Nanochemistry and Sensing in Photonic Crystal Fibers

Photochemie und Spektroskopie im Nanoliter-Bereich in Photonischen Kristallfasern

Der Naturwissenschaftlichen Fakultät der Friedrich-Alexander-UniversitätErlangen-Nürnberg zur Erlangung des Doktorgrades Dr. rer. nat.

vorgelegt von Jocelyn Ssu-Yin Chen aus Taichung

Als Dissertation genehmigt von der Naturwissenschaftlichen Fakultät der Friedrich-Alexander UniversitätErlangen-Nürnberg

Tag der m¨undlichen Pr¨ufung:23 November 2010

Vorsitzender der Promotionskommission: Prof. Dr. Rainer Fink Erstberichterstatter: Prof. Dr. Philip St.J. Russell Zweitberichterstatter: Dr. Clemens F. Kaminski

For my family.

Zusammenfassung

Diese Arbeit handelt von Anwendungsmöglichkeiten photonischer Kristallfasern (PCF) im Bereich der Photochemie und Spektroskopie im Nanoliter-Bereich. Photonische Kristall- fasern haben die Fähigkeit, eine bestimmte Lichtmode über sehr große Distanzen in einem sehr kleinen Probe-Volumen zu führen.Diese einzigartige Eigenschaft photonischer Kristallfasern erlaubt eine drastische Steigerung der erzielbaren Licht-Materie-Wechsel- wirkung und ist Grundlage dieser Arbeit. Die Parameter, von denen optimale Bedingun- gen fürNachweisreaktionen sowie die Ausbeute photochemischer Reaktoren abhängen, werden diskutiert. Außerdem werden verschiedene Verfahren zur Verwendung von PCF- Sensoren in mikrofluidischen Systemen untersucht. Weiterhin wird ein hochgradig kon- trollierbares photochemisches Reaktionsgefäßvorgestellt. Als prinzipieller Beweis seiner Anwendbarkeit zur aktiven Herbeiführungund Beobachtung photochemischer Reaktionen wird die Photolyse wässrigerCobalaminlösungim Kern einer PCF quantitativ gemessen. Wegen der - fürdiese Reaktion typischen - mäßigenQuantenausbeute wäredies mit konventionellen Methoden schwierig oder unmöglich. Die dynamischen Vorgänge während der aktiv herbeigeführten Reaktion konnten mittels Breitband-Absorptionsspektroskopie in der Faser in Echtzeit aufgezeichnet werden. Die Ergebnisse wurden mit denen einer herkömmlichen Küvettenmessung verglichen. Durch das verwendete Reaktionsgefäßkon- nte das benötigteProbevolumen gegenüber konventionellen Techniken stark verkleinert werden (in die Größenordnung von nL/cm). Die starke Licht-Materie-Wechselwirkung in den mikrostrukturierten Fasern ermöglicht es, bei sehr niedrigen Lichtleistungen, kürzere Reaktionszeiten zu erreichen. Weiterhin konnte das schnelle und reversible photoin- duzierte Schalten eines Azobenzol-Derivats nachgewiesen und dadurch die Effizienz und Reproduzierbarkeit des Reaktors bestätigtwerden. Neben dem photochemischen Reaktor wurde ein quantitativer breitbandiger Fasersensor entwickelt, basierend auf der Uberlap-¨

v vi ZUSAMMENFASSUNG pung evaneszenter Felder in den Mantellöchern einer Vollkernfaser. Dabei wurde, trotz des wesentlich verringerten Probevolumens, hervorragende übereinstimmung mit dem unter Verwendung einer gewöhnlichen Küvette erhaltenen Referenzspektrum festgestellt. Zuletzt bieten PCF, neben größererLicht-Materie-Wechselwirkung, auch ein großes Ober- flächen-Volumen-Verhältnis(∼ 105 m−1) fürAnwendungen, in denen Reaktionen mit Oberflächengebundenen Probentypen von Interesse sind. Zu diesem Zweck wurden die Selbstaggregation und das Photobleichen eines Thiazin-Farbstoffs in einer Index-leitenden Faser mit “Mercedesstern”-Querschnitt untersucht. Durch Absorptionsspektroskopie an- hand der evaneszenten Welle, die von der im Kern geleiteten Lichtmode in die Man- tellöcher der Faser ausstrahlt, konnte die Anzahl der, an der Oberfläche der Faseradsor- bierten, Moleküleermitteln werden. Abstract

The work described in this thesis demonstrates the application of photonic crystal fibers in nanochemistry and sensing. In the photonic crystal fiber, a well-defined optical mode can propagate through a sample volume confined within the fiber’s microstructure over very long distances. This property, unique to the photonic crystal fiber, offers greatly enhanced figure of merit for light-matter interactions, and is the basis of this thesis. The parameters governing the optimum sensing conditions and the figure of merit for photochemical reactors are discussed and several fabrication techniques with the objective of combining photonic crystal fiber sensors with microfluidics have also been investigated. A highly-controlled photochemical reactor was proposed and demonstrated. As a proof-of- principle for its application in actively inducing and monitoring photochemical reactions, the photolysis of an aqueous cobalamin was quantitatively measured in a liquid-filled hollow-core photonic crystal fiber. The reaction is characterized by modest quantum yields which would otherwise be difficult or impossible to induce and monitor using conventional methods. The dynamics of the actively induced reaction were monitored in real-time by broadband absorption spectroscopy in the fiber. Results were compared to those obtained using standard techniques in a cuvette. The reactor has greatly reduced the sample volume requirement (in the order of nL/cm) compared to conventional techniques. The strong light-matter interactions in the fiber microstructure allowed shorter reaction times to be achieved at very low optical powers. Additionally, the fast and reversible photoswitching of an azobenzene derivative was demonstrated and confirmed the effectiveness and reproducibility of the photochemical reactor. In addition to the photochemical reactors, a quantitative broadband fiber sensor based on evanescent-field sensing in the cladding holes of a suspended solid-core fiber was demonstrated. Excellent agreement with the reference spectrum measured in a standard cuvette was obtained despite

vii viii ABSTRACT the much reduced sample volume used. Finally, in addition to enhancement in light- matter interactions, the photonic crystal fiber also offers large surface-to-volume ratios (∼ 105 m−1) for experiments in which reactions of surface-bound sample species are of interest. To this end the self-aggregation and photobleaching of a thiazine dye was studied in an index-guiding fiber with suspended solid core. It was shown that the amount of molecules adsorbed onto the inner surfaces of the fiber could be obtained and monitored by absorption spectroscopy via the evanescent wave of the core-guided mode that extends into the cladding holes of the fiber. Acknowledgments

This thesis would not have been possible without the scientiﬁc, technical and friendly support from the following people:

Alexander Nazarkin Greg Pearce Myeong Soo Kang Alexander Podlipensky Gustavo Wiederhecker Nicola Farrer Amir Abdolvand Helga Hussy Nicolai Granzow Amy Wan Hemant Tyagi Nicolas Joly Andre Brenn Howard Lee Patrick Uebel Andreas Walser Jerry Chen Pavel Marchenko Aniruddha Ray Johannes Nold Peter Banzer Anita Jones Konrad Heberlein Peter Sadler Anna Butsch Lam Lee Philip Russell Bastian Etzold Leonhard Heberlein Philipp Hoelzer Bernhard Thomann Leyun Zang Ralf Keding Bettina Schwender Luis Lorenzo Sanchez Soto Robert Fisher Chris Poulton Luis Prill Sempere Robert Gall Christine Kreuzer Marianne Heberlein Sarah Unterkoﬂer Christoph Heberlein Markus Schmidt Sebastian Stark Daniel Ploß Marta Ziemienczuk Silke Rammler Friedrich Heberlein Martin Butryn Stanislaw D¨orschner Gareth Williams Martin Garbos Thomas Spona George Kakarantzas Matthias Schmidt Tijmen Euser Gordon Wong Michael Scharrer Xin Jiang

Cheers, guys!

Contents

Zusammenfassung v

Abstract vii

Acknowledgments ix

List of Figures xv

List of Tables xix

Abbreviations xxi

Preface xxiii

1 Photonic Crystal Fibers 1 1.1 Introduction ...... 1 1.2 Historical Overview ...... 2 1.3 Classiﬁcation and Guidance Mechanisms ...... 4 1.3.1 Index-Guiding PCF ...... 5 1.3.2 Hollow-Core PCF ...... 7 1.4 Fabrication ...... 9 1.5 Optical Sensing with PCF ...... 10 1.5.1 Index-Guiding PCF Sensors ...... 11 1.5.2 Hollow-Core PCF Sensors ...... 13

2 Experimental Considerations and Techniques 15 2.1 Introduction ...... 15

xi xii CONTENTS

2.2 Detection Strategies and Ideal Conditions ...... 17 2.2.1 Ideal Conditions for Absorption-Based Sensors ...... 17 2.2.2 Figure of Merit for Photochemistry ...... 20 2.3 Experimental Setup and Instrumentation ...... 22 2.3.1 Transmission Properties of Liquid-Filled PCF ...... 22 2.3.2 Microﬂuidic Flow in Conﬁned Channels ...... 25 2.3.3 Optical Setup ...... 28 2.3.4 LabVIEW Automation ...... 32 2.4 Fabrication Techniques for PCF Devices ...... 32 2.4.1 Femtosecond Laser Ablation ...... 32 2.4.2 Two-Photon Polymerization ...... 38 2.4.3 Focused Ion Beam Micromachining ...... 40

3 Photochemistry in PCF 43 3.1 Introduction ...... 43 3.2 Fiber Characteristics ...... 44 3.3 Photolysis of Metal Complexes ...... 49 3.3.1 Photoaquation of Cyanocobalamin ...... 50 3.3.2 Experimental Results ...... 51 3.3.3 Reaction Kinetics ...... 53 3.3.4 Discussion ...... 57 3.4 Photoswitching of Azobenzene Molecules ...... 57 3.4.1 Isomerization of Azo Dyes ...... 58 3.4.2 Reversible Isomerization in PCF ...... 62 3.4.3 Reaction Kinetics ...... 66 3.4.4 Discussion ...... 67

4 Spectroscopy in PCF 69 4.1 Introduction ...... 69 4.2 Evanescent-Wave Sensing ...... 70 4.2.1 Fiber Characteristics ...... 70 4.2.2 Results ...... 77 CONTENTS xiii

4.2.3 Discussion ...... 79 4.3 Microscale Surface Chemistry ...... 79 4.3.1 Self-Aggregation and Photobleaching of Methylene Blue ...... 79 4.3.2 Discussion ...... 87

5 Conclusions and Outlook 89 5.1 Optical Tweezers and Photodynamic Therapy ...... 89 5.2 Microﬂuidic Flow Reactor ...... 90 5.3 Mass Spectrometry ...... 90 5.4 Surface Chemistry Using Higher-Order Modes ...... 90 5.5 Final Remarks ...... 91

A Counter-Propagating Pump-Probe Setup 93

List of Publications 97

Curriculum Vitae 121

List of Figures

1.1 Images showing the iridescence in the butterfly Morpho rhetenor and the iridescent setae from polychaete worms...... 3 1.2 Schematic illustration of the cross-section and the refractive index profile for an index-guiding photonic crystal fiber...... 5 1.3 Schematic illustrations of a hollow-core PBG-PCF, a kagome-lattice PCF and a Bragg fiber...... 8 1.4 Images of the cross-section of the cane for a hollow-core PBG-PCF and the fabricated fiber...... 10 1.5 Schematic illustration of the active sensing regions around the core of an index-guiding PCF and a hollow-core PBG-PCF...... 11

2.1 Operational principles of optical sensors in the non-resonant and resonant regimes...... 16 2.2 Ideal sensing parameter diagram for constant absorbance, defining regions in which optimum sensing conditions can be achieved...... 19 2.3 Schematics illustrating and comparing the geometries and sample volumes in a conventional cuvette and a kagome PCF...... 21 2.4 Shift in the central wavelength of the PBG as a result of infiltrating the PBG-PCF with various solvents...... 25 2.5 Cross-section of a capillary tube infiltrated with liquid...... 26 2.6 Simulated water filling time for silica microchannels of bore radii 1, 5 and 10 µm, with an applied pressure head of 1 bar...... 28 2.7 Schematic diagram showing the experimental setup for sensing and photochemistry experiments in PCF...... 29

xv xvi LIST OF FIGURES

2.8 Schematic diagram illustrating the increase in the effective N.A. as a result of change in the interface medium of the objective (air) to that for the fiber (liquid)...... 31 2.9 Schematic diagram of photoionization regimes at low and high frequencies. 33 2.10 Schematic diagram of avalanche ionization...... 34 2.11 Schematic diagram showing the experimental setup for femtosecond laser ablation of side channels in PCF and the two-photon polymerization technique for selective blockage of microstructure holes...... 35 2.12 Diameter of ablated entry hole in the silica fiber as a function of pulse energy incident on the fiber and the number of pulses...... 36 2.13 Schematic showing the dependence of the diameter of laser-ablated entry hole size on the peak irradiance, assuming a Gaussian irradiance distribution. 37 2.14 Examples of ablated side microchannel allowing access to one of the three cladding holes in a suspended-core fiber and a damaged side microchannel after applying 30 bar of water pressure...... 38 2.15 Measured transmission losses as a function of the number of drilled side channels in an ESM-PCF...... 38 2.16 Images of a HC-PCF infiltrated with acrylic resin, a SC-PCF with selectively photopolymerized cladding holes, and a gold nanowire embedded into the cladding hole of a PCF using TPP as the hole-collapsing technique. 39 2.17 SEM of the cross-section of the nanoweb fiber prior to FIB milling and after a hole was milled through the silica jacket of a nanoweb fiber. . . . . 41

3.1 Images showing the cross-section of a kagome HC-PCF...... 45 3.2 Transmission and loss spectra of the kagome HC-PCF...... 46 3.3 Transmission and loss spectra of the kagome HC-PCF ﬁlled with de-ionized water...... 47 3.4 Transmission spectrum of the index-guiding kagome HC-PCF ﬁlled with toluene...... 48

+ 3.5 The photochemical conversion of CNCbl to [H2OCbl] ...... 50 LIST OF FIGURES xvii

3.6 Changes in the absorption spectrum as a result of the photochemical con-

+ version of CNCbl to [H2OCbl] ...... 51 3.7 Spectral and temporal data for the photolysis of CNCbl in a kagome HC-PCF. 53 3.8 Comparison of the temporal evolution of molar absorptivity measured in a kagome HC-PCF and a cuvette...... 54 3.9 Conﬁguration diagram depicting the photoaquation of CNCbl...... 55 3.10 Quantum yields for the photolysis of CNCbl obtained from measurements in a kagome HC-PCF...... 56 3.11 Reversible isomerization between the trans (left) and the cis (right) geometric isomers of azobenzene...... 58 3.12 Spectral and temporal data for the thermal back reaction of disperse orange 1 in toluene...... 59 3.13 Temporal evolution of molar absorptivity for the forward and back reaction of disperse red 1 in cyclohexane...... 61 3.14 Spectral and temporal data for the forward and back isomerization of disperse orange 1 in toluene measured in a kagome HC-PCF...... 62 3.15 Temporal evolution of trans-DO1 in toluene irradiated with broadband xenon lamp...... 65 3.16 Conﬁguration diagram depicting the isomerization paths of trans )* cis... 66

4.1 High resolution SEM of the core region of four different air-suspended solid- core fibers...... 71 4.2 Transmission and loss spectra for air-suspended SC-PCF with air- and water-cladding...... 72 4.3 Normalized mode profiles of an air-suspended SC-PCF with water-filled cladding...... 73 4.4 Dependence of calculated cladding power fraction on the effective core diameter, wavelength and cladding medium...... 75 4.5 Measured and calculated dispersion of air-suspended SC-PCFs...... 76

4.6 Absorption and molar absorptivity spectra of an aqueous NiCl2 solution. . 78 xviii LIST OF FIGURES

4.7 Molar absorptivity spectra of methylene blue in water, and photobleaching of MB in suspended solid-core ﬁber induced by irradiation using the broadband PCF SC source...... 81 4.8 The calculated total surface density of MB along the inner surface of the air-suspended SC-PCF cladding holes...... 83 4.9 Photobleaching and surface adsorption of MB in kagome HC-PCF...... 85

A.1 Schematic diagram of the modiﬁed pump-probe setup with counter-propagating beams for PCF photochemical reactors...... 94 A.2 Schematic diagram showing the eﬀect of refraction due to the tilted liquid cell window...... 95 List of Tables

1.1 Overview of photonic crystal ﬁber development...... 4

2.1 Comparison between various sample cell conﬁgurations...... 23

3.1 Quantum yields for the photolysis of CNCbl at various pH values...... 56

xix

Abbreviations

CCD Charge coupled device CNCbl Cyanocobalamin CW Continuous wave DBI Dimethylbenzimidazole DO1 Disperse orange 1 DR1 Disperse red 1 FC Franck Condon FIB Focused ion beam HC-PCF Hollow-core photonic crystal ﬁber HPLC High performance liquid chromatography

+ [H2OCbl] Hydroxocobalamin, aquacobalamin LMB leuco methylene blue MB Methylene blue MMF Multimode fiber OSA Optical spectrum analyzer PBG Photonic bandgap PCF Photonic crystal fiber SC Supercontinuum SC-PCF Solid-core photonic crystal fiber SEM Scanning electron micrograph TS Transition state UV Ultraviolet ZDW Zero dispersion wavelength

xxi

Preface

From artificial eyes [1] and ears [2] to electronic noses [3] and tongues [4], devices with sensing capabilities mimicking the organs of the most sophisticated living creatures on Earth are paving the ways to improve the living quality of many and simplify routine analyses in industries as well as our daily lives. Such “senses of electronics”, or sensors, are able not only to “feel” the materials but even to distinguish their chemical composition beyond the range of human perception. A substantial amount of effort has been made in the research and development of chemical sensors in recent years [5], incorporating research fields such as electronics, optics, biochemistry, material science, analytical, inorganic and organic chemistry. With technological applications to industrial, medical and environmental needs, this continuously evolving field is pushing for better sensor designs featuring selectivity, low detection limits, reversibility, robustness and portability. However, most of the existing configurations still exhibit clear limitations. Among the world of sensors, a host of sensing modalities exist and are currently being investigated. In particular, sensing devices using optics and photonics have undergone extensive research during the last two decades [6-8] due to the wide variety of optical phenomena that one can exploit as sensing mechanisms. Luminescence, fluorescence, phosphorescence, absorbance, elastic scattering, Raman scattering, surface plasmon resonance, guided-wave resonance, interference, and reflection/transmission microscopy ex- emplify such phenomena. Different detection techniques and setup configurations are constantly being developed and optimized to increase the detection sensitivity. An example is cavity enhanced absorption spectroscopy (CEAS), whereby the probe light makes multiple passes through the same sample, effectively increasing the absorption path by orders of magnitude. While CEAS has proven to be effective in measuring trace samples

xxiii xxiv PREFACE

[9], its intrinsically narrow bandwidth and the requirement for calibration measurements pose limitations on the range of applications for this technique [10]. Furthermore, the conventional monitoring methods based on free-space interferometry and spectroscopy are effective only for the line of sight, and are therefore prone to undesirable misalign- ments and external perturbations. Over the last years, increasing research efforts in fiber- and integrated-optics technologies, which were primarily developed for the telecommunication industry, have been injected into optical sensors research. With advances in the development of high quality fiber-optic components at reasonable costs, the prospect of fiber-optic sensors to replace conventional ones has been realized. Unlike standard communication fibers which act as passive media for signals, the function of a sensing fiber is to produce sensitive responses to various chemical and physical changes that take place in the vicinity of the fiber. Such novel sensing devices are used for routine analyses, with applications in chemical [11-14], biochemical [15-18], biomedical and environmental [18-21] sensing. These fiber-optic chemical and biosensors have shown the potential of a promising technological platform characterized by numerous intrinsic advantages over their conventional electronic counterparts. The principal single attractive feature of fiber-optic sensors is undoubtedly the intrinsic immunity to electromagnetic interferences and the absence of electrical risks which is important for safety in explosive environments. Optical fibers are capable of guiding the light beam in a confined and inaccessible medium over large distances, allowing for more versatile and less perturbed in situ or remote monitoring of environmental or medical parameters. In addition, they also offer the capability for long- range distributed sensing and the ability to be multiplexed, as optical waves of different frequencies do not interfere with one another. The optical fiber is lightweight and its compact geometry implies small volume of analyte consumption, such that low-cost measurements can be achieved with high specificity and sensitivity; its great flexibility also offers the ability to be embedded into various structures and materials, including textiles and fabrics [22]. Fiber-optic sensors have major advantages in many chemically aggres- sive and ionizing environments, and can withstand large physical strain and substantial temperature excursions. They also have the potential to be integrated in rapid, real-time high-throughput analysis and be easily interfaced with optical data communication sys- xxv tems whereby high information density can be achieved. Furthermore, various system configurations demonstrate the accessibility of the fiber sensors for flow cells or pipetting devices, facilitating measurement in the presence of the sample without any rinsing. Emerging technologies in the field of waveguide-based chemical and biosensors are continuously evolving, with new focuses on higher sensitivity and stability. This key demand is stimulating further advancement in the exploration of new material and structural concepts as alternative platforms for standard sensing technologies, which provide better performance with the prospect for novel devices. This thesis focuses on the demonstration of one such novelty in which the concept of optical fiber sensing is further developed to accommodate active in-fiber (photo)chemical reactor employing absorption spectrometry in photonic crystal fibers (PCFs) [23, 24]. These microstructured fibers have revolutionized optical fiber technology by enabling light to be guided and manipulated within the fiber in ways not previously possible; the new degrees of freedom in the fiber design and fabrication have been extensively exploited to considerably improve the sensor performance in terms of accuracy and precision.

Outline of the Thesis

Chapter 1 provides a historical perspective on the progress in the field of photonic crystal fibers, and discusses the classification of PCFs based on their waveguiding mechanisms. Particular attention is paid to the strong light-matter interaction within the fiber’s microstructure because of its importance in the photochemical reactions and sensing experiments performed in this thesis. A short overview of the novel sensing applications based on various types of PCFs and sensing mechanisms to date is also presented. Chapter 2 introduces the various modes of operation for optical snesors and presents the concept of detection sensitivity for a fiber-optic sensor. The parameters dictating the effectiveness of PCF-based sensors are discussed and a diagram relating these parameters is constructed to facilitate the determination of optimum sensing conditions. In addition to passive sensing in PCF, the strong light-matter interaction provided by the PCF can also be utilized as a photochemical reactor to simultaneously induce and monitor reaction kinetics. The figure of merit for photochemical reactors is therefore discussed in detail. xxvi PREFACE

The transmission properties and microfluidic flow through the liquid-filled PCF, together with the main components making up the experimental setup, are also described. The chapter concludes with a presentation of the preliminary results from several fabrication techniques investigated with the objective of combining PCF sensors with microfluidics. Chapter 3 demonstrates the use of a liquid-filled hollow-core PCF (HC-PCF) as a highly-controlled photochemical reactor. Photochemical reactions with very low quantum yields are efficiently induced and monitored within the hollow core of the PCF reactor in real-time. Quantitative absorption spectroscopy of the photo-excited chemical species can be obtained within seconds. Orders of magnitude enhancement in the reaction kinetics is obtained with strongly reduced sample volumes compared to conventional techniques. The second part of the chapter demonstrates the effectiveness of the PCF reactor in monitoring fast, reversible photochemical reactions. The reactions are monitored in real- time and show complete reversibility of the system. Chapter 4 presents quantitative broadband sensing based on the evanescent interaction of the guided core mode in a solid-core PCF (SC-PCF) with the analyte in the fiber cladding. Excellent agreements are obtained between the experimentally measured and numerically calculated fiber sensor characteristics without the use of free parameters. More importantly, the PCF sensor provides stronger signals and excellent agreement with the reference spectrum measured using standard techniques, despite using three orders of magnitude lower sample volume. The second part of the chapter focuses on the interaction between the sample molecules with the inner surfaces of the PCF microchannels. Results from both SC- and HC-PCFs demonstrate the affinity of the molecules to adsorb onto the silica surfaces and self-aggregate, an effect which would otherwise be unobservable in bulk under the same experimental conditions. The high surface-to-volume ratio provided by the microchannels in PCF promises a novel platform for surface chemistry. Chapter 5 concludes this thesis with a discussion on the prospective extensions to the current work. Chapter 1

Photonic Crystal Fibers

1.1 Introduction

Over the past four decades, optical fibers have revolutionized the field of telecommuni- cations [25]. However, despite their excellent performance in the transmission of optical signals, the advances made in fiber optic technology have been driven towards its ultimate limit as the intrinsic properties of silica have imposed fundamental restrictions on the performance of conventional optical fibers. Firstly, standard optical fibers have strict design rules to fulfill, such as limited core diameter for single-mode operation, modal cut-off wavelength, and limitation on material selection as the core and cladding materials must have matching thermal properties. Secondly, restrictions on the geometry and refractive index profile of optical fibers hinder the flexibility in engineering fiber properties such as dispersion, nonlinearity and birefringence for better performance and more specialized applications. Finally, light propagating in an optical fiber suffers from losses. Several factors contribute to the attenuation in optical fibers, with material absorption and Rayleigh scattering being the major contributing sources. Material absorption arises from electronic and vibrational resonances of silica glass or impurities such as the OH− ions in the fiber. Rayleigh scattering refers to the scattering of light from local fluctuations in the refractive index introduced by inhomogeneities in the fiber that are on a scale much smaller than the optical wavelength. These limitations stimulated the development of a new class of optical waveguides known as the photonic crystal fiber (PCF). In a PCF the fiber core is surrounded by a

1 2 CHAPTER 1. PHOTONIC CRYSTAL FIBERS microstructured cladding based on a two-dimensional periodic lattice consisting of air- filled capillaries running parallel to the fiber axis along the entire length of the fiber [26, 27]. Unlike conventional optical fibers, the PCF is usually made of a single material (silica), eliminating the need for two thermally, chemically and optically compatible glasses to form its core and cladding. In addition, the PCF has several structural parameters one can manipulate, offering great design flexibility and a wealth of physical properties for many interesting applications. For example, the PCF allows highly-engineerable refractive index profiles to develop fibers of desirable nonlinearity, birefringence and chromatic dispersion. The possibility for light to be guided in a hollow core also implies that it is possible to fabricate fibers with losses lower than that achievable in conventional fibers at wavelengths limited by high material absorption. In this chapter, a brief historical overview of the development of PCF is given in Section 1.2, followed by the classification and guidance mechanisms of various classes of PCF in Section 1.3. A description of the general fabrication process is outlined in Section 1.4. The chapter concludes with an overview of optical sensing applications in PCF in Section 1.5.

1.2 Historical Overview

Like many scientific inventions which are either inspired by or akin to living beings found in nature, the photonic crystals of PCF also find similarities in the photonic stopband- based nanostructures in the wings of butterflies (Figure 1.1(a)) and the iridescent setae from polychaete worms (Figure 1.1(b)) [28]. The periodic variations in dielectric constant in PCF also find analogy in semiconductor band structures in which electrons interact with the periodic variation in the potential created by the atomic crystal lattice. The idea of using periodic, one-dimensional variations in the dielectric constant to trap light was first presented by Melekhin and Manenkov in 1968 [29] and Yeh et al. in 1978 [30], where it was proposed to clad a fiber core with a multilayer coating similar to that in planar Bragg stacks. Light guidance in the Bragg fiber is therefore the result of a radial stopband. The work by Yablonovitch [31, 32] predicted that certain three-dimensional periodic dielectric structures can have a frequency band in which all propagation modes 1.2. HISTORICAL OVERVIEW 3

(a) (b)

Figure 1.1: (a) Iridescence in the butterfly Morpho rhetenor and transmission electron micrograph (TEM) images showing wing-scale cross-sections. Scale bars: 1.8 µm, 1.3 µm. (b) Iridescent setae from polychaete worms: scanning electron micrograph (SEM) and TEM images of transverse sections through a single iridescent seta. Scale bars: 2 µm, 5 µm, 1 µm, 120 nm [28]. are forbidden, termed the photonic bandgap (PBG). Within the forbidden bandgap range, light can only exist and propagate along defects [33]. This mechanism completely inhibits spontaneous emission (photonic states) in the lattice by having a three-dimensional photonic crystal. In comparison, the Bragg fiber still has photonic states in the cladding and its guidance mechanism can therefore not be classified as that due to the PBG effect. Similar to the three-dimensional photonic crystals, defects can be introduced into photonic crystal slabs consisting of two-dimensional periodicities to form waveguides. The guidance mechanism is based on the photonic bandgap effect in the plane of periodicity, while light in the direction perpendicular to this plane is confined via the index guidance mechanism. In comparison, guided propagation of the electromagnetic field in the PCF is also achieved by the introduction of defects in the two-dimensional microstructured morphology of the fiber. However, unlike the photonic crystal slab, light enters the PCF waveguide normal to the plane of the periodicity. The difference between the photonic crystal slab and the PCF can be understood via a simplified picture involving Bragg’s law for constructive interference:

mλ = 2Λ cos θ, (1.1) 4 CHAPTER 1. PHOTONIC CRYSTAL FIBERS

Year PCF Development 1996 First solid-core PCF [26, 27] 1997 Endlessly single-mode PCF [34] 1998 Ultra-large mode area [35] 1999 PCF with photonic bandgap and air core [36] 2000 Supercontinuum generation with PCF [37] 2001 Four-wave mixing [38] 2002 Laser-tweezer guidance of particles in HC-PCF [39] 2003 Tellurite glass PCF [40] 2005 All-solid photonic bandgap ﬁber at 1% index contrast [41]

Table 1.1: Overview of photonic crystal fiber development. where m is an integer, λ is the wavelength of the incoming light, Λ is the pitch of periodicity and θ is the incident angle the incoming wave. It follows that for θ = 0 (as in the case of the photonic crystal slab), Λ is of the order of λ; while for grazing incidence (as in the case of the PCF), Λ is much larger than the wavelength of the incoming light. The small pitch requirement of the photonic crystal slab waveguides leads to relatively high losses and therefore makes the PCF the superior waveguide. The first PCF reported in 1996 was an index-guiding PCF and utilized a two-dimensional photonic crystal where the structure is periodic in the plane perpendicular to the fiber axis but invariant along the fiber length [26, 27]. A short overview of PCF development is presented in Table 1.1.

