hencea channel dead volume of only 360 nL (01-0163-0142-02, microfluidic Chip-Shop), see Figure 4.3 A. It is fabricated from Topas®, a hard cyclic polyolefin co-polymer with good (bio)chemical compatibility and optical transparency even inthe UV region (α =0.25 dB mm-1 from 300 to 900 nm) [TOPAS ADVANCED POLY-MERS GMBH, 2006]. The chip is tightly bonded to a 140µm thick cover lid fromone side. The channel in- and outlet holes (dhole ≈500µm) can be accessed fromthe other side.

Figure 4.3: Technical drafts of the optofluidic interface mount. One of 16 microfluidic chan-nels on the off-the-shelf chip is used at a time. Dead volume: 360 nL. A) Slantedview from optical access side. B) Side view. Standard PEEK tubing fittingsare screwed into the custom-designed aluminium fibre coupling stage mount andagainst the holes of one of the microfluidic-chip channels to provide a leak-tightconnection (sealed with teflon tape).

An aluminium holder into which this chip can be inserted was specifically de-signed and fabricated as illustrated in Figure 4.3 . One of the 16 channels isaligned in vertical direction, with the covered side of the chip oriented towardsthe aluminium plate which fixes the chip to the holder (A). The fibre is theninserted orthogonally to the chip via the rear, open channel hole (B) to provideoptical access through the chip’s cover lid. The geometry is chosen such that thefibre facet is at appropriate beam height once the holder is bolted down to a fibrecoupling flexure stage.

For a leak-tight connection, the fibre end is encased in a tightly fitting 1/32"tubing sleeve, which in turn is surrounded by a 1/32" PEEK tubing connector(VICI AG). This connector is tapped with external threads to be readily screwedinto the aluminium mount and thereby pressed against the microfluidic chip.Small strips of teflon tape are used for additional sealing.

78

Instrumentation 4.2

The surface roughness of the MFC is on the order of 0.5 to 1µm RMS [MI-CROFLUIDIC CHIPSHOP, 2012], which allows sufficient incoupling efficienciesinto the core mode, typically between 50 and 75 %.

Two of these optofluidic interface mounts are used at either end of the fibre toform a closed circuitry.

SamplesAll of the sample solutions were prepared by my colleague Ruth MCQUITTY,based on doubly de-ionised water (with addition of methanol as a spraying agentin some cases) and at concentrations given in the respective section (typically onthe order of 100µM). When setting the pH care was taken to use strong acidsand bases in order to keep the salt concentration in the samples low, which isimportant not to accumulate debris in the mass spectrometer inlet, which mightlead to blockage.

Cyanocobalamin (CNCbl), aquacobalamin (H2OCbl) and the target moleculesdiscussed later in this chapter (section 4.3.2) are commercially available (CNCbland H2OCbl: SigmaAldrich, target molecules: Acros Orange).

The potential PACT prodrugs were synthesised by members of the SADLER-group:• trans,trans,trans-[Pt(N3)2(OH)2(pyridine)2] – trans-diazido PtIV

by Ruth MCQUITTY and Nicky FARRER (original protocol by Fiona MACKAY),• [(η6-indan)RuCl]2[µ-2,3-bis(2-pyridyl)pyrazine]2+ – dinuclear RuII arene

by Abraha HABTEMARIAM.

4.2.2 Experiment

SetupA schematic diagram of the setup is shown in Figure 4.4 , photographs in Fig-ure 4.5 .

The optical setup is deliberately held as small and simple as possible, see Fig-ure 4.5 A. All the components are mounted to a 60 cm×90 cm breadboard whichis placed onto a trolley just matching the height of the mass spectrometer inletport. As a light source one of two diode lasers can be used, one at wavelength405 nm with a maximum output power of 3 mW (LDM405, Thorlabs), the otherat 488 nm with a maximum output power of 20 mW (iBeam, Toptica Photon-ics AG). The beam is directed via a few mirrors to the fibre incoupling stage(NanoMax, Thorlabs). The outcoupling stage (MBT, Thorlabs) can be fixed invariable distances from the incoupling stage by sliding it along a rail to keep thefibre piece straight independent of its current length. Typically, the fibre wasbetween 15 and 25 cm long. The outcoupled power can then be measured witha power meter (FieldMate, Coherent) or the mode directly imaged by a CCDcamera (uEye, iDS) via another mirror and an attenuation wheel.

79


Figure 4.4: Scheme of the experimental setup (not to scale). The microfluidic circuitry(pink/orange) has a total dead volume of 10-15 µL, depending on the experiment.

Figure 4.4 and Figure 4.5 B focus on the microfluidic circuitry. To obtain high-est sensitivity in the mass spectrometer (maXis, Bruker Daltonics), a continuousflow rate of φtot =27.7 nL s-1 needs to be established. This is done with a syringepump (500µL glass syringe: 1750CX, Hamilton, pump: KDS 100, kDScientific)and measured with a microflow meter (SLG1430-150, Sensirion) just before en-tering the machine. Operated at a readout frequency of 50 Hz, this allows moni-toring the flow rate with a resolution of 0.4 nL s-1 during the entire course of theexperiment. This serves to detect flow instabilities and the occurrence of bub-bles, which would negatively affect data acquisition. The output of the deviceduring the initial filling of the microfluidic circuitry is shown in Figure 4.6 . Thetypical time for the flow to reach its steady state is on the order of 4 min.

While in the first experiment on vitamin B12 [Paper III: UNTERKOFLER et al.,2012] we were using fibre pieces on the order of L ≈25 cm and total circuit deadvolumes of Vdead ≈15µL, these could be reduced to L ≈15 cm and Vdead ≈10µLin the experiments on light-activated drugs [Manuscript IV: MCQUITTY et al.,2013, in preparation] by cutting back both the fibre and the tubing. It appeared

80

Instrumentation 4.2

Figure 4.5: Photographs of the setup. A) Total view of the movable breadboard next to themass spectrometer. The optical path of the 488 nm diode laser is denoted withblue arrows. B) Zoom-in on the microfluidic circuitry (compare to Figure 4.4 ),which is denoted with pink→orange arrows. C) Close-up of the optofluidic inter-face mount. 81


Figure 4.6: Flow-rate monitoring. Flow instabilities and disturbances caused by bubbles areeasily detected (red circles). About 225 s after the pump is started the optimalflow rate for the mass spectrometer φMS=27.7 nL s-1 is reached. The right-handaxis shows the corresponding flow speed vflow in the fibre core.

that the overall time for data acquisition is not solely determined by the deadvolume of the circuitry, but is significantly increased by intrinsic spray stabili-sation and ion travel times in the mass spectrometer used: typically, the signalstarts to build up after approximately 10 minutes and takes another 5 minutesto reach maximum intensity.

The dead volume of the fibre is on the order of 100 nL cm-1, where only3.4 nL cm-1 amount to the volume in the core. Accordingly, the total volumeflow through the fibre consists of a portion flowing through the illuminated re-gion in the core, φcore, and of a portion flowing through the dark cladding holes,φcladd. In this situation, HAGEN-POISEUILLE’s law (see Equation (2.8)) for laminarflow in a parallel circuit of tubes is valid:

φtot =∑

iφi = |∆p| · π

8ηL

∑i

r 4i = |∆p| · π

8ηL

(dH

2

)4

=φcore +φcladd = π

8ηL· |∆p| ·

(r 4

core +∑

ir 4

i ,cladd

),

(4.1)

where η is the viscosity of water, |∆p| the pressure difference between bothends of the fibre, rcore the fibre core radius and

∑i r 4

i ,cladd includes all the dif-ferent cladding hole radii as retrieved from accurate image analysis (via Im-ageJ) of Figure 4.2 A. From this it follows that the hydraulic diameter of the en-tire fibre is dH =31.6µm and φcladd = 5.6φcore. At the given total flow rate ofφtot = 27.7 nL s-1 (set via syringe pump), |∆p| turns out to be less than 4 bar andthe flow velocity in the core is vflow =1.25 cm s-1. This corresponds to a sampletransit time through the core of Ttrans =12 – 20 s, depending on the length L ofthe fibre piece used.

82

Instrumentation 4.2

By knowledge of vflow, the REYNOLDS number can be calculated via Equa-tion (2.3): Recore =0.25¿2000, which is well in the laminar-flow regime, whereno turbulent mixing can occur.

MixingMixing, however, is important for bimolecular reactions in which we try to findout if the photoactivated drugs would attack certain target molecules (see up-coming section 4.3.2). In our particular case the diffusion constant D of thedissolved compounds is not exactly known, but since it is a function of thehydrodynamic radius rh of the molecule (Equation (2.12)), we can compare itto molecules of similar size, e. g. organic fluorescent dyes for which D can bemost accurately determined [DERTINGER et al., 2007]. This gives a number ofD ∼10-10 m2 s-1. In a system where the compound and the target are alreadypremixed when introduced, TAYLOR-ARIS dispersion does not help, therefore:

τD = (∆x)2

2D= d 2

core

2D≈ 2s, (4.2)

which is still sufficiently smaller than the transit time. Moreover, since we aretransforming the species by irradiation with a non-uniform fundamental mode,additional concentration gradients are created to enhance the overall diffusion.Besides, to attach to a nearby molecule, distances are much shorter than sug-gested by the total travel time across the core. The probability for a hit is addi-tionally increased by larger target-to-compound concentration ratios. In conclu-sion, we estimate that mixing is sufficiently taking place in our experiments.

Let me note that for certain drugs the photoreaction scheme might be differentin case they are not premixed with the target, but pre-irradiated and then mixed.To study this, one could think of introducing the target prior to the sample so-lution. If the inner diameter of the outlet tubing (to the MS) is matched to thehydraulic diameter of the fibre we would end up with a P e number > 10 000.This means TAYLOR-ARIS-dispersion would enhance the diffusion coefficient byover 6 orders of magnitude leading to τ ≈5µs, which is much quicker than thetransit time through the outlet tubing, so that compound pre-irradiation studiesare also possible.

Photoconversion ModellingPrior to each experiment, the laser power P0 at a given wavelength λ0 needs tobe set such that full photoconversion is achieved within the sample transit timeTtrans. Referring to what has been said in section 2.1.2 this depends criticallyon the sample-related photochemical parameters (C0, ε, Φ → Ψ), because theconversion yield

κ= 1− C (Ttrans)

C0= 1−e−k Ttrans (4.3)

83


is determined by the conversion rate constant, see Equation (2.30):

k = ln(10) ·1000cm3

L· Imean · λ

NA h c0·Ψ. (4.4)

The respective photoconversion efficienciesΨ of the investigated molecular speciescan be found in Table 4.1.

Taking the absorptivities of all involved chemical species into account (∑εC),

where C = 1z

∫ z0 C (z ′)dz ′, in the fibre the incoupled intensity I0 decays along its

length z according to

I (z) = I0 ·10−

(L /10+∑

εC)

z

= P0

Acore·e

−(L ′/10+∑

σN)

z,

(4.5)

where L ′ = ln(10)L is the fibre loss in e-base and I0 was substituted by P0/Acore

with Acore =330µm2 being the area of the fibre core.In order to obtain the mean irradiance Imean from Equation (4.5) spatio-temporal

integration over the fibre length L and the exposure time Texp would be neces-sary. However, for the general case in which the concentration for the reactingspecies are time-dependent C =C (t ), this is hardly possible in an analytical way5.Therefore, a numerical reaction kinetics model had been introduced in [CHEN,2010], which is applicable to a resting solution.

To retrieve the transmitted intensity as a function of time the total exposuretime is split into 1000 time steps: ∆t = Texp/1000. For each of these i timesteps, the intensity profile is calculated by successive iteration along the fibreover j =1...100 slices of length ∆z = L/100 via

Ii (z j ) = Ii (z j−1) ·exp[−(L ′/10+α′

i−1(z j−1)) ·∆z]

,

∆NA,i (z j ) =−Φ ·(

Ii−1(z j )

hν0σA ·∆t

)︸︷︷︸

= Nabs (Equation (2.26))

·NA,i−1(z j ),

NA,i (z j ) =NA,i−1(z j )+∆NA,i (z j ),

NP,i (z j ) =N0 −NA,i (z j ),

α′i (z j ) =σANA,i (z j )+σPNP,i (z j ),

(4.6)

where the index A refers to the analyte and P to its photoproduct. The overalltransmitted intensity at the i -th time step then corresponds to Ii (z100) = Ii (L).

To demonstrate the viability of the model the temporal photodecay of a well-established photoreaction on a resting sample was recorded by simply monitor-

5Indeed, starting from Equation (4.3), we end up with a nontrivial recursive problem, because C (t ) ∝ e−k

and k ∝ eC (t ).

84

Instrumentation 4.2

Table 4.1: Photochemical parameters of the examined molecular species at λ= 488 nm.

Species ε in M-1 cm-1 Φ Ψ in M-1 cm-1

CNCbl 4089 7 · 10-4 2.86

(pH 2) [CHEN et al., 2010] [CHEN et al., 2010]

dinuclear RuII arene 1880 10-4 0.19

(aquated) [MAGENNIS et al., 2007] [MAGENNIS et al., 2007]

trans-diazido PtIV 28 0.1 2.8

[FARRER et al., 2010] pers. info. Yao ZHAO†

† This value was retrieved from actinometry data by, and with the help of Yao ZHAO.

Figure 4.7: Photoaquation reaction of cyanocobalamin (CNCbl) to aquacobalamin (H2OCbl).A) Reaction scheme. Upon irradiation the cyano group (pink) dissociates and isreplaced by a water molecule (orange). B) Absorption spectra before (CNCbl) andafter (H2OCbl) irradiation (data reproduced from Chen et al. [2010]). C) Pho-toconversion modelling. The transmitted intensity at an excitation wavelength of488 nm was measured for a stationary vitamin B12 sample of C0=5 µM, irradiatedat I0=54Wcm-2.

85


ing the change in transmission at the excitation wavelength as a function of time.We are using this model by inserting all of the input parameters based on the es-tablished photochemical reaction parameters so that there are no free (fitting)parameters. The model can then be directly compared to the data.

A well known natural photoactive transition-metal complex is vitamin B12, alsoknown as adenosylcobalamin. Cobalamins (Cbls) contain a corrin ring witha cobalt(III) centre to which a range of different ligands might be bound. Incyanocobalamin (CNCbl), the synthetic analogue of vitamin B12, the cyanide-ligand is released upon irradiation to be replaced by a solvent molecule, seeFigure 4.7 A. Depending on the pH of the aqueous solution, either aquacobalamin(H2OCbl, low pH) or hydroxocobalamin (OHCbl, high pH) is formed (pK a = 7.8)[PRATT, 1964; AHMAD et al., 1992]. However, since this behaviour is not rele-vant for the vitamin’s biological function (but rather a peculiar side-effect), itdoes not astonish that this reaction is highly inefficient, with a quantum yield onthe order of 10-4 [VOGLER et al., 1976; CHEN et al., 2010]. Figure 4.7 B showshow the absorption peak shifts to shorter wavelength after excitation (the massspectrometric results of this will be discussed later in section 4.3.1).

For comparison with the model, the fibre was filled with sample solution(C0 =5µM) in the dark and then let come to rest. The irradiation with the488 nm laser was started by opening a computer-controlled shutter (SH05, Thor-labs) and the amount of transmitted light monitored with an amplified photo-diode (SC10, Thorlabs) triggered by the shutter. Figure 4.7 C shows that at aninput intensity of 54 W cm-2 the transmission decreases by 30 % within a fewseconds. The decrease in transmission corresponds to the expected value givenby the overall increase in absorption, indicated by the blue line in Figure 4.7 B.The solid yellow curve denotes a single-exponential fit to the data, while the dashedgreen curve corresponds to our reaction kinetics model without freely-adjustableparameters. It shows reasonable agreement with the data6.

Now that the principal validity of our model has been verified, it is possible tosimplify and accelerate the power approximation for all subsequent flow-reactionexperiments. In cases where the absorption coefficient α = εC is small com-pared to the waveguide-loss coefficient αL =L /10 (as for trans-diazido PtIV) orin case the sum of the photoproduct absorption coefficients equals roughly theoriginal value (as for CNCbl and dinuclear RuII arene), the temporal change inanalyte concentration can be neglected and the concentration set to C ≡C0∀t inEquation (4.5). Please note that with these simplifications the estimate for theminimum intensity required for full conversion would come out larger than nec-

6The slight mismatch in rate constant between the model and the data can only be explained by theactual incoupled intensity being larger than the assumed 54 W cm-2. A deviation from the originallydetermined I0-value is possible, because there might have been some bubbles in the cladding during thewater reference measurement (leading to higher fibre loss), which have then been pushed out duringthe introduction of the sample solution.

86

Instrumentation 4.2

essary. This means that total photoconversion in the core is definitely guaranteedby application of the derived irradiance value.

The mean irradiance I mean, which a sample volume element experiences dur-ing the transit time, is then retrieved via simple temporal integration over

I (z = vflow · t ) = P0

Acore·exp

−vflow ln(10) · (L /10+εC0)︸︷︷︸=:α′

tot≈const.

·t

= P0

Acore·e−vflowα

′tot t . (4.7)

As a result we get

Imean = P0

Acore· 1−e−α′

totL

α′tot ·L

. (4.8)

Equations (4.3), (4.4) and (4.8) were combined in an automised calculation im-plemented in a Matlab programme to then set the laser power such that fullphotochemical conversion, κ →1, is achieved for each individual experiment.This takes also into account the respective incoupling efficiency (including thetransmission loss of the objective lens of around 7 %).

Mass SpectrometryAmong the earlier mentioned analytical techniques, electrospray-ionisation massspectrometry is most readily combined with optofluidic devices, because it isbased upon a continuous and fluidic sample supply. It allows to measure themass-to-charge ratio m/z of charged molecules based on which their elementalcomposition can be derived [ASTON, 1919]. State-of-the-art machines allow thestudy of biomacromolecules on the order of 100 kDa in mass7 [MCLAFFERTY,2011; EL-ANEED et al., 2009].

Every mass spectrometer consists of three basic modules, see Figure 4.8 . Firstof all the dissolved molecular species need to be vaporised and charged in anion source. A mass analyser then serves to separate the ionised species based ontheir movement in an electromagnetic field which — breaking it down to thevery basics — combines the effects of LORENTZ’ with NEWTON’s 1st law:

F L = z(E +v ×B )

F N = ma

}⇒ m

za = E +v ×B , (4.9)

where E and B are the external electric and magnetic fields, a and v are theparticle’s acceleration and speed. Finally, the ions hit the mass detector whichintegrates and amplifies their impacts.Figure 4.8 illustrates the instrument used in this study. It is an ultra-high res-olution electrospray-ionisation tandem quadrupole–hexapole time-of-flight in-

71 Dalton (Da) is equivalent to 1 atomic mass unit u = 1.66 · 10-27 kg, which is conventionally used inbiochemistry to denote the molecular mass M . For instance antibodies, which are composed of severalprotein chains, lie in the 100 kDa range.

