Investigation of cryogenic Er:YAG lasers for Gravitational Wave Interferometry. Sophie Hollitt. Master of Philosophy

Transcription

1 Investigation of cryogenic Er:YAG lasers for Gravitational Wave Interferometry by Sophie Hollitt A thesis submitted towards the degree of Master of Philosophy at The University of Adelaide School of Physical Sciences May, 2015

2 Contents Table of Contents i Abstract v Statement of Originality vii Acknowledgements ix List of Figures xi List of Tables xiii List of Symbols xv List of Acronyms xvii 1 Introduction Lasers for Third Generation Gravitational Wave Detectors Available Laser Sources Single-Frequency Sources High Power Erbium-Doped Lasers Minimising Thermal Distortion in High-Power Lasers Current State of Research of Cryogenic Er:YAG Thesis Overview Energy Levels and Spectroscopy of Er:YAG Introduction Energy levels in Er:YAG Choosing a Pump Wavelength Cross Relaxation, Excited State Absorption, and Upconversion Energy Structure at 300K and 77K Absorption Spectroscopy Techniques Description of Measurement System Characterisation of Measurements Summary of Measurement Techniques Results Measured Absorption Cross-Section at 300 K and 77 K Temperature Dependence of Measured Spectra Impact of Doping Concentration i

3 CONTENTS Measurements at Wavelengths >1535 nm Conclusion Laser Diode Cooling Introduction Diode Properties at 300 K Original Specifications Measured Output Power Beam Quality and Focusing Diode Spectral Properties Spectral Width Effect of Diode Temperature on Wavelength Effect of Diode Temperature on Output Power Summary of Room-Temperature Results Cooling the Pump Diode Cooling Method Cooled Diode Results Conclusion Diode-Pumped Er:YAG laser at 300 K Introduction Construction of the Laser Head Interferometry and Thermal Lensing Optimising Lasing Pump Focusing and Output Coupling Pump Absorption Pump Wavelength Tuning Description of Measurement Effect of Diode Wavelength on Laser Power Effect of Diode Wavelength on Laser Efficiency Conclusion Cryogenic Er:YAG lasers Introduction Slab Mounting and Interferometry Mounting the Slab Interferometry Preliminary Cryogenic Laser Pumped at 1470 nm System Configuration Aligning the Slab and Pump Beam ii

4 CONTENTS Lasing Wavelength Laser Performance Pump Wavelength Tuning Summary of Preliminary Experiments Characterising a Lower-Powered Diode Er:YAG Lasing at 77 K System Design Temperature-Tuned Pumping Measuring the Slope Efficiency Effectiveness of Two-Lens Focusing Conclusion Conclusion Thesis Summary Spectroscopy Er:YAG Lasers Future Work Characterisation of Optimal Pumping Development of a Higher-Power Cryogenic Er:YAG laser A Publications 95 A.1 Conference Publications A.1.1 Room temperature and cryogenic operation of an Er:YAG laser using a cooled InGaAsP diode A.1.2 Comparison of diode pumping efficiency of an Er: YAG laser at 300 K and 77 K for Gravitational Wave Interferometry A.1.3 Development of a cryogenic Er:YAG slab laser for Gravitational Wave Interferometry B Chillers 99 B.1 Chiller Specifications B.2 Diode Hysteresis with Water-Cooled Chiller C MATLAB code 103 C.1 Normalising Absorption Cross-Sections C.1.1 backgroundremover.m driver file C.1.2 importtxtfile.m C.1.3 removebg.m C.1.4 makeabsorption.m C.2 Measuring Beam Propagation iii

5 CONTENTS C.2.1 M2 fit.m C.2.2 M2 fitting.m C.3 Calculating the Thermal Lens C.3.1 makemaxcontrast.m driver file C.3.2 importtif.m C.3.3 maxcontrast.m C.3.4 fringefinder.m driver file Bibliography 115 iv