1.3 Classiﬁcation and Guidance Mechanisms

Light guidance in the conventional fiber is based on the slight refractive index difference between the two concentric regions of core and cladding with different doping levels. Photonic crystal fibers, however, can be categorized into different classes, depending on whether the mechanism of optical confinement is based on index guiding or photonic bandgap effects, and whether the periodicity of the structure is one-dimensional or two- dimensional. 1.3. CLASSIFICATION AND GUIDANCE MECHANISMS 5

jacket d L

core Index

Radial distance L cladding n air n n core eff (a) (b)

Figure 1.2: Schematic illustration of (a) the cross-section and (b) the refractive index proﬁle for an index-guiding photonic crystal ﬁber.

1.3.1 Index-Guiding PCF

Index-guiding PCF represents the simplest type of PCF, with its basic cross-sectional structure being that of a solid core surrounded by a two-dimensional photonic crystal consisting of a periodic array of air holes arranged in a hexagonal pattern on a silica background, extending invariantly along the length of the fiber, as illustrated in Figure 1.2(a). In this case the two-dimensional photonic crystal is not utilized for its bandgap, but rather to form a fiber cladding of lower effective index given that the solid core is made up of the same material as the photonic crystal background. Figure 1.2(b) shows a schematic demonstrating the subtle variations in the fiber’s refractive index profile. As a result, light guidance is based on modified total internal reflection, akin to that in a conventional fiber. Due to the range of structures and air-filling fractions one can realize in the microstructured photonic crystal cladding, the PCF offers a number of unique properties that are not attainable in conventional fibers. For example, the index-guiding PCF can be fabricated to exhibit endlessly single-mode behavior. Here, the single lobe of the fundamental mode with a diameter roughly equal to 2Λ is trapped in the core of the index-guiding PCF while the lobes of higher-order modes are smaller and can leak out through the silica gaps between the cladding holes encircling the core. The fiber maintains its endlessly single-mode behavior provided the relative hole size, d/Λ, is small enough; as the air holes 6 CHAPTER 1. PHOTONIC CRYSTAL FIBERS are made larger, successive higher order modes also become trapped in the fiber core. State-of-the-art optical fibers constitute a careful trade-off between optical losses, optical nonlinearity, group velocity dispersion and polarization effects. During the last decade, intense research and fabrication has led to precise control of the PCF characteristics comparable to that of standard fibers. Index-guiding PCF having loss of 0.18 dB/km at 1.55 µm has recently been obtained by reducing the OH− absorption loss and improving the air hole surfaces [42]. By omitting more air holes in the core region of an index-guiding PCF, large mode area single-mode PCF has been fabricated which supports a core diameter of 50 free- space wavelengths [35], a property advantageous for the development of fiber lasers and amplifiers. The birefringence in PCFs can be attributed to either elasto-optical effect induced by the anisotropy of the refractive index in the core due to internal stress, as in conventional PANDA and bowtie fibers, leading to the demonstration of polarization-maintaining PCF with large mode area [43, 44]; or the geometrical asymmetry in the fiber cross-section, as in standard elliptical core fibers. The latter results in fibers with strong form birefringence that are resilient to environmental factors such as temperature, strain and pressure, and can be achieved by using holes with different radii or shape, or by local elongation of the core region. Highly birefringent PCF with birefringence ten times larger than that of conventional fibers has been fabricated [45]. The possibility to engineer the PCF structural parameters such as the cladding air hole size and pitch, and the core diameter, allows one to efficiently manage the fiber chromatic dispersion by changing its waveguide dispersion. Index-guiding PCFs having zero, low or anomalous dispersion at visible wavelengths have been fabricated [37, 45, 46], while ultra-flattened dispersion over a very large wavelength range has been demonstrated by mirroring the PCF waveguide dispersion to the material dispersion [47-50]. The large core-cladding refractive index difference in SC-PCFs can lead to tight modal confinement in the fiber and hence low effective mode area, giving rise to nonlinearities one to two orders of magnitude higher than one can obtain in conventional fibers. This high nonlinearity generally allows reduced interaction length and power requirement for applications based on nonlinear optics, such as four-wave mixing [51, 52], multimode phase 1.3. CLASSIFICATION AND GUIDANCE MECHANISMS 7 matching [53], pulse compression [54] and generation of ultra-broadband supercontinuum (SC) [55]. In addition to strong confinement of the guided mode, PCF nonlinearity can also be enhanced by fabricating fibers from a single material constituent with high intrinsic nonlinearity such as chalcogenide [56], tellurite [57], bismuth silicate [58] and lead silicate [59] glasses.

1.3.2 Hollow-Core PCF

Standard hollow waveguides confine light either by total internal reflection (attenuated total reflection guides) [60, 61] or by reflection off a metallic surface (leaky guides) [62]. These waveguides are inherently weak and highly multimode, allowing the use of only relatively short lengths [63]. In contrast, hollow-core PCFs (HC-PCFs) [36] offer quasi single-mode operation despite supporting multiple optical modes, including guided and surface modes, at any given wavelength, provided careful launching conditions are applied to selectively excite the fundamental mode. Furthermore, higher-order modes usually have much higher confinement and scattering losses compared to the fundamental mode [64], allowing one to effectively achieve single-mode output at the desired wavelength by employing long length of fiber [65] or by bending the fiber. Finally, HC-PCFs facilitate the delivery of light with low attenuation over kilometer length scales, an attribute unachiev- able in conventional hollow waveguides. Losses as low as 1.2 dB/km in HC-PCF has been achieved by enlarging the core from 7 to 19 unit cells to reduce the overlap of the fundamental core mode with the glass-air surface modes [66].

1.3.2.1 Photonic Bandgap PCF

An index-guiding PCF cannot be formed with a hollow core, as total internal reflection requires the effective cladding index to be lower than that of the core index. Light guidance in this case, however, can be realized by coherent Bragg scattering, in which light within finite frequency regions is prohibited from propagating in the photonic crystal cladding and is confined to a defect at the fiber core. Each of these frequency regions corresponds to the existence of a full two-dimensional PBG in the fiber cladding; low-loss guided modes can therefore be formed where a core resonance coincides with a bandgap. These fibers 8 CHAPTER 1. PHOTONIC CRYSTAL FIBERS

(a) (b) (c)

Figure 1.3: Schematic illustrations of (a) a hollow-core PBG-PCF, with a two- dimensionally periodic cladding of air holes, (b) a kagome-lattice PCF, with a periodic cladding structure consisting of fine silica webs forming a kagome lattice and (c) a Bragg fiber, with a one-dimensionally periodic cladding of concentric high and low index layers. are called photonic bandgap PCFs (PBG-PCFs), as depicted in the schematic shown in Figure 1.3(a). The PBG-PCF cladding generally comprises of a honeycomb lattice of air holes and silica struts, with a large air-filling fraction of typically > 80%. In these fibers, losses as low as 1.7 dB/km have been reported [67].

1.3.2.2 Kagome-Lattice PCF

In contrast to guidance via the existence of PBGs in PBG-PCF, another type of HC-PCF has been demonstrated to allow guidance in the air core despite the lack of photonic bandgaps. The cladding microstructure of these fibers consists of an array of thin silica strands that form the kagome lattice, as depicted in Figure 1.3(b). The kagome fiber exhibits much broader optical transmission bandwidth and lower dispersion compared to the PBG-PCF. Several studies have been made towards understanding of the guidance mechanism in the kagome fiber, such as low cladding density of states [68], reduced coupling between the core and cladding mode fields [69, 70] and high-order bandgaps [71]. However complete understanding of the nature of guidance in these fibers is yet to be established. In principle, the kagome fiber is a leaky waveguide in that there are always real photonic states (i.e., propagating fields) in the cladding, consequently Fabry-Pérot- like resonances appear in the cladding. As a result, the leakage rate of the core “mode” depends in a complicated oscillatory manner on the cladding thickness as well as the 1.4. FABRICATION 9 properties of the external medium, making the leaky core mode look more like a Mie resonance than a bound mode [64]. The kagome fibers typically have larger core diameters than the PBG-PCFs, hence allowing them to support several such leaky resonances or “modes”, resulting in higher losses than the PBG-PCFs, with the lowest loss achieved thus far being 0.25 dB/m [72].

1.3.2.3 Bragg Fiber

Instead of employing two-dimensional periodicity in the fiber cladding, a one-dimensional periodicity comprising of alternating multilayer of high and low index glasses (see Figure 1.3(c)) can also be used to confine light within a hollow core, resulting in what is known as Bragg fibers, which were first proposed by Melekhin and Manenkov in 1968 [29] and Yeh et al. in 1978 [30]. Bragg fibers based on omnidirectional mirrors have been demonstrated [73] and utilized for delivery of high power lasers in endoscopic surgeries [74]. Note that although strictly speaking the Bragg fiber cannot be categorized under PCF, it is included here for reference.

1.4 Fabrication

The stack-and-draw technique is the most commonly used process in PCF fabrication. Initially, the stack is manually built on a macroscopic scale using capillaries with a ratio of inner diameter to outer diameter (ID/OD) closely matching the air-filling fraction (d/Λ) of the desired fiber microstructure. The completed stack (typically 1 meter long and a few centimeters in diameter) is then inserted into a jacket tube and drawn into preforms of a few millimeters in diameter. Subsequently, the preform is either drawn directly into fibers using a conventional fiber drawing tower, or drawn into canes before being drawn into fibers in case a large scale reduction factor is required, as shown in Figure 1.4 for a hollow-core PBG-PCF. The newly-drawn optical fibers are then coated with high performance polymers cured by ultraviolet (UV) exposure to improve their mechanical properties. Techniques such as extrusion [59], built-in-casting [75] and drilling [76] for preform production allow fabrication of PCFs using materials with lower melting points. 10 CHAPTER 1. PHOTONIC CRYSTAL FIBERS

(a) (b)

Figure 1.4: (a) Optical micrograph showing the cross-section of the cane for a hollow-core PBG-PCF. (b) SEM showing the cross-section of a hollow-core PBG-PCF. The cane was inserted into a separate silica jacket tube before being drawn into ﬁbers to allow for large scale reduction during drawing. The cladding diameters were 2.6 mm for the cane and 60 µm for the ﬁber.

1.5 Optical Sensing with PCF

The innovation of PCF has proven to be highly valuable for the design of advanced fiber- optic components, enabling new optical phenomena and applications. In the realm of optical fiber sensing, PCF offers a high degree of freedom in design flexibility, facilitating the development of new sensing configurations. Photonic crystal fiber has proven to be effective in enhancing light-matter interactions, offering interaction lengths much longer than those available using conventional techniques, thus dramatically increasing its sensitivity. The possibility for gases and fluids to occupy the holes in the PCF microstructure, thereby utilizing them as a microfluidic channel or gas cell, offers a host of advantages. A well-defined optical mode propagating through the micron-sized sample cell presents a unique approach of monitoring the interaction between the propagating light and the measurand. Furthermore, the micron-sized holes in the PCF microstructure strongly reduce the sample volume required for sensing. The follow sections detail the development of fiber sensors based on index-guiding and HC-PCFs to date. 1.5. OPTICAL SENSING WITH PCF 11

(a) (b)

Figure 1.5: Schematic illustration of the active sensing regions around the core of (a) an index-guiding PCF and (b) a hollow-core PBG-PCF.

1.5.1 Index-Guiding PCF Sensors

Most of the existing optical sensing techniques are based on the evanescent spectroscopic sensor design, whereby the evanescent field associated with the light propagating in the confinement region of the device extends into the region where the analyte to be sensed is located. In the case of optical fiber sensors, this tailing optical field can transfer energy out of the fiber core to the absorbing species in the surrounding medium. Additionally, the evanescent field can also be used to create fluorescence in the surrounding medium, or couple fluorescence into the fiber core. The change in the optical transmission properties of the fiber due to the evanescent absorption of the analyte is then monitored, or “sensed”. This sensor design therefore requires the chemical fingerprint region of the electromagnetic spectrum to lie within the wavelength range of the light guided in the optical fiber core. In order to access the evanescent wave near the boundary of the core and cladding of a conventional fiber, standard evanescent-wave fiber sensors necessitate the complete or partial removal of the fiber cladding by chemical etching [77], precise flame control [78], or polishing [79, 80] to form a D-shape fiber. Alternatively, the evanescent wave of a tapered fiber [81] can also be used to enhance the interaction between the guided light and the sample [82]. 12 CHAPTER 1. PHOTONIC CRYSTAL FIBERS

The sensing mechanism of index-guiding PCF sensors is also based on the evanescent interaction between the guided optical field and the sample, akin to that in the conventional sensors (as shown in Figure 1.5(a)). However, they do not require cumbersome post-processing procedures, since the presence of air-holes in the cladding microstructure allows the accommodation of biological and chemical samples in gaseous or liquid forms in the immediate vicinity of the fiber core. In addition, PCFs naturally integrate optical detection with microfluidics, allowing for continuous on-line monitoring of samples in real-time. The infiltration of sample into the PCF cladding holes also allows the fiber to maintain its original structure, without the need to even remove the polymer coating of the fiber. Consequently the index-guiding PCF provides superior structural robustness compared to the conventional fiber sensors. The evanescent-wave PCF sensor configuration was first theoretically and numerically studied by Monro et al. [83, 84]. In principle, strong light-matter interaction requires a significant modal power overlap with the fiber holes within the wavelength range of the sample absorption spectrum. The power overlap decreases with core size and increases with wavelength, as light of longer wavelength is less tightly confined in the solid core of the index-guiding PCF. Therefore, a larger fraction of the guided mode extends into the cladding holes. The first experimental demonstration of evanescent-wave gas detection with PCF used an index-guiding PCF with a length of 75 cm for the detection of acetylene [85, 86]. The fiber used had a relatively low power overlap (∼ 5.5 % at 1530 nm) of the optical field with the sample; nevertheless the long interaction length provided by the PCF was able to compensate for weak light-matter interaction. Several approaches have been reported in order to improve sensitivity of PCF sensors. For example, dual-cladding PCF in which the solid fiber core was fabricated with additional holes to increase the interaction of the optical field with the sample (e.g. from 0.041% to 4.22% at 633 nm for a water-filled fiber [87]). The relatively simple concept of suspended-core fiber in which a small core is held in air by three thin silica struts was introduced by ref. [83]. These fibers have demonstrated large modal overlap of 29% at 1550 nm, which can find useful applications in gas sensing [88]. In addition to chemical sensing, the evanescent-wave configuration has also been applied to biosensing, whereby fluorescently labeled antibodies in aqueous solution were 1.5. OPTICAL SENSING WITH PCF 13 detected via absorption spectroscopy [89]. Furthermore, SC-PCF has demonstrated superior performance in surface-specific spectroscopy, whereby fluorescence sensing can be optimized with improved detection efficiency of biomolecules compared to conventional single-mode fibers [90, 91]. Additionally, it is worth noting that an axially periodic refractive index variation can be inscribed in the solid core of PCFs, known as long-period gratings (LPGs). These LPGs written in PCF are highly sensitive to refractive index variations of the external medium [92], and have been demonstrated as a label-free technique for detection of biomolecules [93], as well as for temperature and strain measurements [94]. Finally, structural rocking filters can be fabricated by periodically twisting birefringent PCFs [95]. Measurements of the sensitivity of the resonance wavelengths of the rocking filters to temperature, strain and hydrostatic pressure have demonstrated application in hydrostatic pressure sensing with very low cross-sensitivity to temperature [96].

1.5.2 Hollow-Core PCF Sensors

In addition to the various advantages mentioned in the previous sections, HC-PCFs exhibit a significant advantage for sensing applications over evanescent wave PCF sensors in that the modal overlap with the sample is considerably improved, guiding more than 90% of the power in the core defect of the fiber. The direct interaction of the light and the sample within the hollow fiber core is depicted in Figure 1.5(b). Consequently, the strong light confinement provided by the PBG and the possibility of tuning the PBG by tailoring the structural parameters have attracted much attention in the field of fiber sensors. It has been shown that the hollow core of the PBG-PCF selectively filled with a dye solution achieved an almost 100% modal overlap with the sample material, surpassing the performance of index-guiding PCFs [97]. In particular, the detection limit of fluorescence sensing was demonstrated to improve by four orders of magnitude. The study of gas characteristics using PBG-PCF has been performed using a light-emitting diode (LED) to measure the absorption spectra of hazardous gases [98]. The results obtained demonstrated that gas sensing in PCF is feasible using low-power, cost-effective light sources to realize miniaturization of the system setup. In terms of biosensing applications, HC-PCF Bragg fiber has been demonstrated for the detection of single-stranded deoxyribonucleic acid (DNA) by monitoring the changes 14 CHAPTER 1. PHOTONIC CRYSTAL FIBERS in the confinement loss of the Bragg fiber [99]. A Fabry-Pérotstrain sensor based on HC- PCF has also been demonstrated to feature multiplexing capability, wide free-spectral range, and insensitivity to temperature and fiber bending [100]. Chapter 2

Experimental Considerations and Techniques

2.1 Introduction

Photonic crystal fiber sensors offer two modes of operation, namely, the resonant and non- resonant regimes of sensing. In the non-resonant regime, one takes advantage of the large optical modal overlap with the sample. Sensing is realized by monitoring changes in the imaginary part of the sample’s refractive index, i.e., analyte absorption, by detecting the presence and strength of the absorption bands within the fiber transmission spectrum, as depicted in Figure 2.1(a). In this case, the absorption signal strength and sensor sensitivity are directly proportional to the fiber sensor length, as will be shown in Section 2.2. The second mode of operation of PCF-based sensor operates in the resonant regime and can be categorized into two types. The first type relies on monitoring the changes in the real part of the sample’s refractive index, by detecting the variations in the optical confinement of a mode propagating inside a resonant fiber structure such as the PBG-PCF. As the real part of the sample’s refractive index changes, the resonant condition for modal confinement will also change, resulting in a strong variation of the fiber transmission loss, as depicted in Figure 2.1(b). Such sensors can also operate in the non-resonant mode for detection of changes in the imaginary part of the sample’s refractive index. The second type of resonant sensors operate in the vicinity of a phase-matching wavelength between a core-guided mode and a second mode which is sensitive to changes in the real part of

15 16 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES

nn a }-Re{

nia+ Im{D n } na

Modal loss Modal loss na

nna+Re{D }

l l (a) (b)

Figure 2.1: Operational principles of optical sensors in the (a) non-resonant and (b) resonant regimes, whereby changes in the sensor transmission loss due to variations in the (a) imaginary and (b) real part of the analyte’s refractive index are monitored [8]. the sample’s refractive index, such as an absorbing plasmon mode propagating at the interface between the analyte and the metal-coated fiber surface. As the real part of the sample’s refractive index changes, the phase-matching condition between the core and plasmon mode also changes, resulting in strong optical loss of the core mode at a specific resonant wavelength [101, 102]. In this chapter, the figures-of-merit for sensing and photochemical reactions are introduced in Section 2.2 to determine the ideal experimental conditions in PCF. The setup and instrumentation considerations, including the transmission properties of liquid-filled PCF, various components of the optical setup, microfluidic flow through the fiber and computer automation of data acquisition, are summarized in Section 2.3. Finally, various fabrication techniques for PCF devices, especially for sensors and photochemical reactors, are described in Section 2.4. 2.2. DETECTION STRATEGIES AND IDEAL CONDITIONS 17 2.2 Detection Strategies and Ideal Conditions

2.2.1 Ideal Conditions for Absorption-Based Sensors

There are four major optical transduction mechanisms used in fiber-optic sensors, by which the presence of a target analyte induces changes in the transmission of light through the optical fiber, namely, absorbance, fluorescence or chemical luminescence, Raman scattering, and surface plasmon resonance. The PCF sensors described in this thesis are based on the exploitation of changes in the fiber transmission losses as a result of absorption by the sample. The absorption-based sensing methodology can be based on both amplitude and spectral interrogation. In amplitude-based detection methodology, changes in the amplitude of an optical signal at a given wavelength λ are used to deduce the changes in the analyte’s refractive index. An amplitude sensitivity function S(λ, L) can be employed to characterize the sensitivity of a fiber-optic sensor of length L [8]. S(λ, L) represents the relative change in the irradiance P (δ, λ, L) of the transmitted light for an infinitesimal change in the measur- and, δ, which can be any parameter capable of influencing the transmission properties of a fiber sensor, such as the concentration and the real or imaginary parts of the refractive index of the sample, and is defined as P (δ, λ, L) − P (0, λ, L) ∂P (δ, λ, L)/∂δ| S(λ, L) = lim = δ=0 . (2.1) δ→0 δ · P (0, λ, L) P (0, λ, L) The irradiance of light at the fiber output can be written as

P (δ, λ, L) = Pin(λ) exp[−α(δ, λ)L], (2.2) where Pin(λ) is the light irradiance at the ﬁber input and α(δ, λ) is the ﬁber propagation loss. By substituting Equation (2.2) into Equation (2.1), the amplitude sensitivity function can be rewritten as

∂α(δ, λ) S(λ, L) = − · L. (2.3) ∂δ δ=0

According to classic perturbation theory, changes in the effective refractive index ∆neff of a guided mode are related to the changes in the refractive index ∆na of the analyte infiltrating the fiber,

∆neff = ∆na · φ = Re(∆na) · φ + iIm(∆na) · φ, (2.4) 18 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES where φ is the fraction of power overlap in the analyte. An important goal for non-resonant absorption-based sensors is the identification of sample materials by the spectral shape of their absorption. Furthermore, the concentration of certain chemical compounds in the sample can be deduced from the magnitude of the corresponding absorption peak. Defining N to be the number density of the absorbing particles in the analyte, so that δ = N, it follows from Equation (2.4) that the total fiber loss in the presence of absorbing sample can be written as

α(N, λ) = αf (λ) + σ(λ)Nφ, (2.5) based on α(N, λ) ∼ Im(neﬀ) and Im(∆na) ∼ σ(λ)N, where αf (λ) is the ﬁber loss in the absence of the absorbing sample and σ(λ) is the absorption cross-section of a single particle. Substituting Equation (2.5) into Equation (2.3) yields an expression for the amplitude sensitivity function based on the experimental variables,

S(λ, L) = −σ(λ)φL. (2.6)

It follows from Equation (2.6) that the sensitivity of the fiber sensor is proportional to its length and the fractional modal overlap of the guided mode with the sample analyte. In fact, the absorption of light by a sample described in Equation (2.2) is the commonly used Beer-Lambert law which relates the absorption of light to the properties of the material through which it is traveling. Here, the conventional law is slightly modified to take into account the fraction of the light φ that travels through the sample. The resulting absorbance is A(λ) = σ(λ)Nφ(λ)L = ln(10)(λ)cφ(λ)L, (2.7) where (λ) is the molar absorptivity of the sample in Lmol−1cm−1, c is the molar concentration of the sample in molL−1 (or simply in molar, M, as will be used interchangeably throughout the thesis), and the mode-field overlap φ is wavelength dependent as will be shown in Chapter 4. The convention of expressing the absorbance in dB, defined as

AdB(λ) = 10(λ)cφ(λ)L = 10 log10(e)σ(λ)Nφ(λ)L (2.8) will be used throughout the thesis. From Equations (2.6) and (2.7) one sees that the sensitivity is directly linked to the absorbance signal amplitude, while Equation (2.2) signiﬁes that the upper limit of the 2.2. DETECTION STRATEGIES AND IDEAL CONDITIONS 19

100 εc = 0.01 cm εc = 0.001 cm -1 -1

10 εc = 0.1 cm -1

1 1.1 m SC-PCF εc = 1 cm -1 39 cm HC-PCF

Fiber length [m] 0.1

1 cm cuvette 0.01 1 10 100 φ [%]

Figure 2.2: Ideal sensing parameter diagram for constant absorbance of 5 dB, plotted in the φ-L plane, defining regions in which optimum sensing conditions can be achieved. The contour lines are of fixed c. For a given c, any combination of φ and L that lies on the corresponding line will result in a 5 dB absorbance signal. The fine dashed line at 20 m indicates the fiber length above which the fiber loss becomes too high for sensitive measurement due to the power budget (ηdet is assumed to be 20 dB in this case). The solid circle, square and diamond symbols represent the experimental conditions for the sensing and photochemistry experiments in a 1 cm cuvette, 1.1 m of suspended SC-PCF and 39 cm of kagome HC-PCF described in Chapters 3 and 4.

sensor length is defined by the loss of the fiber. Assuming Pdet(λ) to be the lowest output irradiance level of light at which changes can still be detected reliably, the maximum sensor length allowed by the power detection limit can be calculated from Equation (2.2) as η (λ) L = det , (2.9) αf (λ) where ηdet = ln[Pin(λ)/Pdet(λ)] is related to the power budget. Assuming that ηdet(λ) = 1 to allow for the characterization of the inherent sensitivity of the sensor, irrespective of any additional power-dependent sensitivity enhancement, the maximum sensitivity allowed by the power detection limit can be obtained by substituting Equation (2.9) into Equation (2.6): σ(λ) S(λ) = −φ . (2.10) αf (λ) Note that as the absorption cross-section (similarly, the molar absorptivity) is completely independent of the fiber loss, the sensitivity of the fiber sensor can therefore be increased by using longer fibers with low propagation loss. A general parameter diagram is shown in Figure 2.2 to provide further insight into 20 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES the optimum design parameters for fiber sensors. The absorbance signal (and hence the sensitivity) is kept constant at AdB = 5 dB, which is sufficiently large to be detected by any spectrometer of adequate signal-to-noise ratio and dynamic range. The lines in the φ-L plane indicate contours of constant c obtained from Equation (2.8). For a given c, any combination of φ and L that lies on the corresponding line will result in a 5 dB absorbance signal. The lines thus define regions in which optimum sensing conditions can be achieved. The intrinsic fiber loss in the absence of the absorbing particles is assumed

−1 to be 1 dBm in the calculations, while the power budget ηdB is assumed to be 20 dB. The figure clearly demonstrates that the minimum value of a measurand, in this case the concentration of the absorbing particles in the analyte, that can be detected by such a fiber sensor is limited by the fiber loss, as indicated by the fine dashed line (at 20 m of fiber length), since the sensitivity is limited by αf (λ) as shown in Equation (2.10). Depending on the characteristics of the fiber used, the maximum fractional power overlap with the sample can also be limited by the slope dφ(λ)/dλ for large values of φ(λ), resulting in a sensitivity gradient in the measured absorption spectra. While the gradient in φ(λ) can be compensated for, it is preferable to operate at lower values of φ, where it does not vary much with wavelength.

2.2.2 Figure of Merit for Photochemistry

Photonic crystal fibers also provide a platform for performing photochemical reactions within the holes of its microstructure. It is useful to examine critically the advantages that PCF offers over a conventional cuvette-based sample cell (see Figure 2.3). Two important parameters determine the effectiveness of a photochemical experiment. Firstly, the effective path length of the probe light (defined as the length at which the irradiance drops to 1/e of the initial value in the pure liquid host, that is, in the absence of any absorbing particle) should be long enough to allow detection of low concentrations. Sec- ondly, the cross-sectional area of the sample cell should be as small as possible, so as to maximize the irrdiance of the optical pump field; high intensities are required to achieve rapid conversion, in particular for reactions with low quantum yields. Assuming that the objective is to achieve complete photolytic conversion of all the chemicals in the sample, 2.2. DETECTION STRATEGIES AND IDEAL CONDITIONS 21

1 cm

19 µm 1 cm 1 cm

Figure 2.3: Schematics (not to scale) illustrating and comparing the geometries and sample volumes in a conventional cuvette and a kagome PCF. a suitable dimensionless figure of merit (FOM), is L a FOM = eff cuv , (2.11) aeffLcuv where the ratio between the effective interaction length, Leff, and the effective cross- sectional area, aeff, of the sample cell, is normalized to the respective depth, Lcuv, and cross-sectional area, aeff, of a standard sample cuvette. The standard sample cuvette is taken to be a 1 cm2 cross-section filled to a depth of 1 cm with the sample, with a collimated pump beam illuminating the entire cross-section of the sample volume. The FOM can be increased by a factor of 100 by reducing the cross-section of the cuvette from 1 cm × 1 cm to 1 mm × 1 mm, which is close to the smallest practical cuvette size. For a free-space Gaussian beam tightly focused into a sample volume, the small effective area at the focus gives rise to high irradiance of the pump field, but is however counter-balanced by the limitation in the effective interaction length as a result of strong diffraction of the tightly focused beam, which is twice the Rayleigh range, zR, 2πω2 L = 2z = 0 , (2.12) eff,Gaussian R λ where ω0 is the radius of the beam waist and λ is the wavelength of operation. From the 22 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES definition for FOM in Equation (2.11), the FOM for a tightly-focused Gaussian beam in free space is therefore inversely proportional to the wavelength of operation. Hollow capillaries could be used to further decrease the cross-section, allowing FOM to be increased by a further three orders of magnitude. However, the effective length of such a sample cell is limited by optical leakage losses. With the conventional notations for the electric field and irradiance of the EHnm mode (n 6= 0) as Enm(z) = E0 exp(−αnmz) and Inm(z) = I0 exp(−2αnmz), respectively, the attenuation coefficient αnm for a perfectly straight hollow capillary is [103]

2 2 2 unm λ ν + 1 αnm = √ , (2.13) 2π r3 2 ν2 − 1 where r is the bore radius, ν = n2/n1 is the ratio of the refractive indices of the capillary cladding to the material in the hollow bore, and unm is the mth root of the equation

Jn−1(unm) = 0. With unm = 2.405 for the EH11 mode, the 1/e decay length is therefore given by √ r3 ν2 − 1 L = 6.83 . (2.14) eff,capillary λ2 ν2 + 1 The HC-PCF provides a near-ideal sample cell for photochemical reactions in that it allows for single-mode guidance in a hollow core and the FOM is only limited by the fiber loss rate, which dictates Leff. Table 2.1 shows the comparison between hollow capillaries and the kagome HC-PCF used in the experiments. It is worth noting that the losses in hollow capillary waveguides are very sensitive to even slight bends, making the use of long capillaries very difficult. In contrast, hollow-core PCFs are almost completely insensitive to bend losses, and the measured waveguide loss of the PCF used in the experiments is 175 times lower than the calculated loss of a hollow capillary with the same core diameter. As a result, FOM for the kagome PCF is 175 times higher than a capillary with similar dimensions, and more than seven orders of magnitude higher than a standard cuvette.

