Thermal Poling of Twin-hole Fibers. Jiawen Zhang

Transcription

1 Thermal Poling of Twin-hole Fibers by Jiawen Zhang A thesis submitted in conformity with the requirements for the degree of Master of Applied Science Graduate Department of Electrical and Computer Engineering University of Toronto Copyright c 2008 by Jiawen Zhang

2 Abstract Thermal Poling of Twin-hole Fibers Jiawen Zhang Master of Applied Science Graduate Department of Electrical and Computer Engineering University of Toronto 2008 We developed a poling system that allows simultaneous inducing and characterizing second-order nonlinearity (SON) in fibers. The SON in fibers is induced via heating the fiber to an elevated temperature while applying an external voltage across the core of the fiber. The induced SON in fibers is characterized by measuring the electro-optical coefficient of the fiber using a fiber Mach-Zehnder Interferometer (MZI). Using this poling system, we characterized in real time the evolution of the SON induced in a twin-hole fiber during thermal poling and thermal erasure. A SON of 0.1pm/V was recorded in the fiber. We developed a charge dynamics model, involving a fast charge carrier Na + and a slow charge carrier H +, to simulate the mechanism of fiber thermal poling. Good agreement between the simulated results using the model and our experimental observation in poling a twin-hole fiber was achieved. ii

3 Acknowledgements First of all, I would like to express my sincere gratitude and appreciation to my supervisor, Professor Li Qian, for her inspiration, vision, guidance, and encouragement over the last two years. I would also like to thank Professor Peter W.E. Smith, Professor Joyce Poon, Dr. Bing Qi and all the other members in the group for their discussions and suggestions on my project. I am grateful to my friends and colleagues in graduate school for their insights, support and friendship. Special thanks to Fei Ye and Michael Galle for their companies in the laboratory, William Chen for helping design and construct the electric circuit, and Eric Zhu for suggesting creative ideas on the charge dynamics model. Above all, I am deeply grateful to my parents for their love and encouragement throughout all my years in school. I would never have come this far without their inspiration and guidance. iii

4 Contents 1 Introduction Thesis Motivation Thesis Objective Thesis Outline Review of Fiber Poling Introduction to Thermal Poling Fibers for Thermal Poling Methods of SON Characterization Hydrofluoric Acid Etching Second Harmonic Microscopy Electro-optical Characterization Experimental Procedures Twin-hole Fiber Structure Electrode Insertion Fiber Heating/Cooling System Design of the System Performance of the System SON Characterization System Design of the System Performance of the System Summary Experimental Results and Analysis Results of the First Thermal Poling iv

5 4.2 Decay of the SON Results of the Second Thermal Poling Polarization Dependence of the SON Characterization of the Third-order Nonlinearity Summary Charge Dynamics Model Single-charge Dynamics Model General Equations Boundary and Initial Conditions Simulation Results Two-charge Dynamics Model Include the Second Charge Carrier Boundary and Initial Conditions Simulation Results Simulation of the SON Evolution in Experiments Summary Conclusion Significance and Contribution Future Work A Matlab Code for Two-charge Dynamics Model 67 A.1 Body of the Code A.2 Prepare for the Simulation A.3 Run the Simulation A.4 Main Iteration Code Bibliography 74 v

6 List of Tables 4.1 Fitting results of the SON decay at room temperature using the double exponential model Parameters for simulation using the single-charge model Parameters for simulation using the two-charge model Parameters for simulating the first and second thermal poling of a twinhole fiber using the two-charge model vi

7 List of Figures 2.1 SEM image of the cross section of a 125µm fiber[1] Second harmonic images of a poled D-shaped fiber. (a) Two-photon signal. (b) Second harmonic signal. (c) Overlaid image of (a) and (b)[2] Spatial profile of the SON in a poled D-shaped fiber[2] Free-space Mach-Zehnder interferometer for measuring the SON in twinhole fibers[3] Cross-section geometry of the twin-hole fiber with electrodes inserted Twin-hole fiber with side openings[4] Connecting twin-hole fiber with electrodes to single mode fiber via barefiber adapters Cartridge Heater Fiber Holder Cover of the fiber holder Schematics of the temperature controlling system Temperature rise of the fiber holder (from room temperature to 250 o C) Temperature drop of the fiber holder (from 250 o C to room temperature) Fiber MZI for measuring the SON in the twin-hole fiber under thermal poling Comparison between the measured SON when the AC voltage is on and off. The peak-to-peak amplitude of the AC voltage is 400V. The DC voltage is off Signal at the output of the MZI when the external DC is off. The frequency of the phase modulator is 500Hz. The frequency of the AC voltage is 30kHz. The peak-to-peak amplitude of the AC voltage is 400V vii