87


Figure 4.8: Simplified scheme of a UHR ESI Qq-TOF mass spectrometer.

strument, in short: UHR ESI Qq-TOF MS (maXis, Bruker Daltonics) [BRUKER

DALTONICS INC., 2008]. ESI is an especially soft ionisation technique, whichdramatically reduces fragmentation of biomacromolecules during the ionisationprocess. Deliberate fragmentation of a certain m/z-species is however useful toretrieve additional information on its structural composition [GRIFFITHS et al.,2001; COLTON et al., 1995]. To this end, the mass analysers are actuated inMS/MS mode. In this case the first quadrupole-analyser ("Q") serves as ion-filterportion and the second hexapole-analyser ("q")8 serves for fragmentation by col-lisions with a stream of inert gas. A subsequent time-of-flight tube provides addi-tional, high-resolution separation. In combination with a highly-sensitive time-gated detector, this hybrid analyser scheme achieves an extremely high massaccuracy (∼600 ppm) at ultra-high mass resolving power9 of ∼50 000. This isnecessary for the unambiguous assignment of large molecules, because the com-plexity of isotopic composition increases with molecule size. For the compoundsused in this study (molecular masses up to 1355.38 Da) the isotopic fine struc-ture is resolved at peak-to-peak distances of less than 0.03 Da.

Despite the high quality of molecular identification, large-molecule MS is typ-ically a non-quantitative technique. The reason is that the highly inhomoge-neous distribution of masses and charges in the molecules leads to delicate dif-ferences in their spatio-temporal flight properties in the complex analyser po-tential, which are hard to understand analytically. Even variations in sprayingcondition might lead to differences in the relative height of the peaks. Therefore,MS does not tell about the reaction efficiency or kinetics, but it is an excellenttool to qualitatively determine the products of a photochemical reaction.

8Being a hexapole in fact, "q" by convention denotes the analyser associated with the collision cell.9The mass resolving power RM is defined as the ratio of the mass of a species over the minimum resolvable

distance between neighbouring peaks: RM = M∆M .

88


4.3 Results and Discussion

4.3.1 Photoaquation of Vitamin B12

The reaction scheme of the photoaquation reaction of cyanocobalamin (syntheticversion of vitamin B12) to aquacobalamin was already shown in Figure 4.7 A.

Figure 4.9: Results of vitamin B12 photoaquation. Comparison between cuvette (A)and fibre approach (B). For illustrative purposes, only the small region from662m/z to 682m/z is shown. C) Theoretically predicted isotopic pattern (viawww.chemcalc.org) for doubly-protonated cyanocobalamin [CNCbl+ 2H+]2+,which agrees nicely with the experimentally observed.

Figure 4.9 shows the mass spectra obtained. The top row (A) refers to directly-injected samples, where reference irradiation experiments were performed in acuvette; the bottom row (B) shows the outcome of our microfluidic HC-PCF-approach. Molecules of the size of CNCbl (M = 1355.38 Da) typically appearas multiply charged species in the MS. Figure 4.9 left shows the mass spectra ac-quired in the dark for CNCbl, in which the sample is either directly injected intothe mass spectrometer (A) or pumped through the non-irradiated fibre circuitry(B). As expected, they are equivalent, which proves that no unwanted effectstake place in our fibre approach. The peak pattern at around 678.3 m/z (pinkdot), which is a result of the isotopic distribution of carbon, nitrogen and oxygenin the molecule, refers to the doubly-protonated species [CNCbl + 2H+]2+, andfits well to its theoretical prediction (Figure 4.9 C).

The spectra in Figure 4.9 right correspond to samples of an aqueous solution of100µM CNCbl (pH<2), irradiated at 488 nm wavelength. According to the sim-plified reaction kinetics model a moderate input power in the fibre experiment of

89


P HC-PCF0 =2.35 mW is sufficient for >99 % photoconversion. Despite a doubled

input power of P cuv0 =4.5 mW, the intensity of light in the cuvette is more than 4

orders of magnitude lower compared to that in the fibre, i. e. I cuv0 =40 mW cm-2

vs. I HC-PCF0 =700 W cm-2. As a result the cuvette experiment took over 10 hours,

in marked contrast to the fibre approach. Moreover, whereas in a cuvette ∼1 mLof total sample volume is typically used, 25µL was sufficient in the integratedcase.

The mass spectra from both irradiation approaches are in good agreement(Figure 4.9 right). In addition to the CNCbl series of peaks, new peaks are clearlyapparent at 664.8 m/z (blue hexagon), 673.8 m/z (orange star) and 675.8 m/z(green square), assigned as [Cbl+ + H+]2+, [H2OCbl+ + H+]2+ and [Cbl+ + Na+]2+

and indicating successful photoconversion in both cases. While aquacobalamin(H2OCbl) is the only photoproduct to be expected [AHMAD et al., 1992], ligand-free cobalamin (Cbl) is also present. This seems to be an artefact from theionisation process, in which the weakly-bound H2O ligand is lost. We verifiedthat the series of new peaks corresponds to the expected photoproduct by di-rectly injecting an aqueous sample of pure H2OCbl into the mass spectrometer,as shown for reference in Figure 4.9 A centre. Cyanocobalamin is still present inboth the cuvette and the fibre spectra, though for different reasons. In the cu-vette case, the reaction was most probably incomplete. In the fibre case, theflow of analyte through the (dark) fibre cladding holes is about 5.6 times higherthan through the irradiated fibre core, which simultaneously introduces unre-acted species. This is however not of greater concern, since mass spectrometry isintrinsically a non-quantitative technique, so that an intermixture of the productand the analyte do not diminish the power of our approach.

4.3.2 Photoactivation of Potential Anticancer Compounds – AttackingCellular Targets

trans-diazido PtIV compound

Figure 4.10: Proposed primary reaction scheme for the trans-diazido PtIV compound accord-ing to Farrer et al. [2010].

90


The trans-diazido PtIV compound (M = 471.33 Da) shown in Figure 4.10 left wasoriginally synthesised by Fiona MACKAY in the SADLER-lab and its (photo)chem-ical properties were extensively studied by Nicky FARRER et al. [2010]. Forthe reasons explained below, it is the (to date) most promising molecule ofall platinum-based prodrugs which undergo photoactivation into a reactive PtII

species [SADLER & MULLER, 2011].

The compound’s absorption maximum is at 294 nm, but it showed to be ef-fectively excitable also with blue and green light. This fact is supported by abinitio density functional theory calculations which support the presence of weakabsorption bands in the visible region of the spectrum (∼414 nm) [FARRER et al.,2010]. This is a unique feature of this specific compound and important for theintended use in photoactivated therapy, because UV light does not penetrate verydeep into the tissue (∼1 mm) and causes DNA damage in (healthy) cells itself.

Both experimental and theoretical findings suggest that upon irradiation, oneor both azide ligands dissociate to form a reactive species (Figure 4.10 right). Thismethod of activation is not reliant upon the presence of molecular oxygen, unlikeconventional photodynamic therapies, and could therefore be more effective inthe hypoxic regions of tumours than standard PDT. On the contrary, the pyridineligands (py) are rather stable.

At low light doses trans-diazido PtIV was demonstrated to be potently photo-toxic towards several different human cancer cell lines, such as parental (A2780)and cisplatin-resistant (A2780CIS) ovarian carcinoma, oesophageal adenocarci-noma (OE19) and hepatoma (HepG2) cells. Importantly, the viability of thetreated cells was not affected as long as they were kept in the dark, whichproves that the unactivated prodrug is non-toxic. In additional in-vitro experi-ments, the compound has been found to form strong interactions with guanosinemonophosphate10 (GMP) [FARRER et al., 2010]. In a subsequent study, effectivebinding of trans-diazido PtIV to DNA could also be verified, both in the cell andin vitro. The conformational alterations are distinctly different than those in-duced by conventional treatment with cisplatin (and analogues) and the lesionsformed were shown to inhibit DNA transcription more efficiently [PRACHAROVA

et al., 2012].

All these finding do not rule out the possibility for other, potentially short-lived,photoproducts of trans-diazido PtIV to contribute to its phototoxicity. Therefore,additional investigations of this auspicious prodrug are highly recommended.

We therefore conducted photoexcitation experiments with our integrated tech-nique and compared them to previously obtained cuvette-based data. In the cu-

10GMP is one of the nucleotide monomers that constitute nucleic acid strands via phosphodiester bonding.

91


Figure 4.11: Results of trans-diazido PtIV compound (©) photoactivation. A) Cuvette ref-erence measurements. B) PCF nanoflow reactor measurement. C) Theoreticalisotope pattern (via www.chemcalc.org) of the compound’s sodium adduct[©+Na+]+ compared to the experimental peak pattern.

vette, a 10µM sample in 50/50 H2O/methanol11 was irradiated for 30 min witha UV-lamp array (LZC-420, Luzchem Research) at (420±20) nm wavelength andwith a total output irradiance of 23 mW cm-2 [FARRER et al., 2010, SUPPORTING

INFORMATION]. In the fibre we used a 100µM sample in 90/10 H2O/methanol,irradiated at 488 nm and I0 = 243 W cm-2. The photochemical parameters of thecompound were shown in Table 4.1 on page 85.Figure 4.11 shows the mass spectrometric results. Upon direct injection of the

sample in the dark (Figure 4.11A left), peaks corresponding to the compoundmonomer species at 494.1 m/z (sodium adduct: [Pt(N3)2(OH)2(py)2 + Na+]+)and dimer species (protonated: 944.1 m/z, sodium adduct: 966 m/z) are present(all pink). In addition, we find degradation-product peaks, that is, peaks withm/z-ratios below the monomer peak12 (pink dot) and with the typical isotopicpattern of platinum (see Figure 4.11 C). These species are formed due to frag-mentation of the original complex during the sample ionisation process. Ex-cept for obvious ligand losses, they are typically hard to assign, because theunderlying chemistry is extremely intricate. In addition, some peaks overlapand/or have rather low signals. The dominating peak at 387.1 m/z (orange star)

11Methanol is typically used as a co-solvent in ESI MS because it facilitates spraying and helps to increasethe MS signal. So far no deviations from the intermixture of H2O with methanol have been observed incomparison to compound in pure H2O.

12For a small molecule like trans-diazido PtIV only singly-charged ions are produced in the ESI process.

92


is [PtC10N2O2H12]+, while the other peaks remain unassigned. After irradiationin the cuvette (Figure 4.11A right), the primary photoproduct emerges to be thepeak at 387.1 m/z (orange star), which corresponds to the expected primary pho-toproduct (Figure 4.10 right), that is the compound having lost both its azide (-N3)groups to form [PtII(OH)2(py)2]+.

In the fibre approach (Figure 4.11B), several problems occurred. First of all,we were not able to get a sufficiently large signal in the mass spectrometer,which is necessary to unambiguously determine the elemental composition ofthe peaks by the isolation and fragmentation of species. Trying to improve thisby increasing the amount of methanol in the solvent (up to 40 %) or compoundconcentration (up to 500µM) was not successful. In the latter case, instead ofincreasing the monomer peak, an unknown species at around 1050 m/z wouldbuild up, which contains two platinum atoms, but is even larger than the nor-mal compound dimer (data not shown). The spectra shown in Figure 4.11B aretherefore recorded at a rather low compound concentration of 100µM, wherethe 1050 m/z dimer species did not appear. Still, as compared to the cuvette(Figure 4.11A left), the dark fibre spectrum shows a stronger peak for the proto-nated dimer species at 944.1 m/z (pink double hexagon). This means that the darkspectra do not perfectly agree for both approaches, which is a peculiar circum-stance.

Even more interestingly, in the fibre-irradiated spectrum, the primary photo-product is not the same as for the cuvette experiment, but is given by a peak at367.1 m/z (green square), which — due to the low signal — cannot be assigned13.This was reproduced in several other runs (not shown). One might speculatethat this peak corresponds to a photoproduct specific to the decay of the dimerspecies since its peak (green square) increases while the (protonated) dimer peakstrongly decreases (pink double hexagon). However, trans-diazido PtIV dimers havenot been found using other techniques such as NMR, which indicates that theyare presumably not build up before ionisation, that is, when the droplets shrinkand force the molecules together. Regrettably so, no definite conclusion can bedrawn about the compound’s photochemistry in fibra at this point.

A possible explanation for the aberrant fibre spectra could be found in surfaceeffects. The volume-to-surface ratio of the fibre-core boundaries, which gives anidea of how much of the sample gets into contact with the walls, is 1.9 cm2/µL,and hence about 400 times larger than for a standard cuvette with 0.005 cm2/µL.

Having this in mind, the loss of signal in the dark fibre spectrum could be aresult of ion exchange at the silica surfaces. As observed before in experimentswith vitamin B12 (data not shown), it appears that in the fibre spectra sodiumadducts are less prominent than in the cuvette spectra. According to Ralf KED-

13A reasonable guess is [PtC5N5O2H10]+ = [©− py−2N + 3H]+, with an expected peak at 367.048 m/z.

93


ING, glass expert at MPL, this is because silica surfaces tend to adsorb sodiumions under the release of protons. It is possible that only sodium adducts of thecompound have the right flight properties in the mass analyser to be efficientlydetected. This also explains why the signal of the fibre spectra is significantlylower than upon direct injection. Furthermore, we cannot exclude the possibil-ity that the neutral and small inorganic molecules themselves are attracted bythe hydrophilic silica walls. The relatively larger amount of dimer species couldthen either be due to dimers not being as sticky as monomers or due to somequasi-catalytic influence of the wall triggering dimerisation.

With respect to the fibre-irradiated spectrum, it seems moreover reasonablethat the highly reactive radical photoproducts of trans-diazido PtIV irreversiblyreact with the core walls and never make their way to the mass spectrometer.This effect is well known in microflow chemical synthesis and is referred to asradical quenching14 [JÄHNISCH et al., 2004, p. 410].

dinuclear RuII areneThe dead end encountered with the trans-diazido PtIV prodrug forced us to pro-ceed in another direction. In other words: we decided to investigate anotherkind of metal anticancer prodrug for which MS signals are expected to be ade-quately high, again. Following the good example of cyanocobalamin, the dinu-clear RuII arene (M = 743.64 Da) shown in Figure 4.12 is likewise a rather largeorganometallic compound with non-radical photoactivated states.

The compound was synthesised by Abraha HABTEMARIAM in the SADLER-laband its photochemistry studied by MAGENNIS et al. [2007]. When dissolvedin water, the compound’s original chloride ligands (black Xs in Figure 4.12 ) areslowly replaced by water (time constant on the order of hours). This is associ-ated with a change in absorption spectrum. The original species has two absorp-tion peaks, at 340 nm and 440 nm, while in the aquated species these peaks areshifted to 335 nm and 520 nm (at 490 nm there is an isosbestic point, which ispractical with respect to our excitation wavelength of 488 nm). The compoundis therefore easily excitable by visible light. Upon irradiation, it is supposed tolose its arenes which is proposed to be followed by water or chloride attachmentto the free binding sites, as illustrated in the scheme in Figure 4.12 . Yet, the ex-act nature of the compound’s photoproducts is probably more complicated (assuggested by NMR spectroscopic experiments). In their studies, it was assuredhowever by degassing the solution that the reaction mechanism does not dependon oxygen.

MAGENNIS et al. [2007] also found that in the presence of DNA, species ofdinuclear RuII arene attach in either case: in the dark, pre-irradiated (and thenmixed) or irradiated (when pre-mixed). Binding occurred primarily at guanine

14In this framework, however, radical quenching is exploited to throttle radical chain-reactions.

94


Figure 4.12: Proposed reaction scheme for the dinuclear RuII arene compound according toMagennis et al. [2007] (preliminary).

sites and each would inhibit DNA transcription. However, the initial bindingrate of the dark species was lower and its adducts were found to inhibit tran-scription more weakly than the ones performed by pre-irradiated and irradiatedcompound. Additional tests suggest that the weak (dark) adduct corresponds toa monofunctional lesion, while the latter (activated) adducts refer to interstrandcrosslinks. Interestingly, the majority of all the investigated DNA-compoundadducts were only singly ruthenated, implying more than just the loss of anarene.

Despite these promising first results in vitro, the compound would not showefficient toxicity towards cells in culture (unpublished results, personal informa-tion from Abraha HABTEMARIAM).

We have tested the photoexcitation characteristics of the dinuclear RuII arenewith our PCF microflow reactor approach15. Moreover, we have also taken thestep towards bimolecular reactions by letting the compound react with potentialcellular target molecules, see Figure 4.13 , which we had added to the compoundsolution prior to injection into the system. For the sake of brevity, only the resultsfor the pure compound (Figure 4.14A), for compound and GMP (Figure 4.14B)and for the compound with GSH (Figure 4.14 C) shall be discussed here. A moredetailed analysis will be available in [Manuscript IV: MCQUITTY et al., 2013, inpreparation]. The irradiance I0 varied between 1240 and 2500 W cm-2, depend-ing on the experiment.

15Reference experiments in the cuvette (Texp ∼12 h) have also been performed, but since no major devia-tions were found their results are not shown.

95


Figure 4.13: Cellular target molecules. 9-EtG: 9-ethylguanine, GMP: guanosine monophos-phate, AMP: adenosine monophosphate, L-cys: L-cysteine, GSH: glutathione.9-EtG, GMP and AMP are used as mimics for DNA. L-cystein is an amino acidand hence used as a mimic for (sulphur-containing) peptides and proteins. Thetripeptide GSH is an abundant intracellular reducing agent (cGSH ≈mM) andplays an important role in drug resistance.

A dark sample of a 500µM solution of pure compound (Figure 4.14A) showsa range of doubly-charged monomer species (pink box), which corresponds to itsdifferent aquation states (replacement of chlorides by water at the black X sitesin Figure 4.12 ). The pink dot for instance denotes the original species containingtwo chlorides; a close-up can be found in Figure 4.14D for comparison with itsexpected isotopic distribution. In addition, two degradation species are found,which lack one of the two Ru atoms and one arene ligand (blue stars). Uponirradiation the monomer species peaks decrease, while one of the degradationpeaks strongly increases and dominates the spectrum (light-blue star).

To some extent the same happens if the dinuclear RuII arene compound is ir-radiated in the presence of GMP (M = 363.22 Da), at a ratio of 1:2 (250µM to500µM). However, new peaks are also present. Already in the dark, a peak cor-responding to GMP only (violet square) and GMP complexed with a compoundspecies (violet-square/pink-dot doublet) appear. This matches well the finding ofMAGENNIS et al. [2007] that the compound would bind to DNA already in thedark. Upon irradiation, this peak increases even further, but also totally newspecies are found. They all correspond to compound species (all void of Cl)which have lost one arene, while both Ru atoms are still present. Interestingly,both single-GMP adducts (red, orange and dark-yellow hexagons) and double-GMPadducts (light-yellow, white and grey hexagons) are found. This confirms that specif-ically upon irradiation the compound can induce crosslinks via G sites in theDNA.