6 Abstract High power, stable single frequency laser sources are required for gravitational wave interferometry. The next generation of interferometers may require laser sources in the µm band for use with Si test masses and InGaAs photodetectors. We propose a high power cryogenic Er:YAG laser operating at µm for this purpose, adapting existing knowledge about cryogenic Yb:YAG lasers developed at the University of Adelaide. To produce such a laser, further information is required about the viability of Er:YAG in high power, single frequency operation. In this thesis, I report this investigation of the spectroscopy of Er:YAG at room temperature and at cryogenic temperatures ( 77 K) and investigate a variety of wavelengths for diode pumping of an Er:YAG slab laser. Spectroscopy indicates that diode pumping for the 77 K laser slab will be most effective in the nm absorption band, most specifically at the 1453 nm absorption peak. I describe methods for cooling a 1470 nm diode below 0 C to pump this 1453 nm Er:YAG absorption. The cooled diode exhibits up to 9 % increase in slope efficiency and improved beam divergence compared to room temperature operation. I then describe the construction and characterisation of CW Er:YAG lasers at both 300 K and 77 K, tuning the pump wavelength in the nm band. At 300 K, I demonstrate an Er:YAG laser with 4.5 W output power when pumped with 30 W of diode power at 1468 nm, and just under 4 W of output power when pumped with 34 W of diode power at 1456 nm. Both lasers have a threshold of approximately 12 W incident pump power. The laser pumped at 1468 nm also demonstrates a greater slope efficiency relative to incident pump power: 28 % compared to 20 % when pumped at 1456 nm. The development of a preliminary cryogenic Er:YAG laser is also reported. Despite sub-optimal mounting materials and geometries, we demonstrate a cryogenic laser with 5.5 W output power and approximately 6 W threshold under comparable pumping conditions to the 4.5 W 300 K laser pumped at 1468 nm. Unfortunately subsequent studies of the cryogenic slab laser are not comparable to the 300 K Er:YAG laser due to electrical damage to the diode that significantly reduced diode v

7 ABSTRACT power and changed pumping conditions. Nevertheless, these results provide valuable information on the sensitivity of end-pumped cryogenic lasers to mounting conditions and pump focusing that are useful for a future high power design. vi

8 Statement of Originality I certify that this work contains no material which has been accepted for the award of any other degree or diploma in my name in any university or other tertiary institution and, to the best of my knowledge and belief, contains no material previously published or written by another person, except where due reference has been made in the text. In addition, I certify that no part of this work will, in the future, be used in a submission in my name for any other degree or diploma in any university or other tertiary institution without the prior approval of the University of Adelaide and where applicable, any partner institution responsible for the joint award of this degree. I give consent to this copy of my thesis when deposited in the University Library, being made available for loan and photocopying, subject to the provisions of the Copyright Act I also give permission for the digital version of my thesis to be made available on the web, via the Universitys digital research repository, the Library Search and also through web search engines, unless permission has been granted by the University to restrict access for a period of time. SIGNED: DATE: Supervisors: Emeritus Prof. Jesper Munch Associate Prof. Peter Veitch vii

9 viii

10 Acknowledgements First and foremost, I would like to thank my supervisors Jesper Munch and Peter Veitch. Thank you for the time you ve spent helping me find the physics in the midst of the laser engineering work we do, and thank you for your expertise both throughout my project and with editing this thesis. I d also like to thank Miftar Ganija for all he s contributed to this project: for finding a path when I was lost, and for teaching me to squeeze every last drop from our equipment. I d like to thank David Hosken, for his support in helping me build my confidence in the lab, and for all our conversations about the ins and outs of working in research. Life is not a dress rehearsal, so it s been great to have a guide. I m grateful for the assistance of so many other University students and staff throughout my studies, for their contributions great and small. Thank you to Blair Middlemiss and Bob Chivell for helping me find (or make!) equipment, and putting up with me driving the BBQ through the workshop. Thank you to Neville Wild, for helping me troubleshoot all sorts of electronics. And thank you to all of the office staff for the administrative tasks that keep everything running smoothly for research students. Best of luck to all my fellow students undertaking similar studies. Special thanks to Ka, Ori and Elli who ve shared office space, lab experiences and stories, and to Fiorina, Sebastian and Ashby who helped me keep it together to make the OSA and KOALA 2014 a huge success. Finally, thank you to my friends and family who have been supportive of me and tolerant of my long absences from face-to-face contact. Thank you to the Blues Underground (and all of my blues family) for giving me something to look forward to at the end of a long week. Last of all, falling into nearly every category, special thanks to Finn for his tireless support of me through conferences, through intense weeks in the lab, through my long nights editing this thesis, and through anything life could throw at me. ix

11 x

12 List of Figures 2.1 Simplified diagram of energy levels of erbium Ground state and first excited state manifolds for Er:YAG Schematic of measurement apparatus for absorption spectroscopy Schematic of sample mounting method Measured spectra from samples mounted in the cryostat, showing background Comparison of measured spectra from the 1 % doped sample, showing background Cary measurement of the 1 % doped sample Spectra from Figure 2.6 with background removed Absorption cross-section of 1 % doped Er:YAG ( nm) Absorption cross-section of 1 % doped Er:YAG ( nm) Variation of absorption of 1 % doped Er:YAG with temperature ( nm) Variation of absorption of 1 % doped Er:YAG with temperature ( nm) Comparison of measured α for 0.5 % and 1 % samples Comparison of measured σ for 0.5 % and 1 % samples Measurement of 1546 nm absorption at 77 K Measurement of absorption features >1540 nm at 300 K Output power and spectral shape of DILAS E7B W module Output power of DILAS diode at 20 C Power through 25 mm aperture for DILAS diode at 27 C Schematic of beam quality and focusing measurements Image of DILAS diode near focus alongside M 2 x plot Schematic of apparatus for measuring diode spectrum Spectrum of DILAS diode for different supplied currents Repeatability of width measurement Plots of diode tuning behaviour against current and temperature Schematic for measuring diode spectrum and power simultaneously. 44 xi