8 3.13 Signal at the output of the MZI when the external DC is on. The frequency of the phase modulator is 500Hz. The frequency of the AC voltage is 30kHz. The peak-to-peak amplitude of the AC voltage is 400V SON with the external DC off and on Time evolution of the SON induced in a twin-hole fiber during its first thermal poling. The fiber is poled at 250 o C with 3kV DC voltage Decay of the SON at room temperature Thermal erasure of the SON at 250 o C after the room-temperature decay Time evolution of the SON induced in a twin-hole fiber during its second thermal poling. The fiber is poled at 250 o C with 3kV DC voltage Thermal erasure of the SON at 250 o C after the second thermal poling Fiber Mach-Zehnder Interferometer for measuring the SON in fibers under different polarization states SON under different polarization states. The dots in (A) are experimental data. Each straight line in (B) is the average SON under each polarization state SON under different external DC voltages. Dots are the experimental data; solid lines are the fitting results Simulated time evolution of the Na + density during thermal poling using the single-charge model. The depth under the anode is normalized to the lateral distance between the two electrodes. The charge density is normalized to the background negative charge density. The dotted lines represent the location of the core Simulated time evolution of the electrical field during thermal poling using the single-charge model. The depth under the anode is normalized to the lateral distance between the two electrodes. The electrical field is normalized to the initial uniform field throughout the fiber at the beginning of the poling. The dotted lines represent the location of the core Simulated time evolution of the SON at the core of the fiber during thermal poling using the single-charge model viii

9 5.4 Simulated time evolution of the Na + density during thermal poling using the two-charge model. The depth under the anode is normalized to the lateral distance between the two electrodes. The charge density is normalized to the background negative charge density. The dotted lines represent the location of the core Simulated time evolution of the H + density during thermal poling using the two-charge model. The depth under the anode is normalized to the lateral distance between the two electrodes. The charge density is normalized to the background negative charge density. The dotted lines represent the location of the core Simulated time evolution of the net positive charge density during thermal poling using the two-charge model. The depth under the anode is normalized to the lateral distance between the two electrodes. The charge density is normalized to the background negative charge density. The dotted lines represent the location of the core Simulated time evolution of the electrical field during thermal poling using the two-charge model. The depth under the anode is normalized to the lateral distance between the two electrodes. The electrical field is normalized to the initial uniform field throughout the fiber at the beginning of the poling. The dotted lines represent the location of the core Simulated time evolution of the SON at the core of the fiber during thermal poling using the two-charge model Experimental and simulated time evolution of the χ (2) induced in a twinhole fiber during the first thermal poling Experimental and simulated time evolution of the χ (2) induced in a twinhole fiber during the second thermal poling Simulated time evolution of the charge density and electrical field in the twin-hole fiber during thermal poling. The depth under the anode is normalized to the lateral distance between the two electrodes. The charge density is normalized to the background negative charge density. The electrical field is normalized to the initial uniform field throughout the fiber at the beginning of each poling ix

10 List of Acronyms SON Second-order Nonlinearity PPLN Periodically Poled Lithium Niobate MZI Mach-Zehnder Interferometer QPM Quasi-Phase-Matching UV Ultraviolet GVM Group Velocity Mismatch IR Infrared FWM Four Wave Mixing EO Electro-optic SHG Second Harmonic Generation SMF Single Mode Fiber HF Hydrofluoric SHM Second Harmonic Microscopy SSR Solid State Relay PC Polarization Controller x

11 Chapter 1 Introduction Silica fibers are of significant importance in almost all transport links of optical networks. Their high transparency and low cost make fibers excellent media for transmitting optical signals over long distance. However, as silica glass has an amorphous structure with macroscopic inversion symmetry, silica fibers are intrinsically void of second-order nonlinear properties such as the electro-optic (EO) effects. This limitation of fibers forbids their presence in many active applications. In 1991, however, Myers et al. demonstrated that the symmetry of glass can be broken by subjecting the glass to high voltage at elevated temperature, a technique now widely referred to as thermal poling. This treatment of glass creates a permanent second-order nonlinearity (SON) by recording a frozen-in electrical field inside the glass. This phenomenon opens doors for expanding fiber s functionality into many new territories, such as electro-optical modulators and switches, frequency converters, parametric oscillators, and photon pair generators. Furthermore, the excellent attributes of silica glass, including low optical loss, high damage threshold and excellent compatibility into the existing fiber networks, will lead to wide applications of fibers with second-order nonlinear properties. 1.1 Thesis Motivation Second-order nonlinearity (SON) in many nonlinear materials have been extensively used for various optical applications. Frequency conversion is one important and widely explored application. In such applications, nonlinear crystals with large values of SON, such as LiNbO 3, KTP and LBO are usually used as the media for optical nonlinear process. For efficient frequency conversion, phase matching condition must be satisfied, which is 1