96


Figure 4.14: Results of dinuclear RuII arene photoactivation in fibra. A) Pure compound©2X (X=H2O or Cl). B) ©2X + GMP (1:2). C) ©2X + GSH (1:2). D) ©2Xmonomer-species peaks, including theoretically predicted isotopic pattern (viawww.chemcalc.org) for the sodium adduct of ©2Cl.

97


When introducing the compound premixed with GSH (M = 307.32 Da) at aratio of 1:2 (250µM to 500µM) (Figure 4.14 C), the dominating peak is the onefor GSH (light-green square). Already in the dark, the compound, having lost oneof its Ru atoms, forms a complex with GSH plus a GSH-fragment (light-greensquare/pink dot doublet). This peak however vanishes upon irradiation. Instead,new peaks emerge which clearly indicate photoactivated binding of a compoundspecies lacking one arene. Both a single-GSH adduct (dark-green hexagon) anda double-GSH adduct (dark-blue hexagon) are present. This inactivation of thecompound by GSH (both in the dark and upon irradiation) provides a reasonableexplanation why the compound did not show sufficient toxicity in cell culture.

In conclusion, we were able to successfully run a palette of experiments onthe dinuclear RuII arene compound. These demonstrate the capability of ourapproach to also study bimolecular reactions which require sufficient mixing inthe system. With respect to the chemistry, we were able to corroborate previousmajor findings established by other techniques, while at the same time gaininga more detailed insight into the reaction mechanism of this particular Ru-basedagent. More information will be available in [Manuscript IV: MCQUITTY et al.,2013, in preparation].

4.4 Conclusions

We have introduced a novel integrated technique to enhance the efficiency ofanalytical photochemistry. Our particular focus was on investigating the reac-tion mechanisms of photoactivatable metal-based prodrugs, even in the pres-ence of potential cellular target molecules. To this end, a PCF microflow reactorwas successfully connected to conventional microfluidic devices which served assuitable optofluidic interfaces. The combination of a low-volume microfluidic,continuous-flow circuitry with high optical intensities in the HC-PCF achieves afull photoconversion about 1000 times faster as compared to the conventionalcuvette-based approach.

During the study of several different compounds, it appeared that certainmolecules tend to stick to the silica walls of the microreactor16. To prevent this,we could apply an internal hydrophobic silane layer which for silica HC-PCFshas previously been reported and did not introduce excessive extra waveguidelosses [GHOSH et al., 2006b; XIAO et al., 2011]. To circumvent wall binding ofreactive radical species additional means are necessary. In the same way as in[Manuscript V: CUBILLAS et al., 2013, in preparation] and [SCHMIDT et al., 2013],we could grow catalytic nanoparticles on the inner walls serving to transformexpected radicals into nonreactive stable species. These would then no longer

16My colleagues Jocelyn CHEN [2010] and Gareth WILLIAMS et al. [2013] have studied the wall adsorptionof water-soluble dyes before.

98

Conclusions 4.4

sidestep detection. The same is true if one traps the radicals (or dissociated smallmolecules) with the help of scavenger molecules (and other trapping agents).

In combination with a high-resolution mass spectrometric analysis scheme, theoverall technique showed to be up to 50 times less sample- and time-consuming.Despite this remarkable advancement, we noted that the time needed for data ac-quisition was still about two times larger as expected from the total dead volumeof the circuit. As this could not be improved by cutting back the in- and outlettubing we assume that the spraying conditions are not stable enough, which isa very critical point in MS. To achieve our goal to detect shorter-lived reactionspecies, it is therefore necessary to stabilise the fluidic system even further andto tweak the spraying conditions of the mass spectrometer. With this respect,it would bring special benefits to establish the spraying directly from the chip[KOSTER & VERPOORTE, 2007] – in fact, in our case it seems even more palpableto taper the fibre outlet end down to create the spray directly ex fibra.

It seems clear that our approach can be readily combined with other lab-on-a-chip functional devices [DEMELLO, 2006; FAN & WHITE, 2011] and multiplesample in- and outlets. Apart from introducing units for sample preparation andseparation, our device would strongly profit from implementing a multimodalanalysis scheme (e. g. optical/non-optical). The ultimate goal would be a totally-integrated, transportable lab-on-a-chip device for easy handling by non-experts.

99

55Outlook

"Jede Lösung eines Problems ist ein neues Problem."


The two preceding chapters have clearly demonstrated the huge potential ofoptofluidic PCFs for biomedical research. Their unique features allowed us touse them both as a long-range optofluidic cell guide and as an efficiency-boostingmicroflow reactor for bioanalytical photochemistry.

Naturally, over the past few years we had many more ideas and some of themhave already undergone initial testing in the lab. A couple of these ideas shallnow be briefly exposed and — where possible — substantiated by preliminaryfindings. This serves to give an outlook into what else can (and possibly will) bedone for the life sciences in fibra.

5.1 PCF Optofluidic Microparticle and Cell Guides

5.1.1 Guidance of Single Eukaryotic Cells

Red blood cells, which we investigated in chapter 3, are outsiders among thecells within our body. They are not only small, disc-shaped and void of a 3D cy-toskeleton; once matured they also lack a nucleus and are rather passive in theirbehaviour. In contrast, the other cells in our body are larger (typically at least15µm), spherical to ellipsoidal when in suspended state, and they contain both acomplex cytoskeleton and a nucleus. The latter gave them a common name: eu-karyotes1. Cancer cells are eukaryotic cells which proliferate uncontrolled untilthey invade and destroy both nearby and distant tissues (metastasis).

1greek: ευ meaning "proper" and καρυον "kernel"

101

Chapter 5 OUTLOOK

Figure 5.1: A) When trying to launch a 16 µm large HeLa cell into an HC-PBGF withdcore=19 µm, it got stuck at the core entrance. B) Schematic of a HeLa cellwith cycRGD-inhibited surface receptors, and of a K562 cell void of any surfacereceptors.

Given the importance to find means to prevent or to cure cancer, we have alsotried to apply our PCF cell guide to single eukaryotic cells brought into suspen-sion. To this end a collaboration with the Biophysics Group at the Universityof Erlangen-Nuremberg, headed by Prof. Ben FABRY, was initiated. We startedoff with the most prominent of all cancer cell lines, HeLa cervical cancer cells[SCHERER et al., 1953]. However, HeLa cells are typically larger than the coresize of our fibre (dcore = 19µm), characterised in Figure 2.16 on page 42. Eventhough cells of ∼15µm do exist in the population, they could not be launchedbut got stuck, see Figure 5.1 A. It seems that the attracting surface forces bring thecell sufficiently close to the glass to let it attach with its strong surface receptors.

According to Ben FABRY there were two immediate solutions to the problem2.We could either use the cyclic peptide cycRGD to inhibit the cell’s surface recep-tors [AUMAILLEY et al., 1991] or we could try a special cell line which does notcome with surface receptors at all, see Figure 5.1 B. Since this was available intheir lab, we chose the latter option. K562 cells were originally extracted from apatient with chronic myeloid leukaemia [LOZZIO & LOZZIO, 1975].

When trying to launch K562 cells into the PCF at optical powers of up to130 mW at 810 nm wavelength with the titanium-sapphire laser, it emerged thatin most cases this was still not possible. The cells propagated towards the coreentrance, but would then refuse to go inside. In one case, a K562 cell3 could belaunched successfully, however, it would only proceed at extremely low speeds<1µm s-1, see Figure 5.2 A. The good news is that the cells would definitely not

2A third solution is an internal nonstick coating, which will also be discussed further down.3Since this cell was launched by accident its size could not accurately be determined, but was estimated

to be around 12µm.

102

PCF Optofluidic Microparticle and Cell Guides 5.1

Figure 5.2: A) A small K562 cell optically guided at P=130mW in a piece of fibre about100 µm away from the entrance. The cell travelled 82 µm in 90 s (starting positionoverlaid with cutout from a later frame). B) Drag enhancement factor K1 (left)and drag force (at a unit speed of 1 µms-1) (right) for a spherical object, both asa function of cell radius (top) or cell-to-core ratio (bottom).

stick to the walls. They could easily be removed from the facet by blocking thebeam or be rinsed out of the core by a small counterflow.

The reason for the cells’ idleness becomes clear when recalling that the dragforces for spherical objects in the confined space of the core are highly enhancedby a factor K1, see Equation (3.5) on page 61. Figure 5.2 B demonstrates that forthe range of available optical forces (red shaded area) according to our findings inTable 3.1 on page 70, indeed only K562 cells of smallest sizes around 12 – 13µmcan possibly reach non-negligible speeds on the order of 1µm s-1.

For subsequent experiments, Jun.-Prof. Graeme WHYTE (Optofluidics Group,University of Erlangen-Nuremberg) has therefore proposed the use of the chickenlymphoma B-cells line DT40 (a white blood cell type) [WINDING & BERCHTOLD,2001], because they are only about 10µm in size. Concerning the setup, it isadditionally advantageous to either use a fibre with larger core size or to increasethe available laser power in the guidance beam. Both of this will be discussed insection 5.1.4 dealing with intended improvements of the setup.

5.1.2 Biomechanical Probing of Single Cells

So far, we were able to measure the slow deformational response of cells whenthey were steadily progressing through the liquid column in the core, once moreillustrated in Figure 5.3 A. Beyond this, we can think of two alternative ways toprobe the mechanical behaviour of single cells.

For instance, the cell could be held stationary by balancing the optical forcewith a counterflow, as illustrated in Figure 5.3 B. With this flow, the cell feels anactive shear force and will also deform, possibly in a slightly different mannerthan when being pushed through a stationary column of liquid. The experimen-

103

Chapter 5 OUTLOOK

Figure 5.3: The three possible schemes of cellmechanics research in fibra. A) Eukaryoticcell guidance (as introduced for red blood cells in chapter 3). B) Opto-fluidiccounterbalancing. C) Dynamic dual-beam "ping pong" for frequency-dependentcellmechanical studies.

tal implementation would involve an active feedback loop to keep the cell stillat a certain position. The latter could be monitored by the cell’s side scatteringonto a quadrant photodiode and rapid feedback could be given to the beam in-tensity through adjustment of the EOM voltage. From the feedback signal data,information about the deformation can then be obtained. This scheme is inter-esting, because it allows supplying the cell with precise amounts of chemicals toactively affect its mechanics [EL-ALI et al., 2006; VANAPALLI et al., 2009].

In the introduction of the setup in Figure 3.4 /Figure 3.5 on double page 56/57,it was mentioned that we have implemented a dual-beam option (yellow shading).Via periodic modulation of the EOM voltage, and hence of the intensity ratiobetween the beams, this would allow us to play optofluidic ping pong with thecell, as is shown in Figure 5.3 C. This may lead to a new and accurate way ofdetermining the frequency-dependent elastic modulus of a single cell, which isone of the standard parameters to be determined in biomechanical studies.

5.1.3 Investigation of Optically-Bound Cell Chains

Besides the study of single cells alone, one-dimensional cell-cell or cell-particleinteractions between suspended cells might be studied directly in fibra. Thisis especially interesting for tissue engineering, cell signalling and drug deliverystudies. One could for example make cells (or cells and particles) collide orpress them against each other in the dual-beam scheme. Alternatively, one mightstudy the behaviour of guided chains of cells along the core. Figure 5.4 showsan in-fibre chain formation out of three RBCs. RBC 1 approaches two stationarycells, of which RBC 2 is (weakly) attached to the core wall (1). As RBC 1 comescloser, RBC 2 suddenly detaches and is pulled into the centre (2-4). As bothcome close enough to RBC 3, all three proceed forward as a unit (5-6). Similarbehaviour was observed for eukaryotic cells in a free-space beam already in 1987

104


Figure 5.4: Chain formation of guided RBCs. Δt between each frame: 330ms.

by BUICAN et al. Presumably, the underlying physical phenomenon is opticalbinding; for a review see [DHOLAKIA & ZEMÁNEK, 2010]. It refers to the fact thatoptically trapped particles redistribute the incident light, which in turn exertsattractive or repulsive optical forces onto nearby particles. Multiparticle chainformation by longitudinal optical binding in a dual-beam trap (as in Figure 2.7 A)was studied by SINGER et al. [2003]. Interestingly, the authors found that belowa critical particle size, given by the beam diameter (∼2µm in their case), theparticles self-organise into an on-axis chain with equidistant gaps in between,while above the critical size they touch each other. This behaviour is supposedto be largely independent of refractive index and hence assumed to be similarlyvalid for cells. METZGER et al. [2005] have re-adopted the setup to demonstratethe possibility for optical chain formation of three CHO cells. However, to datethe exploitation of optical cell binding is still in its infancy. Our preliminaryfindings propose the use of HC-PCFs for the advancement of this fascinatingarea of research. Moreover, let me note that in earlier studies on solid particleswe have found that the DOPPLER technique facilitates the tracking of multipleindividual particles simultaneously [GARBOS, 2011, ch. 6.5]. We are thereforeconfident that with our approach non-imaging, long-range and dynamical studieson cell chains are feasible.

105

Chapter 5 OUTLOOK

5.1.4 Improvement of the Setup and On-Chip Integration

Cooling SheathIt is clear that for any upcoming in-fibre cell manipulation experiments it is nec-essary to establish a cooling around the fibre. Figure 5.5 shows an attempt toinstall a cooling sheath by putting the fibre inside a water-filled half-channelmade of plastic. Preliminary tests of practicability and mechanical stability weresuccessful, but the device remains to be tested in the actual experiment.

Figure 5.5: Possible solution to the heating issue in chapter 3. The fibre resides on a plastichalf-channel which is filled with water.

Increase in Guidance Beam PowerThe laser used for the RBC guidance measurements was operating at 1064 nmcw. To reduce heating even further, for the eukaryote guidance we intent to useour cw titanium-sapphire laser. At ∼830 nm it has a maximum output power of∼1.7 W. This is not much given that in our current setup we need to split thebeam several times and it travels through many optical components, some ofwhich have non-negligible losses. In essence, only less than 200 mW of powerwere available in the guidance beam so far. Therefore, we plan to decouple thetrapping beam from the guidance beams completely by usage of a separate laser(e. g. small diode laser). Further improvements (replacement of components)should allow us to obtain at least 0.5 W in the guidance beam(s) to achievecomparable optical powers as with the 1064 nm laser.

Fibre DrawingWith what has been said before, an HC-PBGF with a larger core would befavourable so that even larger cells can be studied. Currently, we have severalfibres in stock which are ready to be tested with water-filling. Despite its few-mode guidance, the 27.7µm-core kagomé fibre shown in Figure 2.17 on page 43proved useful in the guidance of RBC cell chains as well. It might hence beworthwhile to more accurately determine the practicability of KFs for cell guid-ance experiments.

106


On-Chip IntegrationStudying cellular mechanics on the basis of single cells and with the demandto retrieve quantitatively-meaningful parameters requires experiments on manycells [ANDERSSON SVAHN & VAN DEN BERG, 2007]. In our current setup thethroughput is rather limited. It would therefore be convenient to miniaturiseour setup by integrating a piece of HC-PCF into a chip to establish the deliveryof the cells via a microfluidic channel. The illustration in Figure 5.6 gives a roughidea of how this might look like. While in chapter 4 we demonstrated the inser-tion of HC-PCF into a microfluidic circuit by integration between two chips andcoupling through the chip in a free-space manner, in this "2nd generation" chipthe optical coupling is intended to be established via precisely pre-aligned single-mode fibres. If necessary, the fluidic gap between the SMF and the HC-PCF mightbe increased by application of lensed fibres. Beam delivery into a microfluidicchannel by capped ESM fibres has been demonstrated by ASHOK et al. [2010]and the integration of an HC-PCF into a microfluidic channel by RINDORF et al.[2006]. Moreover, WHEELER et al. [2003] have developed a microfluidic deviceto selectively introduce single cells into a chip. Our subsequent approaches canbuild on these protocols.

Figure 5.6: Full on-chip integration of a piece of HC-PCF as part of a microfluidic channel.The coupling of light is established via pre-aligned single-mode fibres.

5.1.5 Cell Mechanics Modelling

For RBCs the calculation of the optical forces acting in the fibre is difficult dueto their peculiar shape. For round or ellipsoidal eukaryotes, the calculation ofthe optical forces in a ray optics model will be quite straight forward, see e. g.[GUCK et al., 2001]. Yet, it has emerged that for fibre-guided glass beads there isa significant mismatch between the experiments and ray-optics theory [GARBOS,2011]. This seems reasonable, because in an optical waveguide wave effects canhardly be neglected. In cooperation with our group, MOREIRA et al. [2012] havetherefore implemented a more accurate model, which could be applied to futurecell studies.

107

Chapter 5 OUTLOOK

However, the ultimate goal for our proposed cell mechanics experiments needsto be dynamic modelling which takes into account the cell as a spatio-temporallycomplex material. As alluded to earlier in section 2.1.3, the full theoretical treat-ment of cellular mechanics is extremely difficult so that here a close collaborationwith theorists in the field is highly recommended.

5.1.6 In-Fibre Microenvironment Monitoring with Sensor Beads

Clearly, excessive heating is an unwanted effect for cell experiments, but beingable to controllably set the internal temperature of the fibre opens a whole newrange of possibilities!

First of all, apart from our approach to measure the temperature in an optoflu-idic channel via the viscosity change around a non-deformable moving bead,other people have exploited thermosensitive fluorescent dyes, e. g. rhodamine B(rhoB) [ROSS et al., 2001; EBERT et al., 2007]. The intensity and fluorescencelifetime of rhoB decrease with temperature. Our collaborators at the Universityof Edinburgh, headed by Dr. Anita JONES, have demonstrated that it is possi-ble to combine particle trapping and fluorescence (temperature) spectroscopyby use of temperature-sensitive fluorescent microdroplets (Figure 5.7 A) [BENNET

et al., 2011]. These microdroplets consist of a fluid core of aqueous rhoB solutionwhich is surrounded by a shell of oil. The oil serves to keep the dye entrappedand to establish the refractive index difference needed for optical tweezers mea-surements in water. The droplets are typically between 5 and 20µm in size.Hence, they are not only the right size for particle guidance in the fibre but alsosoft, so that interesting deformational effects might also be studied. This makesthem a suitable mimic for cell deformation experiments. Our colleagues havetherefore kindly provided us with a sample.

Speaking about temperature, a novel material might be of interest for ourexperiments. Hydrogels are typically used as thin layers or as globular micropar-ticles and are considered to be smart materials. That is, depending on the actualenvironment they can significantly swell or shrink4 [TANAKA, 1978; SHIBAYAMA

& TANAKA, 1993]. This behaviour might be induced by a range of stimuli, suchas temperature, pH, ionic strength or a specific kind of biomolecule.