2.3 Experimental Setup and Instrumentation

2.3.1 Transmission Properties of Liquid-Filled PCF

Liquids play a prominent role in the ﬁelds of chemistry and biology, and since the emer- gence of PCF, it has been suggested to utilize these ﬁbers for the miniaturization of 2.3. EXPERIMENTAL SETUP AND INSTRUMENTATION 23

2 Sample cell conﬁguration a [cm ] Leﬀ [cm] Volume [mL] FOM Cuvette 1 cm × 1 cm × 1 cm 1 1 1 1 Cuvette 1 mm × 1 mm × 1 mm 10−2 1 10−2 102 Focused Gaussian beam[a] 4.1 × 104 Straight 19 µm glass capillary 2.8 × 10−6 0.49[b] 1.4 × 10−6 1.8 × 105 Straight 100 µm glass capillary 7.9 × 10−5 71[b] 5.6 × 10−3 9.0 × 105 19 µm core kagome PCF 2.8 × 10−6 86[b] 2.4 × 10−4 3.1 × 107

Table 2.1: Comparison between various sample cell configurations and the kagome PCF. [a] Free-space beam at 488 nm. [b] 1/e length, determined from waveguide losses. chemical and biosensors, simplifying and enhancing the detection of the presence, absence, or properties of liquids and their constituents. Traditionally, the design for liquid- core waveguides is limited by the nature of the waveguiding mechanism. Liquid-core waveguides relying on the principle of total internal reflection allow only certain material combinations, as the refractive indices of solid materials available for making up the cladding are typically higher than most liquids. This design limitation presents a particular challenge for biosensors in which water is typically the basis for most biological analytes, as it has a lower index than most solids used for making hollow capillaries. One way to circumvent the index contrast problem is to use SC-PCF. By infiltrating its cladding air holes with materials of higher refractive indices, the index contrast between the core and the cladding reduces, effectively weakening the strength of optical confinement within the solid core. However, as the core mode remains index-guided within the same silica core, the transmission spectrum of the liquid-filled fiber remains similar to that of the unfilled fiber, with additional absorption dips in the transmission spectrum due to the presence of the infiltrated material in the cladding holes. As the decrease in index contrast also decreases the strength of confinement, the losses for the infiltrated SC-PCF will be higher than the unfilled counterpart, with light at longer wavelengths seeing more effect from the decreased confinement strength. In order to increase the overlap between the probe light and the sample, a liquid- filled hollow core is still the most desirable configuration. Several approaches have been developed to combat the limitations imposed on the analyte index, such as applying a 24 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES layer of fluorinated polymer (e.g. Teflon AF, with an index of 1.29) on the inside of a hollow capillary made of a rigid, higher index material. However, due to technical limitations only large core diameters of 200-500 µm could be fabricated with relatively large fluctuations in the thicknesses of the Teflon AF coating [104]. Furthermore, the large core dimensions imply that the capillary is highly multimode. Alternatively, non-TIR-based waveguides such as capillaries [103] (where the cladding index is higher than the index of the liquid core) and metal-clad waveguides [105] (where the inside of the cladding is coated with a highly reflective metal material) have been used. Nevertheless, these waveguides suffer from attenuation issues (the source of which is intrinsic in the case of capillaries, and surface imperfections in the case of metal-clad waveguides) and are highly sensitive to bending losses. The third and optimal approach would be to use interference-based waveguides such as Bragg fibers and HC-PCFs. When the air holes in HC-PCF are filled with a material of higher refractive index, the change in the refractive index contrast will inevitably change the transmission properties of the fiber. The transmission spectra of the liquid-filled HC-PCFs were found to be shifted in frequency according to the index scaling law for PBG-PCFs which is derived from the scalar approximation of the vector wave equation for the transverse field distribution [106]. When the low index material n2 of the PBG-

PCF is varied while the high index n1 remains unchanged, so that the index contrast of 0 the PCF changes from N0 = n1/n2 to N = n1/n2, the wavelength λ0 of a photonic state (bandgap) will shift to a new wavelength λ given by

1 − N −2 1/2 λ = λ0 −2 . (2.15) 1 − N0 Strictly speaking, Equation 2.15 is only valid for very small index contrasts, but can still provide qualitative results for larger contrasts as the photonic states in PCFs result from interference away from the interfaces where the effect of the vectorial term in the wave equation can be neglected. Additionally, even though the kagome HC-PCF does not guide via the PBG effect, this equation can still give qualitative indication of the shift in the transmission band of the fiber when infiltrated with another material. This equation becomes extremely useful when designing a PBG- or kagome HC-PCF for light transmission that coincides with the absorption spectra of the chemicals to be studied in 2.3. EXPERIMENTAL SETUP AND INSTRUMENTATION 25

0.8

0.7

λ[μm] 0.6 Water Acetone 0.5 Cyclohexane 0.8 1.2 1.6 2 2.4 2.8 λ [μm] 0

Figure 2.4: Shift in the central wavelength of bandgap, from λ0 to λ, as a result of inﬁltrating the PBG-PCF with water, acetone and cyclohexane. The variation in the refractive indices in the range between 476.5 nm to 830 nm are nwater = 1.3380 to 1.3281, nacetone = 1.3644 to 1.3544 and ncyclohexane = 1.4325 to 1.4209. the PCF sensor or reactor. Figure 2.4 shows the shift in the central wavelength of the PBG as a result of inﬁltrating the PBG-PCF with various solvents used in the experiments, including water, acetone and cyclohexane. The variation in the refractive indices in the wavelength range between 476.5 nm to 830 nm are nwater = 1.3380 to 1.3281, nacetone =

1.3644 to 1.3544 and ncyclohexane = 1.4325 to 1.4209. The PBG of the ﬁber to be used for experiments in various solvents can thus be determined. For example, for an experiment where the absorption peak of the sample in water lies at λ = 650 nm, an unﬁlled PBG-PCF with PBG at around λ0 = 1.2 µm is required for the experiment. Clearly, the spectral position of the transmission window of the PBG-PCF is extremely susceptible to changes in the local environmental conditions such as the refractive index, and can therefore be utilized as the transduction signal in PCF sensors.

2.3.2 Microﬂuidic Flow in Conﬁned Channels

Understanding the fiber filling process is important to PCF sensor design. Different geometry of liquid flow pathway may result in different fiber filling behavior such as infiltration time, possibility of entrapping an air bubble, etc. Knowledge of the infiltration process allows for the optimization of liquid cell design and the arrangement of components such as splits and valves to avoid potential filling problem and achieve high filling speed. Consider a capillary immersed in a liquid, as depicted in Figure 2.5, there are four 26 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES

Air

Liquid

◦ Figure 2.5: Cross-section of a capillary tube inﬁltrated with liquid. For θc < 90 , the ◦ force will pull the liquid into the capillary; for θc > 90 , the force will push the liquid out of the capillary. forces acting on the column of liquid inside the capillary, namely the capillary force, the friction force (which is related to the viscosity of the liquid), the force from an applied overhead pressure and the gravitational force [107]. The capillary force for a circular capillary is given by

Fc = 2πaσ cos θc, (2.16) where a is the radius of the capillary, σ is the surface tension and θc is the contact angle ◦ between the liquid and the inner wall of the capillary. For θc below 90 , the force will pull the liquid into the capillary, while for contact angles larger than 90◦ the force will push the liquid out of the capillary. The small dimensions of the fiber holes lead to very small Reynold’s numbers which imply that the liquid flow inside the capillaries will be laminar. The frictional force which results from the Poiseuille flow is

Ff = −8πµxv, (2.17) where µ is the dynamic viscosity of the liquid, x is the length of the liquid column and v is the average velocity of the liquid ﬂow. The force of an applied overhead pressure is given by

2 Fp = ∆P πa , (2.18) 2.3. EXPERIMENTAL SETUP AND INSTRUMENTATION 27 where ∆P is the pressure diﬀerence between the liquid in the capillary and the open end of the ﬁber. Finally, the gravitational force acting on the liquid column is

2 Fg = −ρgπa x, (2.19) where ρ is the density of the liquid and g is the gravitational constant. Balancing all four forces acting on the liquid column yields

d ρπa2xv = 2πaσ cos θ − 8πµxv + ∆P πa2 − ρgπa2x. (2.20) dt c

Noting that v = dx/dt and neglecting the gravitational term for a horizontally oriented ﬁber, the equation can be expressed as

d2 d x2 + B x2 = A, (2.21) dt2 dt where the constants A and B are

4σ cos θ + 2∆P a A = c , (2.22) ρa 8µ B = . (2.23) ρa2 The diﬀerential equation in Equation (2.21) has the solution

A A A 1/2 x(t) = exp(−Bt) + t − . (2.24) B2 B B2

Figure 2.6 shows the simulated infiltration length as a function of filling time for water in capillaries with bore radii of 1, 5 and 10 µm, with an applied pressure head of 1 bar. The results clearly shows the strong dependence of the filling time on the size of the fiber holes, with the larger core in the case of a HC-PCF being filled before the smaller holes in the microstructured fiber cladding. In the case where the hollow core is completely filled for the entire length of the fiber while the cladding holes are empty or only partially filled, the fiber becomes index-guided as the core now has a higher index than the effective index of the unfilled PCF cladding. The large index contrast and core size of the liquid- filled fiber would lead to highly multimode guidance (∼ 103 for operation in the visible wavelength region), resulting in difficult, if not impossible, coupling to the fundamental mode, which is an undesirable effect especially if the measurements are phase-sensitive. 28 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES

1 a= 1µm a= 5µm 0.75 a = 10 µm

0.5

0.25 Infiltration length [m]

0 0 1 2 3 4 5 Infiltration time [min.]

Figure 2.6: Simulated water ﬁlling time for silica microchannels of bore radii 1, 5 and 10 µm, with an applied pressure head of 1 bar. The physical constants for water at 20◦C ◦ −3 used were: θc = 0 , σ = 72.88 mN/m, ρ = 998 kgm and µ = 0.001 Pa·s.

Calculations and design of specific PCF cladding structures for the single-mode guidance of liquid-core PCF have been performed and suggested [108]; however the wavelength and index contrast dependence of the cladding design make the proposed approach an impractical one. It is therefore essential to continue the infiltration process for a longer period of time until all the holes in the cladding have been filled with the liquid so as to maintain single-mode guidance of the HC-PCF, with shift in the transmission window according to the index scaling law discussed in Section 2.3.1.

2.3.3 Optical Setup

The general conﬁguration of the optical setup used in the sensing and photochemical experiments is depicted in the schematic shown in Figure 2.7. The main components constituting the setup are described in the following sections.

2.3.3.1 Broadband Light Sources

A broadband light source is a crucial part of the setup used for broadband absorption spectroscopy. Depending on the wavelength range of interest, either a PCF SC source or a ﬁber-coupled xenon lamp is used. Supercontinuum with emission in the wavelength range from 480 nm to beyond 1750 nm was generated from an ESM-PCF pumped by a Q-switched Nd:YAG (yttrium aluminum garnet) microchip laser at λ = 1064 nm [52]. 2.3. EXPERIMENTAL SETUP AND INSTRUMENTATION 29

Spectro- MMF Computer 10x meter

Excitation

ESM Broad- 10x 20x CCD PCF band BS1 BS2

Sample fiber 4x 10x

Figure 2.7: Schematic diagram showing the experimental setup for sensing and photochemistry experiments in PCF. The broadband light source and excitation laser are co-aligned using a 50:50 beam splitter (BS1) and coupled into 15 cm of ESM-PCF for spatial filtering before being coupled into the fiber filled with the sample chemical. The transmitted light is collected by the spectrometer via a multimode fiber (MMF) and 8% of the output beam is coupled out via a 92:8 beam splitter (BS2) and imaged on a CCD beam profiler. Both light sources and the spectrometer are controlled electronically to automate the data collection process (for the measurement of photochemical reaction kinetics).

The single-mode output of the SC allows for easy coupling into most sample fibers, but has a disadvantage in that it does not provide sufficient output for wavelengths below 480 nm. In this case, a xenon lamp which provides usable wavelength from 380 nm to 1000 nm was used in experiments where the absorption spectrum of the sample lies at the shorter wavelengths in the visible. However, the spatial incoherent nature of the lamp implies that spatial filtering is required. This was achieved by coupling the output of the xenon lamp through 15 cm of ESM-PCF. The coupling efficiency of the xenon lamp through the ESM-PCF was less than 10%, however, the filtered light remained in a single optical mode, allowing coupling to single fiber modes in the sample fiber. 30 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES

2.3.3.2 Excitation Light Sources

Additional excitation sources were required in the photochemical experiments in order to induce photochemical conversions of the chemical samples studied in the PCF nanoreactors, as will be described in Chapter 3. A 20 mW pulsed microchip laser emitting 532 nm pulses at 6 kHz and a 20 mW continuous wave diode laser emitting at 488 nm were used to drive the photochemical reactions. As the samples have wavelength-dependent molar ab- sorptivitiy coefficients, the rate of photochemical conversion would vary depending on the wavelength of excitation. Both excitation sources were coupled into the same ESM-PCF used as a spatial filter for the broadband light source so that excitation lasers of different beam quality and divergence can be coupled into the sample fiber via the same setup. Furthermore, the combination of all the light sources through the ESM-PCF spatial filter simplifies the in-coupling of the light sources into the sample fiber and ensures that they are all launched into the same fiber mode.

2.3.3.3 Liquid Cells

Both ends of the PCF were connected to custom-made liquid cells having thin sapphire windows for in- and out-coupling of light. The aqueous sample was introduced into the PCF by using a single-syringe infusion pump connected to one of the ports of the liquid cells. The dead volume in the liquid cells was limited to 50 µL and the liquid cells could withstand water pressures up to 10 to 500 bars, depending on the thickness of the glass windows used (0.08 to 1 mm).

2.3.3.4 Numerical Aperture

Both the broadband and excitation sources (for photochemical experiments) were combined through the ESM-PCF and coupled into the sample fiber via an objective, allowing more freedom over coupling parameters than fiber butt coupling. The coupling of light into the fibers was optimized by matching the numerical aperture (N.A.) of the coupling objective to that of the fiber. The N.A. of the index-guiding fiber can be approximated by q 2 2 N.A. = n1 − n2, (2.25) 2.3. EXPERIMENTAL SETUP AND INSTRUMENTATION 31

n2 n liquid nair nwindow

Figure 2.8: Schematic diagram (not to scale) illustrating the increase in the effective N.A. as a result of change in the interface medium of the objective (air) to that for the fiber (liquid). The actual fiber N.A. and the effective N.A. are related via Snell’s law.

where n1 and n2 are the refractive indices of the core and cladding material, respectively. The N.A. of the HC-PCF can be approximated by λ N.A. = , (2.26) 2D where λ is the wavelength of operation and D is the diameter of the hollow fiber core. It is worth noting here that HC-PCF generally have very low N.A. in the order of 0.01. The change in the interface medium of the objective (air) to that for the fiber (liquid) will result in a higher effective N.A. (see Figure 2.8) given by

nliquid N.A.eﬀ = N.A.. (2.27) nair

An objective with N.A. in accord with N.A.eff for the fiber/liquid-cell combination should therefore be chosen for efficient coupling of light into the fiber.

2.3.3.5 Spectrum and Mode Proﬁle

The transmitted light at the output facet of the fiber was collected and collimated by a 10×0.25NA objective. A 92:8 beam splitter imaged a small portion (8%) of the transmitted light via a lens onto a charge coupled device (CCD) beam profiler to measure the irradiance profile of the guided mode to ensure coupling to the fundamental core mode, while the rest (92%) of the light was coupled to a multimode fiber (MMF) connected to either an optical spectrum analyzer (OSA) or a USB spectrometer. 32 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES 2.3.4 LabVIEW Automation

A stepwise measurement approach was implemented in the photochemistry experiments described in Chapter 3. This was done to avoid signal saturation in the spectrometer caused by the excitation sources, which had much higher irradiances than the broadband probe light used to measure the absorption spectra. Furthermore, the wavelengths of the excitation sources coincided with the absorption features of interest in the sample absorption spectra, thereby making the use of optical bandpass ﬁlters inappropriate. The broadband and excitation light sources, as well as the spectrometer, have electronic shut- ters which were controlled by a LabVIEW program to allow for automated exposure of the photochemical sample under study and collection of spectra for signal processing. During one iteration of the program, the excitation source is switched on to irradiate the sample for a predeﬁned exposure time, and the broadband light source is then switched on to enable the spectrometer to acquire a spectrum. The whole cycle repeats until the photoreaction is complete or the photostationary state of the reaction is reached.

2.4 Fabrication Techniques for PCF Devices

Several techniques have been developed for the fabrication of PCF devices with applications spanning beyond the field of fiber optic sensors. The techniques described in the following sections have been developed and studied with the main objective of combining PCF sensors with microfluidics.

2.4.1 Femtosecond Laser Ablation

When a pure transparent material is exposed to high laser irradiance, nonlinear material responses can lead to the promotion of electrons from the valence band to the conduction band (photoionization), depositing laser energy into the material in the process (free- carrier absorption by plasma) and ultimately causing damages in the material [110-112]. There are two diﬀerent regimes of photoionization, namely the multipohoton ionization regime and the tunneling ionization regime. For low laser frequencies, nonlinear photoionization is a tunneling process, as depicted in Figure 2.9(a), whereby the Coulomb 2.4. FABRICATION TECHNIQUES FOR PCF DEVICES 33

(a) (b)

Figure 2.9: Schematic diagram of photoionization regimes. (a) At low frequencies, photoionization occurs via tunneling of the valance electron through the suppressed Coulomb well. (b) At high frequencies, photoionization occurs via excitation of the valence electron as a result of multiphoton absorption [109]. well that binds a valance electron to its parent atom is suppressed by the strong applied electric ﬁeld to a level which allows the bound electron to tunnel through and become free. For higher laser frequencies, nonlinear photoionization is a process involving the simultaneous absorption of multiple photons to reach energy higher than the bandgap of the material and excite the electron from the valence to the conduction band, as shown in Figure 2.9(b). For a seed electron already in the conduction band, which can be provided either by thermal excitation, ionization of impurity states, or by multiphoton or tunnelling ionization, it can linearly absorb more photons through free-carrier absorption (see Figure 2.10(a)) and move to an even higher energy state in the conduction band. Once the electron has enough energy it can impact ionize another electron in the valence band, resulting in two electrons near the conduction band minimum, as depicted in Figure 2.10(b). This avalanche process can then continue to impact ionize more valence band electrons. Once the electron plasma density becomes high enough such that the plasma frequency reaches the laser frequency, the absorption of laser energy becomes very eﬃcient and as a result a large fraction of the laser pulse energy will be deposited in the focal volume. It is this deposition of laser energy that leads to permanent damage of the material. For 34 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES

(a) (b)

Figure 2.10: Schematic diagram of avalanche ionization. (a) A seed electron in the conduction band linearly absorbs more photons through free-carrier absorption and moves to a higher energy state in the conduction band. (b) Once the free election has enough energy it impact ionizes another electron in the valence band, resulting in two electrons near the conduction band minimum [109]. longer pulses (> 10’s of ps), the damage mechanism is achieved via thermal diffusion of the energy build-up at the focal volume. Once the temperature becomes high enough the material begins to melt. For pulses shorter than a few picoseconds, the pulse duration is shorter than the time scale for thermal diffusion, and material breakdown is achieved via the build-up of plasma density through self-seeded electrons (via photoionization at the leading edge of the laser pulse) for avalanche ionization. The damage mechanism for short pulses is much less dependent on material defects (for seed electrons) than for longer pulses, and therefore has more determinist damage threshold; as less energy is required for reaching the damage threshold irradiance, less energy is deposited in the material, effectively allowing more precise ablation of the material, which is ideal for controlled micromachinging of PCF devices. Figure 2.11 shows a schematic for the setup used for femtosecond laser ablation of PCF. The goal was to fabricate microfluidic side channels into the PCF cladding holes, through which liquid or gas can be infiltrated to reach the core region of the fiber for detection or reaction. The system used for micromachining is a regeneratively amplified 2.4. FABRICATION TECHNIQUES FOR PCF DEVICES 35

Ti:S 50x

HeNe

CCD 50x

Figure 2.11: Schematic diagram showing the experimental setup for femtosecond laser ablation of side channels in PCF and the two-photon polymerization technique for selective blockage of microstructure holes. The system used for both experiments is a regeneratively ampliﬁed Ti:Sapphire laser producing 150 fs pulses at 800 nm, with a maximum pulse energy of 8 µJ at a repetition rate of 100 kHz.

Ti:Sapphire laser producing 150 fs pulses at 800 nm, with a maximum pulse energy of 8 µJ at a repetition rate of 100 kHz. The laser pulses are co-aligned with a He-Ne laser used for alignment and illumination, and focused through a 50×0.55N.A. Mitutoyo objective lens with a working distance of 13 mm. The backscattered light re-enters the objective lens and passes through a telescope for imaging on a CCD camera, allowing on-line monitoring of the ablation process. Spatial characterization of the pump beam was performed by a knife-edge measurement at the focus of the lens and the focal size of the laser beam was measured to be 2.5 µm. The objective was measured to transmit 30% of the input power for average power levels below 50 mW. In order to determine the threshold energy required to achieve optical breakdown in fused silica, varying laser powers were focused on the surface of bare capillary ﬁber made of fused silica, the same glass as that used in the PCF. The formation of surface void was monitored using the CCD camera. The damage threshold for fused silica was determined to be 50 nJ, corresponding to a peak power of 0.3 MW and an irradiance of 1.7 × 1012 W/cm2. 36 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES

10 10

5 5 E = 0.36 µJ Measured p E = 0.75 µJ Analytical p

Entry hole diameter [µm] 0 0 0 0.2 0.4 0.6 0.8 0 2.5 5 7.5 10 Pulse energy [µJ] Number of pulses [×103] (a) (b)

Figure 2.12: (a) Diameter of ablated entry hole in the silica fiber as a function of pulse energy incident on the fiber. The solid curve shows the analytically calculated hole diameter for a constant damage threshold of Eth = 50 nJ and a Gaussian beam waist of ω0 = 2.5 µm. (b) Diameter of ablated entry hole in the silica fiber as a function of the number of pulses incident on the fiber for constant pulse energies of 0.36 and 0.75 µJ. For exposure exceeding 2000 pulses (20 ms) the hole diameter saturates at 9 µm.

The diameters of the entry holes were subsequently characterized as a function of pulse energy by focusing the laser beam onto the surface of the bare fibers and varying the laser pulse energy while keeping the exposure time constant at 30 s. Figure 2.12(a) shows the measured entry hole diameter as a function of pulse energy incident on the fiber surface. The solid curve shows the analytical estimate of hole diameter calculated assuming a Gaussian irradiance distribution of the beam waist (see Figure 2.13), r ln(E /E ) D = 2ω p th , (2.28) 0 2 where D is the ablated entry hole diameter, ω0 is the beam waist, and Ep and Eth are the pulse energy and the threshold energy (50 nJ). The error bars indicate the standard deviation of the measurements taken at the respective pulse energies. The higher measured values are due to the edges of the entry hole being blown away by the material in the center of the focus which is ablated by the laser. The diameters of the entry holes were then measured as a function of exposure time while keeping the pulse energy constant. Arrays of holes were made with varying exposure time and pulse, and examined post-mortem using an optical microscope. Figure 2.12(b) shows the diameter of the entry holes as a function of the number of pulses for pulse energies at 0.36 and 0.75 µJ. No notable difference was observed for irradiation with 0.36 2.4. FABRICATION TECHNIQUES FOR PCF DEVICES 37

Figure 2.13: Schematic showing the dependence of the diameter of laser-ablated entry hole size on the peak irradiance, assuming a Gaussian irradiance distribution. and 0.75 µJ pulses. Interestingly, for exposure exceeding 2000 pulses, the hole diameter saturates at 9 µm, which is larger than the values measured in Figure 2.12(a). Long microchannels from the surface of the fiber to the core were drilled by translating the fiber through the laser focus using a piezo-controlled stage. Figure 2.14(a) shows SEM of a microchannel drilled on the side of a suspended-core fiber used for evanescent-wave sensing, allowing the lateral access of the cladding hole for introducing gas or liquid samples without influencing the incoupling of light. The resulting microchannel had a diameter of less than 2 µm. In order to verify the robustness of the microchannels, water was pumped through the microchannels (with core diameters of ∼ 1 µm) from the core of the capillary fibers using a liquid pump for high performance liquid chromatography (HPLC) at pump pressures ranging from 6 to 30 bar (corresponding to maximum fluid flow of vmax = 0.67 to 3.33 ms−1). The channels remained unmodified for pressures below 8 bars. At higher pressures, the side walls of the channels were damaged, as shown in Figure 2.14(b). Loss measurements were performed for an ESM-PCF with multiple drilled microchannels. The loss due to individual microchannels drilled through the cladding region (down to the core) was determined to be about 1.1 dB per channel at around λ = 800 nm, and 0.65 dB per channel at around λ = 1550 nm (see Figure 2.15). The loss scales with ω2, 38 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES

(a) (b)

Figure 2.14: (a) Example of ablated side microchannel allowing access to one of the three cladding holes in a suspended-core ﬁber. (b) Ablated side microchannel in a 40 µm capillary ﬁber damaged after applying 30 bar of water pressure using a HPLC pump. 8 8 Measured Measured α = 1.1 dB/channel α = 0.65 dB/channel fit fit 4 4

0 0 α (1.5 to 1.6 μm) [dB] α (750 to 850 nm) [dB] 0 2 4 6 0 2 4 6 Number of drilled microchannels Number of drilled microchannels (a) (b)

Figure 2.15: Measured transmission losses (α) as a function of the number of drilled side channels in an ESM-PCF. The losses are averaged losses in (a) the 750-850 nm wavelength range and (b) the 1500-1600 nm range. and can be attributed to Rayleigh-Gans scattering, which is applicable for scatterers with dimensions approaching that of the incident light and have scattering cross-sections that scale with ω2 [113].

2.4.2 Two-Photon Polymerization

Microfabrication technology utilizing two-photon polymerization (TPP) [114] has been intensively studied towards the development of micromachines [115] and photonic devices [116]. As the probability of n-photon absorption is proportional to the nth power of the photon ﬂux density, high photon ﬂux densities are required to observe this phenomenon 2.4. FABRICATION TECHNIQUES FOR PCF DEVICES 39

(a) (b) (c)

Figure 2.16: (a) Side view of an HC-PCF with all of the cladding holes infiltrated with the acrylic resin via capillary effect (depth of infiltration ≈ 120 µm. (b) TPP-fabricated structure with 10 photopolymerized cladding holes (1 µm diameter) in a SC-PCF. (c) A gold nanowire embedded into one of the sub-micron holes next to the core of a polarization- maintaining PCF using TPP as the hole-collapsing post-processing procedure [118]. Image courtesy of Howard Lee.