12 Chapter 1. Introduction 2 usually achieved through periodical poling of the nonlinear materials, e.g. periodically poled lithium niobate (PPNL). When used in fiber optical systems, the conventional nonlinear materials, like the nonlinear crystals, can suffer from many disadvantages, such as high package cost, high transmission losses, low optical damage threshold and difficult integrability with the incumbent fiber systems. Silica glass, especially silica fibers, if used in fiber systems as the media for applications such as frequency conversion, can overcome many disadvantages brought by the nonlinear crystals. Inducing periodical SON into silica fibers have been demonstrated as a promising technique to achieve quasi-phase-matching (QPM) in fibers, and therefore to realize frequency conversion in fiber-based devices[5]. Although the SON induced in silica glass so far is still relatively low ( 1pm/V), compared to the natural SON of the nonlinear crystals( 30pm/V in LiNbO 3 ), the low dispersion of fibers allows for availability of greater bandwidths and higher conversion efficiency by increasing the interaction length of the fiber device. Furthermore, poled fibers are especially suitable for the applications of frequency conversion of short pulses, as the Group Velocity Mismatch (GVM) between pulses at different frequencies is much smaller than that in nonlinear crystals [6]. As the periodically poled fiber can offer new wavelengths via QPM, it can extend the spectrum range of the conventional fiber lasers and make them suitable for many new applications. For example, emission in mid-far Infrared (IR) (2-25µm), which is not readily available from the conventional fiber lasers can play an important role in chemical and spectroscopic detection of gases, as many chemical gases have strong absorption lines in the IR region, such as CO (2µm) and C 2 H 2 (3µm). This limit of fiber lasers can be overcome by incorporating periodically poled fibers into the incumbent fiber laser systems. In this way, IR spectrum can be obtained by converting the light at fundamental frequencies emitting from the fiber lasers to lower frequencies through a process of frequency down conversion in the poled fibers. An alternative method of realizing frequency conversion in fibers is to use Four Wave Mixing (FWM), a third-order nonlinear process[7]. However, the frequency conversion intensity through three wave mixing in poled fibers, (a second-order nonlinear process), can grow with distance about four times faster than that using FWM[8]. This indicates that, compared to fiber devices using FWM, devices based on poled fibers with second-order nonlinear properties can be much shorter, yet achieving the same conversion efficiency. This allows a more compact structure of fiber-based IR sources.

13 Chapter 1. Introduction 3 In addition to obtaining new wavelengths through frequency conversion, the functionality of poled fibers with a built-in SON can be extended to a much wider range. Fiber modulators and fiber switches are good examples. A research group at the Royal Institute of Technology in Sweden reported an all-fiber switch based on a poled twin-hole fiber. The switch has insertion loss of 1dB and a switching voltage of 100V[9]. Most recently, the same group also demonstrated an all-fiber amplitude modulator in the configuration of Sagnac interferometer incorporating a poled 75-cm twin-hole fiber. The modulator has a V π of 180V at the wavelength of 1.5µm and a rise time of 40ns[10]. Furthermore, the fiber-based phase modulators can be employed as building blocks to construct all-fiber two photon generators for the applications of quantum cryptography[11]. The examples presented above are only part of the broad applications of poled fiber devices. Such applications have generated much attention to the research field of fiber poling, especially to the efforts of achieving a large SON in fibers, as a large SON is the foundation for making efficient and compact fiber devices. The typical value of SON induced in fibers so far is on the order of 0.1pm/V, which is two orders of magnitude smaller than that of nonlinear crystals. We believe that this value of SON can be further increased by optimizing the poling conditions, such as the poling voltage, the poling temperature and the poling duration. Therefore, in this thesis project, we plan to construct a poling system that can simultaneously pole fibers and monitor the induced SON. Such a system will allow us to realize efficient optimization of the poling conditions for obtaining a large SON in fibers. Although many experimental demonstrations of inducing SON in fibers have been reported in the literature, the underlying mechanism of the SON creation in fibers during poling has not been fully understood. Various theoretical models were proposed to simulate the poling process and to explain the SON generation in both bulk glass and glass fibers. Yet, none of them offers a comprehensive understanding of the poling mechanism. Therefore, in this thesis, we will also develop a model for simulating the charge dynamics in fibers and explore its ability of describing the thermal poling process. The model should provide us with insights into the underlying mechanism of thermal poling fibers, and lead us to a more efficient way of poling fibers.

14 Chapter 1. Introduction Thesis Objective The long term objective of the project is to demonstrate frequency conversion in poled twin-hole fibers through quasi-phase-matching. This will be achieved by recording a periodic second-order nonlinearity in the twin-hole fibers. As the first stage of the project, this thesis aims to develop a poling system that can both induce and characterize the SON in twin-hole fibers. First, The performance of the poling system will be investigated and optimized to allow for accurate measurement of the SON induced in fibers. Second, a twin-hole fiber will be poled at certain poling conditions, i.e. poling voltage, temperature and duration, and the time evolution of the induced SON in the fiber will be monitored. The poling length of the fiber should be at least 10cm for efficient frequency conversion through quasi-phase-matching. Third, after the SON is created in the fiber, the stability of the SON at both room temperature and elevated temperatures will be investigated. The lifetime of the SON will be obtained by characterizing the decay of the SON at different temperatures. Because of the limited amount of twin-hole fibers available in the project, the same fiber will be poled, depoled, and repoled several times, and the SON evolution during each depoling and repoling will be measured and studied. Fourth, the theory on thermal poling and the mechanism for inducing SON in fibers will be explored. A charge dynamics model will be developed to simulate the poling process, and the time evolution of several important parameters, such as the charge distribution, the electrical field and the SON during poling will be investigated using the model. The validity of the model will be assessed by fitting the simulation results using the model to our observations in the experiments of thermal poling of twin-hole fibers. 1.3 Thesis Outline This thesis is organized as follows: Chapter 2 briefly reviews the development of thermal poling of silica fibers and the methods of characterizing the induced second-order nonlinearity in fibers. Chapter 3 describes the experimental procedures of poling a twin-hole fiber. In this chapter, we will discuss the preparation of a twin-hole fiber for poling experiments, the design and test of a fiber heating/cooling system and the construction of an interferometric system for characterizing the induced SON in fibers. Chapter 4 presents the results of poling a twin-hole fiber. The time evolution of the SON in the fiber during

15 Chapter 1. Introduction 5 its first poling, the thermal erasure and the subsequent second poling will be discussed. Chapter 5 investigates the mechanism of thermal poling and proposes a charge dynamics model for simulating the poling process. Chapter 6 summarizes the thesis, discusses the significance of the work, and suggests future directions.