13 LIST OF FIGURES 3.11 Plot of diode output power against centre wavelength Schematic and photograph of diode cooling canister Schematic and photograph of the freezer manifold Schematic of apparatus for cooled diode measurements Output power from cooling canister at room temperature and below 0 C Spectrum of room temperature and cooled diode: constant current Spectrum of room temperature and cooled diode: constant power Plot of cooled diode output power against wavelength Diagram of slab dimensions Photograph of the laser head assembly Schematic of the laser head assembly Schematic of Mach-Zehnder interferometer Interferograms of pumped and unpumped slab Schematic of the Er:YAG laser Laser performance for different pump diode temperatures Measured pump absorption against diode current Diagram of apparatus for temperature-tuned pumping Laser output power against diode centre wavelength Laser output power against incident diode power Laser efficiency when optimally pumped by cooled or room temperature diode Schematic of cryogenic laser head assembly Cryogenic laser head assembly in the cryostat Schematic of Mach-Zehnder interferometer Process used to increase contrast on interferograms Interferograms for the mounted slab before and after cryogenic cooling Schematic of cryogenic Er:YAG laser Schematic and photograph of angled window mounts for AR windows Schematic of the laser head alignment system Schematic for the pump diode alignment procedure Graph comparing the cryogenic and room temperature lasers Temperature-tuned pumping of the preliminary laser Comparison of diode output power before and after damage Output power of damaged diode against diode centre wavelength Comparison of diode spectral shape before and after damage Comparison of diode emission spot before and after damage xii

14 LIST OF FIGURES 5.16 Diagram of the pump configuration used for the temperature-tuned laser Laser output power against diode centre wavelength for the 77 K laser Laser output power against incident diode power for various cryogenic Er:YAG lasers As in Figure 5.18 with two additional laser curves Schematic of pinhole experiment Plot of fraction of pump power transmitted against diode wavelength 89 B.1 Plot of laser output power against diode current B.2 Plot of laser output power against diode centre wavelength (fixed set-point) B.3 Plot of laser output power against diode centre wavelength (changing set-point) B.4 Plots of laser output power and diode centre wavelength against time 102 xiii

15 List of Tables 1.1 Requirements for Advanced LIGO Selected single-frequency laser sources Selected high power erbium-doped lasers Thermal properties of YAG at 300 K and 77 K Lasing state occupation properties Peak absorption coefficient α and FWHM from measured spectra of 1 % Er:YAG Specifications of DILAS E7B W module Measured beam parameters of DILAS diode Coolant specifications for DILAS diode Properties of different antifreeze mixtures Slab properties measured using interferometry Output coupler transmission at pump and lasing wavelengths Integrated linear expansion coefficient for various materials Focused 1/e 2 spot size before and after diode damage Focused spot size for one lens and two lens focusing xiv

16 List of Symbols α Absorption coefficient (cm 1 ) α T β λ λ l λ p Coefficienct of linear thermal expansion Inversion fraction Wavelength (general) Laser wavelength Pump wavelength σ or σ a Absorption cross-section (cm 2 ) σ e Emission cross-section (cm 2 ) θ Ω A c dn dt E(x, T ) E 1 and E 2 f Far-field divergence half-angle An arbitrary phase A positive constant The speed of light Change of material refractive index with respect to temperature The energy of an excited state x, at temperature T The electric field amplitude of the reference arm and slab arm of the interferometer respectively Used to represent the focal length of a lens f L and f U Fractional occupation of the lower and upper lasing states F (x) G h I I in I out k k x respectively Boltzmann occupation factor of a state x Laser gain, used in the form I out = I in exp(gz) The reduced Planck s constant Intensity Input intensity, before gain or absorption has been applied. Used to represent the background spectrum. Output intensity, after gain or absorption has been applied. Used to represent the spectrum of a sample. Boltzmann s constant The wave number xv