These properties make hydrogel microglobules or hydrogel-coated solid beadsa well-suited tool for (microfluidic) sensing applications, as well as for drugdelivery [CHATERJI et al., 2007; CALDORERA-MOORE & PEPPAS, 2009].

I have therefore asked experts in the field of macromolecular chemistry to pre-pare hydrogel particles for a joint project. Christopher SYNATSCHKE and his col-league Thomas RUHLAND at the Institute for Macromolecular Chemistry II at theUniversity of Bayreuth have then synthesised a range of these specialty particles

4In fact, not all hydrogels are smart but most of them.

108

PCF Photochemical Flow Reactors 5.2

Figure 5.7: A) Thermosensitive fluorescent "double-bubble" microdroplets. B) PS beadcoated with a thermosensitive layer of NIPAM-based hydrogel. C) Pure hydrogelbead based on pH-sensitive DMAEMA. The hydrogel can also be loaded withcertain agents (orange) for release or uptake.

for me. A well known thermoresponsive hydrogel is poly(N-isopropylacrylamide)(NIPAM) [PELTON & CHIBANTE, 1986], typically copolymerised with another ma-trix material. Since pure microgel particles of this copolymer came out rathersmall (<1µm) in the synthesis, SYNATSCHKE and RUHLAND have also used amine-functionalised PS beads of 2 and 10µm in diameter (PPs-2.0NH2/-10.0NH2,Kisker Biotech) and coated them with a NIPAM-based gel layer (both approachesaccording to the protocol by KARG et al. [2008]). For illustration see Figure 5.7 B.In addition, they have provided me with (pure) microgel particles based onN,N-dimethylaminoethyl methacrylate (DMAEMA), which is pH-sensitive (andat large pH>7 also thermo-sensitive) [YANFENG & MIN, 2001], see Figure 5.7 C.

Equipped by our kind colleagues with these fascinating sensor beads, in-fibremicroenvironment sensing over long distances is within reach.

5.2 PCF Photochemical Flow Reactors

5.2.1 Direct Injection of Online Photoactivated Agents into Tissue Culture

In chapter 4 we could demonstrate the usefulness of PCF microflow reactorsfor the analytical investigation of photoactivatable drugs. For medical applica-tions, an online rapid generation and local delivery of reactive and/or metastablespecies through optofluidic PCF directly into tumour tissue or small blood ves-sels is possible. The combination of an HC-PCF with a microneedle is straight-forward and the idea could rather quickly be tested on tissue cultures in vitro.

109

Chapter 5 OUTLOOK

5.2.2 Microflow Photochemical Synthesis

In fact, the in situ generation of reactive species is of more general use for thechemical sciences and in chemical engineering [HOFFMANN, 2008]. After havingdemonstrated the capabilities of PCF microflow reactors to enhance the photo-chemical analysis with a mass spectrometer [Paper III: UNTERKOFLER et al., 2012;Manuscript IV: MCQUITTY et al., 2013, in preparation], and in the spectroscopicmonitoring of nanoparticle-mediated catalytic reactions [Manuscript V: CUBILLAS

et al., 2013, in preparation], it is a natural next step to perform in-fibre photoas-sisted chemical synthesis.

In fact, microflow reactors are a current hot topic in chemical synthesis [JÄH-NISCH et al., 2004], especially for drug development [WATTS & HASWELL, 2003].It has even been proposed that for industrial purposes they could outperformbatch production in terms of process efficiency, product purity and risk reduction[HESSEL, 2009]. In particular for photoreactors, homogenous and bright illumi-nation in plant-scale reactors is extremely difficult to almost impossible. Homo-geneous illumination has in the meantime been achieved with bulb-illuminatedchemical (serpentine) microreactors [LU et al., 2001; UENO et al., 2002; WOOT-TON et al., 2002; SUGIMOTO et al., 2006; FUSE et al., 2010; LÉVESQUE & SEE-BERGER, 2011], for reviews see [COYLE & OELGEMÖLLER, 2008; OELGEMÖLLER

& SHVYDKIV, 2011]. Yet, external illumination cannot achieve the extremelyhigh irradiances provided by optofluidic PCFs. This unique feature in turn opensthe possibility to tailor the irradiation conditions in a well-controlled way. Forinstance, it is possible to dynamically vary the irradiation conditions or to initi-ate specific photochemical reactions selectively by choice of a single wavelength.In addition, the temperature — another key parameter for chemical synthesis— can be raised by application of a high-power NIR laser beam for accuratethermocontrol.

A suitable example with biological importance would be the synthesis of vi-tamin D3 from provitamin D3 [FUSE et al., 2010], because this involves a two-step process (via previtamin D3), in which a thermal isomerisation follows aphotoactivation. Typically, it is performed by broadband UV(-to-VIS) lamps andsuffers from the fact that other byproducts are also created which do not thermo-isomerise into desired vitamin D3. Hence, it might be possible to find a more ap-propriate irradiation scheme in fibra to enhance the overall reaction yield of thisimportant dietary supplement, currently highly recommended for people with apro-longed sunlight deficiency5.

5Something which I — as an experimental researcher in the field of optics — have now finally overcomeby having written this last sentence. :-)

110

Bibliography

ABDOLVAND A (2011) Coherent RAMAN interaction in gas-filled hollow-core pho-tonic crystal fibers. Ph.D. thesis, Friedrich-Alexander Universität Erlangen-Nürnberg, Naturwissenschaftliche Fakultät. (Cited on pages 39 and 41.)

ABRAHAM M (1909) Zur Elektrodynamik bewegter Körper. Rendiconti del CircoloMatematico di Palermo 28(1):1–28. (Cited on page 18.)

AFSHAR V S, WARREN-SMITH SC & MONRO TM (2007) Enhancement offluorescence-based sensing using microstructured optical fibres. Optics Express15(26):17891–17901. (Cited on page 12.)

AHLUWALIA BS, MCCOURT P, HUSER T & HELLESØ OG (2010) Optical trap-ping and propulsion of red blood cells on waveguide surfaces. Optics Express18(20):21053–21061. (Cited on page 24.)

AHMAD I, HUSSAIN W & FAREEDI AA (1992) Photolysis of cyanocobalamin inaqueous solution. Journal of Pharmaceutical and Biomedical Analysis 10(1):9–15. (Cited on pages 86 and 90.)

AL QUDDUS N, MOUSSA WA & BHATTACHARJEE S (2008) Motion of a sphericalparticle in a cylindrical channel using arbitrary LAGRANGian-EULERian method.Journal of Colloid and Interface Science 317(2):620–630. (Cited on pages 61 and 69.)

AMEZCUA-CORREA R, GÉRÔME F, LEON-SAVAL SG, BRODERICK NGR, BIRKS TA &KNIGHT JC (2008) Control of surface modes in low loss hollow-core photonicbandgap fibers. Optics Express 16(2):1142–1149. (Cited on page 37.)

ANDERSSON SVAHN H & VAN DEN BERG A (2007) Single cells or large popula-tions? Lab on a Chip 7:544–546. (Cited on pages 17 and 107.)

ANTONOPOULOS G, BENABID F, BIRKS TA, BIRD DM, KNIGHT JC & RUSSELL PSTJ(2006) Experimental demonstration of the frequency shift of bandgaps in pho-tonic crystal fibers due to refractive index scaling. Optics Express 14(7):3000–3006. (Cited on page 40.)

111

Bibliography

ARGYROS A & PLA J (2007) Hollow-core polymer fibres with a kagome lattice:potential for transmission in the infrared. Optics Express 15(12):7713–7719.(Cited on page 39.)

ARIS R (1956) On the dispersion of a solute in a fluid flowing through a tube.Proceedings of the Royal Society of London – Series A 235(1200):67–77. (Cited

on page 9.)

ARLT J, GARCÉS-CHÁVEZ V, SIBBETT W & DHOLAKIA K (2001) Optical microma-nipulation using a BESSEL light beam. Optics Communications 197:239–245.(Cited on pages 45 and 46.)

ASHKIN A (1970) Acceleration and trapping of particles by radiation pressure.Physical Review Letters 24(4):156–159. (Cited on pages 19, 20 and 23.)

ASHKIN A (1992) Forces of a single-beam gradient laser trap on a dielectricsphere in the ray optics regime. Biophysical Journal 61(2):569–582. (Cited

on page 19.)

ASHKIN A & DZIEDZIC JM (1987) Optical trapping and manipulation of virusesand bacteria. Science 235:1518–1520. (Cited on pages 19 and 20.)

ASHKIN A, DZIEDZIC JM, BJORKHOLM JE & CHU S (1986) Observation of asingle-beam gradient force optical trap for dielectric particles. Optics Letters11(5):288–290. (Cited on pages 19 and 20.)

ASHKIN A, DZIEDZIC JM & YAMANE T (1987) Optical trapping and manipulationof single cells using infrared laser beams. Nature 330:769–771. (Cited on page

19.)

ASHOK PC, MARCHINGTON RF, MTHUNZI P, KRAUSS TF & DHOLAKIA K (2010)Optical chromatography using a photonic crystal fiber with on-chip fluores-cence excitation. Optics Express 18(6):6396–6407. (Cited on pages 23 and 107.)

ASTON FW (1919) A positive ray spectrograph. Philosophical Magazine (Series 6)38:707–714. (Cited on page 87.)

AUMAILLEY M, GURRATH M, MÜLLER G, CALVETE J, TIMPL R & KESSLER H (1991)Arg-Gly-Asp constrained within cyclic pentapeptides. Strong and selective in-hibitors of cell adhesion to vitronectin and laminin fragment P1. FEBS Letters291(1):50–54. (Cited on page 102.)

BAO G & SURESH S (2003) Cell and molecular mechanics of biological materials.Nature Materials 2:715–725. (Cited on pages 16 and 17.)

BARNETT SM (2010) Resolution of the ABRAHAM-MINKOWSKI dilemma. PhysicalReview Letters 104(7):070401 (4 pp). (Cited on page 18.)

112

Bibliography

BEAUDOU B, COUNY F, BENABID F & ROBERTS PJ (2008) Large pitch hollow corehoneycomb fiber. In Conference on Lasers and Electro-Optics/Quantum Elec-tronics and Laser Science (CLEO/QELS) 2008 (Optical Society of America),San José, CA, USA, paper JFC6. (Cited on page 37.)

BEEBE DJ, MENSING GA & WALKER GM (2002) Physics and applications ofmicrofluidics in biology. Annual Review of Biomedical Engineering 4:261–286.(Cited on page 6.)

BENABID F, KNIGHT JC, ANTONOPOULOS G & RUSSELL PSTJ (2002a) StimulatedRAMAN scattering in hydrogen-filled hollow-core photonic crystal fiber. Science298(5592):399–402. (Cited on pages 15, 29 and 37.)

BENABID F, KNIGHT JC & RUSSELL PSTJ (2002b) Particle levitation and guidancein hollow-core photonic crystal fiber. Optics Express 10(21):1195–1203. (Cited

on page 25.)

BENABID F, ROBERTS PJ, COUNY F & LIGHT PS (2009) Light and gas confinementin hollow-core photonic crystal fibre based photonic microcells. Journal of theEuropean Optical Society – Rapid Publications 4:09004 (9 pp). (Cited on page 76.)

BENNET MA, RICHARDSON PR, ARLT J, MCCARTHY A, BULLER GS & JONES AC(2011) Optically trapped microsensors for microfluidic temperature measure-ment by fluorescence lifetime imaging miroscopy. Lab on a Chip 11:3821–3828. (Cited on page 108.)

BERNINI R, DE NUCCIO E, MINARDO A, ZENI L & SARRO PM (2007) Integratedsilicon optical sensors based on hollow core waveguides. In Proceedings ofSPIE 2007, San José, CA, USA, volume 6477, paper 647714. (Cited on page 49.)

BHARDWAJ A, BEAUDOU B, GÉRÔME F, HUMBERT G, AUGUSTE JL, BLONDY JM &BENABID F (2011) Macro bending losses in kagome-lattice fibers. In 37th Eu-ropean Conference and Exposition on Optical Communications (ECOC) 2011(Optical Society of America), Geneva, Switzerland, paper Mo.2.LeCervin.3.(Cited on page 38.)

BIRKS TA, BIRD DM, HEDLEY TD, POTTAGE JM & RUSSELL PSTJ (2004) Scalinglaws and vector effects in bandgap-guiding fibres. Optics Express 12(1):69–74.(Cited on pages 34 and 40.)

BIRKS TA, KNIGHT JC & RUSSELL PSTJ (1997) Endlessly single-mode photoniccrystal fiber. Optics Letters 22(13):961–963. (Cited on page 29.)

BIRKS TA, ROBERTS PJ, RUSSELL PSTJ, ATKIN DM & SHEPHERD TJ (1995) Full2-d photonic bandgaps in silica/air structures. Electronics Letters 31(22):1941–1943. (Cited on pages 29 and 34.)

113

Bibliography

BLOCH F (1929) Über die Quantenmechanik der Elektronen in Kristallgittern.Zeitschrift für Physik 52(7-8):555–600. (Cited on page 28.)

BOUWMANS G, LUAN F, KNIGHT JC, RUSSELL PSTJ, FARR L, MANGAN BJ &SABERT H (2003) Properties of a hollow-core photonic bandgap fiber at850 nm wavelength. Optics Express 11(14):1613–1620. (Cited on page 36.)

BROWN JM & GIACCIA AJ (1998) The unique physiology of solid tumors: oppor-tunities (and problems) for cancer therapy. Cancer Research 58(7):1408–1416.(Cited on page 75.)

BRUIJNINCX PCA & SADLER PJ (2008) New trends for metal complexes withanticancer activity. Current Opinion in Chemical Biology 12(2):197–206. (Cited

on page 75.)

BRUKER DALTONICS INC (2008) maXis – Maximum Information@MaximumSpeed. Billerica, MA, USA. (Cited on page 88.)

BUICAN TN, SMYTH MJ, CRISSMAN HA, SALZMAN GC, STEWART CC & MARTIN

JC (1987) Automated single-cell manipulation and sorting by light trapping.Applied Optics 26(24):5311–5316. (Cited on pages 21, 104 and 105.)

BURSAC P, LENORMAND G, FABRY B, OLIVER M, WEITZ DA, VIASNOFF V, BUTLER

JP & FREDBERG JJ (2005) Cytoskeletal remodelling and slow dynamics in theliving cell. Nature Materials 4(7):557–561. (Cited on page 17.)

CALDORERA-MOORE M & PEPPAS NA (2009) Micro- and nanotechnologies forintelligent and responsive biomaterial-based medical systems. Advanced DrugDelivery Reviews 61(15):1391–1401. (Cited on page 108.)

CHATERJI S, KWON IK & PARK K (2007) Smart polymeric gels: redefining thelimits of biomedical devices. Progress in Polymer Science 32:1083–1122. (Cited

on page 108.)

CHEN JSY (2010) Nanochemistry and sensing in photonic crystal fibers.Ph.D. thesis, Friedrich-Alexander Universität Erlangen-Nürnberg, Naturwis-senschaftliche Fakultät. (Cited on pages 12, 84 and 98.)

CHEN JSY, EUSER TG, FARRER NJ, SADLER PJ & RUSSELL PSTJ (2010) Photo-chemistry in photonic crystal fiber nanoreactors. Chemistry – A European Jour-nal 16(19):5607–5612. (Cited on pages 73, 85 and 86.)

COLLADON JD (1842) Sur les réflexions d’un rayon de lumière à l’intérieur d’uneveine liquide parabolique. Comptes Rendus de l’Académie des Sciences 15:800–802. (Cited on page 26.)

114

Bibliography

COLTON R, D’AGOSTINA A & TRAEGER JC (1995) Electrospray mass spectrometryapplied to inorganic and organometallic chemistry. Mass Spectrometry Reviews14(2):79–106. (Cited on page 88.)

CONSTABLE A, KIM J, MERVIS J, ZARINETCHI F & PRENTISS M (1993) Demonstra-tion of a fiber-optical light-force trap. Optics Letters 18(21):1867–1869. (Cited

on page 22.)

COUNY F, BENABID F & LIGHT PS (2006) Large-pitch kagome-structured hollow-core photonic crystal fiber. Optics Letters 31(24):3574–3576. (Cited on pages 37

and 39.)

COUNY F, BENABID F, ROBERTS PJ, BURNETT MT & MAIER SA (2007a) Identi-fication of BLOCH-modes in hollow-core photonic crystal fiber cladding. OpticsExpress 15(2):325–338. (Cited on page 33.)

COUNY F, BENABID F, ROBERTS PJ, LIGHT PS & RAYMER MG (2007b) Genera-tion and photonic guidance of multi-octave optical-frequency combs. Science318(5853):1118–1121. (Cited on pages 33, 38 and 39.)

COX FM, ARGYROS A & LARGE MCJ (2006) Liquid-filled hollow core microstruc-tured polymer optical fiber. Optics Express 14(9):4135–4140. (Cited on page 40.)

COYLE EE & OELGEMÖLLER M (2008) Micro-photochemistry: photochemistry inmicrostructured reactors. The new photochemistry of the future? Photochemi-cal & Photobiological Sciences 7(11):1313–1322. (Cited on page 110.)

CRAN-MCGREEHIN S, KRAUSS TF & DHOLAKIA K (2006) Integrated monolithicoptical manipulation. Lab on a Chip 6:1122–1124. (Cited on page 21.)

CREGAN RF, MANGAN BJ, KNIGHT JC, BIRKS TA, RUSSELL PSTJ, ROBERTS PJ& ALLAN DC (1999) Single-mode photonic band gap guidance of light in air.Science 285(5433):1537–1539. (Cited on pages 29 and 36.)

CUBILLAS AM, SCHMIDT M, EUSER TG, ETZOLD BJM, TACCARDI N, UN-TERKOFLER S, WASSERSCHEID P & RUSSELL PSTJ (2013) Liquid-phase hetero-geneous catalysis in photonic crystal fiber microreactor. (in preparation). (Cited

on pages 98 and 110.)

CUBILLAS AM, SCHMIDT M, SCHARRER M, EUSER TG, ETZOLD BJM, TACCARDI N,WASSERSCHEID P & RUSSELL PSTJ (2012) Ultra-low concentration monitoringof catalytic reactions in photonic crystal fiber. Chemistry – A European Journal18:1586–1590. accepted. (Cited on page 73.)

115

Bibliography

CURL CL, BELLAIR CJ, HARRIS T, ALLMAN BE, HARRIS PJ, STEWART AG,ROBERTS A, NUGENT KA & DELBRIDGE LMD (2005) Refractive index mea-surement of viable cells using quantitative phase-amplitude microscopy andconfocal microscopy. Cytometry 65A(1):88–92. (Cited on page 19.)

DEMELLO AJ (2006) Control and detection of chemical reactions in microfluidicsystems. Nature 442:394–402. (Cited on pages 2, 10 and 99.)