[117]. This requirement for high irradiance can again be provided by short-pulse lasers with high pulse peak powers. In two-photon absorption (TPA) a molecule can absorb two photons simultaneously to allow electron transition to the states not attainable with single photon absorption. An UV photopolymerizing material is photo-solidified in a small volume within the depth of focus of a pulsed near-IR laser, while the resin, which is usually transparent in the IR, remain unpolymerized in regions where the laser beam is out of focus due to the optical density being lower than the threshold required for TPA. The polymerization is a process in which monomers or oligomers interconnect to form polymers. Photoinitiators with low photodissociation energies are often added to increase the photosensitivity of the material. Upon absorption of two NIR photons, radicalized photoinitiator is formed via bond cleavage (photodissociation). The radicals break the C=C double bonds in the acrylyl groups of the monomers and oligomers, resulting in radicalized monomers and oligomers. The radicals then combine with other monomers and oligomers and this chain reaction of radical polymerization eventually terminates when a chained radical combines with another chained radical. In our experiments we made use of TPP for the selective collapsing of PCF cladding holes with applications in the fabrication of microfluidic circuits within the PCF as well as in other novel PCF devices. Selective collapsing of PCF cladding holes was initially investigated using the approach of direct mechanical insertion of photosensitive resin into the target holes. A femto-tip and a micromanipulator were used to insert the resin into 40 CHAPTER 2. EXPERIMENTAL CONSIDERATIONS AND TECHNIQUES individual holes with the aid of a CCD camera attached to a microscope. In spite of the high accuracy of the micromanipulator and the small size of the tip, the results were however unsatisfactory due to the time-consuming nature of the process, problems with hardening of the glue on the tip, and destruction of the microstructured cladding caused by the mechanical contact of the tip with the fiber. Using a similar setup configuration as for the femtosecond micromachining process described in the previous section, with the replacement of the fiber holder (by rotating it by 90◦ to that depicted in Figure 2.11), the laser beam now replaces the mechanical tip. A photosensitive acrylic resin fills all the cladding holes of the fiber via capillary effect. Figure 2.16(a) shows an image of the side view of an HC-PCF with all of its cladding holes infiltrated with the acrylic resin. With the laser beam focused on the cleaved facet of the fiber, the individual target holes were selectively exposed to irradiation with an average power of 3 mW and an exposure time of 1 s. The unpolymerized resin in the unexposed holes was then rinsed away with acetone, revealing the fabricated structure as shown in Figure 2.16(b). The polymer-blocked holes were tested to withstand water pressures up to 20 bar. With some further post-processing such as the inflation or collapse of holes using a flame on a fiber tapering rig, the technique presented here has allowed metallic [118] and semiconductor nanowires [119] in the PCF microstructure to be fabricated. Figure 2.16(c) shows an example of the above-mentioned hole-collapsing procedure in which a single gold nanowire is embedded into one of the sub-micron holes next to the core of a polarization-maintaining PCF [118]. As symmetry is not necessary in post-processed fibers, this technique allows for flexible designs of fiber devices. Similar techniques have been used to fabricate an all-fiber mode converter [120].

2.4.3 Focused Ion Beam Micromachining

Another method investigated to allow side access of the longitudinal holes of PCF is via focused ion beam (FIB) micromachining [121, 122]. The use of FIB for the routine fabrication of micro- and nano-structures and devices is based on a sputtering process in which the collision of an energetic ion beam with the target material overcomes its binding energy, resulting in the ejection of the material. Figure 2.17(a) shows a 45 × 45 µm2 hole milled in the silica jacket (approximately 35 µm in thickness) of a nanoweb ﬁber1 [123]

1Fiber fabricated by Michael Scharrer and Alexander Podlipensky. 2.4. FABRICATION TECHNIQUES FOR PCF DEVICES 41

(a) (b)

Figure 2.17: (a) A 45×45 µm2 hole milled through the silica jacket (approximately 35 µm in thickness) of a nanoweb ﬁber. (b) SEM of the cross-section of the nanoweb ﬁber prior to FIB milling.

(Figure 2.17(b)), exposing the nanoweb structure in the fiber. The exposed nanoweb allows for a wide range of experiments to be performed on the in-fiber planar waveguide. In terms of sensing, the fiber geometry offers improved surface-to-volume ratio advantage especially useful for surface enhanced reactions.

Chapter 3

Photochemistry in PCF

3.1 Introduction

There is rapid growth in the applications of photochemistry in many areas, including medicine [124-127], chemical synthesis [128, 129], and the conversion and storage of solar energy [130, 131]. Two important constraints currently limit many photochemical experiments. As discussed in Section 2.2, if spectroscopy (e.g., in the UV/Vis) is to be used to monitor reactions, the product of molar absorptivity and concentration, integrated over the optical path length, must be above a certain value which is limited by the detection limit of the system. For weakly absorbing or sparingly soluble samples, the only way to satisfy this condition is to increase the optical path length; in a conventional cuvette this implies large sample volumes, which is a major drawback for valuable samples such as biological constructs or multistep synthetic products. Furthermore, high pump intensities are required for rapid photochemical conversion, in particular for reactions with low quantum yields. If modest laser powers are to be used, this means that the cross-sectional area of the excitation beam should be as small as possible, and comparable with the cross-sectional area of the sample cell. In systems with low quantum yields and weakly absorbing or sparingly soluble samples, the ideal photochemical cell is therefore one that combines long optical path length with small cross-sectional area which is comparable to the size of the excitation beam. There is no obvious way to satisfy these conditions in existing microfluidic and lab- on-chip systems, in which fluid flow is manipulated and chemical reactions monitored in

43 44 CHAPTER 3. PHOTOCHEMISTRY IN PCF sub-millimeter scale channels [132-135]. Some progress has been made in this direction using SC-PCF, in which the sample is probed by the evanescent tailing field of the guided mode [86, 87, 136, 137]. A more revolutionary approach exploits HC-PCFs for their ability to maximize the interaction between the optical field and the low refractive index sample at path lengths that are much longer than achievable in conventional single-pass sample cells [98, 138]. Hollow-core PCFs have several major advantages over conventional sample cells: the sample volume per optical path length is very small, long optical path lengths are possible as a result of very low intrinsic waveguide loss, and furthermore the light can travel in a diffractionless single mode with a constant transverse irradiance profile. In this chapter, the demonstration of a liquid-filled HC-PCF as a highly-controlled photochemical reactor is reported. In Section 3.2 the fiber characteristics of the HC-PCF used in the experiments are presented. Two photochemical experiments were performed to demonstrate the effectiveness of the PCF reactor in monitoring both irreversible and reversible photochemical reactions, exemplified here by the photolysis of a cobalamin in Section 3.3 and the photoswitching of an azo dye derivative in Section 3.4.

3.2 Fiber Characteristics

In HC-PCFs, guidance is achieved through two mechanisms, as described in Section 1.3. In PBG-PCF, a core resonance coincides with a photonic bandgap in the microstructured cladding, resulting in the formation of ultra-low loss (∼1 dBkm−1 in the best cases) guided modes over restricted bandwidths of typically a few hundred nanometers. In contrast, the guidance mechanism of kagome HC-PCF is based on the reduced coupling between the core mode and the cladding modes [69, 70], resulting in higher losses in the order of 1 dBm−1, nevertheless still better than in a capillary. The kagome HC-PCF has a spectrally much broader guidance band than the PBG-PCF, making it the fiber of choice for broadband spectral measurements. A scanning electron micrograph (SEM) of the kagome HC-PCF used in the experiments is shown in Figure 3.1(a). The fiber was made from fused silica, with a hollow core of 19 µm in diameter, surrounded by a kagome lattice of thin silica webs with a cladding pitch of 10 µm that runs along the entire length of the fiber. The structural parameters of the fiber correspond to an approximate 3.2. FIBER CHARACTERISTICS 45

50 µm 50 µm

(a) (b)

Figure 3.1: (a) Scanning electron micrograph showing the cross-section of the kagome HC- PCF used in the experiments. The fiber was made from fused silica, with a hollow core of 19 µm in diameter, surrounded by a kagome lattice of thin silica webs with a cladding pitch of 10 µm that runs along the entire length of the fiber. (b) Optical micrograph of the light emerging from 3 cm length of the fiber illuminated from below with a halogen lamp. The colors observed in the cladding holes are caused by long-lived Mie-like resonances and are related to the slight nonuniformity in the air hole diameters. sample cell cross-section of 284 µm2, giving an ultra low sample volume of 2.8 nLcm−1. The enhanced sensitivity and pumping efficiency means that even systems with very small quantum yields can be measured much faster than in conventional cuvettes, as will be shown in the later sections. Figure 3.1(b) shows an optical micrograph of the light emerging from 3 cm length of the fiber illuminated from below with a halogen lamp. The colors observed in the cladding holes are caused by long-lived Mie-like resonances and are related to the slight nonuniformity in the air hole diameters [64]. The transmission and attenuation spectra are important in the characterization of the fiber as they provide guidelines for the the wavelength range and maximum fiber length that can be used in the experiments. The spectral attenuation of the fiber was determined via the conventional cut-back technique [139], in which the power transmitted through a long length of fiber is measured and normalized to the power transmitted through a shorter length of the same fiber without changing the incoupling condition. Figure 3.2 shows the transmission spectrum of the fiber, normalized to the spectrum of the SC source described in Section 2.3, showing a transmission band of the fiber extending from the visible to 960 nm. The inset shows the transverse irradiance profiles measured at the 46 CHAPTER 3. PHOTOCHEMISTRY IN PCF

-20

600 nm 800 nm -40

-60 Normalized transmission [dB]

5 Loss [dB/m]

0 600 800 1000 1200 1400 1600 Wavelength [nm]

Figure 3.2: Transmission (normalized to the supercontinuum source) and loss spectra of the kagome HC-PCF. Inset: measured transverse irradiance profiles after 4 m of the fiber at λ = 600 and 800 nm. output of 4 m of the fiber at λ = 600 and 800 nm to confirm that single-mode guidance within the guidance band of the fiber was achieved. The measured loss spectrum of the fiber based on a cut-back from 4 m to 1.1 m indicates a 2 dBm−1 loss region from λ = 510 to 710 nm, while the loss peaks between λ = 800 and 1210 nm correspond to coupling to surface states and resonances in the cladding struts. The kagome HC-PCF was designed and fabricated to have a guidance band in the wavelength range around 450 to 500 nm when filled with water to allow for spectroscopic measurements to be performed on the sample chemicals used in the experiments. As light guidance in the kagome HC-PCF is not via the PBG mechanism, the index scaling law described in Section 2.3 cannot be applied here to quantitatively estimate the guidance band of the unfilled fiber. However, the guidance band of an unfilled kagome HC-PCF is still expected to shift to lower wavelengths upon filling of the air holes. The fabrication of a kagome HC-PCF with guidance band in the required wavelength range can be achieved 3.2. FIBER CHARACTERISTICS 47

-25

-50 400 nm 450 nm 488 nm 500 nm 550 nm 600 nm 700 nm

-75 Normalized transmission [dB]

Loss [dB/m] 10

0 400 500 600 700 800 900 Wavelength [nm]

Figure 3.3: Transmission (normalized to the supercontinuum source) and loss spectra of the kagome HC-PCF filled with de-ionized water. The transmission spectra were measured using two supercontinuum sources (solid and dashed curves). Inset: measured transverse irradiance profiles after 60 cm of the fiber filled with de-ionized water, at λ = 400, 450, 488, 500, 550, 600 and 700 nm. by appropriate scaling of the structural parameters of a fiber with known guidance band during fabrication. The fiber shown above, when completely infiltrated with de-ionized water, has broadband guidance in the visible up to 700 nm as shown in Figure 3.3. Two PCF SC sources were used in the measurement of the transmission spectra; the solid curve was measured with the conventional PCF SC source while the dashed curve was measured using a SC source generated using a tapered conventional fiber1 pumped by a frequency-doubled microchip laser emitting sub-nanosecond pulses at 532 nm. The inset of the figure shows the transverse irradiance profiles measured at the output of 60 cm of the fiber infiltrated with de-ionized water. All the beam profiles were measured using the SC sources in combination with interference filters at the output of the fiber,

1Tapered ﬁber courtesy of Marta Ziemienczuk. 48 CHAPTER 3. PHOTOCHEMISTRY IN PCF

-5

488 nm -15

-25 Normalized transmisssion [dB] 400 500 600 700 800 900 1000 Wavelength [nm]

Figure 3.4: Transmission (normalized to the supercontinuum source) spectrum of the index-guiding kagome HC-PCF filled with toluene. Inset: measured transverse irradiance profile after 39 cm of the fiber filled with toluene at λ = 488 nm. except for the beam profile at 488 nm which was measured using the CW excitation laser described in Section 2.3. The measured irradiance profiles confirmed that the fundamental mode is guided over the whole wavelength range of interest from 450 to 600 nm. The higher-order core mode and light in the cladding region at 400 nm can be attributed to chromatic aberrations in the coupling objective as the optimization wavelength for the loss measurement was at 550 nm. Optimization of the core mode at 400 nm confirmed guidance of the fundamental mode (mode profile not shown). The irradiance profile at 700 nm clearly demonstrates coupling of the core mode to the surface states of the core surround, as indicated by the loss peak at 700 nm in the measured loss spectrum based on a cut-back from 2 m to 0.85 m. The single central lobe of the guided mode interacts strongly with the sample, and the rigid core boundaries restrict diffusion of chemicals away from (or into) the illuminated volume. Quantitative spectroscopic studies of very small sample volumes, selectively introduced into the core, become possible. The transmission property of the same kagome HC-PCF when filled with the solvent toluene was also investigated and the measured normalized transmission spectrum is shown in Figure 3.4 to exhibit broadband guidance in the visible up to 860 nm. The transverse irradiance profile measured using the CW excitation source at 488 nm displayed an asymmetric mode profile indicating coupling into multiple higher order modes and consequently the loss of the fundamental mode could not be measured. The reason for the multimode behavior of the fiber when filled with toluene can be easily understood 3.3. PHOTOLYSIS OF METAL COMPLEXES 49 as the refractive index of toluene is higher than that of fused silica [140], the fiber becomes

p 2 2 index-guided with a V parameter given by VPCF(λ) = 2πΛ/λ ncore − neff, where λ is the wavelength of operation, Λ is the pitch of the cladding structure and ncore and neff are the effective refractive indices of the core and the cladding [141, 142]. The number of

2 guided modes can therefore be approximated by VPCF/2 ≈ 90 modes. With careful optical alignment, keeping the bend radius of the ﬁber large (≥ 15 cm) and avoiding twisting of the ﬁber, it was possible to minimize the number of guided modes and maximize the irradiance concentrated in the core region.

3.3 Photolysis of Metal Complexes

II Platinum-based anticancer dugs such as cisplatin (cis-[Pt Cl2(NH3)2]) are well-established therapeutic compounds. However, as they do not discriminate between cancerous and healthy tissues, their use is constrained by severe dose-limiting side effects [143], as well as acquired resistance to the drug. In order to overcome these problems, research in the field is moving towards the use of inert, nontoxic platinum complexes that can be locally activated by light at the tumor site [127, 144]. Photoactivated drugs are routinely used in photodynamic therapy for the effective treatment of a number of cancers, including those of the skin, brain, lung and esophagus [145-147]. The ongoing research and development of new metal-based complexes aim to provide more potent anticancer drugs which are less oxygen dependent (due to the hypoxic nature of most tumor cells [145, 148]) and less biologically reactive in order to minimize the potential for cytotoxicity of the inactive drug precursors. We demonstrate here a highly-controlled photochemical reactor based on PCF with the prospect of simultaneously activating and monitoring the reaction dynamics of such anitcancer drugs, combined with the advantages offered by PCF-based sensors. The photoaquation of the readily available and nontoxic metal complex vitamin B12 (cyanocobalamin, CNCbl) was studied as a proof of principle for the PCF photochemical reactor. 50 CHAPTER 3. PHOTOCHEMISTRY IN PCF

CONH CONH2 2 CONH CONH2 2 CH3 CH3 H NOC CH3 H2NOC CH3 2

N H3C H3C OH2 CONH CONH2 2 N C N H C N N H3C 3 III Co III hv Co H H N N − N N − CN , + H2O CH CH3 3 H NOC H2NOC 2 CH CH3 3 CH CH CH3 CH3 3 3 O O CONH CONH2 2 NH NH CH N CH3 N 3 H C H3C H 3 H O O N CH N CH3 3 O OH P O OH P O O O O O O HO HO

+ CNCbl [H2OCbl]

+ Figure 3.5: The photochemical conversion of CNCbl to [H2OCbl] .

3.3.1 Photoaquation of Cyanocobalamin

− Irradiation of CNCbl in aqueous solution causes exchange of CN for H2O, forming + [H2OCbl] (aquacobalamin or B12b) (Figure 3.5), a photoreaction with a very low quantum yield (Φ ∼ 10−4 at pH 6) [149, 150]. Cobalamins such as CNCbl possess a low-spin Co3+ conﬁguration with near-octahedral geometry at the center of the corrin ring. The Co3+ is ligated equatorially by four nitrogen atoms of the ring and axially by a nitrogen of the tethered base 5,6-dimethylbenzimidazole (DBI). A number of ligands can occupy the upper axial position [151, 152]. The absorption spectra of many cobalamins are highly similar, since the corrin ring is responsible for the dominant spectral features, namely the α and β bands ( ≈ 8000 − 10000 M−1cm−1) in the visible spectral region and the Soret (γ) band ( ≈ 25000 M−1cm−1) in the UV region [151, 152]. The α and β bands (Figure 3.6) both arise from the same π → π∗ transition, the more intense α band being the electronic origin, and the β band the ﬁrst member of a progression in a vibrational mode, primarily involving C=C stretches of the corrin ring [153]. Within the cobalamins, the wavelength 3.3. PHOTOLYSIS OF METAL COMPLEXES 51

9000 hν CNCbl +

] [H OCbl] 2 -1 α 6000 cm β

-1 γ

3000 ε [Lmol

0 400 450 500 550 600 650 Wavelength [nm]

Figure 3.6: Changes in the absorption spectrum measured in a 1 cm cuvette as a result + of the photochemical conversion of CNCbl to [H2OCbl] in pH 2.5 buﬀer. at which the α band is seen approximately parallels the nephelauxetic eﬀect [154]; in the

− 3+ case of photoaquation, exchange of CN for OH2 decreases the electron density at Co and the α absorption band moves to shorter wavelengths. Although it has been suggested that the DBI group which coordinates CoIII from beneath the ring is readily replaced by H2O in acidic solution [155], spectral analysis of + the Soret (γ) band of CNCbl and [H2OCbl] in extremes of both acid and base has led to the conclusion that DBI remains attached [156]. Dissociation of DBI (and replacement with H2O) is only considered signiﬁcant below pH ≈ 0 for the cobalamins in general [154], and furthermore it is estimated that DBI is bound particularly tightly in CNCbl, three orders of magnitude more than in CH3Cbl, for example [157]. The pK a corresponding to protonation and displacement of the imidazole base of CNCbl has been determined as

0.11 (H2O/H2SO4) [158]. It is reasonable to assume, therefore, that under the conditions − of the experiment, DBI does not dissociate from cobalt in either the CN or H2O adduct.

3.3.2 Experimental Results

Photochemical experiments were performed on solutions of cyanocobalamin in citric acid / phosphate buffer (pH 2.5 to 7.5) using the optical setup described in Section 2.3. The absorption spectra of cobalamins are dominated by the α and β bands ( ≈ 8000 − 10000 M−1cm−1) in the visible spectral region and the Soret (γ) band ( ≈ 25000 M−1cm−1) in the UV region [151, 152], as shown in Figure 3.6. Quantitative absorption spectra were obtained by referencing the spectra to that of the buffer solution. The molar absorptivity 52 CHAPTER 3. PHOTOCHEMISTRY IN PCF spectrum ¯(λ) integrated over the fiber length 0 < z < L follows from the Beer-Lambert law, A = cL, taking the form given by

R L (λ)c (z) + (λ)c (z)dz ¯(λ) = 0 1 1 2 2 , (3.1) c0L where 1,2(λ) are the molar absorptivitity spectra and c1,2(z) the spatial concentration + proﬁles of CNCbl and [H2OCbl] . The initial concentration of the sample is c0. The photochemical reaction was accurately monitored by the spectral changes to the α and β bands. The typical temporal behavior of the absorption spectrum during photolysis is shown in Figure 3.7(c). Figure 3.7(b) shows the absorption spectrum before irradiation

+ (i.e., of pure CNCbl) and 100 s after full conversion to [H2OCbl] . Upon excitation, both α and β bands are shifted to shorter wavelengths, resulting in a decrease in absorption for λ = 530 to 600 nm and an increase in absorption for λ = 450 to 530 nm. Figure 3.7(a) shows the decrease in absorption for the peak of the α band at λ = 550 nm, indicated by the black dashed line on the colormap in Figure 3.7(c). The observed changes are in good agreement with previous work [149, 150].

+ The rate of photoconversion of CNCbl to [H2OCbl] has previously been shown to depend on the pH of the solution. Cyanocobalamin is most stable to photoaquation between pH 7 to 8 and converts more rapidly at both the higher and lower extremes of pH [150]. To investigate this effect, the temporal evolution of absorption at 500 and 550 nm during the photolysis of CNCbl at pH 2.5 and 7.5 was measured in both the cuvette and in the kagome HC-PCF and is shown in Figure 3.8. The cuvette measurements were carried out on 1 mL of 125 µM buffered aqueous CNCbl solution at an excitation power of 9.5 mW. The fiber measurements were performed using a sample volume of approximately 100 nL (4 µM) and only 10 µW of power at pH 2.5 and 20 µW at pH 7.5. The photochemical conversion occurred roughly 1000 times faster in the fiber than in the cuvette, even though the excitation power remained below 20 µW. Previous studies of the photoaquation of CNCbl required much higher lamp powers (> 100 W) [149, 150] and/or acidic conditons (pH 4.75) [159]. 3.3. PHOTOLYSIS OF METAL COMPLEXES 53

7000 (a) 550 nm ]

-1 6000 cm -1 5000 ε [Lmol 4000

600 (b) (c) 6000

5000 550 α ]

hν 4000 -1 cm -1 β 3000 ε [Lmol Wavelength [nm] Wavelength 500 2000

CNCbl 1000 [H OCbl]+ 2 450 7500 5000 2500 0 0 10 20 30 40 50 ε [Lmol-1 cm-1 ] Time of exposure at 488 nm [s]

Figure 3.7: Photolysis of CNCbl at pH 2.5 in 39 cm of kagome HC-PCF. (a) Measured temporal evolution of molar absorptivity at 550 nm taken at 500 ms intervals over a period of 50 s. (b) Molar absorptivity spectra of CNCbl before irradiation (thick curve) and after 100 s of irradiation at 488 nm (ﬁne curve). (c) Colormap showing the measured evolution of molar absorptivity spectrum with time of exposure at 488 nm using 10 µW of optical power.

3.3.3 Reaction Kinetics

The photochemical evolution of the absorption spectra upon excitation can be modeled using the conﬁguration diagram in Figure 3.9. The photophysical transitions and nature of the excited electronic states of CNCbl have been investigated using TD-DFT [160,

161]. Irradiation of CNCbl results in excitation from the ground state (S0) to an initially excited π-π∗ state, which is followed by (sub-picosecond) internal conversion to a lower energy excited singlet state (S1) − a state best characterized as being of π3d character. 54 CHAPTER 3. PHOTOCHEMISTRY IN PCF

Kagome HC-PCF Cuvette 8 10 ] pH 2.5 pH 2.5 -1 500 nm

cm 500 nm

-1 6 8 Lmol

3 4 6 550 nm 10 ´ 550 nm ε[ 2 4 0 20 40 60 0 5 10 15 Exposure time at 488 nm [s] Exposure time at 488 nm [hours]

8 10 ] pH 7.5 pH 7.5 -1 550 nm cm

-1 6 500 nm 8 Lmol

3 4 6

10 500 nm ´ 550 nm ε[ 2 4 0 200 400 600 0 5 10 15 Exposure time at 488 nm [s] Exposure time at 488 nm [hours]

Figure 3.8: Temporal evolution of measured and theoretically-ﬁtted (solid curves) molar absorptivity at 500 (circles) and 550 nm (squares) for photolysis of CNCbl in a cuvette (right column) and a kagome HC-PCF (left column) at pH 2.5 (top row) and 7.5 (bottom row). The quantum yields obtained from the theoretical ﬁts were 6.88 × 10−4 at pH 2.5 and 9.95 × 10−5 at pH 7.5 in the kagome HC-PCF, and 5.46 × 10−4 at pH 2.5 in the cuvette.

The relative photostability of CNCbl compared to alkylcobalamines is attributed to the fast subsequent radiationless decay to the ground state from the S1 excited state (τ ∼ 7 ps in H2O) [161]. If the molecule does not decay from S1 to the ground state, it undergoes intersystem crossing (ISC) to a low-lying ligand-ﬁeld (LF) state, populating − either directly or indirectly − a LF triplet state, in which the Co-CN bond is weakened and from which photoaquation can occur [149]. The quantum yield of the reaction was calculated by modeling the number densities in the three states in Figure 3.9 using the following coupled rate equations,

∂n0(t) Ip = − σ0n0(t) + Γ10n1(t), (3.2) ∂t hνp ∂n (t) 2 = Γ n (t), (3.3) ∂t ISC 1

n0(t) + n1(t) + n2(t) = const., (3.4) 3.3. PHOTOLYSIS OF METAL COMPLEXES 55

π-π* internal conversion (< 1 ps) S1 (π3d) intersystem

ΓISC crossing photochemistry + T1 [Co-OH2] non-radiative LF state Φ ~ 10-4 (~ pH 6) decay (τ ~ 7 ps photoexcitation in H2O)

Γ10

S0 ground state [Co-CN]

Figure 3.9: Conﬁguration diagram depicting the photoaquation of CNCbl ([CoIII-CN] to III [Co -OH2]). Transitions are represented by the dashed (non-radiative) and solid (radiative) lines. The quantum yield for the photochemistry is thought to be low due to competing rapid internal conversion from S1 to S0. Non-radiative decay from T1 to S0 is assumed negligible in the model. Lifetimes are from ref. [161].

where n0,1,2(t) are the number densities at S0,S1 and T1, Ip and hνp the power density and photon energy of the pump light, σ0 the absorption cross-section at the pump wavelength ∗ for excitation from the S0 to the π-π state, Γ10 is the non-radiative decay rate from S1 back to S0 and ΓISC is the rate for intersystem crossing. The third equation follows from the conservation law. The quantum yield, which describes the fraction of CNCbl excited

+ to S1 converted to [H2OCbl] upon irradiation, has the form given by Γ Φ = ISC . (3.5) Γ10 + ΓISC The differential equations (Equations 3.2 to 3.4) were numerically solved and fitted to the experimental data, taking into account the exponential decay of the 488 nm pump irradiance (caused by fiber loss and absorption) by integrating along the length of the fiber,

− R z α(ξ)dξ Ip(z) = I010 0 (3.6) where I0 is the initial pump irradiance and α(ξ) is the combined position-dependent 56 CHAPTER 3. PHOTOCHEMISTRY IN PCF

pH Quantum yield (×10−4)[a] 2.5 6.73 ± 0.33 3.5 6.56 ± 0.15 4.5 5.30 ± 0.27 5.5 3.87 ± 0.23 6.5 1.59 ± 0.33 7.5 0.99 ± 0.01

Table 3.1: Quantum yields from theoretical ﬁts of data for the photolysis of CNCbl at pH 2.5 to 7.5 in a kagome HC-PCF. [a] Mean ± SE. 8 ] -4

10 6 ´

2 Quantum yield [ 0 2.5 3.5 4.5 5.5 6.5 7.5 pH

Figure 3.10: Theoretically fitted quantum yields obtained from measurements of photolysis in kagome HC-PCF (squares). The bars indicate the standard error and the dashed curve is intended as guide for the eye only. attenuation due to fiber loss and absorption. It is important to note that only one free parameter, namely the quantum yield, was used to fit the experimental data to the theoretical model. We find excellent agreement between our model and the data (Figure

3.8). The quantum yields for aqueous vitamin B12 determined at pH 2.5 to 7.5 are listed in Table 3.1. The measurements were repeated for six pH values at diﬀerent ﬁber lengths, sample concentrations and excitation power, and the results are summarized in Figure 3.10. The small standard deviation demonstrates the reproducibility of the method. The results showed that the quantum yield increases with decreasing pH, as expected [150]. 3.4. PHOTOSWITCHING OF AZOBENZENE MOLECULES 57 3.3.4 Discussion

In summary, the use of HC-PCF for the simultaneous quantitative assay of cyanocobalamin and aquacobalamin in aqueous solution has been demonstrated on a nanoliter scale. Laser-driven changes in the absorption spectrum are monitored within the fiber. This new method requires not only 104 times less sample volume compared to conventional techniques, but also greatly reduced excitation power, allowing system minimization using cheap on-chip diode lasers. Furthermore, the reaction is 1000 times faster as a result of the strong confinement of both sample and light in the hollow core. The procedure should therefore find wide application, enabling rapid investigation of photochemical reactions with modest quantum yields. Implementation of PCFs as flow reactors would allow continuous optimization of exposure conditions and reagent parameters, and integration of the reactor into an optical tweezer/particle guidance setup [162] would open up new opportunities for in vitro investigation of photoactive anticancer complexes [126]. The exploitation of PCFs as optofluidic devices offers significant advantages including minimal consumption of reagents and flexibility for integration into other microfluidic circuitry for improved performance.

3.4 Photoswitching of Azobenzene Molecules

Azobenzene chromophores are widely recognized as one of the most important and versatile classes of synthetic organic compounds, and have received much attention in both fundamental and applied research. With two phenyl rings separated by an azo (−N=N−) bond, azobenzene serves as the parent molecule for a host of aromatic azo compounds. The strong electronic absorption maximum can be tuned via the combination of the properties of the azo group and the substitution of the aromatic ligands, resulting in intense colors of dye over the whole visible range. Furthermore, the thermal and chemical robustness of these azo compounds, combined with non-complex synthetic methodologies and low production costs, has prompted extensive study of azobenzene-based structures as dyes and colorants [163, 164]. The mesogenic shape of the molecule also ﬁnds holographic applications in which azobenzene chromophores embedded in polymers (azo polymers) are used in gratings and liquid crystalline media [165-167]. When azobenzene is push-pull sub- 58 CHAPTER 3. PHOTOCHEMISTRY IN PCF

ν N N N h N hν', Δ

Figure 3.11: Reversible isomerization between the trans (left) and the cis (right) geometric isomers of azobenzene. stituted (i.e. when it has strong electron-donating and electron-attracting substituents), a very large permanent electrical dipole moment is formed which can yield high optical nonlinearity with extensive nonlinear optical applications [168-170]. One of the most interesting properties of the azobenzene chromophores, and the focus of this section, is the switching between two geometric isomers upon UV-vis irradiation. This readily induced photoisomerization is rapid, reversible and of high quantum yield, allowing large host systems incorporating azobenzenes to be used as photoswitches [171, 172]. The photoreaction studied in Section 3.3, namely the photolysis of cyanocobalamin, is a relatively slow and irreversible process under the experimental conditions imposed on the sample. This section demonstrates that the PCF reactor can also be used to study fast, reversible photoswitching processes in real-time, exempliﬁed here by the photoisomerization of two azobenzene derivatives.