16 Chapter 2 Review of Fiber Poling 2.1 Introduction to Thermal Poling Compared to other poling techniques, such as CO 2 laser-assisted poling and ultraviolet (UV) poling, thermal poling offers a repeatable and reliable method to produce a large second-order nonlinearity (SON) and linear electro-optic (EO) coefficient in bulk silica and silica fibers[12]. In the process of thermal poling, a fiber is usually heated to temperatures of o C, while a strong external electrical field 10 7 V/m is applied across the fiber. In this temperature range, alkali ions inside the fiber, such as K +, Li + and especially Na +, become thermally activated and free to move. Under the influence of the external field, these ions migrate from the anode toward the cathode through the glass matrix[13, 14]. The ionic current in the glass due to the charge migration is on the order of magnitude of µa upon the application of the external field. After tens of minutes, the current decreases and reaches a steady state value [15]. At this time, the fiber is cooled down to room temperature with the external electric field still applied. Once the fiber reaches room temperature, the external field is removed. Because the mobilities of the alkali ions at room temperature are several orders of magnitude smaller than at elevated temperatures, the alkali ions tend to be frozen inside the glass, which results in an internal space electrical field. This internal electrical field, coupled with the intrinsic third-order nonlinear susceptibility of the glass, gives an effective second-order nonlinearity. The electrical field created inside the glass after poling has a spatial profile determined by the internal charge distribution. As the SON results from the electrical field, it also display such a spatial profile. This SON profile is mainly distributed within several 6

17 Chapter 2. Review of Fiber Poling 7 micrometers beneath the anode[15, 16], and it is nonuniform throughout the glass. If the portion of the profile with the maximum values of SON has a good overlap with the core region of the fiber, a large effective SON can be experienced by the optical wave propagating in the fiber. Under such a condition, the poling is efficient and the poled fiber can be used for efficient nonlinear process. The first experiment of inducing large SON in bulk glass using thermal poling technique was demonstrated by R.A. Myers et al. in 1991[15]. In their work, SON of 1pm/V was achieved by poling commercial fused-silica optical flats at temperatures of o C with an electrical field of V/m. Since this demonstration, many researchers have been attempting to obtain a large SON in various silica samples through thermal poling. Silica fibers are the most widely sought medium for recording the SON due to their excellent capability of transmitting optical signals. In 1994, P.G. Kazansky et al. poled a fiber in vacuum at 280 o C with 4.3kV voltage for 15 minutes, and obtained an electrical-optical coefficient of 0.05pm/V. This value is 25 times larger than previously reported in silica fibers[17]. In 1995, T. Fujiwara et al. obtained electro-optical coefficients of 0.36pm/V and 0.84pm/V for TE and TM mode, respectively, in a germanosilicate fiber thermally poled at 250 o C for minutes with an electrical field over V/cm[18]. Besides the scientific explorations, poled fibers were also explored for practical applications. P.G. Kazansky et al. demonstrated that Second Harmonic Generation (SHG) could be realized in a thermally poled fiber by erasing the induced SON periodically to satisfy phase-matching condition[19]. Research groups in Royal Institute of Technology in Sweden successfully poled meter-long twin-hole fibers with Au-Sn alloy electrodes, and demonstrated their applications in an all-fiber electro-optical switch[20]. 2.2 Fibers for Thermal Poling Unlike bulk silica glass, which can be readily used for thermal poling experiments, as the electrodes can be easily attached to the two flat surfaces of the glass, silica fibers impose many difficulties in the poling experiments. The fibers must meet several requirements for efficient thermal poling. First, as a strong external electrical field is required to break the symmetry of the silica material, high external voltage (5kV) is usually applied along the fiber length. This requires special design of the fiber geometry to let electrodes be incorporated into the fibers. Second, the insertion loss of the fiber might be increased, as the optical mode could be disturbed due to the presence of the electrodes. Thus the