17 LIST OF SYMBOLS K Thermal conductivity L Length L 1 to L 4 The lowest four energy bands in Er:YAG, see Figure 2.2 m An integer M 2 n Beam quality, times diffraction limited Refractive index N Number density of ions in the material(in cm 3 ) N L and N U Number density of ions in the lower and upper lasing states respectively P change P signal t T TEM 00 w 0 The change in power from the mean value P signal, used to calculate intensity stability The mean value of the laser output power, used to calculate intensity stability Time Temperature The fundamental Transverse ElectroMagnetic mode Radial beam width xvi

18 List of Acronyms aligo AR ASE CR CW DI ESA FBG GRIN GWI HR LIDAR LIGO MOPA NA ND NPRO OSA OPL QD RF SBS Advanced LIGO Anti-Reflection Amplified Spontaneous Emission Cross-Relaxation Continuous Wave De-Ionised Excited State Absorption Fibre Bragg Grating GRaded INdex Gravitational Wave Interferometers High Reflectivity LIght Detection and Ranging Laser Interferometer Gravitational wave Observatory Master Oscillator Power Amplifier Numerical Aperture Neutral Density Non Planar Ring Oscillator Optical Spectrum Analyser Optical Path Length Quantum Defect Radio Frequency Stimulated Brillouin Scattering xvii

19 Chapter 1 Introduction 1.1 Lasers for Third Generation Gravitational Wave Detectors Gravitational Wave Interferometers (GWI) expect to detect gravitational waves by measuring the relative motion of large, almost-free test masses several kilometres apart. In current systems, fused silica test masses are used as mirrors in kilometreslong Michelson interferometers, where test-mass displacements as small as m may be detected. [1] So far, this level of sensitivity has been insufficient to detect gravitational waves. The enhanced LIGO interferometer together with the VIRGO collaboration GWI was used to search for gravitational waves from binary systems of stars, neutron stars or black holes, but the sensitivity and range of detection was insufficient to detect any during their shared science run. [2 4] This is thought to be a result of a lack of sufficiently strong gravitational wave sources within the current range of detection. Thus, advanced-generation interferometers are currently being installed and commissioned, and third-generation interferometers are being planned [5]. Both short-term and long-term advances will require modifications to the laser used in the Michelson Interferometer, or will require a completely new laser system design. A laser for GWI must satisfy stringent requirements for intensity noise, frequency noise, pointing stability, and mode quality, while also having sufficiently high power to reduce the effect of shot noise. The requirements for Advanced LIGO (aligo) are listed in Table 1.1, in lieu of any new stability requirements for third generation GWI at present. Note that these requirements are for the free-running laser, before additional stabilisation [6 8]. For the Voyager third generation LIGO project, cryogenic silicon test masses are being considered to reduce acoustic losses and thermal noise [5] while the laser power required increases to W [8]. This change to silicon masses will, 1

20 CHAPTER 1. INTRODUCTION Table 1.1: Requirements for aligo [6] Property TEM 00 power Non- TEM 00 power Frequency Noise Amplitude Noise RF intensity noise Value 165 W <5 W 1 Hz/ Hz at 10 Hz / Hz at 10 Hz 1 db above shot noise of 100 ma above 9 MHz however, require a new laser wavelength. The absorption of bulk silicon in the µm low-absorption band and particularly at 1.55 µm is already under investigation [9]. Possible laser wavelengths in this low-absorption band include 1319 nm from Nd:YAG, 1550 nm from erbium-doped fibre, and 1617 or 1645 nm from Er:YAG. The choice of wavelength in this band depends on several factors. While preliminary measurements of absorption in cryogenic silicon suggest that it is minimised near 1.3 µm [10], selecting a longer laser wavelength reduces the significance of polishing errors in the optical surface. As the GWI measurement is interferometric, the significance of roughness and deviations from the required curve depends on the height of the deviation compared to the wavelength [11]. Longer wavelengths result in a lower significance for any deviations on the surface. Another consideration is the type of detector used for the interferometer. Commercially available InGaAs detectors have a typical spectral sensitivity approximately 10 % greater at µm than at 1.3 µm [12], though InGaAs detectors with a peak sensitivity closer to 1.3 µm also exist. Lastly, the choice of wavelength in the µm band is dependent on the availability of laser technology to meet the strict stability and noise requirements, discussed further in the next section. 1.2 Available Laser Sources One of the simplest ways to satisfy the high-power low-noise requirements for thirdgeneration GWI is to use a low-noise, low-power single-frequency master laser source to injection-lock a higher-power oscillator that is carefully designed to prevent adding noise to the laser. The output of the higher-power oscillator could then be used to either to injection-lock a very high power oscillator, or amplified in a Master Oscillator Power Amplifier (MOPA) configuration. [13] Single-Frequency Sources The two most common low power single frequency laser geometries are the Non- Planar Ring Oscillator (NPRO) and the fibre ring laser, though narrow-linewidth 2