3.4.1 Isomerization of Azo Dyes

The reversible isomerization between the trans and cis geometric isomers of azobenzene is depicted in Figure 3.11. Azo aromatic chromophores can be classified based on the energetic ordering of their n-π∗ and π-π∗ electronic states as azobenzene, aminoazobenzene or pseudo-stilbene. The azobenzene-type molecules, which are similar to the unsubsti- tuted azobenzene, exhibit a low irradiance n-π∗ absorption band in the visible, and a high irradiance π-π∗ band in the UV. The n-π∗ and π-π∗ bands of the aminoazobenzenes are closely-space in the violet or near-visible UV. In the pseudo-stilbene class, the substitution of electron donor and acceptor substituents (push-pull configuration) shifts the π-π∗ and the n-π∗ bands such that they effectively overlap. The three classes therefore display the colors of yellow, orange and red, respectively. The readily induced and reversible isomerization about the azo bond between the trans- and cis-isomers can occur via pho- 3.4. PHOTOSWITCHING OF AZOBENZENE MOLECULES 59

Absorption spectrum PSS → trans (back) 1.7 1.5 (a) Trans (b) PSS 1.55 1

A(λ) k = 0.0011 ± 0.0001 s-1 fit 0.5 1.4 λ = 450 nm max λ = 450 nm pump 0 1.25 400 500 600 0 1200 2400 3600 Wavelength [nm] Time [s] 1.7 1.5 (c)Trans (d) PSS 1.65 1 -1 A(λ) k = 0.0015 ± 0.0007 s fit 0.5 1.6 λ = 450 nm max λ = 532 nm pump 0 1.55 400 500 600 0 600 1200 1800 Wavelength [nm] Time [s]

Figure 3.12: Changes measured in the absorption spectra and the temporal evolution of the absorbance for the thermal back reaction of DO1 in toluene in a 1 cm cuvette, excited at λ = 450 nm ((a) and (b)) and 532 nm ((c) and (d)), as indicated by the dashed lines on the absorption spectra. The dashed curves on the temporal evolution of the absorbance are theoretical ﬁts to the experimental data, yielding thermal rate constants of 0.0011 ± 0.0001 and 0.0015 ± 0.0007 s−1 for excitation at λ = 450 and 532 nm, respectively. The measurements were taken at the University of Edinburgh, United Kingdom. tochemical and thermal processes. In this section, the ’forward’ reaction is used to refer to the trans → cis isomerization, while the ’back’ or ’reverse’ reaction refers to the cis → trans isomerization. The trans-isomer has a planar elongated form, while the cis-isomer assumes a bent geometry with the phenyl rings twisted at right angles to the C−N=N−C plane [173]. Upon irradiation with light, the thermally stable trans molecules are converted to the cis form, while the cis molecules can be converted back to the trans form either photochemically or thermally. The isomerization process is completely reversible and free from secondary reactions. After a certain irradiation time, the equilibrium state of the three competing conversion processes, known as the photostationary state (PSS), is reached. The rates and extent (determined by the concentration of cis-isomer in the PSS) of isomerization depend on several factors including the irradiance, wavelength of irradiation, temperature, substituents and the solvent. 60 CHAPTER 3. PHOTOCHEMISTRY IN PCF

In order to demonstrate the eﬀect of some of these factors, a series of forward and thermal back reactions were undertaken for 4-(4-Nitrophenylazo)diphenylamine (disperse orange 1, DO1) and N-Ethyl-N-(2-hydroxyethyl)-4-(4-nitrophenylazo)aniline (disperse red 1, DR1) in the solvents cyclohexane and toluene, irradiated with λ = 450, 488 and 532 nm radiation. Figure 3.12 shows the changes in the absorption spectra and the temporal evolution of the absorbance for the thermal back reaction of DO1 in toluene, measured in a 1 cm cuvette. The sample was excited at λ = 450 nm and 532 nm to demonstrate the dependence of the concentration of the cis-isomer in the PSS on excitation wavelength. The advantage of pumping near the absorption maximum of the trans-isomer was immediately evident: the absorbance decreased by 20% for the PSS induced by 450 nm radiation, indicating that at least 20% of the molecules had been converted to the cis form; on the other hand, for PSS induced by 532 nm radiation, the absorbance has only decreased by 5%. As the thermal isomerization was the sole process taking place, it can be described by a simple ﬁrst-order kinetic equation written as dc (t) dc (t) − cis = trans = kc (t) (3.7) dt dt cis where ccis(t) and ctrans(t) are the concentrations of the cis- and trans-isomers after time t of thermal back reaction, and k is the thermal rate constant. The temporal evolution of the absorption spectrum for the thermal back reaction can therefore be derived by using the Beer-Lambert law in Equation 3.7 as

A(t) = [A(0) − A(∞)] exp(−kT ) + A(∞), (3.8) where A(0) and A(∞) are the absorbance at the start and end of the thermal back reaction, respectively. The dashed curves on the temporal evolution of the absorbance (Figure 3.12(b) and (d)) are the theoretical fits to the experimental data using Equation 3.7 with fitted thermal rate constants of 0.0011 ± 0.0001 and 0.0015 ± 0.0007 s−1 for the excitation at λ = 450 and 532 nm. The agreement in the fitted thermal rate constants within the error margin is expected as the thermal back reaction rate should be independent of the wavelength at which the sample had been excited to reach the PSS. Figure 3.13 shows the measured temporal evolution of molar absorptivity for the forward and thermal back reactions of 15 µM of DR1 in cyclohexane and toluene in a 1 cm 3.4. PHOTOSWITCHING OF AZOBENZENE MOLECULES 61 trans → PSS (forward) PSS → trans (back) 2.8 2.8 ] (a) (b)

-1 DR1 in cyclohexane λ = 488 nm cm -1 2.4 2.4

Lmol DR1 in cyclohexane 4 λ = 488 nm k = 0.0082 ± 0.0007 s-1 fit ε[×10 2 2 0 2 4 6 8 0 3 6 9 12 Time of exposure at 488 nm [min.] Time [min.]

] (c) (d) -1 3 DR1 in toluene 3

cm λ = 473 nm -1 2.5 2.5

Lmol DR1 in toluene 4 λ = 473 nm -1 2 2 k = 0.020 ± 0.003 s

ε[×10 fit 0 0.5 1 1.5 2 0 0.6 1.2 1.8 2.4 Time of exposure at 488 nm [min.] Time [min.]

Figure 3.13: Temporal evolution of molar absorptivity for the forward and thermal back reaction of 15 µM of DR1 in cyclohexane ((a) and (b)) and toluene ((c) and (d)) in a 1 cm cuvette, excited with 200 µW at λ = 488 nm. The dashed curves on the temporal evolution of the molar absorptivity for the thermal back reactions are theoretical ﬁts to the experimental data, yielding the thermal rate constants 0.0082 ± 0.0007 and 0.020 ± 0.003 s−1 for DR1 in cyclohexane and toluene, respectively. cuvette, excited with 200 µW of optical power at 488 nm. Disperse red 1, with the azobenzene unit substituted with an electron-donating group on one benzene ring and an electron-withdrawing group on the other, belongs to the group of push-pull azobenzenes (which places the molecule in the pseudo-stilbene spectra class) [174]. Push-pull azobenzene derivatives have a permanent dipole moment and the thermal cis-trans isomerization is much faster than that for the nonpolar azobenzenes. The isomerization is strongly solvent dependent as shown in Figure 3.13, whereby DR1 in the polar solvent (toluene) has a much higher thermal rate constant compared to that in the nonpolar solvent (cyclohexane). This eﬀect can be understood by considering the planar structural geometry of the trans-isomer, which provides a greater de-localization of the π electrons. A polar solvent aids in this de-localization of the π electrons to reduce the energy further (in comparison to the less polar cis form), thereby increasing the cis → trans isomerization 62 CHAPTER 3. PHOTOCHEMISTRY IN PCF

trans → PSS (forward) PSS → trans (back) 3.5 λ = 450 nm max 2.5 k = 0.012 ± 0.001 s-1

4 -1 -1 fit λ = 450 nm 1.5 max ´ λ = 488 nm (a) (d) pump

ε [ 10 Lmol cm ] 500 Trans PSS 3

475 2.5

450 2

4 -1 -1

Wavelength [nm] Wavelength 1.5

425 ε [ 10 Lmol cm ]

(b) (c) (e) 1 4004 2 0 0 8 16 24 0 280 560 840 ε [´ 10 4 Lmol -1 cm -1 ] Time of exposure at 488 nm [s] Time [s]

Figure 3.14: Photoisomerization of DO1 (0.75 µM) in toluene in 39 cm of kagome HC- PCF. (a) Measured temporal evolution of molar absorptivity at 455 nm taken at 1 s intervals over a period of 15 s. (b) Molar absorptivity spectra of DO1 before irradiation (thick curve, trans-isomer) and after 65 s of irradiation at 488 nm (ﬁne curve, PSS). (c) Colormap showing the measured evolution of molar absorptivity spectrum with time of exposure at 488 nm using 3 µW of optical power. (d) Measured temporal evolution of molar absorptivity at 455 nm over a period of 14 minutes of cis-trans thermal isomerization. (e) Color map showing the measured evolution of molar absorptivity spectrum with time of thermal back reaction at 450 nm. rate. The eﬀect is most pronounced for push-pull azobenzenes due to their intrinsically higher polarity.

3.4.2 Reversible Isomerization in PCF

The photoisomerization of DO1 in toluene was performed in the kagome HC-PCF shown in Figure 3.1 at ambient temperature in dark room conditions using the optical setup described in Section 2.3. Quantitative absorption spectra were obtained by referencing the spectra to that of the solvent and the molar absorptivity spectrum obtained at the output of the ﬁber was given by Equation 3.1, where the subscripts 1 and 2 now represent the trans- and cis-isomers. The forward photoisomerization of DO1 in toluene was accurately monitored by the decrease in the absorption maximum in the visible, corresponding to 3.4. PHOTOSWITCHING OF AZOBENZENE MOLECULES 63

∗ the n-π (S1 state) transition. The spectral changes were monitored for a period at least four times longer than that necessary to reach the PSS. The temporal behavior of the absorption spectrum during photoisomerization is shown in Figure 3.14(c). Figure 3.14(b) shows the absorption spectrum before irradiation (thick curve), assumed to be that of pure trans molecules as the trans-isomer is the energetically favored configuration in the ground electronic state due to greater π electron de-localization and steric interaction. The spectrum at the end of the experiment (fine curve) after 65 s of irradiation at λ = 488 nm corresponds to the absorption spectrum of the PSS. Upon excitation, the decrease in the absorption maximum of the n-π∗ chromophore for the trans-isomer is coupled to the increase in the absorption maximum of the π-π∗ chromophore for the cis-isomer located further into the UV, which was beyond the available wavelength range of the probe beam used in the setup; however, the onset of the isosbestic point, that is, the wavelength at which both the trans- and cis-isomers have the same molar absorptivity (visible in Figure 3.12 at around 388 nm) can be observed near 400 nm. The temporal evolution of the absorption peak of the trans-isomer at 450 nm (indicated by the dashed line on the colormap in Figure 3.14(c)) is shown in Figure 3.14(a). The results show that despite the very low excitation power of 3 µW used, the PSS was readily reached within 10 s. In the PSS, the absorptivity has decreased by 40%, indicating that at least 40% of the molecules are in the cis form. In comparison, 1 W of excitation power (five orders of magnitude higher than that used in the fiber) would be required to achieve the same irradiance level and hence reaction dynamics in a 1 cm cuvette. The excitation source was switched off after 65 s of irradiating the sample to allow the sample to thermally relax back to the trans form. The temporal evolution of the molar absorptivity was monitored over 14 minutes as shown in Figure 3.14(e). The measured evolution of the trans absorption maximum at 450 nm is shown in Figure 3.14(d), demonstrating that the reaction was completely reversible, and the absorbance measured at the end of the experiment coincided with that measured initially, indicating that no irreversible secondary reactions took place. The thermal rate constant for the cis-trans thermal isomerization of DO1 in toluene measured in the HC-PCF was obtained by fitting the experimental data use the rate equation in Equation 3.7 to be k = 0.012 ± 0.001 s−1. The much higher apparent thermal rate constant measured in the fiber compared to that 64 CHAPTER 3. PHOTOCHEMISTRY IN PCF obtained from measurements in the cuvette shown in Figure 3.12 was unexpected as the thermal back reaction is a first order process, independent of the concentration and volume of sample used in the experiments. A possible factor which could induce the difference in the thermal rate constant measured is the change in temperature. From the Arrhe- nius equation, k(T ) = A exp(−EA/RT ), where R is the Boltzmann constant, and using an activation energy EA of 72 kJ/mol and ln(A) of 22 [175], the temperature difference between the two different laboratories in which the cuvette and fiber measurements were performed would have to be ∼ 25 K to induce the difference in the measured thermal rate constants. This seems unrealistically high. This discrepancy stimulated further in- vestigations to establish plausible causes for the marked differences in the experimental results. Another possible source of discrepancy could come from the probe light. Without making changes to the configuration of the experimental setup and dark room conditions, the absorption spectrum of the azo molecule in its thermally stable trans form was monitored using the broadband light source, namely the xenon lamp, at various average lamp powers. Figure 3.15(a) shows the measured temporal evolution of the absorbance at the n-π∗ absorption peak of the trans-isomer, for average lamp powers of 0.02, 0.55 and 1.15 µW between λ = 400 and 500 nm. It was observed that despite the very low average lamp power used, the resulting irradiance in the 19 µm hollow fiber core was still high enough to induce photoisomerization, ranging from I = 70.5 W/m2 (for P = 0.02 µW) to 4050 W/m2 (for P = 1.15 µW). The n-π∗ absorption of the cis-isomer, which is non-zero in the wavelength range of 400 to 500 nm, would similarly cause the reverse cis-trans photoisomerization. For the thermal isomerization measurement performed, the xenon lamp was required to be on for the duration of the spectrometer integration time, which was 500 ms in this case, during which sufficient (and undesirable) effect occurred and could have led to the discrepancies in the rate constants measured in the cuvette and the fiber. Another noticeable effect observed was the dependence of the extent of photoisomerization on the irradiance. As shown in Figure 3.15(b), the relative change in the absorbance at the PSS, given by [A(0) − A(PSS)]/A(0), increased with irradiance, indicating increased fraction of cis-isomer in the PSS. Due to the broadband nature of the pump source, which was originally intended as 3.4. PHOTOSWITCHING OF AZOBENZENE MOLECULES 65

1.15 µW 0.95 0.55 µW 0.02 µW 0.8 0.85

Absorbance 0.65

0.5 0 4 8 12

Absorbance 0.7 Exposure time [s]

0.55 (a) 0 40 80 120 Exposure time [s] 0.49 (b)

0.46 [A(0)-A(PSS)]/A(0) 0.43 0 0.4 0.8 1.2 Average power of xenon lamp [µW]

Figure 3.15: (a) Measured temporal evolution of absorbance at the absorption peak of trans-DO1 in toluene irradiated with a broadband xenon lamp with average powers of 1.15, 0.55 and 0.02 µW. Inset: zoomed-in figure for the first 12 s. (b) Measured dependence of the relative change in the absorbance at the PSS on irradiance, extracted from the data in (a). The dashed curves are intended as guide for the eye only. the probe beam, further analysis of the data would require knowledge of the cis-isomer absorption spectrum and the spectral density of the xenon lamp in this wavelength range. An accurate cis absorption spectrum could be extrapolated by the Fischer method [176], which requires the measurements for temporal absorption dynamics at two different excitation wavelengths. Due to the influence of the probe beam on the rate of photoisomerization in both the forward and reverse directions, in addition to the stepwise pump-probe cycle implemented for the measurements, solution to the coupled rate equations describing the reaction kinetics becomes complicated and requires tedious numerical computation, 66 CHAPTER 3. PHOTOCHEMISTRY IN PCF

S1 FC (~ps)

S1 TS

photoexcitation

ΓTC photoexcitation

ΓCT S ground state 0 k cis thermal reverse isomerization (ms – s) S 0 ground state trans

Figure 3.16: Configuration diagram depicting the isomerization paths of trans )* cis. The trans- and cis-isomers can theoretically excite to different transition states (TS) simultaneously but are indistinguishable in the experiments. A simplified model assuming excitation to the same Franck-Condon (FC) excited state from the ground states is employed. and is the subject of on-going progress.

3.4.3 Reaction Kinetics

The photoisomerization of the cis- and trans-isomers can be modeled using the conﬁgura- tion diagram in Figure 3.16. For the experiments described here, photoisomerization was

∗ induced via direct excitation to the S1 (n-π transition) state in the visible. The trans- and cis-isomers can theoretically excite to different transition states (TS), however, as both processes occur simultaneously and are indistinguishable in the experiments, the model has been simplified to assume a configuration in which both isomers are photo-excited to the same Franck Condon (FC) excited state before relaxing into the metastable TS. From the TS a fraction of the excited molecules revert back to the initial trans form, while the rest undergo photoisomerization to the cis-isomer. In addition, the thermal reverse isomerization takes place from the cis ground state to the trans ground state. In the experiments, the sum of the decays from the excited state, and hence the overall quantum yield for trans-cis isomerization, was measured in the photo-excitation experi- 3.4. PHOTOSWITCHING OF AZOBENZENE MOLECULES 67 ments, while the additional back reaction measurements allow the thermal rate constant to be determined (see Equation 3.7). The reaction kinetics can be described by modeling the number densities in the three states in Figure 3.16 using the following coupled rate equations,

∂nT(t) Ip = − σTnT(t) + ΓCTnTS(t) + knC(t), (3.9) ∂t hνp

∂nC(t) Ip = − σCnC(t) + ΓTCnTS(t) − knC(t), (3.10) ∂t hνp

nT(t) + nTS(t) + nC(t) = 1, (3.11) where nT,C,TS(t) are the normalized number densities in the trans, cis and transition states, Ip and hνp the power density and photon energy of the pump light, σTC and σCT the absorption cross-section at the pump wavelength for excitation from the trans and cis states to the TS, ΓCT is the decay rate from TS to trans and ΓTC is the rate for photoisomerization. The third equation follows from the conservation law. The quantum yield, which describes the fraction of trans excited to TS which are converted to cis upon irradiation, has the form given by Γ Φ = TC . (3.12) ΓTC + ΓCT Under the assumption that the molecules do not spend any time in TS, Equation 3.10 can be re-written as:

∂nC(t) Ip Ip = − σCnC(t) + Φ nT(t) − knC(t). (3.13) ∂t hνp hνp

In the photostationary state, ∂nC(t)/∂t = 0 and the fraction of molecules in the cis form is given by ΦIp/hνp nC = . (3.14) (σT + σC)Ip/hνp + k

It can be seen from Equation 3.14 that in the limit of high irradiance Ip and excitation at the isosbestic point, so that σT = σC, the quantum yield for the forward reaction can be obtained from the fraction of cis-isomer in the photostationary state.

3.4.4 Discussion

The results have demonstrated real-time monitoring of fast, reversible photoswitching processes induced in PCF photoreactors. The strong enhancement of light-matter interaction 68 CHAPTER 3. PHOTOCHEMISTRY IN PCF in the tiny hollow fiber core has led to undesirable effects in which the broadband probe beam also induces photoisomerization in both direction, therefore interfering with the actual pump-probe measurement and hindering the quantitative determination of reaction rate constants. Implementation of the Fischer method to extrapolate the full absorption spectrum of the cis-isomer and integration of the spectrum into the numerical computation of the solution to the coupled rate equations is in progress and should lead to further insight into the reaction kinetics. Furthermore, by incorporating the experiment in a femtosecond spectroscopy setup, the enhanced temporal resolution and the possibility for two-photon photoisomerization could help clarify the on-going debate on the mechanism of isomerization [177, 178]. In addition, the stepwise measurement approach has proven to complicate the analyses of reaction dynamics in which thermally reversible back reactions retard the efficiency and rate of the photo-induced forward reaction. The huge computational time required for modeling the stepwise reaction kinetics can be resolved by implementing the experimental setup in a counter-propagating pump-probe configuration to allow for continuous sample irradiation and spectra collection, and is detailed in Appendix A. Chapter 4

Spectroscopy in PCF

4.1 Introduction

As already addressed in Section 1.5, the advances in PCF design have generated much interest in exploring its use as vehicles for optical sensors, in particular as the post- processing step is no longer required thanks to the fiber microstructure. Additionally, PCF sensors can strongly reduce the sample volume required for measurements and provide the robustness and flexibility needed for fiber sensors. In the context of absorption- based PCF sensor designs, hollow-core PBG-PCF offers an ideal environment for optical spectroscopy by virtue of its ability to maximize light-matter interaction at path lengths that are much longer than achievable in conventional sample cells. However, practical application is limited to narrow-band spectroscopic gas sensing measurements [138, 179], as the transmission bandwidth of the hollow-core PBG-PCF is typically narrower than 100 nm, impeding its competitiveness as liquid-based chemical sensors requiring detection of broad spectral features. In this chapter, a quantitative broadband fiber sensor based on evanescent-wave sensing in the cladding holes of an air-suspended SC-PCF is demonstrated in Section 4.2. As the evanescent-wave sensors preferentially probe surface effects, significant differences between bulk and in-fiber measurements can result. Results from the investigation of surface interactions between the fiber surfaces and the sample are presented and discussed in Section 4.3, concluded with suggestions for further investigative ventures.

69 70 CHAPTER 4. SPECTROSCOPY IN PCF 4.2 Evanescent-Wave Sensing

In order to overcome the limited bandwidth of hollow-core PBG-PCF, sensors based on SC-PCFs have been proposed and demonstrated in the literature [87, 89]. The most common design of an index-guiding PCF consists of a solid silica core surrounded by a periodic array of silica webs and air holes that make up the cladding. The propagating light in the solid core of the index-guiding PCF probes the sample in the cladding holes via an evanescent field. By manipulating the core size and the pitch of the cladding air holes, the amount of evanescent field available can be varied. However, the maximum achievable fraction of power overlap in the cladding holes in these fibers is usually too low (typically 5% or less) to allow ultra-high sensitivity sensing [85, 86]. An index-guiding PCF design that strongly enhances the power overlap in the cladding holes consists of a solid silica rod held in air by three silica nanowebs. This allows direct access to the fiber core for sensing applications [83, 180]. By varying a single structural parameter, the fraction of evanescent field available for sensing can be controlled while maintaining the broad transmission window of silica. Experiments have been reported in which the narrow spectral lines of acetylene were resolved using the evanescent field of light propagating in such fibers [88, 136]; however, sensing experiments in these fibers have only been non-quantitative and limited to a narrow frequency range, failing to exploit the full potential of these fibers. In this section, quantitative detection of an environmen- tally hazardous industrial chemical is demonstrated in these fibers, with the capability of accurately resolving the sub-peaks of the broad absorption spectrum.

4.2.1 Fiber Characteristics

The air-suspended SC-PCFs used in the experiments were fabricated using the conventional stack-and-draw process described in Section 1.4. The preform of the fiber contains only three capillaries. This simple preform illustrates another advantage of this fiber over hollow-core PBG-PCF, the preform of which typically contains over 300 capillaries and rods. During the fiber drawing process, the core size of the fiber, which determines the amount of evanescent field available for sensing, can be controlled via the scale-down ratio of the preform. Using this technique, kilometers of fibers with ten different core diameters 4.2. EVANESCENT-WAVE SENSING 71

Fiber 1 Fiber 2 Fiber 3 Fiber 4

Figure 4.1: High resolution SEM of the core region of four different air-suspended solid- core fibers. The effective core diameter of the fiber is defined as the diameter of the largest circle that can be drawn in the core region. The effective core diameter for the fibers are 0.87 µm, 1.03 µm, 2.32 µm and 2.98 µm for fibers 1 to 4, respectively. The inset shows the hollow cladding region of fiber 2 with a diameter of 64 µm. The thicknesses of the nanowebs that hold the fiber cores in place vary between 160 and 500 nm. in the range of 0.8 to 3.0 µm were fabricated. Figure 4.1 shows typical SEM images of the fibers drawn. Three nanowebs with thicknesses between 160 and 550 nm hold the central silica core in place. The effective core diameter, deff, of the air-suspended SC-PCF is defined as the diameter of the largest circle which can be inscribed in the core region. For fibers 1 to 4 shown in Figure 4.1, the core diameters are 0.873, 1.03, 2.32 and 2.98 µm, respectively. The hollow cladding region (see the inset of fiber 2) acts as an easily accessible sample chamber and has a typical diameter of 30 to 70 µm. The uniformity of the structural parameters along the fiber was verified from high resolution SEM images. For most samples, variations of less than

3.5% over tens to 100 m were detected. The largest variation in deff observed was less than 7%. For the fibers used in the experiments, the observed variations were below 2%. This analysis demonstrates that the stack-and-draw process allows highly reproducible and flexible fabrication of air-suspended SC-PCFs.

4.2.1.1 Transmission and Losses

The transmission and loss spectra are important because they provide guidelines to the wavelength range and maximum ﬁber length that can be used in the sensing experiments, as discussed in Section 2.2. Figure 4.2 (a) shows the broadband transmission windows for 2.9 m of ﬁber 2 (solid curve) with air cladding from λ = 500 to 1350 nm and from λ = 1450 to beyond 1750 nm. The absorption line near λ = 1400 nm is attributed to 72 CHAPTER 4. SPECTROSCOPY IN PCF

(a) 0

-20

-40 Air cladding D O cladding -60 2

Normalized transmission [dB]

18 (b)

Loss [dB/m]

0 600 800 1000 1200 1400 1600 Wavelength [nm]

Figure 4.2: (a) Transmission (normalized to the supercontinuum source) and (b) loss spectra for 2.9 m of fiber 2 with air cladding (solid curves) and 1.0 m of fiber 4 infiltrated with heavy water (D2O, dashed curves). The transmission spectra show that the fibers guide light over a broad wavelength range, allowing sensing measurements between 500 and 1750 nm. The loss spectra for both the air- and D2O-filled fibers are flat over a broad wavelength range, implying that the length of a fiber can be adjusted without changing the shape of the transmission spectrum.

OH− contamination during the fiber drawing process. This absorption can be reduced by drying the silica preform or by pre-treatment with chlorine gas. The dashed curve in Figure 4.2(a) shows the transmission window for 1 m of fiber 4 infiltrated with heavy water

(D2O). The refractive index of D2O is similar to that of H2O. However, all absorption bands are shifted to longer wavelengths due to the almost doubled moment of inertia of

D2O compared to H2O, which reduces the vibrational frequencies by a factor of about √ 2. The measured transmission of the D2O-filled fiber indeed displayed an absorption band at 1600 nm, which accords with the H2O absorption band at 1190 nm, shifted by √ ∼ 2 [181]. The transmission spectra showed that the fibers guide light over a broad wavelength range, allowing sensing measurements between 500 and 1750 nm. 4.2. EVANESCENT-WAVE SENSING 73 l = 500 nm l = 700 nm l = 800 nm l = 975 nm l = 1000 nm 1

measured

0.5

calculated

Figure 4.3: Normalized mode profiles (time-averaged z-component of the Poynting vector, 2 Sz) of fiber 2 with H2O-filled cladding. All images are 2.5 × 2.5 µm . The white curves overlaying the images indicate the contours of the fiber structure obtained from SEM. The images show the measured (top row) and calculated (bottom row) Sz profiles at wavelengths 500, 700, 800, 975 and 1000 nm. The contour lines are 0.1 apart (in normalized units).

A typical loss spectrum for ﬁber 2 with air cladding is shown in Figure 4.2 (b). The spectrum reveals a low-loss region with losses below 0.2 dB/m between λ = 500 and 900 nm. The maximum observed loss within the transmission windows of this ﬁber was

4 dB/m. The loss spectrum was also measured in fiber 4 filled with D2O. The losses are slightly higher than in the unfilled fiber 2 but remain below 3 dB/m over the wavelength range between 500 and 1220 nm. For typical sensing measurements, fiber lengths of less than 20 m suffice. It can therefore be concluded that losses do not limit the performance of these fiber sensors. The loss spectra for both the air- and D2O-filled fibers are flat over a broad wavelength range, which leads to the conclusion that the length of a fiber can be adjusted without changing the shape of the transmission spectrum.