DERTINGER T, PACHECO V, VON DER HOCHT I, HARTMANN R, GREGOR I & ENDER-LEIN J (2007) Two-focus fluorescence correlation spectroscopy: a new tool foraccurate and absolute diffusion measurements. ChemPhysChem 8(3):433–443.(Cited on page 83.)

DESOIZE B (2004) Metals and metal compounds in cancer treatment. AnticancerResearch 24(3A):1529–1544. (Cited on page 75.)

DHOLAKIA K & CIŽMÁR T (2011) Shaping the future of manipulation. NaturePhotonics 5:335–342. (Cited on page 21.)

DHOLAKIA K & ZEMÁNEK P (2010) Colloquium: Gripped by light: optical binding.Review of Modern Physics 82(2):1767–1791. (Cited on pages 21 and 105.)

DI CARLO D & LEE LP (2006) Dynamic single-cell analysis for quantitative biol-ogy. Analytical Chemistry 78(23):7918–7925. (Cited on page 17.)

DIGONNET MJF, KIM HK, KINO GS & FAN S (2005) Understanding air-corephotonic-bandgap fibers: analogy to conventional fibers. Journal of LightwaveTechnology 23(12):4169–4177. (Cited on page 34.)

DRESS P & FRANKE H (1996) A cylindrical liquid-core waveguide. Applied PhysicsB: Lasers and Optics 63(1):12–19. (Cited on page 49.)

DUGUAY MA, KOKUBUN Y, KOCH TL & PFEIFFER L (1986) Antiresonant reflect-ing optical waveguides in SiO2-Si mulitlayer structures. Applied Physics Letters49(1):13–15. (Cited on page 49.)

DURNIN J (1987) Exact solutions for nondiffracting beams. I. The scalar theory.Journal of the Optical Society of America A, Optics and Image Science 4(4):651–654. (Cited on page 45.)

DURNIN J, MICELI, JR JJ & EBERLY JH (1987) Diffracion-free beams. PhysicalReview Letters 58(15):1499–1501. (Cited on page 45.)

EBERT S, TRAVIS K, LINCOLN B & GUCK J (2007) Fluorescence ratio thermometryin a microfluidic dual-beam laser trap. Optics Express 15(23):15493–15499.(Cited on pages 22 and 108.)

116

Bibliography

EHRLICH P (1913) Über Chemotherapie: Die Behandlung der Syphilis mitSalvarsan und verwandten Stoffen. Münchener medizinische Wochenschrift1959:345–356. (Cited on page 74.)

EINSTEIN A (1905) Über die von der molekularkinetischen Theorie der Wärmegeforderte Bewegung von in ruhenden Flüssigkeiten suspendierten Teilchen.Annalen der Physik 322(8):549–560. (Cited on page 8.)

EL-ALI J, SORGER PK & JENSEN KF (2006) Cells on chips. Nature 442:403–411.(Cited on page 104.)

EL-ANEED A, COHEN A & BANOUB J (2009) Mass spectrometry, review of thebasics: electrospray, MALDI, and commonly used mass analyzers. Applied Spec-troscopy Reviews 44:210–230. (Cited on page 87.)

EUSER TG, CHEN JSY, SCHARRER M, RUSSELL PSTJ, FARRER NJ & SADLER PJ(2008a) Quantitative broadband chemical sensing in air-suspended solid-corefibers. Journal of Applied Physics 103(10):103108 (7 pp). (Cited on page 12.)

EUSER TG, GARBOS MK, CHEN JSY & RUSSELL PSTJ (2009) Precise balancingof viscous and radiation forces on a particle in liquid-filled photonic bandgapfiber. Optics Letters 34(23):3674–3676. (Cited on pages 25, 51 and 62.)

EUSER TG, GARBOS MK, CHEN JSY & RUSSELL PSTJ (2010) Precise balancingof viscous and radiation forces on a particle in liquid-filled photonic-bandgapfiber: erratum. Optics Letters 35(13):2142. (Cited on pages 25, 51 and 62.)

EUSER TG, WHYTE G, SCHARRER M, CHEN JSY, ABDOLVAND A, NOLD J, KAMIN-SKI CF & RUSSELL PSTJ (2008b) Dynamic control of higher-order modes inhollow-core photonic crystal fibers. Optics Express 16(22):17972–17981. (Cited

on page 34.)

EVANS E & FUNG YC (1972) Improved measurements of the erythrocyte geome-try. Microvascular Research 4(4):335–347. (Cited on pages 52, 66 and 69.)

FABRY B & FREDBERG JJ (2007) Mechanotransduction, asthma and airwaysmooth muscle. Drug Discovery Today: Disease Models 4(3):131–137. (Cited

on page 16.)

FAINMAN Y, LEE LP, PSALTIS D & YANG C (2010) Optofluidics – Fundamentals,Devices and Applications. 1st edition, Biophotonics Series (The McGraw-HillCompanies, Inc., New York, NY, USA). (Cited on page 2.)

FAN X & WHITE IM (2011) Optofluidic microsystems for chemical and biologicalanalysis. Nature Photonics 5:591–597. (Cited on pages 2 and 99.)

117

Bibliography

FARRER NJ, SALASSA L & SADLER PJ (2009) Photoactivated chemotherapy(PACT): the potential of excited-state d-block metals in medicine. DaltonTransactions (48):10690–10701. (Cited on pages 12 and 75.)

FARRER NJ, WOODS JA, SALASSA L, ZHAO Y, ROBINSON KS, CLARKSON G,MACKAY FS & SADLER PJ (2010) A potent trans-diimine platinum anticancercomplex photoactivated by visible light. Angewandte Chemie – InternationalEdition 49(47):8905–8908. (Cited on pages 85, 90, 91 and 92.)

FAZAL FM & BLOCK SM (2011) Optical tweezers study life under tension. NaturePhotonics 5:318–321. (Cited on page 21.)

FÉVRIER S, BEAUDOU B & VIALE P (2010) Understanding origin and loss in largepitch hollow-core photonic crystal fibers and their design simplification. OpticsExpress 18(5):5142–5150. (Cited on page 38.)

FICK AE (1855) Über Diffusion. Annalen der Physik und Chemie 94:59–86. (Cited

on page 8.)

FLETCHER DA & MULLINS RD (2010) Cell mechanics and the cytoskeleton. Na-ture 463:485–492. (Cited on page 16.)

FRAZÃO O, SANTOS JL, ARAÚJO FM & FERREIRA LA (2008) Optical sensing withphotonic crystal fibers. Laser & Photonics Reviews 2(6):449–459. (Cited on page

12.)

FRICKER SP (2007) Metal based drugs: from serendipity to design. Dalton Trans-actions (43):4903–4917. (Cited on page 75.)

FROSZ MH, NOLD J, WEISS T, STEFANI A, RAMMLER S, BABIC F & RUSSELL

PSTJ (2012) Five-ring hollow-core photonic bandgap fiber with 1.8 dB/kmloss. In Frontiers in Optics (FiO) Conference 2012 (Optical Society of Amer-ica), Rochester, NY, USA, paper FW6C.5. (Cited on page 37.)

FUSE S, TANABE N, YOSHIDA M, YOSHIDA H, DOI T & TAKAHASHI T (2010)Continuous-flow synthesis of vitamin D3. Chemical Communications 46:8722–8724. (Cited on page 110.)

GARBOS MK (2011) LASER-propulsion of microparticles in liquid-filled hollow-core photonic-crystal fibers -Going against the flow-. Ph.D. thesis, Friedrich-Alexander Universität Erlangen-Nürnberg, Naturwissenschaftliche Fakultät.(Cited on pages 51, 71, 105 and 107.)

GARBOS MK, EUSER TG & RUSSELL PSTJ (2011a) Optofluidic immobility of par-ticles trapped in liquid-filled hollow-core photonic crystal fiber. Optics Express19(20):19643–19652. (Cited on pages 25 and 51.)

118

Bibliography

GARBOS MK, EUSER TG, SCHMIDT OA, UNTERKOFLER S & RUSSELL PSTJ(2011b) Doppler velocimetry on microparticles trapped and propelled by laserlight in liquid-filled photonic crystal fiber. Optics Letters 36(11):2020–2022.(Cited on pages 51, 58, 59, 61 and 62.)

GARCÉS-CHÁVEZ V, MCGLOIN D, MELVILLE H, SIBBETT W & DHOLAKIA K (2002)Simultaneous micromanipulation in multiple planes using a self-reconstructinglight beam. Nature 419(6903):145–147. (Cited on page 46.)

GAUGIRAN S, GÉTIN S, FEDELI JM, COLAS G, FUCHS A, CHATELAIN F &DÉROUARD J (2005) Optical manipulation of microparticles and cells on sil-icon nitride waveguides. Optics Express 13(18):6956–6963. (Cited on page 24.)

GHOSH N, BUDDHIWANT P, UPPAL A, MAJUMDER SK, PATEL HS & GUPTA PK(2006a) Simultaneous determination of size and refractive index of red bloodcells by light scattering measurements. Applied Physics Letters 88:084101 (3pp). (Cited on page 19.)

GHOSH S, BHAGWAT AR, RENSHAW CK, GOH S, GAETA AL & KIRBY BJ (2006b)Low-light-level optical interactions with rubidium vapor in a photonic band-gap fiber. Physical Review Letters 97(2):023603 (4 pp). (Cited on page 98.)

GILLESPIE PG & WALKER RG (2001) Molecular basis of mechanosensory trans-duction. Nature 413:194–202. (Cited on page 16.)

GRIER DG (2003) A revolution in optical manipulation. Nature 424:810–816.(Cited on page 21.)

GRIFFITHS WJ, JONSSON AP, LIU S, RAI DK & WANG Y (2001) Electrospray andtandem mass spectrometry in biochemistry. Biochemical Journal 355:545–561.(Cited on page 88.)

GRUJIC K & HELLESØ OG (2007) Dielectric microsphere manipulation and chainassembly by counter-propagating waves in a channel waveguide. Optics Express15(10):6470–6477. (Cited on page 24.)

GUCK J, ANANTHAKRISHNAN R, MAHMOOD H, MOON TJ, CUNNINGHAM CC &KÄS J (2001) The optical stretcher: a novel laser tool to micromanipulatecells. Biophysical Journal 81(2):767–784. (Cited on pages 22, 23, 66, 70 and 107.)

GUCK J, ANANTHAKRISHNAN R, MOON TJ, CUNNINGHAM CC & KÄS J (2000)Optical deformability of soft biological dielectrics. Physical Review Letters84(23):5451–5454. (Cited on pages 22 and 23.)

GUCK J, SCHINKINGER S, LINCOLN B, WOTTAWAH F, EBERT S, ROMEYKE M,LENZ D, ERICKSON HM, ANANTHAKRISHNAN R, MITCHELL D, KÄS J, ULVICK

119

Bibliography

S & BILBY C (2005) Optical deformability as an inherent cell marker for test-ing malignant transformation and metastatic competence. Biophysical Journal88(5):3689–3698. (Cited on page 22.)

GUO Z & SADLER PJ (1999) Metals in medicine. Angewandte Chemie – Interna-tional Edition 38:1512–1531. (Cited on page 74.)

HAGEN GHL (1839) Über die Bewegung des Wassers in engen cylindrischenRöhren. Poggendorfs Annalen der Physik und Chemie 46:423–442. (Cited on

page 7.)

HANSEN TP, BROENG J, JAKOBSEN C, VIENNE G, SIMONSEN HR, NIELSEN MD,SKOVGAARD PMW, FOLKENBERG JR & BJARKLEV A (2004) Air-guiding pho-tonic bandgap fibers: spectral properties, macrobending loss, and practicalhandling. Journal of Lightwave Technology 22(1):11–15. (Cited on page 37.)

HARRIS AL (2002) Hypoxia – a key regulatory factor in tumour growth. NatureReviews Cancer 2:38–47. (Cited on page 75.)

HAWKINS AR & SCHMIDT H (2010) Handbook of Optofluidics (CRC Press, BocaRaton, FL, USA). (Cited on page 2.)

HEDLEY TD, BIRD DM, BENABID F, KNIGHT JC & RUSSELL PSTJ (2003) Mod-elling of a novel hollow-core photonic crystal fibre. In Conference on Lasersand Electro-Optics/Quantum Electronics and Laser Science (CLEO/QELS)2003 (Optical Society of America), Baltimore, MD, USA, paper QTuL4. (Cited

on page 38.)

HÉNON S, LENORMAND G, RICHERT A & GALLET F (1999) A new determina-tion of the shear modulus of the human erythrocyte membrane using opticaltweezers. Biophysical Journal 76(2):1145–1151. (Cited on page 21.)

HENRY W (1994) Evanscent field devices: a comparison between tapered opticalfibres and polished or D-fibres. Optical and Quantum Electronics 26(3):S261–S272. (Cited on page 12.)

HERAEUS QUARZGLAS GMBH & CO KG (2010) Lamp Materials – Heraeus Quarz-glas. Kleinostheim, Germany. (Cited on page 32.)

HERMAN RM & WIGGINS TA (1991) Production and uses of diffractionlessbeams. Journal of the Optical Society of America A, Optics and Image Science8(6):932–942. (Cited on page 45.)

HESSEL V (2009) Novel process windows – gate to maximizing process intensi-fication via flow chemistry. Chemical Engineering & Technology 32(11):1655–1681. (Cited on page 110.)

120

Bibliography

HIRSCHOWITZ BI, PETERS CW & CURTISS LE (1957) Preliminary reports on along fiberscope for examination of the stomach and duodenum. University ofMichigan Medical Bulletin 23:178–180. (Cited on page 26.)

HOFFMAN BD & CROCKER JC (2009) Cell mechanics: Dissecting the physicalresponses of cells to force. Annual Review of Biomedical Engineering 11:259–288. (Cited on page 17.)

HOFFMANN N (2008) Photochemical reactions as key steps in organic synthesis.Chemical Reviews 108(3):1052–1103. (Cited on pages 12 and 110.)

HOMOLA J, YEE SS & GAUGLITZ G (1999) Surface plasmon resonance sensors:review. Sensors and Actuators B: Chemical 54:3–15. (Cited on page 12.)

HOPKINS HH & KAPANY NS (1954) A flexible fibrescope, using static scanning.Nature 173:39–41. (Cited on page 26.)

HUMBERT G, KNIGHT JC, BOUWMANS G, RUSSELL PSTJ, WILLIAMS DP, ROBERTS

PJ & MANGAN BJ (2004) Hollow core photonic crystal fibers for beam delivery.Optics Express 12(8):1477–1484. (Cited on page 37.)

HUNSBERGER RG (2009) Integrated Optics: Theory and Technology. 6th edition(Springer, New York, NY, USA). (Cited on page 2.)

HUNT HC & WILKINSON JS (2008) Optofluidic integration for microanalysis.Microfluidics and Nanofluidics 4:53–79. (Cited on page 2.)

IMASAKA T, KAWABATA Y, KANETA T & ISHIDZU Y (1995) Optical chromatography.Analytical Chemistry 67(11):1763–1765. (Cited on page 23.)

IPG PHOTONICS CORP (2010) Fiber Laser Model: YLR-50-1064-LP-SF (Manual).Oxford, MS, USA. Issue 1. (Cited on pages 40 and 55.)

JÄHNISCH K, HESSEL V, LÖWE H & BAERNS M (2004) Chemistry in microstruc-tured reactors. Angewandte Chemie – International Edition 43(4):406–446.(Cited on pages 94 and 110.)

JANMEY PA & WEITZ DA (2004) Dealing with mechanics: mechanisms of forcetransduction in cells. Trends in Biochemical Sciences 29(7):364–370. (Cited on

page 16.)

JENSEN JB, PEDERSEN LH, HOIBY PE, NIELSEN LB, HANSEN TP, FOLKENBERG

JR, RIISHEDE J, NOORDEGRAAF D, NIELSEN K, CARLSEN A & BJARKLEV A(2004) Photonic cystal fiber based evanscent-wave sensor for detection ofbiomolecules in aqueous solution. Optics Letters 29(17):1974–1976. (Cited

on page 12.)

121

Bibliography

JIANG P, SCHROEDER T, BATH M, LESNYAK V, GAPONIK N, EYCHMÜLLER A & BEN-SON O (2012) Incoherent photon conversion in selectively infiltrated hollow-core photonic crystal fibers for single photon generation in the near infrared.Optics Express 20(10):11537–11547. (Cited on page 49.)

JOANNOPOULOS JD, JOHNSON SG, WINN JN & MEADE RD (2008) Photonic Crys-tals – Molding the flow of light. 2nd edition (Princeton University Press, Prince-ton, NJ, USA). (Cited on page 29.)

JOHN S (1987) Strong localization of photons in certain disordered dielectricsuperlattices. Physical Review Letters 58(23):2486–2489. (Cited on page 27.)

JÄCKLE J & KAWASAKI K (1995) Intrinsic roughness of glass surfaces. Journal ofPhysics: Condensed Matter 7(23):4351–4358. (Cited on page 37.)

KAISER P & ASTLE HW (1974) Low-loss single-material fibers made from purefused silica. The Bell System Technical Journal 53(6):1021–1039. (Cited on page

31.)

KANETA T, MAKIHARA J & IMASAKA T (2001) An "optical channel": a tech-nique for the evaluation of biological cell elasticity. Analytical Chemistry73(24):5791–5795. (Cited on pages 21 and 66.)

KAO KC & HOCKHAM GA (1966) Dielectric-fibre surface waveguides for opticalfrequencies. Proceedings of the IEE 113(7):1151–1158. (Cited on page 27.)

KAOUI B, BIROS G & MISBAH C (2009) Why do red blood cells have asymmetricshapes even in a symmetric flow? Physical Review Letters 103(18):188101(4 pp). (Cited on page 72.)

KARG M, PASTORIZA-SANTOS I, RODRIGUEZ-GONZÁLEZ B, VON KLITZING R,WELLERT S & HELLWEG T (2008) Temperature, pH, and ionic strength inducedchanges of the swelling behavior of PNIPAM-poly(allylacetic acid) copolymermicrogels. Langmuir 24(12):6300–6306. (Cited on page 109.)

KASZA KE, ROWAT AC, LIU J, ANGELINI TE, BRANGWYNNE CP, KOENDERINK GH& WEITZ DA (2007) The cell as a materials. Current Opinion in Cell Biology19(1):101–107. (Cited on page 17.)

KAWATA S & TANI T (1996) Optically driven MIE particles in an evanescent fieldalong a channeled waveguide. Optics Letters 21(21):1768–1770. (Cited on page

24.)

KECK DB & SCHULTZ PC (1973) Method of producing optical waveguide fibers.US patent No. 3,711,262. (Cited on page 27.)

122

Bibliography

KESTIN J, IMAISHI N, NOTT SH, NIEUWOUDT JC & SENGERS JV (1985) Viscosityof light and heavy water and their mixtures. Physica A: Statistical Mechanicsand its Applications 134(1):38–58. (Cited on pages 64 and 65.)