18 Chapter 2. Review of Fiber Poling 8 relative positions of the fiber core and the electrodes need to be carefully designed to mitigate loss. Third, the core of the fiber must be properly positioned relative to the two electrodes to achieve a good overlap with the induced SON profile for a large effective SON. Several new fibers have been designed to meet these requirements. In this section, we will introduce and compare several special fibers designed for thermal poling, such as D-shaped fibers, twin-hole fibers and single-hole fibers with conductive coatings. The pros and cons of each fiber for the thermal poling experiments will be discussed. The D-shaped fibers are used by many research groups for its ease of attaching electrodes to their plane surface. The D-shaped fiber used by P.G. Kazansky et al had a numerical aperture of 1.91, a core diameter of 5µm and a cladding diameter of 300µm[21]. This unique fiber geometry allows a patterned aluminum electrode to be fabricated on plane surface of the fiber using standard planar lithography, and therefore, periodic SON can be written directly in the fiber during poling for quasi-phase-matching. However, as the distance between the core and the plane surface in commercial D-shaped fibers is too large for efficient poling, etching is usually needed to reduce the distance, which complicated the experiments. Moreover, D-shaped fibers can not be spliced with standard single mode fibers (SMF), which introduces large insertion losses if they are incorporated into standard fiber systems. Twin-hole fibers, unlike D-shaped fibers, have two holes parallel to the core along the fiber length. Each core can accommodate an electrode, to which, high external voltage can be applied. Fig.2.1 shows the cross-section image of a twin-hole fiber fabricated by Acreo in Sweden[1]. As Twin-hole fibers can be easily spliced to SMF, they only add small coupling losses (<3dB) to the fiber systems. However, direct periodic poling is not readily available for twin-hole fibers. Alternatively, the quasi-phase matching is achieved by erasing the poled region periodically using UV light after a uniform SON is recorded in the fiber. Furthermore, in order to connect external electrode wires to the electrodes inside the hole of the fiber, side-polishing is usually employed to open holes on two sides of the fiber, which introduced extra sophistication into the experiments. Electrodes can be either manually inserted into the fiber hole, which requires hours of careful work under a microscope [22], or pumped into the hole as liquid alloy at high temperatures [23]. The former approach limits the length of the electrodes that can be inserted into the hole, while the latter can provide internal electrodes up to 1m long. Besides D-shaped fibers and twin-hole fibers, other fiber geometries have also been designed and fabricated for thermal poling experiments. One example is the single-hole

19 Chapter 2. Review of Fiber Poling 9 Figure 2.1: SEM image of the cross section of a 125µm fiber[1] fiber with conductive coatings [24]. This fiber only has one hole, into which, an electrode is inserted as the anode. The conductive coating deposited onto the outer surface of the fiber is used as cathode. However, the difficulty of depositing conductive coatings on the fiber and the non-standard cross-section of the fiber made this fiber less favorable than the D-shaped fibers and the twin-hole fibers. 2.3 Methods of SON Characterization In order to understand the origin of the SON in silica glass and fibers, and to optimize the fiber designs for achieving a large SON in the fiber core, there is a strong need to characterize both the time evolution and spatial distribution of the SON induced in the fiber under poling. So far, there are three common methods for doing so, hydrofluoric (HF) acid etching method[25, 26], second harmonic microscopy (SHM)[2] and electrooptical characterization[17, 3] Hydrofluoric Acid Etching Etching a poled glass with hydrofluoric acid (HF) has been proved to reveal the spatial profile of the induced SON[25, 26]. The ion depleted region of silica glass is more resistant

20 Chapter 2. Review of Fiber Poling 10 to the etching by HF than the untreated region[27]. Therefore, the etching rate in the depleted region is lower than that of elsewhere in the sample. As the depleted region in the poled sample is also the region containing the SON, the variance in the etching rate therefore indicates the strength and location of the SON in the sample. The method of etching samples using HF has several disadvantages, especially in measuring the SON in a poled twin-hole fiber. The HF will flow into the hole of the fiber and destroy the fiber by etching it from inside. In this case, fibers need to be embedded into HF-resistant epoxies to prevent the flow of the HF into the fiber holes, which complicates the measurement process. Other disadvantages include: The HF destroys the poled sample during etching, which may affect the accuracy of the measurement, and therefore the measured SON profile may not truly represent the original profile; it cannot measure the absolute value of the SON in the sample; It is time consuming in optimizing the poling conditions to achieve a large SON, as many samples need to be poled under different poling voltages, temperatures and durations, and then etched and observed under a microscope to determine the best conditions for achieving a large SON Second Harmonic Microscopy Second harmonic microscopy (SHM) has wide applications in characterizing biological materials[28], periodically ferroelectric domain structures[29], and thermally poled silica planar waveguides[30]. In characterizing the SON profile in the poled fibers, it can provide nondestructive measurements with high resolution, which is not available in the HF etching method. A typical SHM consists of a microscope with a confocal system and a tunable pulse laser. The excitation wavelength of the laser pulse is around 800nm. Two detectors are usually used to detect signals in different spectrum ranges. One detector receives two-photon signals with spectrum ranges in nm, while the other detector, equipped with a narrow bandpass filter, receives only the second harmonic signals with spectrum around 400nm. Typical SHM images of the cross section of a poled fiber is shown in Fig.2.2[2]. As the received intensity of the second harmonic signal at each point in the image is related to χ (2) through P (2ω) = (P (ω)χ (2) ) 2, the magnitude of the SON can be calculated by taking the square root of the signal intensity (shown in Fig.2.3). Therefore, the spatial distribution of the induced SON can be obtained by scanning across the SON layer from the hole to the silica cladding. Compared to the HF etching method, SHM provides