4.2.1.2 Mode Field Distribution

The sensing mechanism in air-suspended SC-PCFs is based on the overlap between the evanescent field of the guided mode and the sample. In quantitative sensing experiments it is essential to know the fraction of power, φ, in the cladding holes that is available for interaction with the sample. Measurements of the mode profiles of the fibers at various wavelengths were taken with a CCD beam profiler (WinCamD-UHR-1310) by imaging 74 CHAPTER 4. SPECTROSCOPY IN PCF the output facet of the fiber onto the CCD with a 60×0.85NA microscope objective. The CCD camera was placed at a distance of about 2 m from the imaging objective to ensure that only the core region was imaged by the beam camera. The scale of the images obtained was calibrated by translating the fiber coupling stage over known distances. The resulting normalized beam profiles for H2O-filled fiber 2 at λ = 500, 700, 800, 975 and 1000 nm are shown in the top row of Figure 4.3. The measured profiles show that the mode is confined to the core region at shorter wavelengths, and extends further into the cladding holes as the wavelength increases, implying that a larger power fraction in the cladding holes is available for sensing. The mode profiles were also calculated using the finite element method (FEM). The calculations were based on contours extracted from SEM images of the measured fibers and consequently do not contain any freely adjustable parameters. The fiber structure is discretized using triangular elements of 0.05 µm in the core region and larger elements in the cladding region to achieve a realistic discretization of the fiber structure. The bottom row of Figure 4.3 shows the calculated time-averaged z-component of the Poynting vector,

Sz, of fiber 2 with H2O-filled cladding at the same wavelengths as the measured beam profiles. The measured and calculated mode profiles are in good agreement, and both display an increase in the amount of evanescent field in the cladding region with increasing wavelength. Some irradiance profiles calculated using the FEM revealed discontinuities on a sub- 100 nm scale across the glass-air (core-cladding) boundary. These discontinuities are attributed to field enhancement effects caused by the discontinuity in the normal component of the electric field given by the ratio between the dielectric constants of the two media [182]. These near-field features do not appear in the measured beam profiles in Figure 4.3 since they have dimensions that are well below the free space diffraction limit of the light. Scanning near-field optical microscopy can be used to resolve such features [182]. The dependence of the calculated power fraction in the cladding holes on the core diameter is shown in Figure 4.4(a) for a H2O-filled fiber at λ = 700 (solid curve) and 1000 nm (dashed curve). The value of φ increases with decreasing deff as more light becomes available in the cladding holes for sensing. Figure 4.4(b) shows the dependence of φ on the 4.2. EVANESCENT-WAVE SENSING 75

40 40 H O cladding 700 nm 2 1000 nm Air cladding H O cladding Core diameter = 1.045 µm 20 2 20

% power in cladding 0 0 0.6 1.2 1.8 2.4 450 650 850 1050 Effective core diameter [µm] Wavelength [nm] (a) (b)

Figure 4.4: (a) Dependence of calculated cladding power fraction in a H2O-filled fiber on effective core diameter at λ = 700 (solid curve) and 1000 nm (dashed curve). The power fraction is shown to increase with decreasing core diameter. The square symbols are power fractions for fiber 2 obtained from the measured mode profiles. (b) Calculated wavelength dependence of cladding power fraction for fiber 2 with both H2O- (solid curve) and air-filled (dashed curve) cladding. Both curves show that φ increases with wavelength. The data points are experimental values obtained from measured mode profiles and are in quantitative agreement with theory (within 3%). cladding medium and wavelength of the light propagating in the core. It is shown that by inserting an aqueous sample into the holes, the field extends further into the cladding due to the decreased index contrast. The power fraction also increases with wavelength, as light with longer wavelengths is less tightly confined to the solid core. The experimental power fractions in the cladding were extracted from the measured mode profiles shown in Figure 4.3. The beam profiles were multiplied with masks generated from the SEM images. The optimum position and orientation of the masks were determined by an automated cross-correlation routine, in which the amount of light in the glass core was optimized. The data points (square symbols) in Figures 4.4(a) and (b) show the resulting measured power fractions for a range of wavelengths and demonstrate quantitative agreement with the calculated power fractions to within 3% over the entire wavelength range.

4.2.1.3 Dispersion

Another important characteristic of the fiber is the fiber dispersion, as shown in Figure 4.5(a). The dispersion for fibers 3 (circles) and 4 (squares) was measured with white light interferometry [183] using the PCF SC source. Results from the measurements were compared to FEM calculations taking into account the dispersion of silica. The calculated 76 CHAPTER 4. SPECTROSCOPY IN PCF ]

-1 100 (a) 900 (b) nm -1 0

km 800 ⋅ Fiber 3 (measured) -100 Fiber 3 (calculated) 700 Fiber 4 (measured) Fiber 4 (calculated) Calculated ZDW [nm] 600 -200 Dispersion [ps 600 800 1000 1200 0.5 1 1.5 2 2.5 3 Wavelength [nm] Effective core diameter [µm]

Figure 4.5: (a) Measured and calculated dispersion of fiber 3 and 4 in the wavelength range between 600 and 1200 nm. The measured data for fiber 3 (circles) and fiber 4 (squares) show a ZDW at 846 and 887 nm, respectively. The solid and dashed curves represent dispersion curves obtained from FEM calculations without free parameters, and agree very well within 2 ps·km−1nm−1 with a polynomial fit (not shown) of the measured data points over the entire wavelength range. The inset shows the dependence of the calculated ZDW on the effective core diameter (diamonds). The dotted line is a linear fit through the calculated data points. dispersion of fibers 3 and 4 between λ = 600 and 1200 nm is shown as the solid and dashed curves in Figure 4.5(a), respectively. The calculated dispersion of both fibers agrees within 2 ps·km−1nm−1 with a polynomial fit (not shown) through the measured data points over the entire wavelength range. The excellent agreement between the measurements and the theory is remarkable since no parameters were freely adjustable, exemplifying the accuracy of the FEM calculations. The region near the zero dispersion wavelength (ZDW) is particularly interesting for nonlinear optical experiments. Fibers 3 and 4 have ZDWs at 846 and 887 nm, respectively. The ZDW can be tailored for nonlinear experiments in either the solid core of the fiber or in the cladding holes. The ZDWs of the fibers are controlled by deff. Figure 4.5(b) shows the dependence of the first calculated ZDW (diamonds) on deff. The first ZDW shifts toward the blue as the core size decreases, in agreement with the silica strand model [184]. Supercontinuum generation has been demonstrated by launching regeneratively amplified Ti:sapphire pulses into a 10 cm long air-suspended solid-core fiber [87]. While the generated SC was 730 nm broad, the pulse irradiance in their experiment exceeded 1 TWcm−2 (assuming a conservatively estimated coupling efficiency of 1%). The high peak irradiance and short fiber length suggest that the ZDW did not play a dominant 4.2. EVANESCENT-WAVE SENSING 77 role in the experiment. Clearly, by carefully tuning the ZDW to lie close to the pump wavelength, the required pulse irradiance for SC generation can be dramatically reduced, allowing the use of standard Ti:sapphire or microchip lasers as pump sources. We propose that such efficient SC generation could be used in single-fiber sensors in which both the SC source and sample chamber are combined. In such systems, the dispersion of the sample should also be taken into account. As an example: from FEM modeling of the dispersion of fiber 3, we have obtained that the ZDW changes from 846 to 1090 nm upon infiltration with water. This redshifted ZDW is close to the wavelength of the Nd:YAG microchip laser (1064 nm), typically used for SC generation in ESM-PCF.

4.2.2 Results

2+ An aqueous NiCl2 solution (in which nickel is present largely as [Ni(H2O)6] ) was chosen as the analyte to demonstrate broadband chemical sensing in the air-suspended SC-PCF.

NiCl2 is a compound that is commonly used for electroplating and also in batteries. It is hazardous for the environment and particularly toxic to aquatic organisms. The

1 −4 LC50/96 h for water organisms is about 100 mg/L, corresponding to c = 4.2 × 10 M.

Unfortunately, eﬃcient monitoring of NiCl2 concentrations is hampered by the low molar 2+ absorptivity of [Ni(H2O)6] . Thus, NiCl2 is a compount highly suitable for testing the performance of the proposed PCF sensor.

In order to compare the measurement to standard spectroscopic techniques, a NiCl2 concentration of 2.1 × 10−2 M was chosen, which is just detectable in a L = 1 cm cuvette measurement. To obtain the reference molar absorptivity spectrum, the transmission of a collimated halogen light source through the cuvette with the sample was measured using a USB spectrometer. The ﬁne curve in Figure 4.6(a) shows the absorbance in the λ =

550 to 875 nm range, obtained by normalizing to the transmission through a H2O-ﬁlled cuvette. The absorbance reaches a maximum value of 0.4 dB at λ = 720 nm. The ﬁne curve in Figure 4.6(b) shows the resulting molar absorptivity spectrum.

2+ The absorption spectrum of [Ni(H2O)6] is known to exhibit three broad absorption bands between 350 and 1400 nm, arising from spin-allowed d-d electronic transitions.

1Lethal Concentration 50: concentration in water having 50% chance of causing death to aquatic life after 96 h exposure. 78 CHAPTER 4. SPECTROSCOPY IN PCF

(a) PCF 2 (b) PCF 4 cuvette ] cuvette -1 cm -1 1 2 ε [Lmol Absorbance [dB] 0 0 550 650 750 850 550 650 750 850 Wavelength [nm] Wavelength [nm]

−2 Figure 4.6: (a) Absorbance spectra of an aqueous 2.1×10 M NiCl2 solution, normalized to H2O reference, measured in a 1.1 m long piece of fiber 2 (thick curve) and in a 1 cm standard cuvette (fine curve). Two subpeaks at 660 and 720 nm could be resolved in both spectra. The absorbance signal measured has strongly increased from 0.4 dB in the cuvette to 4.7 dB in the fiber. (b) Molar absorptivity spectra obtained from the modified Beer-Lambert law, taking into account the power fraction in the fiber cladding. The excellent agreement is striking since no parameters were freely adjusted.

The central absorbance band splits into two maxima at 660 and 720 nm (in accordance with the literature values of 656 and 720 nm [185, 186]), exhibited molar absorptivities of 1.5 and 2.1 Lmol−1cm−1, respectively. This central absorbance corresponds to the

3 3 3 A2g → T1g( F) electronic transition. In the absence of coupling, this would be expected to give a single maximum. However, due to the presence of strongly coupled electronic

3 1 states (in this case T1g and Eg), a superposition of several transitions is detected, giving rise to the two maxima observed [187].

For the fiber-based measurement, fiber 2 with deff = 1.05 µm and φ = 10.4% at

700 nm, the center of the absoprtion band for NiCl2, was chosen. According to the ideal sensing parameter diagram (see Figure 2.2), the optimum fiber length required for a 5 dB absorbance signal is 1.1 m (displayed as a square in Figure 2.2). The fiber was connected to liquid cells and initially filled with de-ionized H2O to obtain a reference spectrum. The sample volume in the fiber (1 µL) is reduced by three orders of magnitude compared to the cuvette measurement (1 mL). The transmission through the fiber was recorded and subsequently, the H2O in the fiber was replaced by an aqueous NiCl2 solution (21 mM). The thick curve in Figure 4.6(a) shows the resulting absorbance spectrum, obtained by normalizing the NiCl2 data to the transmission through the same fiber filled with H2O. The same broad absorption band as that measured in the cuvette was observed between 4.3. MICROSCALE SURFACE CHEMISTRY 79

λ = 600 and 800 nm. Importantly, the fiber spectrum also resolves the two peaks at 660 and 720 nm, illustrating that the fiber does not introduce spectral artifacts. The maximum measured absorbance of 4.7 dB is in good agreement with the prediction. A direct quantitative comparison between the fiber data and the reference sample is made in

Figure 4.6(b). Here, the (λ) spectrum of NiCl2 was extracted by applying the modified Beer-Lambert law on the absorbance data, with φ = 10.4% at 700 nm also taken into account. The striking agreement between the in-fiber measurement (thick curve) and the reference spectrum measured in the the standard cuvette (fine curve) demonstrates that the air-suspended SC-PCF can be used in quantitative broadband chemical-sensing measurements.

4.2.3 Discussion

A quantitative broadband fiber sensor based on evanescent-wave sensing in the cladding holes of an air-suspended SC-PCF has been demonstrated. The measured mode profiles were in good agreement with numerical calculations based on the finite element method made without free parameters. The fraction of light in the hollow cladding can be tuned via the core diameter of the fiber. Dispersion measurements were in excellent agreement with the theory and demonstrated tuning of the zero dispersion wavelength via the core diameter. The applicability of the proposed evanescent-wave PCF sensor was demonstrated by measuring the broad absorption peak of an aqueous NiCl2 solution and showing excellent agreement with the reference spectrum measured in a standard cuvette despite three orders of magnitude lower sample volume used.

4.3 Microscale Surface Chemistry

4.3.1 Self-Aggregation and Photobleaching of Methylene Blue

Methylene blue, MB, is a cationic thiazine dye with a broad spectrum of applications ranging from antidote for cyanide poisoning [188], antiseptic in veterinary medicine, to in vitro diagnostic in biology, cytology, hematology and histology [189, 190]. Furthermore, its photochemical activity is well-established, given its common role as a sensitizer in 80 CHAPTER 4. SPECTROSCOPY IN PCF various areas of photochemistry including photogalvanic cells [191, 192], singlet oxygen production [193] and reductive electron transfer [194], as a result of the relatively long- lived triplet state (450 µs in the triplet state, compared to 30 to 390 ps of the singlet state) and high quantum yield (φT = 0.52) [194, 195]. Like many thiazine dyes, MB readily undergo self-aggregation to form dimers (and higher aggregates) [196] in aqueous solution in spite of like-charge repulsion. The reaction for dimerization takes the following form

KD 2MB )* (MB)2, (4.1) where KD is the equilibrium constant of the dimerization process given by [(MB) ] K = 2 . (4.2) D [MB]2

[MB] and [(MB)2] are the concentrations of the MB monomer and dimer, and KD is reported in literature to vary between 2000 and 6000 Lmol−1 [197-200], the variation in the values obtained was mainly due to the diﬀerences between the actual experimental conditions under which the experiments were performed, such as the temperature and the pH of the buﬀer. The main forces responsible for the aggregation of the dye molecules are hydrogen bonding, van der Waals forces and the predominant force due to hydrophobic interactions (i.e. water acts as a catalyst in inducing aggregation) [201]. We assumed that dimerization is the only self-aggregation process MB can undergo under the experimental conditions here. Therefore [MB]total = [MB] + 2[(MB)2], and from Equation 4.2 it follows that the concentration of the monomer, [MB], can be obtained by solving the following quadratic equation,

2 2KD[MB] + [MB] − [MB]total = 0. (4.3)

Solutions of methylene blue in de-ionized water were prepared with concentrations ranging from 0.415 to 20.8 µM by serial dilutions from a stock solution. From Equation 4.2

−1 and taking the median of KD = 3000 Lmol reported in various literature sources [197- 200], approximately 87% of the 20.8 µM sample is in its monomer form. Figure 4.7(a) shows the molar absorptivity spectrum of a 20.8 µM MB sample in water, measured in a standard 1 cm cuvette. The absorption peaks at λ = 612 and 665 nm have the coeﬃcients = 2.46 × 104 and 4.48 × 104 Lmol−1cm−1, respectively. The peak near λ = 660 nm is attributed to the monomer, while the peak near λ = 610 nm is due to the dimer. 4.3. MICROSCALE SURFACE CHEMISTRY 81

4.8 24 ] (a) (b)

-1 t = 0

cm 3.2 16 -1 3 min. 20.8 μM 2.08 μM Lmol

4 1.6 8 5 min. 6 min. ε[×10 0 0 480 560 640 720 800 480 560 640 720 800 Wavelength [nm] Wavelength [nm] 42 120 ] (c) (d) -1 t = 0

cm 10 min. t = 0 28 80 -1 1.04 μM 4 min. 0.415 μM

Lmol 13 min. 4 14 40 8 min. 16 min. ε[×10 12 min. 0 0 480 560 640 720 800 480 560 640 720 800 Wavelength [nm] Wavelength [nm]

Figure 4.7: (a) Molar absorptivity spectra of methylene blue in water, measured in a standard 1 cm cuvette. For a 20.8 µM sample (solid curve), the absorption peak near 660 nm is attributed to the monomer, while the peak near 610 nm is due to the dimer. (b) Photobleaching of MB (2.08 µM) in suspended solid-core fiber (fiber 2, 107 cm) induced by irradiation using the broadband PCF SC source. Absorption of the sample in the wavelength range of 480 to 800 nm completely vanished after 10 minutes of irradiation with an average irradiance ∼ 100 kWcm−2. Subsequent measurements using (c) 1.04 and (d) 0.415 µM of MB in 117 and 100 cm of fiber 2 demonstrated similar photobleaching effects. All measurements were performed until absorption in the wavelength range of 480 to 800 nm has completely vanished.

The dye can undergo photoreduction process by visible light [201], resulting in the doubly reduced form of MB, leuco-MB (LMB), which is colorless. This photobleaching phenomenon of MB from its bright blue color is due to the covalent modification of the molecule upon excitation from an excited singlet state to the excited triplet state. Exper- iments with the aim of monitoring the photoreduction process in the PCF photochemical reactor were performed using fiber 2 of the air-suspended SC-PCF shown in Figure 4.1. Figure 4.7(b) shows the changes in the molar absorptivity spectrum of a 2.08 µM MB sample in 107 cm of fiber 2, as a result of photobleaching induced by irradiation using the broadband PCF SC source. It was observed that the absorption band in the wavelength range of 480 to 800 nm completely vanished after 10 minutes of irradiation with an 82 CHAPTER 4. SPECTROSCOPY IN PCF average optical power of several mW, corresponding to an irradiance of ∼ 100 kW/cm2. Subsequent measurements, shown in Figures 4.7(c) and (d), using 1.04 and 0.415 µM of MB in 117 and 100 cm of fiber 2, demonstrated similar photobleaching effects. All measurements were performed until the absorption band in the wavelength range of 480 to 800 nm has completely vanished. A much lower power was used initially for optimization of fiber alignment for all three measurements to avoid photobleaching effects. Guidance of the fundamental mode in the solid fiber core was confirmed for the wavelength range of interest before each measurement. The higher-than-expected molar extinction coefficients obtained in all three in-fiber measurements (compared to that obtained in the bulk measurement in cuvette as shown in Figure 4.7(a)) suggest aggregation of the dye molecules on the silica surface, leading to the higher apparent concentration of the sample. Note that while no quantitative comparison between the initial spectra can be made for the different sample concentrations due to the different initial infiltration conditions (namely the duration of infiltration before the first “t = 0” spectrum was taken), further exami- nation of the spectra revealed that the absorption peak near 610 nm increases (relative to the amplitude of the sub-peak near 640 nm) with increasing sample concentration, indicating the increase in the formation of dimer aggregates on the silica surfaces. Such surface-induced aggregation of the dye molecules have also been observed and reported in a microstructured-core fiber for evanescent-wave sensing [87]. Further insight into the extent of adsorption of MB on the fused silica nanowebs can be obtained by deducing the surface density of the sample. The absorption coefficient, α(λ), is defined by the fraction of total optical power absorbed per unit length, and can be written in the form of H C ρSσ(λ)I(r) · dr α(λ) = R , (4.4) S I(r)ds where ρS is the surface number density of adsorbed molecules, σ(λ) is the absorption cross-section of one molecule and I(r) is the the irradiance of the light propagating in the core of the fiber at the position vector r on the contour C. The concept of the calculation is as depicted in the inset of Figure 4.8. In the numerator of Equation 4.4, the absorbed optical power is obtained by performing closed line integral along the contours of all the cladding holes immediately adjacent to the core of the fiber. The total power is evaluated in the denominator by integrating I(r) across the entire cross-section of the fiber, 4.3. MICROSCALE SURFACE CHEMISTRY 83 1.2

] 1 2

0.9 mol/cm

-12 dr 0.5 0.6

C 0.3 0 Surface density [×10 0 0 3 6 9 12 15 18 Exposure time [minutes]

Figure 4.8: The variation in the calculated total surface density of MB along the inner surface for the photoreduction of 0.415 µM of MB in 100 cm of fiber 2 with irradiation time. Approximately 6% of the molescules that passed through the fiber are estimated to have remained in the fiber due to adsorption. The dotted curve is a fitted Gaussian curve with a decay constant of 0.27 ± 0.0019 min−1. The concept of the closed contour line integral performed in the calculations can be visualized in the inset; the contour lines for Sz are 0.1 apart in normalized units. including the cladding region. I(r) was obtained from the irradiance profile calculated using FEM for the entire cross-section of the fiber, while the absorption cross-sections σ(λ) and attenuation coefficients α(λ) were obtained from the experimentally measured data shown in Figures 4.7(a) and (d), respectively. The calculated surface density of MB along the inner surface of the fiber cladding holes is shown in Figure 4.8 for the photoreduction of 0.415 M of MB in 100 cm of fiber 2 with irradiation time (see Figure 4.7(d)). As shown, the calculated surface density for the initial measurement at t = 0 is 10−12 mol/cm2. The fiber used was 1 m of fiber 2 with a cladding diameter of 64 µm, which corresponds to a total inner fiber surface area of approximately 4 cm2. Therefore approximately 4 ×10−12 mol of MB molecules were adsorbed onto the inner fiber surface, with each molecule occupying approximately 13 nm2 of the inner fiber surface. For comparison, the three-dimensional molecular size of MB is 1.43 nm × 0.61 nm × 0.4 nm [202]. In addition, the fiber was infiltrated at a rate of approximately 1 mLh−1 for at least 10 minutes before the first spectrum (the t = 0 spectrum) was taken. One can therefore assume that at least 0.17 mL of the 84 CHAPTER 4. SPECTROSCOPY IN PCF sample has passed through the fiber. This corresponds to approximately 7 ×10−11 mol of total MB molecues that have “seen” the surface of the fiber. From the 4 ×10−12 mol of MB molecules which were adsorbed onto the inner fiber surface, one can estimate that approximately 6% of the molecules that passed through the fiber have remained in the fiber. It is therefore possible to quantify the decrease in the surface density of MB due to photoreduction as a function of exposure time. The MB was determined to reduce to the colorless LMB form (undetectable via absorption spectroscopy within the operating wavelength range) at a rate of 0.27 ± 0.0019 min−1. Further quantitative analysis of the adsorption dynamics would require controlled infiltration conditions. However, the preliminary result has clearly demonstrated the affinity of the MB molecules to adsorb onto the fiber surfaces. In order to further investigate the affinity of the MB molecules to adsorb on the silica surfaces, the photobleaching experiments were performed in a hollow-core kagome PCF. Figure 4.9(a) shows the photobleaching of a 20.8 µM MB sample in ∼ 30 cm of the kagome fiber induced by irradiation using the broadband PCF SC source. The measured spectral shape agrees with that of the same sample measured in the cuvette (solid curve in Figure 4.7(a)). Initial inspection of the result reveals that the molar absorptivity of the dye molecule measured in the HC-PCF is lower than that measured in bulk. This suggests that some molecules may have moved out of the active sensing region of the fiber core and adsorbed onto the surface surrounding the core where they are only very weakly detectable by light. Figure 4.9(b) shows the variation in the molar absorptivity spectrum demonstrating effect of adsorption in 36 cm of the kagome fiber. The spectrum taken after the initial infiltration of 0.415 µM of MB sample (at t = 0, curve 1) shows approximately 78% lower absorption at the absorption peak than expected from bulk measurement in the cuvette. This indicates that at least 78% of the dye molecules were adsorbed on the silica surface surrounding the hollow fiber core, where the irradiance of the core mode is low and could not detect the sample efficiently. The high concentration of adsorbed molecules is a result of the large surface-to-volume ratio (which is inversely proportional to the core diameter and in this case ∼ 105 m−1, three orders of magnitude higher than the conventional 1 cm cuvette) provided by the microstructured fiber. Continuous infiltration 4.3. MICROSCALE SURFACE CHEMISTRY 85

9 42 ] 6 min. 1: t = 0 1 (a) (b)

- 3 24 min. 2: 29 min. cm 2 1 6 28 3: 33 min. - 4: 88 min. irrad. 4 Lmol

3 3 14 1 ε[×10 0 0 550 600 650 700 550 600 650 700 Wavelength [nm] Wavelength [nm] 90 33 ] First infiltration 1: First infiltration 1 (c) (d)

- 3 Second infiltration 2: 4 min. irrad.

cm 1

1 60 22 3: Second infiltration - 2 Lmol

3 30 11 ε[×10 0 0 550 600 650 700 550 600 650 700 Wavelength [nm] Wavelength [nm]

Figure 4.9: (a) Photobleaching of MB (20.8 µM) in kagome HC-PCF (∼ 107 cm) induced by irradiation using the broadband PCF SC source. (b) Increase in the molar absorptivity spectrum as a result of continuous infiltration of MB (0.415 µM) in 36 cm of the kagome PCF (curves 1 to 3). Irradiation of the sample without the infiltration of new sample into the fiber displayed photobleaching effect similar to that observed in the previous experiments (curve 4). (c) Multiple discontinuous infiltration of MB (0.415 µM) in 34.5 cm of the kagome PCF showed an increase in the measured molar absorptivity beyond that measured in the bulk. (d) Multiple infiltration and photobleaching experiments performed in 50 cm of silanized kagome PCF. The measurements showed a molar absorptivity spectrum (initially lower than the bulk values) that increased with infiltration attempts, and decreased upon irradiation without infiltration of new sample into the fiber, displaying similar dynamics of the results compared to that of the non-silanized fibers. of the sample through the fiber revealed an increase in the measured absorption, and the expected value (bulk value measured in the cuvette) was obtained after 33 minutes of continuous infiltration, the spectral shape of which agreed with that measured in the cuvette. It is postulated that the continuous infiltration of the sample had eventually saturated the inner silica surface of the hollow fiber core with the dye molecules. A coating was effectively formed on the inner surface of the fiber core and forced the newly- infiltrated molecules to the active sensing region defined by the mode propagating in the fiber core. The PCF SC source remained on during the 33 minutes of infiltration at a low average power level of 17 µW, corresponding to an irradiance level of 2.4 W/cm2 86 CHAPTER 4. SPECTROSCOPY IN PCF

(four orders of magnitude lower than that used in the suspended SC-PCF), to prevent possible/significant counter-effect of photobleaching on the absorption spectra measured. The infiltration of the sample was subsequently stopped while the SC source was left on to irradiated the sample in the fiber for 88 minutes. The spectrum obtained after 88 minutes of irradiation showed that photobleaching took place at a much reduced rate compared to those performed in the air-suspended SC-PCF shown in Figures 4.7(b) to (d). As the rate of photoreduction is independent of the concentration of the solution, the reduced photobleaching rate observed in the HC-PCF can be attributed to the much lower power level used to induce the reduction. A separate experiment in which infiltration of the 0.415 µM MB sample through 34.5 cm of the kagome fiber was performed in a stop-and-go configuration (i.e. the sample infiltration was stopped for an undefined period of time during which the absorption spectrum was recorded, after which the infiltration was restarted). The results shown in Figure 4.9(c) showed that the measured molar absorptivity increased with increasing infiltration attempts. This increase in absorption, and hence the apparent sample concentration within the fiber core, was possibly due to the physisorbed dye molecules coming off the surfaces in the subsequent infiltration, which allowed them to become free and detectable in the core volume again. Fringe-like features were also observable in both spectra in Figure 4.9(c). The spectral position of these features coincide on both spectra, and were at a spacing of 3.66 nm. The irradiance profile of the beam exiting the fiber revealed a double-lobe structure, indicating the propagation of higher order modes in the fiber core. Using the following equation for the propagation constants of straight dielectric guides [103]: ( ) 2π 1 u λ2 β = 1 − nm (4.5) nm λ 2 2πa where unm is the mth root of the equation Jn−1(unm) = 0, λ is the wavelength and a is the radius of the guide. The propagation constants, and therefore the beat length, for the EH11 and TE01 modes can be calculated to be 3.47 nm, which is close to the experimentally observed value. Consequently, intermodal dispersion can give rise to the fringes observed. In order to combat the undesirable effects of dye molecule adsorption on the silica sur- 4.3. MICROSCALE SURFACE CHEMISTRY 87 face surrounding the hollow fiber core, surface-treated fiber samples were prepared by infiltrating the fiber with a 2% v/v solution of dimethyldichlorosilane in 1,1,1-trichloroethane and left to stand for several hours to achieve hydrophobic surfaces. The fiber was rinsed with isopropyl alcohol and then dried by purging with air before use. The photobleaching and multiple sample infiltration experiments were repeated in 50 cm of the silanized fiber. Figure 4.9(d) shows that irradiation of the sample (without infiltration of new sample into the fiber) decreased the absorption signal (curve 1→2), while multiple infiltration of the 0.415 µM MB sample increased the molar absorptivity towards the expected value measured in bulk, as indicated by the progression from curve 2 to 3. The results displayed similar dynamics to that of the unsilanized fiber, indicating that the initial attempt in surface treament of the fiber was unsuccessful. Several ideas have been recently proposed to reduce or completely eliminate the effect of adsorption on the silica surface: as the polarity of the aqueous sample and the surface is the key factor, it has been suggested that the interaction potential between the molecules and the silica surface can be tuned by replacing water with an inorganic salt as the buffer solution.

4.3.2 Discussion

Efficient adsorption of methylene blue dye molecules on the inner silica surfaces was observed. Opposite effects on the magnitude of the measured molar absorptivity spectra were demonstrated in the solid and hollow-core of two types of PCF. In particular, an index- guiding PCF design with moderate evanescent field extending into the cladding region can provide novel platforms for interrogating surface-bound molecules on the nanoscale. De- velopment of such devices necessitates further quantitative analyses of the adsorption and desorption processes, including well-controlled sample infiltration and residence time. It is instructive to perform bulk measurements using planar fused silica substrate as reference for comparison with the in-fiber measurements. It has been demonstrated that several phenothiazines such as methylene blue were able to form oligomeric tau species upon binding to inhibit neurodegeneration in vitro [203, 204]. The development of methylene blue as an Alzheimer drug to prevent the aggregation of tau and amyloid proteins can be efficiently studied in PCF. Furthermore, the possibility to use methylene blue to generate singlet oxygen which then photochemically changes the 88 CHAPTER 4. SPECTROSCOPY IN PCF vitamin B12 [205] can aid towards optimization of photoactivated anticancer drugs. Chapter 5

Conclusions and Outlook

The enhanced light-matter interaction made possible by the photonic crystal fiber offers a multitude of applications in various branches of chemistry. In particular, the novel demonstration of utilizing the photonic crystal fiber as a highly efficient photochemical reactor opens up new regimes of “doing chemistry”, many of which may one day find their permanent place in common measurement equipments or as routine laboratory techniques. Here we discuss possible applications of the work presented in this thesis.