21 1.2. AVAILABLE LASER SOURCES fibre-bragg-grating-stabilised (FBG stabilised) diode lasers are also common. The NPRO configuration is considered the most robust solid state single frequency laser [14], and has an output power of approximately 1 W. NPRO sources have been demonstrated to have low frequency and intensity noise, and satisfy LIGO requirements [13], though fibre ring lasers (also known as fibre loop lasers) have been demonstrated to be very stable [15]. Fibre ring lasers typically have lower output power, but can be combined with fibre amplifiers to produce all-fibre high power lasers. FBG-stabilised diode lasers are also very suitable for all-fibre amplifiers, and they are much more cost-effective than DFB lasers which operate on a similar principle and allow a much higher output power than fibre ring lasers. At present, very few suitable single-frequency FBG-stabilised diode lasers have been demonstrated in the wavelength region of interest, although narrow-linewidth FBG-stabilised diode lasers in this wavelength band do exist [16, 17]. NPRO sources have been demonstrated to have low frequency and intensity noise, and satisfy LIGO requirements [13], though fibre ring lasers (also known as fibre loop lasers) have been demonstrated to be very stable [15]. FBG-stabilised diode lasers may also be used in narrow linewidth applications [?, 16], and they are much more cost-effective than DFB lasers which operate on a similar principle and allow a much higher output power than fibre ring lasers. The development of single-frequency FBG-stabilised diode lasers is, however, limited by available diode technologies in the wavelength band of interest. The relevant characteristics of several single-frequency master oscillators with wavelengths appropriate for third-generation GWI are given in Table 1.2. Unfortunately, high-power diffraction-limited 1319 nm lasers are probably difficult to realise due to their large thermal loading from the large difference in energy between pump and laser photons. This difference is known as the quantum defect, calculated using the formula QD = hc λ p hc λ l where h is Planck s constant, c is the speed of light, and λ p and λ l are the wavelengths of the pump and laser respectively. For an Nd:YAG laser at 1319 nm pumped at 808 nm as in [23], the quantum defect is 590 mev compared to mev for resonantly-pumped erbium lasers. Thus, for example, a 1319 nm NPRO achieved an output power of 1.55 W only by using an undoped end-cap to reduce thermal lensing in the Nd:YAG. [23] Without the end-cap, the same NPRO design could only achieve 0.6 W before being limited by distortion of the lasing mode. Although a >100 W 1319 nm Nd:YAG laser has been developed [24], the beam quality is not stated and so we assume that the output is highly multimode. 3

22 CHAPTER 1. INTRODUCTION Table 1.2: Selected single-frequency laser sources. The intensity stability is presented as a percentage deviation from the stated output power ( P change P signal ) over the timescale presented in the reference, while the wavelength stability λ is given in picometres. λ (nm) Laser architecture Power (W) 1645 Resonantly pumped Er:YAG NPRO M 2 P change P signal (%) λ (pm) Ref % 0.09 [18] 1645 Resonantly pumped Er:YAG NPRO 1645 Resonantly pumped Er:YAG NPRO 1550 Co-doped ErYb fibre ring laser 1550 InGaAlAs/InP Diode with FBG external cavity 1535 Er-doped fibre ring laser 1319 Nd:YAG NPRO with undoped end cap % - [19] % 1.4 [20] % - [21] < 10 4 % - [22] < 10 6 % 1.3 [15] [23] 4

23 1.2. AVAILABLE LASER SOURCES In contrast, a large number of stable single-frequency sources in the µm wavelength band appropriate for an erbium-doped slave laser stage have been developed. We conclude that an erbium-doped laser is more promising for thirdgeneration GWI, and continue to review the development of high power erbiumdoped lasers in the next section High Power Erbium-Doped Lasers Erbium-doped gain media at room temperature are well described in the literature. [25 40] Erbium-doped glasses have wide absorption and emission bands, so Erbium-doped glass fibres are typical in the communications industry. Additional applications in eye-safe LIDAR have driven development of both high power fibre devices [25, 27 31] and also erbium-doped crystalline gain media [35, 36, 38, 40] Fibre lasers are often considered for high power applications as their large surface area to volume ratio allows efficient cooling, while still producing near-diffractionlimited output. Despite the prevalence of single frequency master oscillator sources and high power fibre amplifiers, high power single-frequency operation is often difficult in fibre. Amplified Spontaneous Emission (ASE) acts as broadband noise on the signal and can saturate the gain in the fibre [41], while the power output of the single frequency laser is ultimately limited by Stimulated Brillouin Scattering (SBS) [42]. SBS is a significant problem for high-power narrow-linewidth fibre lasers and though SBS can be mitigated, the techniques required often mean that the laser design is not easily scalable to higher power and single frequency operation at the same time [43]. A single-frequency erbium-doped fibre laser may be able to fulfil the requirements of third-generation GWI if sufficient SBS suppression and appropriate materials can be developed. Investigation of 1.6 µm emission in crystalline Er-doped gain media started with studies of flashlamp-pumped crystals, and moved toward resonant pumping as pump sources at 1.47 µm and 1.53 µm became available. The first resonantly-pumped cryogenic Er:YAG laser was reported by Killinger in 1987 [44], followed in 1992 by the first resonantly-pumped room temperature Er:YAG laser, pumped by an Er:glass laser [45]. In contrast, resonant pumping of Er:YAG for 1.6 µm emission using diodes as pump sources is a relatively recent development. The first such resonantly diode pumped laser result was published in 2005 [46] as diode sources at 1470 nm and 1530 nm were first maturing [32]. A selection of high power CW erbium-doped lasers are listed in Table 1.3. In general, very few high power CW erbium-doped lasers exist compared to rare earth dopants such as Nd and Yb. For fibre lasers, thermally induced damage of the fibre s outer coating is one of the limiting factors of a high power design. Co-doped 5