21 Chapter 2. Review of Fiber Poling 11 Figure 2.2: Second harmonic images of a poled D-shaped fiber. (a) Two-photon signal. (b) Second harmonic signal. (c) Overlaid image of (a) and (b)[2] Figure 2.3: Spatial profile of the SON in a poled D-shaped fiber[2]

22 Chapter 2. Review of Fiber Poling 12 more accurate measurement of the SON. However, it suffers the same drawbacks as the HF etching method: unable to measure the absolute value of the SON and inefficient in optimizing the poling conditions Electro-optical Characterization The third method of characterizing the SON in poled fibers is to measure the linear electro-optical (EO) coefficient of the fiber. The EO coefficient r is related to the SON through χ (2) = n4 r, where n is the refractive index of the fiber[31]. Characterization of the 2 EO coefficient therefore gives the information of the SON as well. A common method of characterizing the linear EO effect of an optical medium is to use interferometry, which can also be used to measure the EO coefficient in a poled fiber. Both free-space and fiber-based Mach-Zehnder interferometers have been developed to characterize the EO coefficient in fibers[32, 3, 20]. Fig.2.4 shows the free-space Mach-Zehnder interferometer developed by D.Wong et al. at the University of Australia. In this setup, a linear phase ramp is introduced in one arm of the interferometer by a phase modulator, while a small phase variation is induced in the twin-hole fiber in the other arm by applying a probe AC voltage to the electrodes in the fiber. The electro-optic response of the poled fiber is then characterized by monitoring the interference signal at the output of the interferometer. Figure 2.4: Free-space Mach-Zehnder interferometer for measuring the SON in twin-hole fibers[3] The benefits of the EO characterization are obvious. First, it can measure the time

23 Chapter 2. Review of Fiber Poling 13 evolution of the SON induced in fibers while the fibers are under poling. This real-time measurement can greatly facilitate the optimization of the poling experiments: once the optimal SON is achieved during poling, the poling process can be terminated to retain this optimum SON. Second, it is nondestructive, leaving the poled samples intact for future applications. Third, it measures the absolute magnitude of the induced SON, which gives insights into the potential values of the poled devices for practical use. The disadvantages of the EO method, compared with the other two discussed before, is its incapability of acquiring the spatial profile of the SON within the poled sample. In fiber poling experiments, where the magnitude of the effective SON in the core of the fiber is more important than the overall profile of the SON throughout the fiber, this disadvantage is less significant. Considering all the advantages of the EO method discussed above, we decide to employ the EO method in our experiments for characterizing the SON induced in fibers. The EO measurements developed so far have their own limitations. First, the alignment in freespace setups, such as the one in Fig.2.4, is complex, time-consuming, and susceptible to instabilities. Therefore, we plan to construct an all-fiber interferometer for measuring the SON during poling. Fiber-based interferometer can avoid the difficulties associated with the free-space approach, such as alignment and mechanical instabilities due to air optical path. Second, in order to integrate the poling fiber to one arm of the interferometer with low insertion loss, while still getting access to the electrodes inside the fiber hole, side polish technique is required[12, 23], which compromises the integrity of the poled fiber. In our experiment, we propose to insert the electrodes from the cleaved end facets of the fiber, thus eliminate the polishing process. The fiber with electrodes inside can be connected to standard SMF through bare fiber adapters. The new method of inserting electrodes are expected to be more efficient and less lossy than the conventional one. In Chapter 3, we will discuss in detail about our all-fiber interferometer and new method of inserting electrodes.

24 Chapter 3 Experimental Procedures In this chapter, we will describe in detail the experimental procedure of thermal poling of a twin-hole fiber. In section 3.1, we will introduce the structure of the twin-hole fiber we used in our poling experiments. In section 3.2, we will present a new approach to incorporate electrodes into the twin-hole fibers, and discuss its advantages compared to two other existing methods of electrode insertion. In section 3.3, we will discuss the design and performance of a custom-made fiber heating/cooling system, as well as its capability of providing the desired thermal environment for poling fibers. Finally, in section 3.4, we will present an interferometric system for real-time characterization of the SON induced in fibers. We will discuss the measurement principle of the system and its accuracy in measuring the SON in fibers. 3.1 Twin-hole Fiber Structure The twin-hole fiber (Fig.3.1) used in the thermal poling experiment was fabricated at the University of Southampton. It was specially designed with two air holes of 50µm diameter in its cladding region on each side of the core. Gold-coated tungsten wires of 25µm diameter were inserted into the holes (one wire in each hole) as electrodes for the application of an external DC voltage. Along the length of the fiber core, the hole that accommodates the anode is deliberately positioned closer to the core region to achieve a better overlap between the core and the SON profile induced during the poling. The diameter of the twin-hole fiber is 125µm, the same as that of the standard single mode fiber, which allows the twin-hole fiber to be easily spliced to the standard fiber optical system with low coupling loss. The cross-section dimensions are given in Fig