5.1 Optical Tweezers and Photodynamic Therapy

The work on the photoaquation of vitamin B12 presented in Chapter 3 was intended to study the feasibility of efficiently inducing and monitoring photochemical processes in the photonic crystal fiber chemical reactor. These reactions are similar to that of the photoactivated anticancer drugs currently under intensive research and development. The success in this initial case study suggests that the photonic crystal fiber is a well-suited candidate for testing the effectiveness of these drugs. The next stage would involve testing these drugs on cancer cells in a well-controlled environment. Recently, precise control over a particle against fluidic counterflow was demonstrated in a liquid-filled photonic bandgap fiber [162]. By combining a photonic crystal fiber chemical reactor with an in-fiber cell guidance setup, the optical and mechanical response of a cancer cell to the synthetic drugs under test can immediately be obtained via a range of sensing modalities available for photonic crystal fiber sensors.

89 90 CHAPTER 5. CONCLUSIONS AND OUTLOOK 5.2 Microﬂuidic Flow Reactor

The small dimensions of the microfluidic channels in the fiber’s microstructure provide unique opportunities for the implementation of photonic crystal fibers as miniature flow reactors. Such flow reactors would allow continuous on-line optimization of the exposure conditions and reagent parameters in a reaction. In addition, most biochemical experiments are performed in fluidic environments. The exploitation of photonic crystal fibers as optofluidic devices offers significant advantages including minimal consumption of reagents and mechanical flexibility. Furthermore, recent experiments have shown that photonic crystal fibers can be integrated into existing planar microfluidic circuitry [206].

5.3 Mass Spectrometry

Mass spectrometry is a common analytical technique used in the laboratories to determine the elemental composition of a sample, and is an integral routine used in chemical synthesis, including the development of therapeutical anticancer complexes. It is therefore of great interest to incorporate microfluidic photonic crystal fiber circuitry in a mass spectrometer setup to accommodate direct measurements of the products of photochemical reactions. Such a setup could also allow for the quantitative spectroscopic assay of reaction products and monitoring of the reaction kinetics. Preliminary results from a separate project have already confirmed the feasibility of such a setup configuration.

5.4 Surface Chemistry Using Higher-Order Modes

The experiments on the self-aggregation and photobleaching of methylene blue in Chap- ter 4 have demonstrated that the reaction kinetics of molecules with high affinity to adsorb onto the fiber surfaces can be efficiently studied by taking the advantage of the high surface-to-volume ratio offered by the microstructure of the photonic crystal fibers. A well- defined optical mode propagating in the fiber is capable of detecting minute changes in the composition of the sample via, for example, absorption spectroscopy. It has been demonstrated that higher-order modes propagating in hollow-core photonic crystal fibers can be selectively excited by using holograms electronically generated by a spatial light modu- 5.5. FINAL REMARKS 91 lator [64]. By careful excitation of both the fundamental mode and a surface-sensitive higher order mode in a hollow-core photonic crystal fiber, it is possible to measure the local concentration gradients across the core volume of the fiber.

5.5 Final Remarks

The examples outlined above give a clear indication to the breadth of research possibilities in photonic crystal fibers, utilizing the unprecedented interplay between light and matter within the fiber’s microstructure. The novelty in the development of photonic crystal fiber lies in its cross-disciplinary applications, bridging different fields in science and technology. With no foreseeable limitation to the future prospects in sight, the continual development in the novel applications of photonic crystal fiber devices is to be anticipated.

Appendix A

Counter-Propagating Pump-Probe Setup

In Section 3.4 the HC-PCF photochemical reactor was demonstrated to track the fast and reversible photoswitching process of an azobenzene derivative. While the experiments were able to show the complete reversibility of the photochemical reaction, quantitative analyses of the reaction dynamics has proven to be a computation intensive task. There are two interconnecting factors contributing to the need for tedious numerical computation. First, the wavelengths of the excitation sources used to induce the photochemical reactions were chosen to coincide with the main absorption feature of interest, so that efficient photochemical conversion can be achieved. However, as the pump beam is at a much higher irradiance than the broadband probe beam used to measure the absorption spectra, the excitation signal would completely saturate the USB spectrometer (which is a grating spectrometer based on CCD arrays). The use of a narrow band-pass filter at the entrance of the spectrometer could help solve the issue of the limited dynamic range provided by the spectrometer, however, the filter would also completely block out signals within the wavelength range of interest. It was therefore proposed that a modified pump-probe setup with counter-propagating beams be implemented for the PCF-based photochemical reactors. A schematic diagram is shown in Figure A.1 depicting the main configuration of the setup. In this configuration, the broadband probe beam is the only beam propagating along the path towards the spectrometer. The excitation light still propagates through the entire length of the

93 94 APPENDIX A. COUNTER-PROPAGATING PUMP-PROBE SETUP

ESM 10x 20x Excitation PCF

Broad- ESM CCD 10x 20x band PCF BS2 BS1 MMF Spectro- 10x Computer CCD meter BS1 BS2 Sample fiber 4x 4x

Figure A.1: Schematic diagram of the modified pump-probe setup with counter- propagating beams for PCF photochemical reactors. sample fiber, however, only along the opposite direction to the propagating probe beam. Both the pump and probe beams are spatially filtered by a separate ESM-PCF to ensure optimum coupling efficiency into the sample fiber, with the aid from CCD beam cameras on both sides of the setup. The beam splitters BS1 have a splitting ratio of 50:50, while the beam splitters BS2 which are placed in front of the CCD cameras have splitting ratios of 92:8, as only a small fraction of the light is required for imaging. Preliminary implementation of this setup configuration revealed that upon incidence of the pump beam on the uncoated liquid cell window, strong reflection results which propagate in the reverse direction (i.e., towards the spectrometer), and “aligns” itself efficiently into the multimode fiber connected to the spectrometer. A further improvement has thus been made to the setup, by introducing a 10◦ wedge in the liquid cells to divert any reflection off the liquid cell window, as shown in the setup schematic. As the incoming beam is no longer at normal incidence to the liquid cell window, it is instructive to determine whether refraction of the broadband source at different wavelengths would hamper the coupling efficiency into the fiber. Consider the schematic

◦ diagram in Figure A.2 showing the glass window at 10 tilt, with n1, n2 and n3 denoting the regions of air, window and liquid, respectively. By applying Snell’s law for an incoming beam at 10◦ incidence, and using the dispersion relation for fused silica to obtain the refractive indices of fused silica at λ = 380 and 750 nm (i.e. the visible wavelength 95 d

d·tanq d·tanq 750 q750 380 q380 10°

nair nwindow

Figure A.2: Schematic diagram showing the eﬀect of refraction due to the tilted liquid cell window. Diagram exaggerated and not to scale.

range), nfused silica(380 nm) = 1.4725 and nfused silica(750 nm) = 1.4542. The refracted angles are calculated to be at 2.1611◦ and 2.1883◦. For window thicknesses ranging from 0.08 to 1 mm used in the experiments, the separation of the longer and shorter wavelength components varies between 38 to 475 nm and should pose no problem in the coupling of the broadband source into the hollow ﬁber core.

List of Publications

1. J. S. Y. Chen, T. G. Euser, N. J. Farrer, P. J. Sadler, M. Scharrer, and P. St.J. Russell, “Photochemistry in Photonic Crystal Fiber Nanoreactors,” Chemistry - A European Journal, 16(19), 5607-5612 (2010).

2. T. G. Euser, M. K. Garbos, J. S. Y. Chen, and P. St.J. Russell, “Precise balancing of viscous and radiation forces on a particle in a liquid-ﬁlled photonic bandgap ﬁber,” Optics Letters, 34(23), 3674-3676 (2009).

3. T. G. Euser, G. Whyte, M. Scharrer, J. S. Y. Chen, A. Abdolvand, J. Nold, C. F. Kaminski, and P. St.J. Russell, “Dynamic control of higher-order modes in hollow- core photonic crystal ﬁbers,” Optics Express 16(22), 17972-17981 (2008).

4. T. G. Euser, J. S. Y. Chen, M. Scharrer, P. St.J. Russell, N. J. Farrer, and P. J. Sadler, “Quantitative broadband chemical sensing in air-suspended solid-core ﬁbers,” Journal of Applied Physics 103, 103108 (2008).

5. J. S. Y. Chen, T. G. Euser, G. O. S. Williams, A. C. Jones, and P. St.J. Russell, “Photoswitching in Photonic Crystal Fiber,” in Advanced Photonics: OSA Optics & Photonics Congress (EurOPC), SThB3. Karlsruhe, Germany. 21 - 24 June 2010.

6. J. Chen, A. Hangauer, R. Strzoda, T. G. Euser, J. S. Y. Chen, M. Scharrer, P. St.J. Russell, and M. Amann, “Sensitivity Limits for Near- Infrared Gas Sensing with Suspended-core PCFs directly coupled with VCSELs,” in Conference on Lasers and Electro-Optics (CLEO), JThB7. San Jose, USA. 16 - 21 May 2010.

7. J. Chen, A. Hangauer, R. Strzoda, M. Amann, T. Euser, J. S. Y. Chen, M. Schar- rer, P. Russell, “Near-infrared gas sensing using hollow waveguides and PCFs di-

97 98 LIST OF PUBLICATIONS

rectly coupled to VCSELs,” in Field Laser Applications in Industry and Research (FLAIR). Grainau, Germany. 6 - 11 September 2009.

8. J. S. Y. Chen, T. G. Euser, N. J. Farrer, P. J. Sadler, and P. St.J. Russell, “Photo- chemistry in photonic crystal ﬁbers,” in European Conference on Lasers and Electro- Optics (CLEO-Europe), CH1.3. Munich, Germany. 14 - 19 June 2009.

9. M. K. Garbos, T. G. Euser, J. S. Y. Chen, and P. St.J. Russell, “Controlled particle guidance in a liquid-ﬁlled single-mode hollow-core photonic crystal ﬁber,” in Optical Trapping Applications (OTA), OMA6. Vancouver, Canada. 26 - 30 April 2009.

10. T. G. Euser, J. S. Y. Chen, M. Scharrer, and P. St.J. Russell, “Quantitative broadband chemical sensing in air-suspended solid-core ﬁbers,” in Conference on Lasers and Electro-Optics (CLEO), CMZ6. San Jose, USA. 4 - 9 May 2008. Bibliography

[1] J. D. Loudin, D. M. Simanovskii, K. Vijayraghavan, C. K. Sramek, A. F. Butterwick, P. Huie, G. Y. McLean and D. V. Palanker, “Optoelectronic retinal prosthesis: system design and performance,” Journal of Neural Engineering 4, S72-S84 (2007).

[2] J. K. Niparko, Cochlear Implants: Principles & Practices, Second ed. (Lippincott Williams & Wilkins, Philadelphia, 2009).

[3] J. W. Gardner and P. N. Bartlett, Electronic Noses: Principles and Applications, First ed. (Oxford University Press, New York, 1999).

[4] F. Winquist, P. Wide and I. Lundstr¨om,“An electronic tongue based on voltam- metry,” Analytica Chimica Acta 357, 21-31 (1997).

[5] R. W. Cattrall, Chemical Sensors, Illustrated ed. (Oxford University Press, New York, 1997).

[6] G. Orellana and M. C. Moreno-Bondi, Frontiers in chemical sensors: novel principles and techniques, (Springer, Berlin, 2005).

[7] M. Zourob and A. Lakhtakia, Optical Guided-Wave Chemical and Biosensors I, (Springer, Berlin, 2010).

[8] M. Zourob and A. Lakhtakia, Optical Guided-Wave Chemical and Biosensors II, (Springer, Berlin, 2010).

[9] S.-S. Kiwanuka, T. Laurila and C. F. Kaminski, “Sensitive Method for the Kinetic Measurement of Trace Species in Liquids Using Cavity Enhanced Absorption Spec- troscopy with Broad Bandwidth Supercontinuum Radiation,” Analytical Chemistry 82, 7498-7501 (2010).

99 100 BIBLIOGRAPHY

[10] T. Laurila, I. S. Burns, J. Hult, J. H. Miller and C. F. Kaminski, “A calibration method for broad-bandwidth cavity enhanced absorption spectroscopy performed with supercontinuum radiation,” Applied Physics B 1-8 (2010).

[11] K. Chan, H. Ito and H. Inaba, “An optical-ﬁber-based gas sensor for remote absorption measurement of low-level CH4 gas in the near-infrared region,” Journal of Lightwave Technology 2, 234-237 (1984).

[12] G. Stewart, W. Jin and B. Culshaw, “Prospects for ﬁbre-optic evanescent-ﬁeld gas sensors using absorption in the near-infrared,” Sensors and Actuators B: Chemical 38, 42-47 (1997).

[13] O. S. Wolfbeis, “Fiber-Optic Chemical Sensors and Biosensors,” Analytical Chem- istry 76, 3269-3284 (2004).

[14] O. S. Wolfbeis and H. E. Posch, “Fibre-optic ﬂuorescing sensor for ammonia,” An- alytica Chimica Acta 185, 321-327 (1986).

[15] S. L. R. Barker, R. Kopelman, T. E. Meyer and M. A. Cusanovich, “Fiber-Optic Nitric Oxide-Selective Biosensors and Nanosensors,” Analytical Chemistry 70, 971- 976 (1998).

[16] B. G. Healey, L. Li and D. R. Walt, “Multianalyte biosensors on optical imaging bundles,” Biosensors and Bioelectronics 12, 521-529 (1997).

[17] J. A. Ferguson, T. C. Boles, C. P. Adams and D. R. Walt, “A ﬁber-optic DNA biosensor microarray for the analysis of gene expression,” Nature Biotechnology 14, 1681-1684 (1996).

[18] J. M. L´opez-Higuera, Handbook of Optical Fibre Sensing Technology, (Wiley, Chich- ester, 2002).

[19] A. M. Dietrich, J. N. Jensen and W. F. da Costa, “Chemical Species,” Water Environment Research 68, 391-406 (1996).

[20] B. Mizaikoff, K. Taga and R. Kellner, “Infrared fiber optic gas sensor for chloroflu- orohydrocarbons,” Vibrational Spectroscopy 8, 103-108 (1995). BIBLIOGRAPHY 101

[21] G. Schweizer, I. Latka, H. Lehmann and R. Willsch, “Optical sensing of hydro- carbons in air or in water using UV absorption in the evanescent ﬁeld of ﬁbers,” Sensors and Actuators B: Chemical 38, 150-153 (1997).

[22] M. A. El-Sherif, “Smart Textiles Created with Embedded Sensors,” MRS Bulletin Technology Advances 28, 101-102 (2003).

[23] P. Russell, “Photonic Crystal Fibers,” Science 299, 358-362 (2003).

[24] P. S. J. Russell, “Photonic-Crystal Fibers,” Journal of Lightwave Technology 24, 4729-4749 (2006).

[25] K. C. Kao and G. A. Hockham, “Dielectric-ﬁbre surface waveguides for optical frequencies,” Proceedings of the IEEE 113, 1151-1158 (1966).

[26] J. C. Knight, T. A. Birks, P. S. J. Russell and D. M. Atkin, “All-silica single-mode optical ﬁber with photonic crystal cladding,” Optics Letters 21, 1547-1549 (1996).

[27] J. C. Knight, T. A. Birks, P. S. J. Russell and D. M. Atkin, “All-silica single-mode optical ﬁber with photonic crystal cladding: errata,” Optics Letters 22, 484-485 (1997).

[28] P. Vukusic and J. R. Sambles, “Photonic structures in biology,” Nature 424, 852-855 (2003).

[29] V. N. Melekhin and A. B. Manenkov, “Dielectric tube as a low-loss waveguide,” Zhurnal Tekhnicheskoi Fiziki 38, 2113-2115 (1968).

[30] P. Yeh, A. Yariv and E. Marom, “Theory of Bragg ﬁber,” Journal of the Optical Society of America 68, 1196-1201 (1978).

[31] E. Yablonovitch, “Inhibited Spontaneous Emission in Solid-State Physics and Elec- tronics,” Physical Review Letters 58, 2059-2059 (1987).

[32] E. Yablonovitch, “Photonic band-gap structures,” Journal of the Optical Society of America B 10, 283-295 (1993). 102 BIBLIOGRAPHY

[33] S. John, “Strong localization of photons in certain disordered dielectric superlat- tices,” Physical Review Letters 58, 2486-2486 (1987).

[34] T. A. Birks, J. C. Knight and P. S. J. Russell, “Endlessly single-mode photonic crystal ﬁber,” Optics Letters 22, 961-963 (1997).

[35] J. C. Knight, T. A. Birks, R. F. Cregan, P. S. J. Russell and P. D. de Sandro, “Large mode area photonic crystal ﬁbre,” Electronics Letters 34, 1347-1348 (1998).

[36] R. F. Cregan, B. J. Mangan, J. C. Knight, T. A. Birks, P. S. J. Russell, P. J. Roberts and D. C. Allan, “Single-Mode Photonic Band Gap Guidance of Light in Air,” Science 285, 1537-1539 (1999).

[37] J. K. Ranka, R. S. Windeler and A. J. Stentz, “Visible continuum generation in air- silica microstructure optical ﬁbers with anomalous dispersion at 800 nm,” Optics Letters 25, 25-27 (2000).

[38] J. E. Sharping, M. Fiorentino, A. Coker, P. Kumar and R. S. Windeler, “Four-wave mixing in microstructure ﬁber,” Optics Letters 26, 1048-1050 (2001).

[39] F. Benabid, J. Knight and P. Russell, “Particle levitation and guidance in hollow- core photonic crystal ﬁber,” Optics Express 10, 1195-1203 (2002).

[40] V. V. R. K. Kumar, A. George, J. Knight and P. Russell, “Tellurite photonic crystal ﬁber,” Optics Express 11, 2641-2645 (2003).

[41] A. Argyros, T. Birks, S. Leon-Saval, C. M. Cordeiro, F. Luan and P. S. J. Russell, “Photonic bandgap with an index step of one percent,” Optics Express 13, 309-314 (2005).

[42] K. Kurokawa, K. Nakajima, K. Tsujikawa, T. Yamamoto and K. Tajima, “Ultra- Wideband Transmission Over Low Loss PCF,” Journal of Lightwave Technology 27, 1653-1662 (2009).

[43] J. Folkenberg, M. Nielsen, N. Mortensen, C. Jakobsen and H. Simonsen, “Polar- ization maintaining large mode area photonic crystal ﬁber,” Optics Express 12, 956-960 (2004). BIBLIOGRAPHY 103

[44] J. R. Folkenberg, M. D. Nielsen and C. Jakobsen, “Broadband single-polarization photonic crystal ﬁber,” Optics Letters 30, 1446-1448 (2005).

[45] A. Ortigosa-Blanch, J. C. Knight, W. J. Wadsworth, J. Arriaga, B. J. Mangan, T. A. Birks and P. S. J. Russell, “Highly birefringent photonic crystal ﬁbers,” Optics Letters 25, 1325-1327 (2000).

[46] J. C. Knight, J. Arriaga, T. A. Birks, A. Ortigosa-Blanch, W. J. Wadsworth and P. S. J. Russell, “Anomalous dispersion in photonic crystal ﬁber,” Photonics Tech- nology Letters, IEEE 12, 807-809 (2000).

[47] A. Ferrando, E. Silvestre, J. J. Miret, J. A. Monsoriu, M. V. Andres and P. S. J. Russell, “Designing a photonic crystal ﬁbre with ﬂattened chromatic dispersion,” Electronics Letters 35, 325-327 (1999).

[48] A. Ferrando, E. Silvestre, J. J. Miret and P. Andrés,“Nearly zero ultraflattened dispersion in photonic crystal fibers,” Optics Letters 25, 790-792 (2000).

[49] A. Ferrando, E. Silvestre, P. Andres, J. Miret and M. Andres, “Designing the properties of dispersion-ﬂattened photonic crystal ﬁbers,” Optics Express 9, 687-697 (2001).

[50] W. Reeves, J. Knight, P. Russell and P. Roberts, “Demonstration of ultra-ﬂattened dispersion in photonic crystal ﬁbers,” Optics Express 10, 609-613 (2002).

[51] A. Cerqueira Sodre, J. M. Chavez Boggio, A. A. Rieznik, H. E. Hernandez-Figueroa, H. L. Fragnito and J. C. Knight, “Highly eﬃcient generation of broadband cascaded four-wave mixing products,” Optics Express 16, 2816-2828 (2008).

[52] W. Wadsworth, N. Joly, J. Knight, T. Birks, F. Biancalana and P. Russell, “Super- continuum and four-wave mixing with Q-switched pulses in endlessly single-mode photonic crystal ﬁbres,” Optics Express 12, 299-309 (2004).

[53] J. K. Ranka, R. S. Windeler and A. J. Stentz, “Optical properties of high-delta air silica microstructure optical ﬁbers,” Optics Letters 25, 796-798 (2000). 104 BIBLIOGRAPHY

[54] J. C. Travers, J. M. Stone, A. B. Rulkov, B. A. Cumberland, A. K. George, S. V. Popov, J. C. Knight and J. R. Taylor, “Optical pulse compression in dispersion decreasing photonic crystal ﬁber,” Optics Express 15, 13203-13211 (2007).

[55] W. J. Wadsworth, A. Ortigosa-Blanch, J. C. Knight, T. A. Birks, T. P. M. Man and P. S. J. Russell, “Supercontinuum generation in photonic crystal ﬁbers and optical ﬁber tapers: a novel light source,” Journal of the Optical Society of America B 19, 2148-2155 (2002).

[56] T. M. Monro, Y. D. West, D. W. Hewak, N. G. R. Broderick and D. J. Richardson, “Chalcogenide holey ﬁbres,” Electronics Letters 36, 1998-2000 (2000).

[57] J. Y. Y. Leong, P. Petropoulos, J. H. V. Price, H. Ebendorff-Heidepriem, S. Asi- makis, R. C. Moore, K. E. Frampton, V. Finazzi, X. Feng, T. M. Monro and D. J. Richardson, “High-nonlinearity dispersion-shifted lead-silicate holey fibers for efficient 1-µm pumped supercontinuum generation,” Lightwave Technology, Journal of 24, 183-190 (2006).

[58] H. Ebendorﬀ-Heidepriem, P. Petropoulos, S. Asimakis, V. Finazzi, R. Moore, K. Frampton, F. Koizumi, D. Richardson and T. Monro, “Bismuth glass holey ﬁbers with high nonlinearity,” Optics Express 12, 5082-5087 (2004).

[59] V. V. R. Kumar, A. George, W. Reeves, J. Knight, P. Russell, F. Omenetto and A. Taylor, “Extruded soft glass photonic crystal ﬁber for ultrabroad supercontinuum generation,” Optics Express 10, 1520-1525 (2002).

[60] J. A. Harrington and C. C. Gregory, “Hollow sapphire ﬁbers for the delivery of CO2 laser energy,” Optics Letters 15, 541-543 (1990).

1 [61] T. Hidaka, “GeO2-ZnO-K2O glass as the cladding material of 940-cm CO2 laser- light transmitting hollow-core waveguide,” Journal of Applied Physics 53, 5484-5484 (1982).

[62] T. Abel, J. Hirsch and J. A. Harrington, “Hollow glass waveguides for broadband infrared transmission,” Optics Letters 19, 1034-1036 (1994). BIBLIOGRAPHY 105

[63] J. Dai and J. A. Harrington, “High-peak-power, pulsed CO2 laser light delivery by hollow glass waveguides,” Applied Optics 36, 5072-5077 (1997).

[64] T. G. Euser, G. Whyte, M. Scharrer, J. S. Y. Chen, A. Abdolvand, J. Nold, C. F. Kaminski and P. S. J. Russell, “Dynamic control of higher-order modes inhollow- core photonic crystal ﬁbers,” Optics Express 16, 17972-17981 (2008).

[65] M. N. Petrovich, F. Poletti, A. van Brakel and D. J. Richardson, “Robustly single mode hollow core photonic bandgap ﬁber,” Optics Express 16, 4337-4346 (2008).

[66] P. Roberts, F. Couny, H. Sabert, B. Mangan, D. Williams, L. Farr, M. Mason, A. Tomlinson, T. Birks, J. Knight and P. St. J. Russell, “Ultimate low loss of hollow- core photonic crystal ﬁbres,” Optics Express 13, 236-244 (2005).

[67] B. J. Mangan, L. Farr, A. Langford, P. J. Roberts, D. P. Williams, F. Couny, M. Lawman, M. Mason, S. Coupland, R. Flea, H. Sabert, T. A. Birks, J. C. Knight and P. S. J. Russell, “Low loss (1.7 dB/km) hollow core photonic bandgap ﬁber,” in Optical Fiber Communication Conference and Exposition (OFC), PD24. Los Angeles, USA. 22-27 February 2004.

[68] T. D. Hedley, D. M. Bird, F. Benabid, J. C. Knight and P. S. J. Russell, “Modelling of a novel hollow-core photonic crystal ﬁbre,” in Quantum Electronics and Laser Science (QELS), 2. Baltimore, USA. 1-6 June 2003.

[69] G. J. Pearce, G. S. Wiederhecker, C. G. Poulton, S. Burger and P. S. J. Russell, “Models for guidance in kagome-structured hollow-core photonic crystal ﬁbres,” Optics Express 15, 12680-12685 (2007).

[70] A. Argyros and J. Pla, “Hollow-core polymer ﬁbres with a kagome lattice: potential for transmission in the infrared,” Optics Express 15, 7713-7719 (2007).

[71] F. Couny, F. Benabid and P. S. Light, “Large-pitch kagome-structured hollow-core photonic crystal ﬁber,” Optics Letters 31, 3574-3576 (2006).

[72] Y. Y. Wang, F. Couny, P. J. Roberts and F. Benabid, “Low loss broadband transmission in optimized core-shape Kagome hollow-core PCF,” in Conference on Lasers and Electro-Optics (CLEO), CPDB4. San Jose, USA. 18-20 May 2010. 106 BIBLIOGRAPHY

[73] Y. Fink, D. J. Ripin, S. Fan, C. Chen, J. D. Joannepoulos and E. L. Thomas, “Guiding optical light in air using an all-dielectric structure,” Journal of Lightwave Technology 17, 2039-2041 (1999).

[74] D. Torres, O. Weisberg, G. Shapira, C. Anastassiou, B. Temelkuran, M. Shurgalin, S. A. Jacobs, R. U. Ahmad, T. Wang, U. Kolodny, S. M. Shapshay, Z. Wang, A. K. Devaiah, U. D. Upadhyay and J. A. Koufman, “OmniGuide photonic bandgap

ﬁbers for ﬂexible delivery of CO2 laser energy for laryngeal and airway surgery,” Proceedings of SPIE 5686, 310-321 (2005).

[75] X. Feng, A. K. Mairaj, D. W. Hewak and T. M. Monro, “Towards high-index glass based monomode holey ﬁbre with large mode area,” Electronics Letters 40, 167-169 (2004).

[76] A. Mori, K. Shikano, K. Enbutsu, K. Oikawa, K. Naganuma, M. Kato and S. Aozasa, “1.5 µm band zero-dispersion shifted tellurite photonic crystal ﬁbre with a nonlinear coeﬃcient γ of 675 W−1km−1,” in European Conference on Optical Communication (ECOC), TH3.3.6. Stockholm, Sweden. 5-9 September 2004.

[77] M. Kanso, S. Cuenot and G. Louarn, “Sensitivity of Optical Fiber Sensor Based on Surface Plasmon Resonance: Modeling and Experiments,” Plasmonics 3, 49-57 (2008).

[78] Z. Zhang, P. Zhao, F. Sun, G. Xiao and Y. Wu, “Self-Referencing in Optical- Fiber Surface Plasmon Resonance Sensors,” Photonics Technology Letters, IEEE 19, 1958-1960 (2007).

[79] S.-M. Tseng and C.-L. Chen, “Side-polished ﬁbers,” Applied Optics 31, 3438-3447 (1992).

[80] S. M. Chandani and N. A. F. Jaeger, “Fiber-optic temperature sensor using evanescent ﬁelds in D ﬁbers,” Photonics Technology Letters, IEEE 17, 2706-2708 (2005).

[81] J. Villatoro, D. Monz´on-Hern´andezand E. Mej´ıa, “Fabrication and Modeling of Uniform-Waist Single-Mode Tapered Optical Fiber Sensors,” Applied Optics 42, 2278-2283 (2003). BIBLIOGRAPHY 107

[82] D. Monzón-Hernándezand J. Villatoro, “High-resolution refractive index sensing by means of a multiple-peak surface plasmon resonance optical fiber sensor,” Sensors and Actuators B: Chemical 115, 227-231 (2006).

[83] T. M. Monro, D. J. Richardson and P. J. Bennett, “Developing holey ﬁbres for evanescent ﬁeld devices,” Electronics Letters 35, 1188-1189 (1999).

[84] T. M. Monro, W. Belardi, K. Furusawa, J. C. Baggett, N. G. R. Broderick and D. J. Richardson, “Sensing with microstructured optical ﬁbres,” Measurement Science and Technology 12, 854-854 (2001).

[85] Y. L. Hoo, W. Jin, H. L. Ho, D. N. Wang and R. S. Windele, “Evanescent wave gas sensing using microstructure ﬁbre,” in Paciﬁc Rim Conference on Lasers and Electro-Optics (CLEO/PACIFIC RIM), 8-9. Chiba, Japan. 15-19 July 2001.