KHETANI A, LAFERRIÈRE M, ANIS H & SCAIANO JC (2008) Laser flash photolysiswith nanoliter samples: photonic crystal fibers as ultrasmall smart test tubes.Journal of Materials Chemistry 18:4769–4774. (Cited on page 73.)

KIM DH, WONG PK, PARK J, LEVCHENKO A & SUN Y (2009) Microengineeredplatforms for cell mechanobiology. Annual Review of Biomedical Engineering11:203–233. (Cited on page 21.)

KIM HK, DIGONNET MJF, KINO GS, SHIN J & FAN S (2004) Simulations of theeffect of the core ring on surface and air-core modes in photonic bandgapfibers. Optics Express 12(15):3436–3442. (Cited on page 37.)

KNIGHT JC, ARRIAGA J, BIRKS TA, ORTIGOSA-BLANCH A, WADSWORTH WJ &RUSSELL PSTJ (2000) Anomalous dispersion in photonic crystal fiber. IEEEPhotonics Technology Letters 12(7):807–809. (Cited on page 30.)

KNIGHT JC, BIRKS TA, RUSSELL PSTJ & ATKIN DM (1996) All-silica single-modeoptical fiber with photonic crystal cladding. Optics Letters 21(19):1547–1549.(Cited on pages 29 and 31.)

KNIGHT JC, BIRKS TA, RUSSELL PSTJ & ATKIN DM (1997) All-silica single-modeoptical fiber with photonic crystal cladding: errata. Optics Letters 22(7):484–485. (Cited on pages 29 and 31.)

KNIGHT JC, BROENG J, BIRKS TA & RUSSELL PSTJ (1998) Photonic band gapguidance in optical fibers. Science 282(5393):1476–1478. (Cited on page 29.)

KOLLMANNSBERGER P & FABRY B (2011) Linear and nonlinear rheology of livingcells. Annual Review of Materials Research 41:75–97. (Cited on page 17.)

KÖNIG K, LLANG H, BERNS MW & TROMBERG BJ (1995) Cell damage by near-IRmicrobeams. Nature 377:20–21. (Cited on page 21.)

KOSTER S & VERPOORTE E (2007) A decade of microfluidic analysis coupled withelectrospray mass spectrometry: an overview. Lab on a Chip 7(11):1394–1412.(Cited on page 99.)

KÜHN S, LUNT EJ, PHILLIPS BS, HAWKINS AR & SCHMIDT H (2009a) Optoflu-idic particle concentration by a long-range dual-beam trap. Optics Letters34(15):2306–2308. (Cited on pages 25 and 49.)

123

Bibliography

KÜHN S, MEASOR P, LUNT EJ, PHILLIPS BS, DEAMER DW, HAWKINS AR &SCHMIDT H (2009b) Loss-based optical trap for on-chip particle analysis. Labon a Chip 9:2212–2216. (Cited on pages 25 and 49.)

LADENBURG R (1907) Über den Einfluß von Wänden auf die Bewegung einerKugel in einer reibenden Flüssigkeit. Annalen der Physik 328(8):447–458.(Cited on page 61.)

LAMM H (1930) Biegsame optische Geräte. Zeitschrift für Instrumentenkunde50:579–81. (Cited on page 26.)

LEBEDEW PN (1901) Untersuchungen über die Druckkräfte des Lichtes. Annalender Physik 311:433–458. (Cited on page 19.)

LEE SR, KIM J, LEE S, JUNG Y, KIM JK & OH K (2010) All-silica fiber BESSEL-like beam generator and its applications in longitudinal optical trapping andtransport of multiple dielectric particles. Optics Express 18(24):25299–25305.(Cited on page 46.)

VAN LEEUWENHOEK A (1675) Other Microscopical Observations, made by thesame, about the texture of the Blood, the Sap of some Plants, the Figure ofSugar and Salt, and the probable cause of the difference of their Tasts. Philo-sophical Transactions (1665-1678) 10:380–385. (Cited on page 52.)

LEONHARDT U (2006) Momentum in an uncertain light. Nature 444:823–824.(Cited on page 18.)

LÉVESQUE F & SEEBERGER PH (2011) Highly efficient continuous flow reactionsusing singlet oxygen as a "green" reagent. Organic Letters 13(19):5008–5011.(Cited on page 110.)

LIFANTE G (2005) Integrated Photonics: Fundamentals (John Wiley & Sons, Ltd,Chichester, UK). (Cited on page 2.)

LIN SY, FLEMING JG, HETHERINGTON DL, SMITH BK, BISWAS R, HO KM,SIGALAS MM, ZUBRZYCKI W, KURTZ SR & BUR J (1998) A three-dimensionalphotonic crystal operating at infrared wavelengths. Nature 394:251–253.(Cited on page 28.)

LINCOLN B, ERICKSON HM, SCHINKINGER S, WOTTAWAH F, MITCHELL D, ULVICK

S, BILBY C & GUCK J (2004) Deformability-based flow cytometry. Cytometry59A(2):203–209. (Cited on pages 22 and 23.)

LIU Y, SONEK GJ, BERNS MW & TROMBERG BJ (1996) Physiological monitoringof optically trapped cells: assessing the effects of confinement by 1064-nmlaser tweezers using microfluorometry. Biophysical Journal 71(4):2158–2167.(Cited on page 21.)

124

Bibliography

LOZZIO CB & LOZZIO BB (1975) Human chronic myelogeneous leukemia cell-line with positive Philadelphia chromosome. Blood 45(3):321–334. (Cited on

page 102.)

LU H, SCHMIDT MA & JENSEN KF (2001) Photochemical reactions and on-lineUV detection in microfabricated reactors. Lab on a Chip 1:22–28. (Cited on page

110.)

LUKOSZ W (1991) Principles and sensitivities of integrated optical and surfaceplasmon sensors for direct affinity sensing and immunosensing. Biosensors andBioelectronics 6(3):215–225. (Cited on page 12.)

M SQUARED LASERS LTD (2011) SolsTis User Manual v4 & Test Data Sheet. Glas-gow, UK. (Cited on pages 40 and 55.)

MACDONALD MP, SPALDING GC & DHOLAKIA K (2003) Microfluidic sorting in anoptical lattice. Nature 426:421–424. (Cited on page 21.)

MAGENNIS SW, HABTEMARIAN A, NOVAKOVA O, HENRY JB, MEIER S, PARSONS

S, OSWALD IDH, BRABEC V & SADLER PJ (2007) Dual triggering of DNA bind-ing and fluorescence via photoactivation of a dinuclear ruthenium(II) arenecomplex. Inorganic Chemistry 46(12):5059–5068. (Cited on pages 85, 94, 95 and 96.)

MANDAL S & ERICKSON D (2007) Optofluidic transport in liquid core waveg-uiding structures. Applied Physics Letters 90:184103 (3 pp). (Cited on pages 24

and 49.)

MANGAN BJ, FARR L, LANGFORD A, ROBERTS PJ, WILLIAMS DP, COUNY F, LAW-MAN M, MASON M, COUPLAND S, FLEA R, SABERT H, BIRKS TA, KNIGHT JC& RUSSELL PSTJ (2004) Low loss (1.7 dB/km) hollow core photonic bandgapfiber. In Optical Fiber Communication (OFC) Conference 2004 (Optical Soci-ety of America), Los Angeles, CA, USA, paper PDP24. (Cited on page 37.)

MARCATILI EAJ & SCHMELTZER RA (1964) Hollow metallic and dielectric waveg-uides for long distance optical transmission and lasers. Bell Systems TechnicalJournal 43:1783–1809. (Cited on pages 33 and 47.)

MARKLE DR, EVANS EA & HOCHMUTH RM (1983) Force relaxation and perma-ment deformation of erythrocyte membrane. Biophysical Journal 42(1):91–98.(Cited on page 72.)

MCLAFFERTY FW (2011) A century of progress in molecular mass spectrometry.Annual Review of Analytical Chemistry 4:1–22. (Cited on page 87.)

MCLEOD JH (1954) The axicon: a new type of optical element. Journal of theOptical Society of America 44(8):592–597. (Cited on page 45.)

125

Bibliography

MCQUITTY RJ, UNTERKOFLER S, HABTEMARIAN A, EUSER TG, FARRER NJ, RUS-SELL PSTJ & SADLER PJ (2013) Mass spectrometric analysis of light-sensitivedrugs activated via inline photonic crystal fiber microflow reactor. (in prepara-tion). (Cited on pages 74, 80, 95, 98 and 110.)

MEASOR P, KÜHN S, LUNT EJ, PHILLIPS BS, HAWKINS AR & SCHMIDT H (2008)Hollow-core waveguide characterization by optically induced particle trans-port. Optics Letters 33(7):672–674. (Cited on pages 25 and 49.)

METZGER NK, JESS PRT, PATERSON L, WRIGHT EM & DHOLAKIA K (2005) Op-tical binding of chinese hamster ovary cells. In Proceedings of SPIE 2005,Bellingham, WA, USA, volume 5930, paper 59300M. (Cited on page 105.)

MICROFLUIDIC CHIPSHOP (2012) Lab-on-a-Chip Catalogue. Jena, Germany. (Cited

on page 79.)

MILONNI PW & BOYD RW (2010) Momentum of light in a dielectric medium.Advances in Optics and Photonics 2(4):519–553. (Cited on page 19.)

MINKOWSKI H (1908) Die Grundgleichungen für die elektromagnetischenVorgänge in bewegten Körpern. Nachrichten von der Gesellschaft der Wis-senschaften zu Göttingen, Mathematisch-Physikalische Klasse :53–111. (Cited

on page 18.)

MOFRAD MRK (2009) Rheology of the cytoskeleton. Annual Review of Fluid Me-chanics 41:433–453. (Cited on page 17.)

MOHANDAS N & GALLAGHER PG (2008) Red cell membrane: past, present, andfuture. Blood 112(10):3939–3948. (Cited on pages 16 and 52.)

MONAT C, DOMACHUK P & EGGLETON BJ (2007) Integrated optofluidics: A newriver of light. Nature 1:106–114. (Cited on page 2.)

MOREIRA WL, NEVES AAR, GARBOS MK, EUSER TG, RUSSELL PSTJ & CE-SAR CL (2012) Expansion of arbitrary electromagnetic fields in terms ofvector spherical wave functions. URL http://arxiv.org/abs/1003.2392,arXiv:1003.2392v3 [physics.optics]. (Cited on pages 69 and 107.)

MUSIELAK M (2009) Red blood cell-deformability measurement: review of tech-niques. Clinical Hemorheology and Microcirculation 42(1):47–64. (Cited on pages

52, 64 and 72.)

NAGAYAMA K, KAKUI M, MATSUI M, SAITOH T & CHIGUSA Y (2002) Ultra-low-loss (0.1484 dB/km) pure silica core fibre and extension of transmission dis-tance. Electronics Letters 38(20):1168–1169. (Cited on page 37.)

126

http://arxiv.org/abs/1003.2392

Bibliography

NEUMAN KC & BLOCK SM (2004) Optical trapping. Review of Scientific Instru-ments 75(9):2787–2809. (Cited on page 19.)

NÈVE N, KOHLES SS, WINN SR & TRETHEWAY DC (2010) Manipulation of sus-pended single cells by microfluidics and optical tweezers. Cellular and Molecu-lar Bioengineering 3(3):213–228. (Cited on pages 21 and 72.)

NÈVE N, LINGWOOD JK, ZIMMERMAN J, KOHLES SS & TRETHEWAY DC (2008)The µPIVOT: an integrated particle image velocimeter and optical tweezers in-strument for mircoenvironment investigations. Measurement Science and Tech-nology 19(9):095403 (23 pp). (Cited on page 21.)

NICHOLS EF & HULL GF (1903a) The pressure due to radiation. AstrophysicalJournal 17(5):315–351. (Cited on page 19.)

NICHOLS EF & HULL GF (1903b) The pressure due to radiation. (Second paper.).Physical Review (Series I) 17(2):91–104. (Cited on page 19.)

NIELSEN K, NOORDEGRAAF D, SØRENSEN T, BJARKLEV A & HANSEN TP (2005)Selective filling of photonic crystal fibres. Journal of Optics A: Pure and AppliedOptics 7:L13–L20. (Cited on pages 10 and 49.)

NOGUCHI H & GOMPPER G (2005) Shape transitions of fluid vesicles and redblood cells in capillary flows. Proceedings of the National Academy of SciencesUSA 102(40):14159–14164. (Cited on page 72.)

ODDE DJ & RENN MJ (2000) Laser-guided direct writing of living cells. Biotech-nology and Bioengineering 67(3):312–318. (Cited on pages 23, 24, 63 and 70.)

OELGEMÖLLER M & SHVYDKIV O (2011) Recent advances in microflow photo-chemistry. Molecules 16(9):7522–7550. (Cited on page 110.)

OKAMOTO K (2006) Fundamentals of Optical Waveguides. 2nd edition (Aca-demic Press – Elsevier, Burlington, MA, USA). (Cited on page 15.)

PADGETT M & BOWMAN R (2011) Tweezers with a twist. Nature Photonics 5:343–348. (Cited on page 21.)

PADGETT M & DI LEONARDO R (2011) Holographic optical tweezers and theirrelevance to lab on chip devices. Lab on a Chip 11(7):1196–1205. (Cited on page

21.)

PANG L, CHEN HM, FREEMAN LM & FAINMAN Y (2012) Optofluidic devices andapplications in photonics, sensing and imaging. Lab on a Chip 12(19):3543–3551. (Cited on page 2.)

127

Bibliography

PATERSON L, PAPAGIAKOUMOU E, MILNE G, GARCÉS-CHÁVEZ V, TATARKOVA SA,SIBBETT W, GUNN-MOORE FJ, BRYANT PE, RICHES AC & DHOLAKIA K (2005)Light-induced cell separation in a tailored optical landscape. Applied PhysicsLetters 87(12):123901 (3 pp). (Cited on page 46.)

PATMORE NJ (2008) Photophysical properties of metal complexes. Annual Re-ports on the Progress of Chemistry, Section A: Inorganic Chemistry 104:498–528.(Cited on page 75.)

PAYNE DN & GAMBLING WA (1972) New low-loss liquid-core fibre waveguide.Electronics Letters 8(15):374–376. (Cited on pages 27 and 49.)

PEARCE GJ (2006) Plane-wave methods for modelling photonic crystal fibre.Ph.D. thesis, University of Bath, Department of Physics. (Cited on page 35.)

PEARCE GJ, HEDLEY TD & BIRD DM (2005) Adaptive curvilinear coordinates ina plane-wave solution of MAXWELL’s equations in photonic crystals. PhysicalReview B 71(19):195108 (10 pp). (Cited on page 35.)

PEARCE GJ, WIEDERHECKER GS, POULTON CG, BURGER S & RUSSELL PSTJ(2007) Models for guidance in kagome-structured hollow-core photonic crys-tal fibres. Optics Express 15(20):12680–12685. (Cited on pages 38 and 39.)

PELTON RH & CHIBANTE P (1986) Preparation of aqueous latices with N -Isopropylacrylamide. Colloids and Surfaces 20(3):247–256. (Cited on page 109.)

PETERMAN EJG, GITTES F & SCHMIDT CF (2003) Laser-induced heating in opti-cal traps. Biophysical Journal 84(2):1308–1316. (Cited on page 21.)

PETROVICH MN, POLETTI F, VAN BRAKEL A & RICHARDSON DJ (2008) Robustlysingle mode hollow core photonic bandgap fiber. Optics Express 16(6):4337–4346. (Cited on page 34.)

PFEIFER RNC, NIEMINEN TA, HECKENBERG NR & RUBINSZTEIN-DUNLOP H(2011) Optical tweezers and paradoxes in electromagnetism. Journal of Op-tics 13(4):044017 (7 pp). (Cited on page 18.)

PHILLIPS D (1995) The photochemistry of sensitisers for photodynamic therapy.Pure and Applied Chemistry 67(1):117–126. (Cited on page 75.)

PINTO AMR & LOPEZ-AMO M (2012) Photonic crystal fibers for sensing applica-tions. Journal of Sensors 2012:598178 (21 pp). (Cited on page 12.)

POISEUILLE JLM (1840) Recherches expérimentales sur le mouvement des liq-uides dans les tubes de très petits diamètres. I. Influences de la pression surla quantité de liquide qui traverse les tubes de très petits diamètres. ComptesRendus de l’Académie des Sciences 11:961–967. (Cited on page 7.)

128

Bibliography

POLETTI F (2010) Hollow core fiber with an octave spanning bandgap. OpticsLetters 35(17):2837–2839. (Cited on page 36.)

POLLARD TD & BORISY GG (2003a) Cellular motility driven by assembly anddisassembly of actin filaments. Cell 112(4):453–465. (Cited on page 16.)

POLLARD TD & BORISY GG (2003b) Cellular motility driven by assembly anddisassembly of actin filaments - erratum. Cell 113(4):549. (Cited on page 16.)

POLLOCK CR & LIPSON M (2003) Integrated Photonics (Kluwer Academic Pub-lishers, Norwell, MA, USA). (Cited on page 2.)

POTTAGE JM, BIRD DM, HEDLEY TD, BIRKS TA, KNIGHT JC & RUSSELL PSTJ(2003) Robust photonic band gaps for hollow core guidance in PCF made fromhigh index glass. Optics Express 11(22):2854–2861. (Cited on page 35.)

POZRIKIDIS C (2005) Axisymmetric motion of a file of red blood cells throughcapillaries. Physics of Fluids 17(3):031503 (14 pp). (Cited on page 72.)

PRACHAROVA J, ZERZANKOVA L, STEPANKOVA J, NOVAKOVA O, FARRER NJ,SADLER PJ, BRABEC V & KASPARKOVA J (2012) Interactions of DNA with a newplatinum(IV) azide dipyridine complex activated by UVA and visible light: re-lationship to toxicity in tumor cells. Chemical Research in Toxicology 25:1099–1111. (Cited on page 91.)

PRAPROTNIK M, DELLE SITE L & KREMER K (2008) Multiscale simulation of softmatter: from scale bridging to adaptive resolution. Annual Review of PhysicalChemistry 59:545–571. (Cited on page 17.)

PRATT JM (1964) The chemistry of vitamin B12. Part II. Photochemical reactions.Journal of the Chemical Society (988):5154–5160. (Cited on page 86.)

PSALTIS D, QUAKE SR & YANG C (2006) Developing optofluidic technologythrough the fusion of microfluidics and optics. Nature 442:381–386. (Cited

on page 2.)

RAMSER K & HANSTORP D (2010) Optical maniplation for single-cell studies.Journal of Biophotonics 3(4):187–206. (Cited on page 17.)

RANKA JK, WINDELER RS & STENTZ AJ (2000) Visible continuum generation inair-silica microstructure optical fibers with anomalous dispersion at 800 nm.Optics Letters 25(1):25–27. (Cited on page 30.)