25 Chapter 3. Experimental Procedures μm 50μm 4μm 3μm 7μm Figure 3.1: Cross-section geometry of the twin-hole fiber with electrodes inserted. 3.2 Electrode Insertion Inserting electrodes into the holes of the twin-hole fiber is the first step of the poling experiment. In the literature, two methods have been reported for inserting electrodes: pumping conductive molten alloys into the holes[23, 20, 1] and inserting metal wires into the holes manually[12, 3]. The former was first demonstrated by a research group at the Royal Institute of Technology in Sweden[23]. In this method, a pressure cell was used to pump the molten Au-Su alloy at 280 o C into the holes of a twin-hole fiber. Once the fiber is cooled down to room temperature, the molten alloy becomes solid and remains inside the fiber as conductive electrodes. The electrodes in the fiber were then bonded to external electrodes through polished side-openings on the fibers, so that an external poling voltage could be applied. Using this method, electrodes over one meter long were successfully produced inside twin-hole fibers[23, 20]. The second method is more straightforward. Fibers are usually fixed under a microscope, and metal wires are threaded directly through the side-openings on the fiber into the holes by hand, as shown in Fig.3.2[4]. As the diameter of the wires is only about 25µm, the wires are easily bent during the insertion. Therefore, long electrodes are difficult to obtain using the second method. Typically, the length of the electrodes that can

26 Chapter 3. Experimental Procedures 16 be inserted into the fiber manually is limited to several centimeters. In both methods, the fibers with electrodes inside are spliced to standard single mode fiber (SMF) for future applications. Figure 3.2: Twin-hole fiber with side openings[4] In our experiment, we choose the second method, manual insertion, as the desired poling length is about 10cm for efficient frequency conversion through quasi-phase-matching[21]. Gold-coated tungsten wires of 25µm diameter are used as electrodes. Unlike the conventional method, however, we do not require side-polished openings for the access of electrodes to the fiber holes. Instead, we insert the wires into the holes through the cleaved end facet of the twin-hole fiber, and connect the fiber with SMFs via bare-fiber adapters. The joint between the twin-hole fiber and the SMF using this method is illustrated in Fig.3.3. This new method eliminates the procedure of side polishing a fiber for two slots, as shown in Fig.3.2, and the subsequent splicing process. Therefore, it reduces the time of preparing twin-hole fibers for poling experiments. The absence of slots also makes the fiber robust and easy to handle in the experiment. Using this method, We have achieved a maximum 15cm overlap between the two electrodes in a twin-fiber. The coupling loss between our twin-hole fiber and the SMF is mainly due to mode mismatch and an air gap between the two fibers. The difference between the mode field diameter of the SMF and the twin-hole fiber can result in a coupling loss 4 L mode = 10log ( ) 2, (3.1) ω 1 ω 2 + ω 2 ω 1

27 Chapter 3. Experimental Procedures 17 Figure 3.3: Connecting twin-hole fiber with electrodes to single mode fiber via bare-fiber adapters where ω 1 and ω 2 are the mode field diameter of the SMF and the twin-hole fiber, respectively. With ω 1 = 10.5µm and ω 2 = 6.4µm, L mode = 1.0dB is obtained. As in our method, the electrode is inserted into the twin-hole fiber from its end facet, when the twin-hole fiber is connected to the SMF in the adapter, the electrode can cause a longitudinal displacement between the end facets of the SMF and the twin-hole fiber. The displacement, which is equal to the diameter of the electrode, can also lead to a coupling loss where ( ) 1 L long = 10log, (3.2) Z Z = λ z, (3.3) 2πn 0 ω2 where λ = 1550nm is the wavelength of the light, n 0 = 1.0 is the refractive index of air, z = 25µm is the longitudinal displacement and ω is the mode field diameter of the fiber. Here we use ω = 6.4µm to estimate the maximum loss, and L long = 0.6dB is obtained. From the above calculations, we can know that the total coupling loss between the SMF and the twin-hole fiber is about L mode + L long = 1.6dB. Notice that this total loss is calculated in the ideal condition, where other less significant sources of loss, such

28 Chapter 3. Experimental Procedures 18 as the angular and lateral misalignment between the two fibers, are ignored. In the experiments, we have achieved a minimum coupling loss of about 3dB between the SMF and the twin-hole fiber. This loss is comparable to the loss using fusion splicing[9]. 3.3 Fiber Heating/Cooling System Design of the System A customized fiber heating/cooling system is designed and fabricated to provide the twin-hole fiber with the desired temperature dynamics for the thermal poling experiment. The heating of the fiber is realized using a fiber heater, which comprises three parts: a commercial cartridge heater (Fig.3.4), a custom-made fiber holder (Fig.3.5) and its cover (Fig.3.6). The cartridge heater is 0.635cm in diameter and 12.7cm long. It is constructed of coiled resistance wires and an incoloy sheath. With a voltage applied, the resistance generates joule heat, and the heat increases the temperature of the heater. The maximum work temperature of the cartridge heater is around 760 o C. We designed the fiber holder Figure 3.4: Cartridge Heater for accommodating the twin-hole fiber and heating the fiber during the thermal poling experiment. The holder is an aluminum block with a cylindrical hole through the center of its body and a V-groove on its top surface. The center hole accommodates the cartridge