[86] Y. L. Hoo, W. Jin, H. L. Ho, D. N. Wang and R. S. Windeler, “Evanescent-wave gas sensing using microstructure ﬁber,” Optical Engineering 41, 8-9 (2002).

[87] C. M. B. Cordeiro, M. A. R. Franco, G. Chesini, E. C. S. Barretto, R. Lwin, C. H. Brito Cruz and M. C. J. Large, “Microstructured-core optical ﬁbre for evanescent sensing applications,” Optics Express 14, 13056-13066 (2006).

[88] A. S. Webb, F. Poletti, D. J. Richardson and J. K. Sahu, “Suspended-core holey ﬁber for evanescent-ﬁeld sensing,” Optical Engineering 46, 010503-010503 (2007).

[89] J. B. Jensen, L. H. Pedersen, P. E. Høiby, L. B. Nielsen, T. P. Hansen, J. R. Folkenberg, J. Riishede, D. Noordegraaf, K. Nielsen, A. Carlsen and A. Bjarklev, “Photonic crystal ﬁber based evanescent-wave sensor for detection ofbiomolecules in aqueous solutions,” Optics Letters 29, 1974-1976 (2004).

[90] M. T. Myaing, J. Y. Ye, T. B. Norris, T. Thomas, Jr. Baker, W. J. Wadsworth, G. Bouwmans, J. C. Knight and P. S. J. Russell, “Enhanced two-photon biosensing with double-clad photonic crystal ﬁbers,” Optics Letters 28, 1224-1226 (2003).

[91] S. Afshar V, S. C. Warren-Smith and T. M. Monro, “Enhancement of ﬂuorescence- based sensing using microstructured optical ﬁbres,” Optics Express 15, 17891-17901 (2007). 108 BIBLIOGRAPHY

[92] L. Rindorf and O. Bang, “Highly sensitive refractometer with a photonic-crystal- ﬁber long-period grating,” Optics Letters 33, 563-565 (2008).

[93] L. Rindorf, J. B. Jensen, M. Dufva, L. H. Pedersen, P. E. Høiby and O. Bang, “Pho- tonic crystal ﬁber long-period gratings for biochemical sensing,” Optics Express 14, 8224-8231 (2006).

[94] Y.-G. Han, S. Song, G. H. Kim, K. Lee, S. B. Lee, J. H. Lee, C. H. Jeong, C. H. Oh and H. J. Kang, “Simultaneous independent measurement of strain and temperature based on long-period ﬁber gratings inscribed in holey ﬁbers depending on air-hole size,” Optics Letters 32, 2245-2247 (2007).

[95] G. Kakarantzas, A. Ortigosa-Blanch, T. A. Birks, P. S. J. Russell, L. Farr, F. Couny and B. J. Mangan, “Structural rocking ﬁlters in highly birefringent photonic crystal ﬁber,” Optics Letters 28, 158-160 (2003).

[96] G. Statkiewicz-Barabach, A. Anuszkiewicz, W. Urbanczyk and J. Wojcik, “Sensing characteristics of rocking ﬁlter fabricated in microstructured birefringent ﬁber using fusion arc splicer,” Optics Express 16, 17249-17257 (2008).

[97] S. Smolka, M. Barth and O. Benson, “Highly efficient fluorescence sensing with hollow core photonic crystal fibers,” Optics Express 15, 12783-12791 (2007).

[98] T. Ritari, J. Tuominen, H. Ludvigsen, J. Petersen, T. Sørensen, T. Hansen and H. Simonsen, “Gas sensing using air-guiding photonic bandgap ﬁbers,” Optics Express 12, 4080-4087 (2004).

[99] D. Passaro, M. Foroni, F. Poli, A. Cucinotta, S. Selleri, J. Lægsgaard and A. O. Bjarklev, “All-Silica Hollow-Core Microstructured Bragg Fibers for Biosensor Ap- plication,” Sensors Journal, IEEE 8, 1280-1286 (2008).

[100] Q. Shi, Z. Wang, L. Jin, Y. Li, H. Zhang, F. Lu, G. Kai and X. Dong, “A Hollow- Core Photonic Crystal Fiber Cavity Based Multiplexed FabryP´erotInterferometric Strain Sensor System,” Photonics Technology Letters, IEEE 20, 1329-1331 (2008). BIBLIOGRAPHY 109

[101] A. Hassani and M. Skorobogatiy, “Design of the microstructured optical ﬁber-based surface plasmon resonance sensors with enhanced microﬂuidics,” Optics Express 14, 11616-11621 (2006).

[102] B. Gauvreau, A. Hassani, M. Fassi Fehri, A. Kabashin and M. A. Skorobogatiy, “Photonic bandgap ﬁber-based Surface Plasmon Resonance sensors,” Optics Ex- press 15, 11413-11426 (2007).

[103] E. A. J. Marcatili and R. A. Schmeltzer, “Hollow Metallic and Dieletric Waveguides for Long Distance Optical Transmission and Lasers,” Bell System Technical Journal 43, 1783-1809 (1964).

[104] A. Datta, I.-Y. Eom, A. Dhar, P. Kuban, R. Manor, I. Ahmad, S. Gangopadhyay, T. Dallas, M. Holtz, H. Temkin and P. K. Dasgupta, “Microfabrication and characterization of teﬂon AF-coated liquid core waveguide channels in silicon,” Sensors Journal, IEEE 3, 788-795 (2003).

[105] M. Mohebbi, R. Fedosejevs, V. Gopal and J. A. Harrington, “Silver-coated hollow- glass waveguide for applications at 800 nm,” Applied Optics 41, 7031-7035 (2002).

[106] G. Antonopoulos, F. Benabid, T. A. Birks, D. M. Bird, J. C. Knight and P. S. J. Russell, “Experimental demonstration of the frequency shift of bandgaps in photonic crystal ﬁbers due to refractive index scaling,” Optics Express 14, 3000-3006 (2006).

[107] K. Nielsen, D. Noordegraaf, T. Sørensen, A. Bjarklev and T. P. Hansen, “Selective ﬁlling of photonic crystal ﬁbres,” Journal of Optics A: Pure and Applied Optics 7, L13-L20 (2005).

[108] J. M. Fini, “Microstructure ﬁbres for optical sensing in gases and liquids,” Mea- surement Science and Technology 15, 1120-1128 (2004).

[109] C. B. Schaﬀer, Interaction of femtosecond laser pulses with transparent materials,(Harvard University, 2001).

[110] N. Bloembergen, “Laser-induced electric breakdown in solids,” Quantum Electron- ics, IEEE Journal of 10, 375-386 (1974). 110 BIBLIOGRAPHY

[111] M. H. Niemz, “Threshold dependence of laser-induced optical breakdown on pulse duration,” Applied Physics Letters 66, 1181-1181 (1995).

[112] M. Lenzner, J. Krger, S. Sartania, Z. Cheng, C. Spielmann, G. Mourou, W. Kautek and F. Krausz, “Femtosecond Optical Breakdown in Dielectrics,” Physical Review Letters 80, 4076-4076 (1998).

[113] C. F. Bohren and D. R. Huﬀman, Absorption and scattering of light by small particles, (Wiley, Weinheim, 1983).

[114] S. Maruo, O. Nakamura and S. Kawata, “Three-dimensional microfabrication with two-photon-absorbed photopolymerization,” Optics Letters 22, 132-134 (1997).

[115] S. Kawata, H.-B. Sun, T. Tanaka and K. Takada, “Finer features for functional microdevices,” Nature 412, 697-698 (2001).

[116] B. H. Cumpston, S. P. Ananthavel, S. Barlow, D. L. Dyer, J. E. Ehrlich, L. L. Erskine, A. A. Heikal, S. M. Kuebler, I. Y. S. Lee, D. McCord-Maughon, J. Qin, H. Rockel, M. Rumi, X.-L. Wu, S. R. Marder and J. W. Perry, “Two-photon polymerization initiators for three-dimensional optical data storage and microfabrication,” Nature 398, 51-54 (1999).

[117] A. Korkin and F. Rosei, Nanoelectronics and photonics: from atoms to materials, devices, and architectures, 1 ed. (Springer, New York, 2009).

[118] H. W. Lee, M. A. Schmidt, H. K. Tyagi, L. P. Sempere and P. S. J. Russell, “Polarization-dependent coupling to plasmon modes on submicron gold wire in photonic crystal ﬁber,” Applied Physics Letters 93, 111102-111102 (2008).

[119] H. K. Tyagi, M. A. Schmidt, L. Prill Sempere and P. S. Russell, “Optical properties of photonic crystal ﬁber with integral micron-sized Ge wire,” Optics Express 16, 17227-17236 (2008).

[120] K. Lai, S. G. Leon-Saval, A. Witkowska, W. J. Wadsworth and T. A. Birks, “Wavelength-independent all-ﬁber mode converters,” Optics Letters 32, 328-330 (2007). BIBLIOGRAPHY 111

[121] C. M. B. Cordeiro, C. J. S. de Matos, E. M. dos Santos, A. Bozolan, J. S. K. Ong, T. Facincani, G. Chesini, A. R. Vaz and C. H. Brito Cruz, “Towards practical liquid and gas sensing with photonic crystal fibres: side access to the fibre microstructure and single-mode liquid-core fibre,” Measurement Science and Technology 18, 3075- 3081 (2007).

[122] C. Martelli, P. Olivero, J. Canning, N. Groothoﬀ, B. Gibson and S. Huntington, “Micromachining structured optical ﬁbers using focused ion beam milling,” Optics Letters 32, 1575-1577 (2007).

[123] N. Y. Joly, T. A. Birks, A. Yulin, J. C. Knight and P. S. J. Russell, “Linear and nonlinear guidance in an ultralow loss planar glass membrane,” Optics Letters 30, 2469-2471 (2005).

[124] G. Stochel, A. Wanat, E. Kulis and Z. Stasicka, “Light and metal complexes in medicine,” Coordination Chemistry Reviews 171, 203-220 (1998).

[125] Z. Huang, “A review of progress in clinical photodynamic therapy,” Technology in Cancer Research & Treatment 4, 283-293 (2005).

[126] N. J. Farrer, J. A. Woods, V. P. Munk, F. S. Mackay and P. J. Sadler, “Photocyto- toxic trans-Diam(m)ine Platinum(IV) Diazido Complexes More Potent than Their cis Isomers,” Chemical Research in Toxicology 23, 413-421 (2010).

[127] F. S. Mackay, J. A. Woods, P. Heringová,J. Kaˇspárková,A. M. Pizarro, S. A. Moggach, S. Parsons, V. Brabec and P. J. Sadler, “A potent cytotoxic photoactivated platinum complex,” Proceedings of the National Academy of Sciences 104, 20743-20748 (2007).

[128] N. Hoﬀmann, “Photochemical Reactions as Key Steps in Organic Synthesis,” Chem- ical Reviews 108, 1052-1103 (2008).

[129] J. Iriondo-Alberdi and M. F. Greaney, “Photocycloaddition in natural product synthesis,” European Journal of Organic Chemistry 2007, 4801-4815 (2007).

[130] M. Gr¨atzel,“Solar Energy Conversion by Dye-Sensitized Photovoltaic Cells,” Inor- ganic Chemistry 44, 6841-6851 (2005). 112 BIBLIOGRAPHY

[131] N. Robertson, “Catching the rainbow: Light harvesting in dye-sensitized solar cells,” Angewandte Chemie International Edition 47, 1012-1014 (2008).

[132] A. Huebner, S. Sharma, M. Srisa-Art, F. Hollfelder, J. B. Edel and A. J. deMello, “Microdroplets: A sea of applications?,” Lab on a Chip 8, 1244-1254 (2008).

[133] C. Monat, P. Domachuk and B. J. Eggleton, “Integrated optoﬂuidics: A new river of light,” Nat Photon 1, 106-114 (2007).

[134] P. Anzenbacher and M. A. Palacios, “Polymer nanoﬁbre junctions of attolitre volume serve as zeptomole-scale chemical reactors,” Nat Chem 1, 80-86 (2009).

[135] S. M. Matthews, A. D. Elder, K. Yunus, C. F. Kaminski, C. M. Brennan and A. C. Fisher, “Quantitative Kinetic Analysis in a Microﬂuidic Device Using Frequency- Domain Fluorescence Lifetime Imaging,” Analytical Chemistry 79, 4101-4109 (2007).

[136] C. M. Cordeiro, M. A. Franco, C. J. Matos, F. Sircilli, V. A. Serro and C. H. Cruz, “Single-design-parameter microstructured optical ﬁber for chromatic dispersion tailoring and evanescent ﬁeld enhancement,” Optics Letters 32, 3324-3326 (2007).

[137] T. G. Euser, J. S. Y. Chen, M. Scharrer, P. S. J. Russell, N. J. Farrer and P. J. Sadler, “Quantitative broadband chemical sensing in air-suspended solid-core ﬁbers,” Jour- nal of Applied Physics 103, 103108-103108 (2008).

[138] L. W. Kornaszewski, N. Gayraud, J. M. Stone, W. N. MacPherson, A. K. George, J. C. Knight, D. P. Hand and D. T. Reid, “Mid-infrared methane detection in a photonic bandgap ﬁber using a broadband optical parametric oscillator,” Optics Express 15, 11219-11224 (2007).

[139] D. Marcuse, Principles of Optical Fiber Measurements, (Academic Press, New York, 1981).

[140] H. El-Kashef, “Study of the refractive properties of laser dye solvents: toluene, carbon disulphide, chloroform, and benzene,” Optical Materials 20, 81-86 (2002). BIBLIOGRAPHY 113

[141] N. A. Mortensen, J. R. Folkenberg, M. D. Nielsen and K. P. Hansen, “Modal cut- oﬀ and the V parameter in photonic crystal ﬁbers,” Optics Letters 28, 1879-1881 (2003).

[142] M. D. Nielsen, N. A. Mortensen, J. R. Folkenberg and A. Bjarklev, “Mode-ﬁeld radius of photonic crystal ﬁbers expressed by the V parameter,” Optics Letters 28, 2309-2311 (2003).

[143] Z. H. Siddik, “Cisplatin: mode of cytotoxic action and molecular basis of resistance,” Oncogene 22, 7265-7279 (2003).

[144] W. M. Sharman, C. M. Allen and J. E. van Lier, “Photodynamic therapeutics: basic principles and clinical applications,” Drug Discovery Today 4, 507-517 (1999).

[145] S. B. Brown, E. A. Brown and I. Walker, “The present and future role of photodynamic therapy in cancer treatment,” The Lancet Oncology 5, 497-508 (2004).

[146] S. H. Ibbotson, H. Moseley, L. Brancaleon, M. Padgett, M. O’Dwyer, J. A. Woods, A. Lesar, C. Goodman and J. Ferguson, “Photodynamic therapy in dermatology: Dundee clinical and research experience,” Photodiagnosis and Photodynamic Ther- apy 1, 211-223 (2004).

[147] R. A. Hsi, D. I. Rosenthal and E. Glatstein, “Photodynamic therapy in the treatment of cancer: current state of the art,” Drugs 57, 725-734 (1999).

[148] A. L. Harris, “Hypoxia - a key regulatory factor in tumour growth,” Nat Rev Cancer 2, 38-47 (2002).

[149] A. Vogler, R. Hirschmann, H. Otto and H. Kunkely, “Photochemistry of Biologi- cally Important Transition Metal Complexes. I. Cyanocobalamin and Related Cor- rin Complexes of Rhodium(III),” Berichte der Bunsengesellschaft fr physikalische Chemie 80, 420-424 (1976).

[150] I. Ahmad, W. Hussain and A. A. Fareedi, “Photolysis of cyanocobalamin in aqueous solution,” Journal of Pharmaceutical and Biomedical Analysis 10, 9-15 (1992). 114 BIBLIOGRAPHY

[151] K. L. Brown, “Chemistry and Enzymology of Vitamin B12,” Chemical Reviews 105, 2075-2150 (2005).

[152] J. M. Pratt, Inorganic Chemistry of Vitamin B12, (Academic Press Inc, New York, 1972).

[153] T. A. Stich, A. J. Brooks, N. R. Buan and T. C. Brunold, “Spectroscopic and Computational Studies of Co3+-Corrinoids: Spectral and Electronic Properties of

the B12 Cofactors and Biologically Relevant Precursors,” Journal of the American Chemical Society 125, 5897-5914 (2003).

[154] R. A. Firth, H. A. O. Hill, J. M. Pratt, R. J. P. Williams and W. R. Jackson,

“The Circular Dichroism and Absorption Spectra of Some Vitamin B12 Derivatives,” Biochemistry 6, 2178-2189 (1967).

[155] T. Andruniow, P. M. Kozlowski and M. Z. Zgierski, “Theoretical analysis of elec-

tronic absorption spectra of vitamin B12 models,” The Journal of Chemical Physics 115, 7522-7522 (2001).

[156] J. A. Hill, J. M. Pratt and R. J. P. Williams, “The corphyrins,” Journal of Theo- retical Biology 3, 423-445 (1962).

[157] K. L. Brown, J. M. Hakimi and D. W. Jacobsen, “Heteronuclear NMR studies of cobalamins. 3. Phosphorus-31 NMR of aquocobalamin and various organocobal- amins,” Journal of the American Chemical Society 106, 7894-7899 (1984).

[158] K. L. Brown and J. M. Hakimi, “Heteronuclear NMR studies of cobalamins. 2. Carbon-13 and phosphorus-31 NMR studies of [13C]cyanocobalamin,” Inorganic Chemistry 23, 1756-1764 (1984).

[159] J. M. Pratt, “The chemistry of vitamin B12. Part II. Photochemical reactions,” Journal of the Chemical Society 5154-5154 (1964).

[160] J. J. Shiang, A. G. Cole, R. J. Sension, K. Hang, Y. Weng, J. S. Trommel, L. G.

Marzilli and T. Lian, “Ultrafast Excited-State Dynamics in Vitamin B12 and Related Cob(III)alamins,” Journal of the American Chemical Society 128, 801-808 (2006). BIBLIOGRAPHY 115

[161] D. A. Harris, A. B. Stickrath, E. C. Carroll and R. J. Sension, “Inﬂuence of Envi- ronment on the Electronic Structure of Cob(III)alamins: Time-Resolved Absorption Studies of the S1 State Spectrum and Dynamics,” Journal of the American Chemical Society 129, 7578-7585 (2007).

[162] T. G. Euser, M. K. Garbos, J. S. Y. Chen and P. S. J. Russell, “Precise balancing of viscous and radiation forces on a particle in liquid-ﬁlled photonic bandgap ﬁber,” Optics Letters 34, 3674-3676 (2009).

[163] R. M. Christie, Colour Chemistry, (Royal Society of Chemistry, Cambridge, 2001).

[164] H. Zollinger, Color Chemistry, (Wiley-VCH, Weinheim, 2001).

[165] F. Lagugn´eLabarthet, J. L. Bruneel, T. Buﬀeteau and C. Sourisseau, “Chromophore Orientations upon Irradiation in Gratings Inscribed on Azo-Dye Polymer Films: A Combined AFM and Confocal Raman Microscopic Study,” The Journal of Physical Chemistry B 108, 6949-6960 (2004).

[166] C. B. McArdle, Side chain liquid crystal polymers, (Chapman and Hall, New York, 1989).

[167] T. Todorov, L. Nikolova and N. Tomova, “Polarization holography. 1: A new high- eﬃciency organic material with reversible photoinduced birefringence,” Applied Op- tics 23, 4309-4312 (1984).

[168] D. M. Burland, R. D. Miller and C. A. Walsh, “Second-order nonlinearity in poled- polymer systems,” Chemical Reviews 94, 31-75 (1994).

[169] L. R. Dalton, A. W. Harper, R. Ghosn, W. H. Steier, M. Ziari, H. Fetterman, Y. Shi, R. V. Mustacich, A. K. Y. Jen and K. J. Shea, “Synthesis and Processing of Improved Organic Second-Order Nonlinear Optical Materials for Applications in Photonics,” Chemistry of Materials 7, 1060-1081 (1995).

[170] S. K. Yesodha, C. K. Sadashiva Pillai and N. Tsutsumi, “Stable polymeric materials for nonlinear optics: a review based on azobenzene systems,” Progress in Polymer Science 29, 45-74 (2004). 116 BIBLIOGRAPHY

[171] G. F¨uchsel, T. Klamroth, J. Doki´cand P. Saalfrank, “On the Electronic Structure of Neutral and Ionic Azobenzenes and Their Possible Role as Surface Mounted Molecular Switches,” The Journal of Physical Chemistry B 110, 16337-16345 (2006).

[172] M. J. Comstock, N. Levy, A. Kirakosian, J. Cho, F. Lauterwasser, J. H. Harvey, D. A. Strubbe, J. M. J. Frchet, D. Trauner, S. G. Louie and M. F. Crommie, “Reversible Photomechanical Switching of Individual Engineered Molecules at a Metallic Surface,” Physical Review Letters 99, 038301-038301 (2007).

[173] T. Naito, K. Horie and I. Mita, “The Eﬀect of Polymer Rigidity on Photoisomeriza- tion of 4-Dimethylamino-4-nitroazobenzene,” Polymer Journal 23, 809-813 (1991).

[174] M. Poprawa-Smoluch, J. Baggerman, H. Zhang, H. P. A. Maas, L. De Cola and A. M. Brouwer, “Photoisomerization of Disperse Red 1 Studied with Transient Absorption Spectroscopy and Quantum Chemical Calculations,” The Journal of Physical Chemistry A 110, 11926-11937 (2006).

[175] S. R. Hair, G. A. Taylor and L. W. Schultz, “An easily implemented ﬂash photolysis experiment for the physical chemistry laboratory: The isomerization of 4-anilino- 4’-nitroazobenzene,” Journal of Chemical Education 67, 709-709 (1990).

[176] E. Fischer, “Calculation of photostationary states in systems A )* B when only A is known,” The Journal of Physical Chemistry 71, 3704-3706 (1967).

[177] T. Asano and T. Okada, “Thermal Z-E isomerization of azobenzenes. The pressure, solvent, and substituent eﬀects,” The Journal of Organic Chemistry 49, 4387-4391 (1984).

[178] J. Dokić,M. Gothe, J. Wirth, M. V. Peters, J. Schwarz, S. Hecht and P. Saal- frank, “Quantum Chemical Investigation of Thermal Cis-to-Trans Isomerization of Azobenzene Derivatives: Substituent Effects, Solvent Effects, and Comparison to Experimental Data,” The Journal of Physical Chemistry A 113, 6763-6773 (2009).

[179] F. Benabid, F. Couny, J. C. Knight, T. A. Birks and P. S. J. Russell, “Compact, stable and efficient all-fibre gas cells using hollow-core photonic crystal fibres,” Nature 434, 488-491 (2005). BIBLIOGRAPHY 117

[180] K. M. Kiang, K. Frampton, T. M. Monro, R. Moore, J. Tucknott, D. W. Hewak, D. J. Richardson and H. N. Rutt, “Extruded singlemode non-silica glass holey optical ﬁbres,” Electronics Letters 38, 546-547 (2002).

[181] G. M. Hale and M. R. Querry, “Optical Constants of Water in the 200-nm to 200-µm Wavelength Region,” Applied Optics 12, 555-563 (1973).

[182] G. S. Wiederhecker, C. M. B. Cordeiro, F. Couny, F. Benabid, S. A. Maier, J. C. Knight, C. H. B. Cruz and H. L. Fragnito, “Field enhancement within an optical ﬁbre with a subwavelength air core,” Nature Photonics 1, 115-118 (2007).

[183] S. C. Rashleigh and R. Ulrich, “Polarization mode dispersion in single-mode ﬁbers,” Optics Letters 3, 60-62 (1978).

[184] T. A. Birks, D. Mogilevtsev, J. C. Knight and P. St. J. Russell, “Dispersion com- pensation using single-material ﬁbers,” Photonics Technology Letters, IEEE 11, 674-676 (1999).

[185] H. C. Egghart, “A study of nickel-azido complex formation in aqueous solution,” Journal of Inorganic and Nuclear Chemistry 31, 1538-1541 (1969).

[186] G. Bussi`ereand C. Reber, “Coupled Excited States in Nickel(II) Complexes Probed by Polarized Absorption Spectroscopy,” Journal of the American Chemical Society 120, 6306-6315 (1998).

[187] M. Triest, G. Bussi`ere,H. B´elisleand C. Reber, “Why Does the Middle Band in

2+ the Absorption Spectrum of Ni(H2O)6 Have Two Maxima?,” Journal of Chemical Education 77, 670-670 (2000).

[188] M. M. Brooks, “Methylene Blue as an Antidote for Cyanide and Carbon Monoxide Poisoning,” The Scientiﬁc Monthly 43, 585-586 (1936).

[189] R. E. Lee, R. H. Young and B. Castleman, “James Homer Wright: A Biography of the Enigmatic Creator of the Wright Stain on the Occasion of Its Centennial,” The American Journal of Surgical Pathology 26, (2002). 118 BIBLIOGRAPHY

[190] J. H. Bruce-Gregorios, Histophatologic Techniques, (Goodwill Trading Co., Inc., Quezon City, 2006).

[191] A. S. N. Murthy and K. S. Reddy, “Photochemical energy conversion studies in systems containing methylene blue,” International Journal of Energy Research 3, 205-210 (1979).

[192] I. S. Bayer, I. Ero˘gluand L. T¨urker, “Photogalvanic eﬀect in aqueous Methylene blue nickel mesh systems: Conversion of light into electricity,” International Journal of Energy Research 25, 207-222 (2001).

[193] D. R. Kearns, “Physical and chemical properties of singlet molecular oxygen,” Chemical Reviews 71, 395-427 (1971).

[194] S. L. Murov, I. Carmichael and G. L. Hug, Handbook of Photochemistry, (CRC Press, New York, 1993).

[195] E. M. Tuite and J. M. Kelly, “New trends in photobiology: Photochemical interactions of methylene blue and analogues with DNA and other biological substrates,” Journal of Photochemistry and Photobiology B: Biology 21, 103-124 (1993).

[196] J. F. Padday, “The halide ion activity of dyes and organic salts in aqueous solution,” The Journal of Physical Chemistry 71, 3488-3493 (1967).

[197] E. Rabinowitch and L. F. Epstein, “Polymerization of Dyestuﬀs in Solution. Thio- nine and Methylene Blue 1,” Journal of the American Chemical Society 63, 69-78 (1941).

[198] K. Bergmann and C. T. O’Konski, “A Spectroscopic Study of Methylene Blue Monomer, Dimer, and Complexes with Montmorillonite,” The Journal of Physi- cal Chemistry 67, 2169-2177 (1963).

[199] E. Braswell, “Evidence for trimerization in aqueous solutions of methylene blue,” The Journal of Physical Chemistry 72, 2477-2483 (1968). BIBLIOGRAPHY 119

[200] P. Mukerjee and A. K. Ghosh, “Multiple association equilibria in the self-association of methylene blue and other dyes,” Journal of the American Chemical Society 92, 6408-6412 (1970).

[201] K. Patil, R. Pawar and P. Talap, “Self-aggregation of Methylene Blue in aqueous medium and aqueous solutions of Bu4NBr and urea,” Physical Chemistry Chemical Physics 2, 4313-4317 (2000).

[202] S. Lei, J. Miyamoto, H. Kanoh, Y. Nakahigashi and K. Kaneko, “Enhancement of the methylene blue adsorption rate for ultramicroporous carbon ﬁber by addition of mesopores,” Carbon 44, 1884-1890 (2006).

[203] S. Taniguchi, N. Suzuki, M. Masuda, S. i. Hisanaga, T. Iwatsubo, M. Goedert and M. Hasegawa, “Inhibition of Heparin-induced Tau Filament Formation by Phe- nothiazines, Polyphenols, and Porphyrins,” Journal of Biological Chemistry 280, 7614-7623 (2005).

[204] M. Necula, L. Breydo, S. Milton, R. Kayed, W. E. van der Veer, P. Tone and C. G. Glabe, “Methylene Blue Inhibits Amyloid A Oligomerization by Promoting Fibrillization,” Biochemistry 46, 8850-8860 (2007).

[205] B. Kr¨autlerand R. e. Step´anek, “Photooxygenolytic Degradation of the Vitamin-

B12 Derivative Heptamethyl Coα,Coβ-Dicyanocobyrinate. Eﬃcient Preparation of Bicyclic Fragments,” Helvetica Chimica Acta 68, 1079-1088 (1985).

[206] P. C. Ashok, R. F. Marchington, P. Mthunzi, T. F. Krauss and K. Dholakia, “Optical chromatography using a photonic crystal ﬁber with on-chip ﬂuorescence excitation,” Optics Express 18, 6396-6407 (2010).

Curriculum Vitae

Personal Information

Name: Jocelyn Ssu-Yin Chen Date of Birth: 8 July 1983 Place of Birth: Taichung City, Taiwan Gender: Female Nationality: New Zealander

Education

Max-Planck Institute for the Science of Light, Erlangen, Germany University of Erlangen-Nuremberg, Erlangen, Germany

Ph.D., Physics September 2006 – August 2010 Thesis: Nanochemistry and Sensing in Photonic Crystal Fibers Advisors: Professor Philip St.J. Russell and Dr. Tijmen G. Euser

University of Auckland, Auckland, New Zealand

M.Sc. (Hons.), Physics March 2005 – February 2006 Thesis: Optical Parametric Ampliﬁcation in Photonic Crystal Fibers Advisors: Professor John D. Harvey and Dr. Stuart G. Murdoch

University of Auckland, Auckland, New Zealand

B.Tech. (Hons.), Optoelectronics March 2001 – December 2004 Thesis: High-Speed Infrared Laser Hygrometer Advisors: Professor Rainer Leonhardt and Dr. Igor Shvarchuck

121