24 CHAPTER 1. INTRODUCTION Laser architecture Cryogenic side-pumped 2% Er:YAG slab End-pumped 0.5% Er:YAG slab Cryogenic end-pumped 2% Er:YAG slab Cryogenic end-pumped 2% Er:YAG rod End-pumped 0.25% Table 1.3: Selected high power erbium-doped lasers. λ l (nm) λ p (nm) Quantum defect (mev) Power (W) M 2 Ref [47] [26] [48] [49] [36] Er:YAG rod Er Fibre [27] ErYb Fibre MOPA [25] Er Fibre MOPA [28] Er,Yb fibre lasers have a large quantum defect, resulting in a typical value of about 100 W predicted output power limited by thermal damage, compared to 0.55 kw for an equivalent Yb-only fibre with a much lower quantum defect [50]. High power operation of an Er,Yb fibre laser at 1.5 µm also requires simultaneous Yb lasing to prevent catastrophic self-pulsing power fluctuations [25, 51]. Interestingly, the large thermal load in the 151 W ErYb Fibre MOPA provides some SBS suppression, allowing single frequency (linewidth 1 MHz) operation [25], but this linewidth is too large and the SBS suppression effect will not necessarily be scalable to power levels appropriate for third-generation GWI. Thermal distortion effects in solid state lasers are also one of the largest barriers to high-power low-noise operation. The 400 W cryogenic Er:YAG laser [47] has a highly multimode output, which could be an indication of large thermal wavefront distortions or a result of non-uniform cooling and heating in the laser slab. Wavefront distortion may be reduced by adopting the strategy used for the aligo laser, in which four laser heads are grouped in two pairs to compensate for depolarisation loss and thermal distortion [52]. Further reduction in thermal distortion will be needed for third-generation lasers to counteract the increased thermal loading due to higher power operation. 6

25 1.3. MINIMISING THERMAL DISTORTION IN HIGH-POWER LASERS 1.3 Minimising Thermal Distortion in High-Power Lasers As previously discussed, quantum defect heating is a significant contributor to thermal distortion, degrading the beam quality of a high-power laser. Apart from resonant pumping, another way to reduce thermal distortion is to look for materials with a higher conductivity and lower thermal expansion and dn/dt than room temperature YAG, in order to reduce the distortion for the same level of heating. Significant improvements to these thermal properties of YAG can also be obtained via cryogenic cooling, as shown in Table 1.4. Table 1.4: Thermal properties of YAG at 300 K and 77 K Property YAG at 300 K YAG at 77 K Ref. Thermal conductivity K (W cm 1 K 1 ) [53] Thermal expansion α T ( 10 6 K [53] ) [54] dn/dt ( 10 6 K 1 ) 9 < 1.5 [54] As a result of these improvements, heat deposited into a cryogenic laser slab is more easily conducted away from the laser path, and the reduced temperature gradient results in less thermal stress and less thermo-optic distortion. In addition to this effect, a given temperature profile also produces less thermo-optic distortion due to the reduced dn/dt and α T. Combining these effects, there is much less thermal distortion for a cryogenic YAG slab than for the same laser slab and pump heat load at 300 K. As an additional improvement, quasi-three-level gain media become four level gain media when cooled, due the reduced thermal occupation of electronic states (see Chapter 2). This change lowers the threshold of the laser, which further reduces thermal loading of the gain medium and thus reduces the potential for distortion. It has been shown through extensive modelling [55] and later shown experimentally [56] that higher average power is possible at cryogenic temperatures than at room temperature for a given YAG host. This experimental result was demonstrated at the University of Adelaide for a cryogenic Yb:YAG laser, which emits over 200 W of output power in an almost diffraction-limited TEM 00 beam at nm. A number of other erbium-doped gain media have also been investigated in recent years to find the material with the lowest quantum defect for lasing in the µm band, or to find a better thermal conductivity and dn/dt. The majority of the quantum defect work is conducted by M. Dubinskii and N. Ter-Gabrielyan [57 59]. 7