29 Chapter 3. Experimental Procedures 19 Figure 3.5: Fiber Holder. Figure 3.6: Cover of the fiber holder

30 Chapter 3. Experimental Procedures 20 heater, which heats the whole aluminum block. The V-groove is for accommodating the fiber snugly with large surface contact with the holder. The two threaded holes on one side of the holder are used to fix a thermal coupler for monitoring the temperature of the holder. The six threaded holes on the top of the holder are used to mount the cover. The cover presses the fiber in the V-groove against the holder to ensure a tight contact. The holder is 1.27cm wide, 1.27cm high and 12.7cm long. The cover is 1.27cm wide, 0.65cm high and 12.7cm long. Temperature Contorller Fiber Holder Thermal Coupler Fuse SSR 120 V AC Figure 3.7: Schematics of the temperature controlling system The temperature of the fiber heater is adjusted using a temperature controlling system shown in Fig.3.7. The system is comprised of five parts: a temperature controller, a solid state relay (SSR), an AC power supply, a thermal coupler, and a fiber heater (discussed in the previous paragraph). During the heating process, the relay is turned on, and the heater is connected to the power supply; therefore, the temperature of the heater increases. The thermal coupler monitors the change of the temperature and sends signals to the controller for controlling the temperature. When the temperature rises to the desired level, it is maintained at a constant level by switching the relay on and off periodically Performance of the System Fig.3.8 shows the rise of the temperature of the fiber heater when the desired temperature is set to 250 o C. The desired temperaute is achieved within 10 minutes, after which, the temperature is fairly stable, with variance no more than ±1 o C. Temperature uniformity along the fiber length is crucial in obtaining a uniform SON during poling. As the fiber is pressed tight against the surface of the V-groove during

31 Chapter 3. Experimental Procedures 21 Time (s) Figure 3.8: Temperature rise of the fiber holder (from room temperature to 250 o C) poling, and the size of the fiber ( an outer diameter of 250µm ) is much smaller than the size of the holder (1.27cm wide and 1.27cm high), we believe that the transfer of heat from the holder to the fiber is efficient, and therefore the temperature along the fiber length is equal to the temperature along the V-groove. We measured the temperature along the V-groove using a thermal coupler when the heater is maintained at 250 o C, and the variation is within ±1 o C. The good temperature uniformity ensures uniform heating of the fiber, and thus uniform SON during poling. In order to record an internal SON in the fiber, the space distribution of the charge carriers responsible for the creation of the internal electrical field should be frozen in the glass at the end of the poling. Cooling the fiber to low temperatures in a short time is therefore crucial, or the induced internal SON will be erased. An electrical fan is used to blow the heater and to speed up the cooling. The decrease in the temperature of the fiber heater from 250 o C to room temperature with the assistance of the fan is shown in Fig.3.9. It takes only about 300 seconds for the temperature to drop below 50 o C, under which, the mobility of the charge carriers is several orders of magnitude smaller than that at 250 o C, and the charge distribution is frozen in the fiber. During the poling, when the SON reaches the maximum value, it usually stays at that value for several minutes before starting to decay. Therefore, 300 second cooling time is sufficiently short

32 Chapter 3. Experimental Procedures 22 to terminate the poling and retain a large SON. Figure 3.9: Temperature drop of the fiber holder (from 250 o C to room temperature) 3.4 SON Characterization System Design of the System A fiber Mach-Zehnder Interferometer (MZI) is constructed for real-time measuring of the SON induced in the twin-hole fiber during thermal poling, as shown in Fig During the measurement, an external DC voltage V dc of several kilovolts is applied to the electrodes in the twin-hole fiber. This high DC voltage is obtained from a DC- DC converter, which converts an input signal of 1-5V to an output signal of 1-5kV. A sinusoidal AC voltage V ac with a peak-to-peak amplitude in the range of 200V-800V is superimposed on the DC voltage for probing the electro-optic response of the twin-hole fiber. The AC voltage is obtained by stepping up a low AC voltage signal of several volts from a function generator through an electrical transformer. The frequency of the AC signal is set to 30kHz, away from any mechanical resonance of the twin-hole fiber. In the reference arm of the interferometer, a phase modulator is used to introduce a linear phase ramp from 0-2π. The frequency of the phase ramp is 1kHz.

33 Chapter 3. Experimental Procedures 23 V V Max Oscilloscope Figure 3.10: Fiber MZI for measuring the SON in the twin-hole fiber under thermal poling The output of the MZI is the interference between the two signals from its two arms, which can be expressed as V 0 (φ) = V max (1 + cosφ), (3.4) 2 where φ is the phase difference between the two arms, and V max is the peak-to-peak amplitude of the signal resulting from the linear phase ramp in the reference arm ( the signal with lower frequency in the inset of Fig.3.10). Due to the presence of the linear electro-optic coefficient induced in the core of the twin-hole fiber, the refractive index of the core changes with the applied voltage, which varies at the frequency of the AC signal. introduced into the twin-hole fiber arm, and Eq.(3.4) becomes: When φ = π, Eq.(3.5) can be rewritten as: 2 As a result, a small phase variation φ is V 0 (φ φ 2 ) = V [ max 1 + cos(φ φ ] 2 2 ). (3.5) V 0 ( π 2 φ 2 ) = V [ max 1 + sin( φ ] 2 2 ) = V max 2 = V max 2 (3.6) + V max 2 sin( φ 2 ) (3.7) + V 2, (3.8)