REMMERBACH TW, WOTTAWAH F, DIETRICH J, LINCOLN B, WITTEKIND C & GUCK

J (2009) Oral cancer diagnosis by mechanical phenotyping. Cancer Research69(5):1728–1732. (Cited on page 22.)

129

Bibliography

RENN MJ, PASTEL R & LEWANDOWSKI HJ (1999) Laser guidance and trappingof mesoscale particles in hollow-core optical fibers. Physical Review Letters82(7):1574–1577. (Cited on page 24.)

RINDORF L, HØIBY PE, JENSEN JB, , HAGSHOLM PEDERSEN L, BANG O &GESCHKE O (2006) Towards biochips using microstructured optical fiber sen-sors. Analytical and Bioanalytical Chemistry 385:1370–1375. (Cited on pages 12

and 107.)

RISK WP, KIM HC, MILLER RD, TEMKIN H & GANGOPADHYAY S (2004) Opticalwaveguides with an aqueous core and a low-index nanoporous cladding. OpticsExpress 12(26):6446–6455. (Cited on page 49.)

ROBERTS PJ, COUNY F, SABERT H, MANGAN BJ, WILLIAMS DP, FARR L, MASON

MW, TOMLINSON A, BIRKS TA, KNIGHT JC & RUSSELL PSTJ (2005a) Ultimatelow loss of hollow-core photonic crystal fibres. Optics Express 13(1):236–244.(Cited on page 37.)

ROBERTS PJ, WILLIAMS DP, MANGAN BJ, SABERT H, COUNY F, WADSWORTH

WJ, BIRKS TA, KNIGHT JC & RUSSELL PSTJ (2005b) Realizing low loss aircore photonic crystal fibers by exploiting an antiresonant core surround. OpticsExpress 13(20):8277–8285. (Cited on page 37.)

ROBERTSON CA, HAWKINS EVANS D & ABRAHAMSE H (2009) Photodynamic ther-apy (PDT): a short review on cellular mechanisms and cancer research applica-tions for PDT. Journal of Photochemistry and Photobiology, B: Biology 96(1):1–8. (Cited on pages 12 and 75.)

ROSS D, GAITAN M & LOCASCIO LE (2001) Temperature measurement in mi-crofluidic systems using a temperature-dependent fluorescent dye. AnalyticalChemistry 73(17):4117–4123. (Cited on page 108.)

RUAN Y, SCHARTNER EP, EBENDORFF-HEIDEPRIEM H, HOFFMANN P & MONRO

TM (2007) Detection of quantum-dot labelled proteins using soft glass mi-crostructured optical fibers. Optics Express 15(26):17819–17826. (Cited on page

12.)

RUSSELL PSTJ (2003) Photonic crystal fibers. Science 299:358–362. (Cited on pages

27 and 29.)

RUSSELL PSTJ (2006) Photonic-crystal fibers. Journal of Lightwave Technology24(12):4729–4749. (Cited on pages 3, 29, 30, 34 and 35.)

RUSSELL PSTJ (2007) Photonic crystal fiber: Finding the holey grail. Optics andPhotonics News 18(7):26–31. (Cited on page 30.)

130

Bibliography

RUSSELL PSTJ, BIRKS TA & LLOYD-LUCAS FD (1993) Photonic BLOCH wavesand photonic band gaps. In E BURSTEIN & C WEISBUCH (Editors), ConfinedElectrons and Photons (Plenum Press, New York, NY, USA), (pages 585–633).(Cited on page 29.)

SADLER PJ & MULLER P (2011) Photoreactive compounds and compositions. USpatent No. 7,928,097 B2. (Cited on page 91.)

SAITOH K, MORTENSEN NA & KOSHIBA M (2004) Air-core photonic band-gapfibers: the impact of surface modes. Optics Express 12(3):394–400. (Cited on

page 37.)

SAWATARI T & KAPANY NS (1970) Wave propagation along hollow dielectricwaveguides. Journal of the Optical Society of America 60(1):132–133. (Cited

on page 46.)

SCHERER WF, SYVERTON JT & GEY GO (1953) Studies on the propagation invitro of poliomyelitis viruses. IV. Viral multiplication in a stable strain of humanmalignant epithelial cells (strain HeLa) derived from an epidermoid carcinomaof the cervix. Journal of Experimental Medicine 97(5):695–710. (Cited on page

102.)

SCHMIDT BS, YANG AHJ, ERICKSON D & LIPSON M (2007) Optofluidic trappingand transport on solid core waveguides within a microfluidic device. OpticsExpress 15(22):14322–14334. (Cited on page 24.)

SCHMIDT H & HAWKINS AR (2008) Optofluidic waveguides: I. Concepts andimplementations. Microfluidics and Nanofluidics 1(4):3–16. (Cited on page 49.)

SCHMIDT H & HAWKINS AR (2011) The photonic integration of non-solid mediausing optofluidics. Nature Photonics 5:598–604. (Cited on pages 2 and 49.)

SCHMIDT M, CUBILLAS AM, TACCARDI N, EUSER TG, CREMER T, MAIER F, STEIN-RÜCK HP, RUSSELL PSTJ, WASSERSCHEID P & ETZOLD BJM (2013) Chemi-cal and (photo)-catalytical transformations in photonic crystal fibers. Chem-CatChem 5(3):641–650. (Cited on page 98.)

SCHMIDT OA, GARBOS MK, EUSER TG & RUSSELL PSTJ (2012a) Metrology oflaser-guided particles in air-filled hollow-core photonic crystal fiber. Optics Let-ters 37(1):91–93. (Cited on page 25.)

SCHMIDT OA, GARBOS MK, EUSER TG & RUSSELL PSTJ (2012b) Reconfigurableoptothermal microparticle trap in air-filled hollow-core photonic crystal fiber.Physical Review Letters 109(2):024502 (5 pp). (Cited on pages 25 and 63.)

131

Bibliography

SECOMB TW, SKALAK R, ÖZKAYA N & GROSS JF (1986) Flow of axisymmetricred blood cells in narrow capillaries. Journal of Fluid Mechanics 163:405–423.(Cited on page 72.)

SEIYAMA A, SUZUKI Y & MAEDA N (1993) Increased viscosity of erythrocyte sus-pension upon hemolysis. Colloid & Polymer Science 271:63–69. (Cited on pages 64

and 65.)

SHIBAYAMA M & TANAKA T (1993) Volume phase transition and related phe-nomena of polymer gels. Advances in Polymer Science 109:1–62. (Cited on page

108.)

SIDDIK ZH (2003) Cisplatin: mode of cytotoxic action and molecular basis ofresistance. Oncogene 22(47):7265–7279. (Cited on page 74.)

SINGER W, FRICK M, BERNET S & RITSCH-MARTE MAM (2003) Self-organizedarray of regularly spaced microbeads in a fiber-optical trap. Journal of the Op-tical Society of America B: Optical Physics 20(7):1568–1574. (Cited on page 105.)

SKALAK R & BRÅNEMARK PI (1969) Deformation of red blood cells in capillaries.Science 164(3880):717–719. (Cited on page 72.)

SMITH CM, VENKATARAMAN N, GALLAGHER MT, MÜLLER D, WEST JA, BORRELLI

NF, ALLAN DC & KOCH KW (2003) Low-loss hollow-core silica/air photonicbandgap fibre. Nature 424:657–659. (Cited on page 37.)

SMOLKA S, BARTH M & BENSON O (2007) Highly efficient fluorescence sensingwith hollow core photonic crystal fibers. Optics Express 15(20):12783–12791.(Cited on pages 12 and 49.)

SNITZER E (1961) Cylindrical dielectric waveguide modes. Journal of the OpticalSociety of America 51(5):491–498. (Cited on page 33.)

SPANGGAARD H & KREBS FC (2004) A brief history of the development of organicand polymeric photovoltaics. Solar Energy Materials and Solar Cells 83:125–146. (Cited on page 12.)

SQUIRES TM & QUAKE SR (2005) Microfluidics: Fluid physics at the nanoliterscale. Reviews of Modern Physics 77(3):977–1026. (Cited on pages 2 and 6.)

STEVENSON DJ, GUNN-MOORE F & DHOLAKIA K (2010) Light forces thepace: optical manipulation for biophotonics. Journal of Biomedical Optics15(4):041503 (21 pp). (Cited on page 21.)

STONE J (1972) Optical transmission in liquid-core quartz fibers. Applied PhysicsLetters 20(7):239–240. (Cited on pages 27 and 49.)

132

Bibliography

SUGIMOTO A, SUMINO Y, TAKAGI M, FUKUYAMA T & RYU I (2006) The BARTON

reaction using a microreactor and black light. Continuous-flow synthesis of akey steroid intermediate for an endothelin receptor antagonist. TetrahedronLetters 47(35):6197–6200. (Cited on page 110.)

SUGIMOTO Y, TANAKA Y, IKEDA N, NAKAMURA Y & ASAKAWA K (2004)Low propagation loss of 0.76 dB/mm in GaAs-based single-line-defect two-dimensional photonic crystal slab waveguides up to 1 cm in length. Optics Ex-press 12(6):1090–1096. (Cited on page 28.)

SURESH S (2007) Biomechanics and biophysics of cancer cells. Acta Biomaterialia3(4):413–438. (Cited on page 16.)

SVANBERG K, BENDSOE N, AXELSSON J, ANDERSSON-ENGELS S & SVANBERG S(2010) Photodynamic therapy: superficial and interstitial illumination. Jour-nal of Biomedical Optics 15(4):041502 (10 pp). (Cited on page 75.)

SVOBODA K & BLOCK SM (1994) Biological applications of optical forces. AnnualReview of Biophysics and Biomolecular Structure 23:247–285. (Cited on page 21.)

TAM AC & PATEL CKN (1979) Optical absorptions of light and heavy water bylaser optoacoustic spectroscopy. Applied Optics 18(19):3348–3358. (Cited on

page 77.)

TANAKA T (1978) Dynamics of critical concentration fluctuations in gels. PhysicalReview A 17(2):763–766. (Cited on page 108.)

TAYLOR GI (1953) Dispersion of a soluble matter in solvent flowing slowlythrough a tube. Proceedings of the Royal Society of London – Series A219(1137):186–203. (Cited on pages 9 and 10.)

TONUCCI RJ, JUSTUS BL, CAMPILLO AJ & FORD CE (1992) Nanochannel arrayglass. Science 258(5083):783–785. (Cited on page 31.)

TOPAS ADVANCED POLYMERS GMBH (2006) Topas® Cyclic Olefin Copolymer(COC) - Product Brochure. Frankfurt am Main, Germany. (Cited on page 78.)

TORCHILIN VP (2000) Drug targeting. European Journal of Pharmaceutical Sci-ences 11(Suppl. 2):S81–S91. (Cited on page 74.)

UENO K, KITAGAWA F & KITAMURA N (2002) Photocyanation of pyrene across anoil/water interface in a polymer microchannel chip. Lab on a Chip 2(4):231–234. (Cited on page 110.)

UNTERKOFLER S, GARBOS MK, EUSER TG & RUSSELL PSTJ (2013) Long distanceoptical guidance and deformation monitoring of cells in optofluidic photoniccrystal fiber. Journal of Biophotonics X:Y. (Cited on page 51.)

133

Bibliography

UNTERKOFLER S, MCQUITTY RJ, EUSER TG, FARRER NJ, SADLER PJ & RUSSELL

PSTJ (2012) Microfluidic integration of photonic crystal fibers for online pho-tochemical reaction analysis. Optics Letters 37(11):1952–1954. (Cited on pages

72, 74, 80 and 110.)

VANAPALLI SA, DUITS MHG & MUGELE F (2009) Microfluidics as a functionaltool for cell mechanics. Biomicrofluidics 3:012006 (15 pp). (Cited on pages 16, 21

and 104.)

VANNEMAN M & DRANOFF G (2012) Combining immunotherapy and targetedtherapies in cancer treatment. Nature Reviews Cancer 12(4):237–251. (Cited on

page 74.)

CIŽMÁR T, GARCÉS-CHÁVEZ V, DHOLAKIA K & ZEMÁNEK P (2005) Optical con-veyor belt for delivery of submicron objects. Applied Physics Letters 86:174101(3 pp). (Cited on page 46.)

VILLENEUVE PR & PICHÉ M (1992) Photonic band gaps in two-dimensionalsquare and hexagonal lattices. Physical Review B 46(8):4969–4972. (Cited on

page 29.)

VOGLER A, HIRSCHMANN R, OTTO H & KUNKELY H (1976) Photochemistry ofbiologically important transition metal complexes. I. Cyanocobalamin and re-lated corrin complexes of rhodium(III). Berichte der BUNSEN-Gesellschaft fürphysikalische Chemie 80(5):420–424. (Cited on page 86.)

VUKUSIC P & SAMBLES JR (2003) Photonic structures in biology. Nature424:852–855. (Cited on page 27.)

VUKUSIC P & SAMBLES JR (2004) Photonic structures in biology – corrigendum.Nature 429:680. (Cited on page 27.)

WADSWORTH WJ, JOLY N, KNIGHT JC, BIRKS TA, BIANCALANA F & RUSSELL PSTJ(2004) Supercontinuum and four-wave mixing with Q-switched pulses in end-lessly single-mode photonic crystal fibres. Optics Express 12(2):299–309. (Cited

on page 31.)

WANG MM, TU E, RAYMOND DE, YANG JM, ZHANG H, HAGEN N, DEES B, MER-CER EM, FORSTER AH, KARIV I, MARCHAND PJ & BUTLER WF (2005) Microflu-idic sorting of mammalian cells by optical force switching. Nature Biotechnol-ogy 23(1):83–87. (Cited on page 21.)

WANG YY, PENG X, ALHARBI M, FOURCADE DUTIN C, BRADLEY TD, GÉRÔME F,MIELKE M, BOOTH T & BENABID F (2012) Design and fabrication of hollow-core photonic crystal fibers for high-power ultrashort pulse transportation andpulse compression. Optics Letters 37(15):3111–3113. (Cited on page 39.)

134

Bibliography

WARBURG O (1956) On the origin of cancer cells. Science 123(3191):309–314.(Cited on page 75.)

WATTS P & HASWELL SJ (2003) Continuous flow reactors for drug discovery.Drug Discovery Today 8(13):586–593. (Cited on page 110.)

WAUGH RE, NARLA M, JACKSON CW, MUELLER TJ, SUZUKI T & DALE GL (1992)Rheologic properties of senescent erythrocytes: loss of surface area and vol-ume with red blood cell age. Blood 79(5):1351–1358. (Cited on page 66.)

WELLS J (1989) Hair light guide. Nature 338(6210):23. (Cited on page 26.)

WEST JA, SMITH CM, BORRELLI NF, ALLAN DC & KOCH KW (2004) Surfacemodes in air-core photonic band-gap fibers. Optics Express 12(8):1485–1496.(Cited on page 37.)

WHEELER AR, THRONDSET WR, WHELAN RJ, LEACH AM, ZARE RN, LIAO YH,FARRELL K, MANGER ID & DARIDON A (2003) Microfluidic device for single-cell analysis. Analytical Chemistry 75(14):3581–3586. (Cited on page 107.)

WHITESIDES GM (2006) The origins and the future of microfluidics. Nature442:368–373. (Cited on page 2.)

WILLIAMS GOS, CHEN JSY, EUSER TG, RUSSELL PSTJ & JONES AC (2012) Pho-tonic crystal fibre as an optofluidic reactor for the measurement of photochem-ical kinetics with sub-picomole sensitivity. Lab on a Chip 12(18):3356–3361.(Cited on page 73.)

WILLIAMS GOS, EUSER TG, RUSSELL PSTJ & JONES AC (2013) Spectrofluorime-try with attomole sensitivity in photonic crystal fibres. Methods and Applica-tions in Fluorescence (in print). (Cited on pages 12 and 98.)

WINDING P & BERCHTOLD MW (2001) The chicken B cell line DT40: a novel toolfor gene disruption experiments. Journal of Immunological Methods 249:1–16.(Cited on page 103.)

WOLFE DB, CONROY RS, GARSTECKI P, MAYERS BT, FISCHBACH MA, PAUL KE,PRENTISS M & WHITESIDES GM (2004) Dynamic control of liquid-core/liquid-cladding optical waveguides. Proceedings of the National Academy of SciencesUSA 101(34):12434–12438. (Cited on page 49.)

WOOTTON RCR, FORTT R & DE MELLO AJ (2002) A microfabricated nanoreac-tor for safe, continuous generation and use of singlet oxygen. Organic ProcessResearch & Development 6(2):187–189. (Cited on page 110.)

XIAO L, BIRKS TA & LOH WH (2011) Hydrophobic photonic crystal fibers. OpticsLetters 36(23):4662–4664. (Cited on page 98.)

135

YABLONOVITCH E (1987) Inhibited spontaneous emission in solid-state physicsand electronics. Physical Review Letters 58(20):2059–2062. (Cited on page 27.)

YABLONOVITCH E (1993) Photonic band-gap structures. Journal of the OpticalSociety of America B 10(2):283–295. (Cited on page 29.)

YANFENG C & MIN Y (2001) Swelling kinetics and stimuli-responsiveness ofpoly(DMAEMA) hydrogels prepared by UV-irradiation. Radiation Physics andChemistry 61(1):65–68. (Cited on page 109.)

YEH P, YARIV A & MAROM E (1978) Theory of BRAGG fiber. Journal of the OpticalSociety of America 68(9):1196–1201. (Cited on pages 30 and 39.)

YEH Y & CUMMINS HZ (1964) Localized fluid flow measurements with a He-Nelaser spectrometer. Applied Physics Letters 4(10):176–178. (Cited on page 58.)

YIN D, DEAMER DW, SCHMIDT H, BARBER JP & HAWKINS AR (2006) Single-molecule detection sensitivity using planar integrated optics on a chip. OpticsLetters 31(14):2136–2138. (Cited on page 49.)

YIN D, SCHMIDT H, BARBER JP, LUNT EJ & HAWKINS AR (2005) Optical charac-terization of arch-shaped ARROW waveguides with liquid cores. Optics Express13(26):10564–10570. (Cited on page 49.)

ZAMANI AGHAIE K, DANGUI V, DIGONNET MJF, FAN S & KINO GS (2009) Classifi-cation of the core modes of hollow-core photonic-bandgap fibers. IEEE Journalof Quantum Electronics 45(9):1192–1200. (Cited on page 33.)

ZHANG H & LIU KK (2008) Optical tweezers for single cells. Journal of the RoyalSociety Interface 5(24):671–690. (Cited on page 17.)

Publications

Journal Articles

• R. J. MCQUITTY, S. UNTERKOFLER, A. HABTEMARIAN, T. G. EUSER, N. J. FAR-RER, P. St.J. RUSSELL and P. J. SADLER

Rapid reaction analysis of photoactivatable anticancer prodrugs by integrationof photonic crystal fiber microflow reactors to mass spectrometryin preparation.