26 CHAPTER 1. INTRODUCTION Although they have had success with tens-of-watts class lasers at room temperature and at 77 K, they have not yet demonstrated a 100 W class laser. Other early research into gadolinium vanadate (GdVO 4 ) host material indicates that it may have a higher thermal conductivity than YAG [58], though this is not yet conclusive. Due to existing experience at the University of Adelaide in both cryogenic lasers and Er:YAG laser development, we choose to focus on cryogenic Er:YAG as an appropriate third-generation laser candidate rather than beginning an investigation of new materials. 1.4 Current State of Research of Cryogenic Er:YAG Resonant pumping of cryogenic Er:YAG is important to reduce the quantum defect and decrease the heat load. The absorption cross section of Er:YAG in the resonant pumping bands is thus important in deciding on a pump wavelength, though other factors such as Excited State Absorption (ESA) may also be important. Fortunately, ESA and other processes which reduce lasing efficiency at 1.6 µm are well-studied in Er:YAG at room temperature and at 77 K. Early studies of upconversion and energy transfer dynamics in erbium-doped materials focused on high doping concentrations (> 25 %) for 2.94 µm lasing, revealing strong energy transfer processes which prevent 1.6 µm lasing. These processes are concentration dependent, however, and early study of upconversion dynamics at low doping concentrations (< 5 %) in 1994 [60] made use of the first resonantly pumped room temperature Er:YAG laser at 1.6 µm. A variety of recent papers investigate the effects of upconversion on Er:YAG lasing at different (low) doping concentrations and/or at different temperatures [61 65]. The absorption spectrum of Er:YAG was first recorded at cryogenic temperatures in the 1960s, and at room temperature in the 1970s, but these were often compiled numerically in tables rather than graphically [66, 67]. While these results were crucial for determining the energy level structure of Er:YAG, they are less useful for characterisation of the comparative absorption strengths and spectral widths in Er:YAG, as required for choosing the pump wavelength for a cryogenic laser. One of the earliest graphically represented cryogenic absorption spectra of Er:YAG was reported in 1965 [68], although only a few of the absorptions in the µm band are visible. A large amount of more recently published work on the spectroscopy of cryogenic erbium-doped lasers is attributed to Larry Merkle, Nikolay Ter-Gabrielyan, and Mark Dubinskii [48, 57]. They have described spectroscopy of Er:YAG and other erbium-doped materials at room temperature and cryogenic temperatures, in search of the erbium-doped material with the smallest quantum defect for cryogenic 8

27 1.5. THESIS OVERVIEW lasing. Unfortunately, many of their papers report results with the stimulated emission cross section covering the absorption cross section [57], or Er:YAG results from 1440 nm to 1550 nm in one plot [48], concealing the width of the absorption peaks. The clearest published absorption spectrum of cryogenic Er:YAG is in [47] by Setzler et al. The doping concentration of the sample measured is unspecified, but likely to be 2 or 3 %, as that doping concentration was used for lasing in the same paper. Although the presence of absorption peaks > 1535 nm is of great interest for minimising the quantum defect or determining the expected laser re-absorption, none of these published absorption cross sections clearly present this information. Furthermore, no comparison of the absorption cross section for different low erbium doping concentrations or at different temperatures between 77 K and 300 K exist. While the absorption cross section is important for determining pump wavelengths for cryogenic Er:YAG, the behaviour and efficiency of the laser when pumped with both spectrally broad and spectrally narrow sources should also be examined. The behaviour of cryogenic Er:YAG when pumped at 1532, 1534 or 1546 nm with a narrow source has been characterised in [48], and a study of room temperature Er:YAG pumped by broad diode sources is presented in [69], though a characterisation of cryogenic Er:YAG absorption and laser behaviour when pumped with a spectrally broad diode does not yet exist. 1.5 Thesis Overview This thesis describes a preliminary investigation of diode-pumped Er:YAG lasers at cryogenic temperatures and at 300 K, in order to assess the viability of constructing a stable, high power TEM 00 cryogenic Er:YAG laser suitable for third generation requirements. The energy level structure and absorption of Er:YAG at 300 K and at 77 K is described in Chapter 2. Possible pumping and lasing schemes are described in detail, and loss mechanisms expected to negatively affect lasing are discussed. This chapter also presents spectroscopic measurements of the absorption cross-section of Er:YAG at 300 K and 77 K in order to find spectral regions suitable for diode pumping or fibre laser pumping. The characterisation of the laser diode used to pump the Er:YAG slab is reported in Chapter 3. A method for cooling the diode below 0 C in order to extend its wavelength tuning range is described. Chapter 4 describes the construction and characterisation of an end-pumped room temperature Er:YAG laser used to determine the pump focusing and output coupling used for later cryogenic lasers. The pump wavelength of the laser is tuned to allow comparison of the effectiveness of diode pumping in the nm and 9