• A. M. CUBILLAS, M. SCHMIDT, T. G. EUSER, B. J. M. ETZOLD, N. TACCARDI,S. UNTERKOFLER, P. WASSERSCHEID and P. St.J. RUSSELL

Liquid-phase heterogeneous catalysis in photonic crystal fiber microreactorin preparation.

• A. M. CUBILLAS, S. UNTERKOFLER, T. G. EUSER, B. J. M. ETZOLD, A. C. JONES,P. J. SADLER, P. WASSERSCHEID and P. St.J. RUSSELL

Photonic crystal fibres for chemical sensing and photochemistryChemical Society Reviews (in print) → doi: 10.1039/C3CS60128E.

• S. UNTERKOFLER*, M. K. GARBOS*, T. G. EUSER and P. St.J. RUSSELL

Long-distance laser propulsion and deformation-monitoring of cells in optoflu-idic photonic crystal fiberJournal of Biophotonics (in print) → doi: 10.1002/jbio.201200180.

• S. UNTERKOFLER, R. J. MCQUITTY, T. G. EUSER, N. J. FARRER, P. J. SADLER,and P. St.J. RUSSELL

Microfluidic integration of photonic crystal fibers for online photochemicalreaction analysisOptics Letters 37(11):1952-1954 (2012) → doi: 10.1364/OL.37.001952.

• M. K. GARBOS, T. G. EUSER, O. A. SCHMIDT, S. UNTERKOFLER and P. St.J. RUS-SELL

Doppler velocimetry on microparticles trapped and propelled by laser light inliquid-filled photonic crystal fiberOptics Letters 36(11):2020-2022 (2011) → doi: 10.1364/OL.36.002020.

137

http://dx.doi.org/10.1039/C3CS60128E

http://dx.doi.org/10.1002/jbio.201200180

http://dx.doi.org/10.1364/OL.37.001952

http://dx.doi.org/10.1364/OL.36.002020

• S. UNTERKOFLER, T. PFLOCK, J. SOUTHALL, R. J. COGDELL, J. KÖHLER

Fluorescence blinking of the RC-LH1 complex from Rhodopseudomonas palus-trisChemPhysChem 12:711-716 (2011) → doi: 10.1002/cphc.201000588.

Conference Contributions

• S. UNTERKOFLER, M. K. GARBOS, F. TÜMER†, T. G. EUSER, G. WHYTE and P. St.J.RUSSELL

Poster: Long-distance laser propulsion and deformation-monitoring of cells inhollow-core photonic crystal fiberBiophotonics ’13 – 6th International Graduate Summer School, Isle of Ven, Swe-den (June 2013).

• A. M. CUBILLAS†, M. SCHMIDT, T. G. EUSER, B. J. M. ETZOLD, N. TACCARDI,S. UNTERKOFLER, P. WASSERSCHEID and P. St.J. RUSSELL

Talk: Optically monitored catalytic photonic crystal fibre microreactorCLEO/Europe – IQEC 2013, Munich, Germany (Optical Society of America), pa-per CH-6.1_THU (May 2013).

• T. G. EUSER†, O. A. SCHMIDT, S. UNTERKOFLER and P. St.J. RUSSELL

Invited Talk: Laser-propulsion of particles and cells in hollow-core photoniccrystal fiberOptical Trapping Applications (OTA), Waikoloa Beach (HI), USA (Optical Societyof America), paper TT3D.1 (April 2013) → doi: 10.1364/OTA.2013.TT3D.1.

• T. G. EUSER†, O. A. SCHMIDT, M. K. GARBOS, S. UNTERKOFLER and P. St.J. RUS-SELL

Invited Talk: Laser-propulsion of microparticles in hollow-core photonic crys-tal fibers: a review of recent developmentsIEEE 3rd International Conference on Photonics ICP2012, Penang, Malaysia,paper 316/317 (October 2012) → doi: 10.1109/ICP.2012.6379528.

• S. UNTERKOFLER†, M. K. GARBOS, T. G. EUSER and P. St.J. RUSSELL

Poster: Optofluidic folding of red blood cells in liquid-filled hollow-core pho-tonic crystal fiberEMBL Microfluidics 2012, Heidelberg, Germany (May 2012).

• S. UNTERKOFLER†, R. J. MCQUITTY, T. G. EUSER, N. J. FARRER, P. J. SADLER,and P. St.J. RUSSELL

Talk: Optofluidic hollow-core photonic crystal fiber coupled to mass spectrom-etry for rapid photochemical reaction analysisCLEO: Science and Innovations, San José (CA), USA (Optical Society of America),paper CW1G.1 (May 2012) → doi: 10.1364/CLEO_SI.2012.CW1G.1.

http://dx.doi.org/10.1002/cphc.201000588

http://dx.doi.org/10.1364/OTA.2013.TT3D.1

http://dx.doi.org/10.1109/ICP.2012.6379528

http://dx.doi.org/10.1364/CLEO_SI.2012.CW1G.1

• M. K. GARBOS†, T. G. EUSER, S. UNTERKOFLER and P. St.J. RUSSELL

Talk: Doppler velocimetry of microspheres laser-propelled in liquid-filled hol-low-core photonic crystal fiberin CLEO/Europe 2011 Conference, Munich, Germany (Optical Society of Amer-ica), paper CLEB5_1 (May 2011) → doi: 10.1109/CLEOE.2011.5943237.

• S. UNTERKOFLER†, M. K. GARBOS, T. G. EUSER, B. FABRY and P. St.J. RUSSELL

Poster: Single cell mechanics in liquid-filled hollow-core photonic crystal fibersBiophotonics ’11 – 5th International Graduate Summer School, Isle of Ven, Swe-den (May 2011).

• S. UNTERKOFLER, T. PFLOCK†, R. J. COGDELL, J. KÖHLER

Poster: Fluorescence blinking of the RC-LH1 core complex from Rhodopseu-domonas palustrisDPG Frühjahrstagung 2010, Fachverband Molekülphysik, Hannover, Germany,MO 26.4 (March 2010).

*) equal contribution†) presenterpublications not related to this thesis

139

http://dx.doi.org/10.1109/CLEOE.2011.5943237

Acknowledgements

"Leider läßt sich eine wahrhafte Dankbarkeit mit Worten nicht ausdrücken."


My groupThe biggest thank you I undoubtedly owe Prof. Philip RUSSELL for giving me theopportunity of performing research in this truly interdisciplinary and fascinatingPh.D. project. It is a privilege to work in such a group full of motivated peopleand in such an institute full of outstanding scientists. I am deeply grateful thatyou made this possible for me. I admire your visionary enthusiasm and intuitionfor science and hope I have learned how to develop this myself.

Secondly, my thanks go to my advisor Tijmen EUSER for letting me share hisprofound knowledge and broad experience all around PCFs. Every time I was"stuck", you would borrow me your bright mind and — which for me is evenmore important — your patience. I couldn’t imagine a better supervisor and Iwish you all the best on your way of becoming a professor. You are made for it.

Supervisors are indispensable, but so are reliable and cheerful lab mates! Ihrwisst, meine Lieben, dass ich Euch daher auch lieber Doktorbrüder und -schwes-tern nenne! ;-) Vielen Dank, großer Bruder Martin GARBOS... für sooo vieleDinge! Zuvörderst natürlich dass Du mich in alle Tricks und Geheimnisse des"PCF particle guidance setup" eingeweiht hast, dessen Gründer Du schließlichbist. In diesem Zusammenhang sei auch zu erwähnen, dass es angesichts unsererArbeitsstile wohl keine bessere (da gegensätzlichere) Kombination hätte gebenkönnen. Aber Arbeit ist ja nicht alles und ich werde es Dir nicht vergessen,dass Du mich immer wieder aufgemuntert und ermutigt hast, wenn die Dinge(welche auch immer) schlecht liefen (was auch für Annie gilt!). Thank you bigsister Jocelyn CHEN, pioneer of PCF photoreactor science, for all the work thatyou have done and that I could build my own project on... and for helping mesearch these boots all over Edinburgh! Danke Danke Danke, lieber kleiner Bruder,Oliver SCHMIDT. Wobei "klein" angesichts Deiner körperlichen sowie geistigenGröße hier wie der blanke Hohn daherkommt! Auch Du hast mich immer wiederaufgebaut oder — falls Aufbauen zu unrealisch war — zumindest ordentlichmit draufgekloppt. Ich denke "Tante" Ana CUBILLAS hatte mit uns im Labor die

141

besten Voraussetzungen, um ihr Vokabular an deutschen Schimpfwörtern kräftigaufzustocken. In der Tat sagt keiner schöner "Scheiße" als Du, liebe Ana! Vielenlieben Dank auch Dir für Deine Unterstützung und unsere gemeinsamen Papers.Good luck to baby sister (... just arrived ...) Fatma TÜMER. You seem to beequipped with a strong will and good nerves, which is the perfect prerequisitecombination to successfully run the cell guidance experiments.

Ohne passende Hohlkern-PCFs bestünde diese Arbeit lediglich aus Schall undRauch. Man braucht sehr viel Erfahrung, sowie viel Schweiß und Spucke (nurbildlich versteht sich), um gescheite Fasern zu ziehen. Dafür liebes fibre fabri-cation team habt ihr meinen vollen Respekt! Herzlichen Dank daher an MichaelSCHARRER, Fehim BABIC und Michael FROSZ, aber am allermeisten an SilkeRAMMLER, mit der ich so viel Zeit (quatschenderweise) im Reinraum verbrachthabe. Gut, dass Frauen multitasking-fähig sind...

Vielen Dank an meine Bürokumpanen Patrick UEBEL, (Mohiudeen) AZHAR undNicolai GRANZOW dafür dass sie angesichts meiner Endlosschleife: "...ich hasseWindows, ich hasse Windows, ich hasse Windows..." nicht die Nerven verlorenund trotz der sich stapelnden Schuhkartons immer wieder zu ihrem Schreibtischgefunden haben. Bitte verzeiht mir, dass ich die LabSnacks-Box Competitiondann doch vorzeitig und mit haushohem Abstand gewonnen habe! ;-P (und das,obwohl ich gar keine Süßigkeiten mag!)

Dear remaining group members (former and present), you have become toomany to have grateful words for every single one of you. I honestly enjoyedworking in such a helpful and inspiring environment and I am happy and proudthat you guys were my colleagues. I wish you all the best for the future (in anti-alphabetical order): Marta ZIEMIENCZUK, Leyun "Laurel" ZANG, Xiaoming XI,Gordon WONG, Anja WINTERSTEIN, Thomas WEISS, Qi "Aaron" WANG, AndreasWALSER, Frederik "Freddy" VINZENT, Shaleindra VARSHNEY, Hemant TYAGI, JohnTRAVERS, Barbara TRABOLD, Francesco TANI, Alessio STEFANI, S. Paul STARK,Patricia SCHREHARDT, Markus SCHMIDT, Michael SCHMIDBERGER, Luis PRILL

SEMPERE, Alexander PODLIPENSKY, Johannes NOLD, Thang NGUYEN, AlexanderNAZARKIN, KaFai MAK, Howard LEE, Günther KRON, Johannes KÖHLER, Chris-tine KREUZER, Ralf KEDING, (MyeonSoo) KANG, Nicolas "Nick" JOLY, Xin "Jimmy"JIANG, Namvar JAHANMEHR, Holger HUNDERTMARK, Philipp HÖLZER, MartinFINGER, Georg EPPLE, Stan DÖRSCHNER, Wonkeun CHANG, Anna BUTSCH, Mar-tin BUTRYN, Federico BELLI, Sebastian BAUERSCHMIDT, Amir ABDOLVAND.

... und natürlich danke ich Bettina "Tina" SCHWENDER, die die ganze Russell-bande zur Zucht und Ordnung ruft und damit am Laufen hält!

Im Zusammenhang mit Zusammenhalten gilt mein Dank auch allen Leutenaus der Verwaltung, einschließlich IT, Elektronikwerkstatt, aber vor allem dermechanischen Werkstatt: Vielen Dank Bernhard THOMANN und Robert GALL fürdie allzeit bereite Bearbeitung meiner "Gleindeile"!

My collaboratorsInterdisciplinary research is only made possible by collaborations with experts inother fields.

Daher möchte ich zu allererst Prof. Ben FABRY danken, der zudem auchmein 2nd supervisor innerhalb der IMPRS war. Danke für Dein Interesse, Deineguten Ideen und Deine allgemeine Unterstützung. Speziell natürlich dafür, dassDu mir uneingeschränkten Zutritt zu Deinen Labors gewährt hast. In diesemZusammenhang danke ich Deinen Leuten Christine ALBERT, Astrid MAINKA, VeraAUERNHEIMER und meiner IMPRS-Kollegin Lena LAUTSCHAM für die Zellenprä-paration, sowie Caroline GLUTH für ihre Einführung in die PDMS microfluidic-chip fabrication. Ebenso danke ich Christian WEIS, allerdings eher in persön-licher (und musikalischer!) Hinsicht.

Newcomer at the FAU, but at the same time good friend of our group alreadybefore his arrival: Jun.-Prof. Graeme WHYTE. Thanks for the many good ideasand discussions. I wish you all the best for your research and scientific career.

I have greatly enjoyed my stays with my collaborators at the Chemistry-depart-ment of the University of Warwick! Prof. Peter SADLER, I thank you for givingme the chance to work with and learn from you and your group! Foremost, Ithank Ruth MCQUITTY and Nicky FARRER for setting up the "PCF-MS" with meand spending so many hours in the dark little maXis-lab. By the way: Thanksalso to the mass spec staff, Lijiang SONG and Philip ASTON, for their support,patience and understanding. Moreover, I am grateful to Abraha HABTEMARIAM

for his advice and nicely behaving Ru-compound and to Yao ZHAO for advise onthe Pt-compound. I would also like to mention the rest of the SADLER-group fortheir extreme friendliness and the welcoming atmosphere during my stays. Itwas my pleasure!

In addition let me thank the rest of the KÖRBER-project consortium for inspi-ring discussions at the meetings; especially Anita JONES and Gareth WILLIAMS

for mutual exchange of knowledge, experience and equipment concerning ourclosely-related project topics.

Und wo ich gerade all meinen eher chemisch-bewanderten Kollegen danke, sodarf ich auch die nachfolgenden nicht vergessen: Dank an Matthias SCHMIDT

(Chemische Reaktiontechnik, FAU) für die (für mich spontan zustande gekom-mene) Kollaboration zum Thema "Katalyse in PCFs", und an Christopher SY-NATSCHKE (Makromolekulare Chemie II, Uni Bayreuth) für die Synthese der Hy-drogelpartikel (die wiederum für ihn recht "spontan zustande gekommen" gewe-sen sein dürfte)!

143

My friendsAuch außerhalb meiner Gruppe habe ich am MPL sehr viele tolle Menschen ken-nen gelernt, die mir unglaublich viel Freude bereitet haben – sei es beim Kaf-feetrinken, Sport oder Musik machen etc. pp. – Ich werde Euch vermissen! EinDankesgruß geht raus an: Christoffer WITTMANN (mit dem alles angefangenhat), Aaron WEBSTER (ich hoffe, dass für Dich alles bei "die Üblich" bleibt ;-) ),Vincent SCHULTHEISS, Andreas "And" SCHREIBER (für Dein allzeit offenes Ohrund dafür, dass ich persönlich jemanden mit einem Science-Paper kenne), To-bias "Rötzi" RÖTHLINGSHÖFER (für Deine Fürsorge, speziell was meine Koffein-versorgung anbelangt), Christian "Peunti" PEUNTINGER, Christian MÜLLER, JosipMILANOVIC (die Legende), Jan KORGER, Imran "Imi" KHAN (E–A–D–G!), Bettina"Betty" HEIM, Andrea GOLLA (... zuständig für alles was Spaß macht: Haare-färben, Schlemmen, Sauna, Quatschen...), Josef FÜRST (der auch bei Nachtauf unwegsamem und unbekanntem Terrain den richtigen Weg findet), Michael"Michel(angelo)" FÖRTSCH (für erheiternde Stunden), Andreas ECKSTEIN, Tho-mas "Tom" BAUER, Peter BANZER (die Welt kann ab und zu auch ohne Dich ;-) ),Malte AVENHAUS.

My familyMeiner Familie möchte ich dafür danken, dass sie mich zu dem freidenkendenMenschen hat werden lassen, der ich bin. Ihr habt mir nie neigschwätzt, sondernmich meine Entscheidungen selbst treffen lassen und mich darin immer unter-stützt (finanziell, seelisch, moralisch). Dafür bin ich Euch unendlich dankbar.Ich hab Euch lieb!

Auch Du Christoph gehörst für mich mittlerweile zu diesem erlauchten Kreise.In Worten kann ich Dir beim besten Willen aber nur für zwei Dinge danken. Er-stens für das unermüdliche Korrekturlesen, das Du wie immer fabelhaft gemachthast – und diese Dissertation um Welten besser, nicht nur des Englisches wegen.Und zweitens dafür, dass Du mich daran erinnert hast, dass "24/7" 24 Tage imMonat bei 7 Stunden am Tag bedeutet. ;-) Ansonsten gilt, Du ahnst es schon,"I have found beyond all doubt, we say more by saying nothing at all." ;-*

Funding and Support

Curriculum VitaeSarah Unterkofler Dipl.-Phys. Univ.

born 22nd September 1984in Schwäbisch Gmünd

2010 – 2013 Doctorate in Physics (Biophotonics)at the Friedrich-Alexander-University Erlangen-Nuremberg,the Max Planck Institute for the Science of Light andthe International Max Planck Research School ’Physics of Light’ in Erlangen

• Thesis title: Optofluidic Photonic Crystal Fibres for Biomedical Researchin fibra

• Advisor: Prof. Philip St.J. RUSSELL (Photonic Crystal Fibre Division)

2008 – 2012 Postgraduate Study Programme ’Macromolecular Science’at the University of Bayreuth,within the Elite Network of Bavaria (ENB)

• Certificate

• Grade: very good

2004 – 2009 Diploma in Physics (Biophysics)at the University of Bayreuth

• Degree: Dipl.-Phys. Univ.

• Grade: with distinction

• Thesis title: Room and Low Temperature Single-Molecule Studies ofLight-Driven Processes in Pigment-Protein Complexes

• Advisor: Prof. Jürgen KÖHLER (Experimetal Physics IV & Bayreuth In-stitute of Macromolecular Research)

2007 – 2008 Studies Abroad in Engineering Physics (Biophysics & Medical Engineering)at the Royal Institute of Technology, Stockholm (Sweden)

• 1-year scholarship: ’Excellence Programme’ of the German AcademicExchange Service (DAAD)

• Lab study: Bacteriorhodopsin Reconstituted into Large UnilamellarVesicles – Protonation Studies based on Fluorescence Correlation Spec-troscopy

• Advisor: Prof. Jerker WIDENGREN (Biomolecular Physics)

optofluidic photonic crystal fibres for biomedical research in fibra

Documents