28 CHAPTER 1. INTRODUCTION nm bands. Chapter 5 describes the preliminary operation of a cryogenic Er:YAG laser at 77 K and at approximately 120 K. Characterisation of laser output power for diode pumping in the nm band is presented for both the 120 K and 77 K laser, as well as a preliminary investigation of the 77 K laser pumped in the nm band. Finally, the work presented is summarised in Chapter 6 and potential future developments are discussed. 10

29 Chapter 2 Energy Levels and Spectroscopy of Er:YAG 2.1 Introduction As previously discussed, cryogenic spectroscopy of Er:YAG published in literature often neglects spectral regions of interest such as the 1546 nm absorption suitable for low-quantum-defect pumping, and does not present the absorption spectrum for temperatures between 77 K and 300 K. While the pump wavelength is important in designing a laser, analysis of the energy level structure of Er:YAG also allows us to determine the expected laser wavelength at 77 K or at room temperature, and characterise the laser behaviour using the number of energy levels involved in pumping and lasing. In this Chapter, we describe the energy level structure of Er:YAG and measure the absorption spectrum of Er:YAG at 300 K and at 77 K. In Section 2.2, the energy transfer processes relevant to Er:YAG lasing are discussed, and likely laser wavelengths at 300 K and at 77 K are described. In Section 2.3, the process of measuring the absorption spectrum of Er:YAG is described. Finally, the absorption cross-section of Er:YAG at 77 K and at 300 K is presented in Section 2.4, which also includes measurements of the absorption spectrum at intermediate temperatures between 300 K and 77 K, and additional measurements of the long-wavelength absorptions ( nm). 2.2 Energy levels in Er:YAG Choosing a Pump Wavelength A diagram of the energy levels of Er 3+ ions is presented in Figure 2.1. The 1.5 to 1.65 µm lasing emissions in erbium-doped materials are from the 4 I 13/2 upper mani- 11

30 ESA3 980 nm CHAPTER 2. ENERGY LEVELS AND SPECTROSCOPY OF ER:YAG fold, with the 4 I 15/2 ground state manifold as the lower state of the laser transition. Erbium has a variety of higher energy levels that can be pumped for lasing on this transition, or that degrade lasing efficiency through excited state absorption (ESA) and other processes, which are also shown in this Figure and are described further in later sections. Energy levels and common transitions in Er:YAG 20 4 F 7/2 2 H 11/2 4 S 3/2 Energy (10 3 cm -1 ) F 9/2 4 I 9/2 4 I 11/2 4 I 13/2 4 I 15/2 pump µm lasing µm ESA µm ESA µm CR1 fluorescence 980 nm CR2 fluorescence 520 or 540 nm Figure 2.1: Simplified diagram of energy levels of erbium including cross-relaxation (CR) energy transfer processes [62, 70] and excited state absorptions [71] obtained from various sources. For an efficient laser and effective cooling, it is useful to reduce the laser s quantum defect: the energy difference between the pump photon and lasing photons. The energy from the quantum defect is stored in the laser slab as heat, so to reduce heating and thermal distortion the quantum defect should be as small as possible. The quantum defect can be reduced through resonant pumping: choosing the pump wavelength of the laser to pump into the upper lasing manifold, rather than a higher state. For laser emission in the 1.5 to 1.65 µm band, this means pumping into the 4 I 13/2 upper manifold rather than a higher level. 12

31 2.2. ENERGY LEVELS IN ER:YAG Cross Relaxation, Excited State Absorption, and Upconversion Many of the higher states in erbium have energy level differences corresponding to the energy of the 4 I 13/2 to 4 I 15/2 transition or the 4 I 11/2 to 4 I 15/2 transition. This results in a number of cross-relaxation processes, where two excited erbium ions share or transfer energy between them. The most common of these processes is where two neighbouring erbium ions excited to the 4 I 13/2 state cross-relax to leave one ion in the ground state, and one excited to the 4 I 9/2 state. This cross-relaxation is marked CR1 in Figure 2.1. Many other cross-relaxations at other wavelengths exist, but for this thesis the only other relevant process is CR2 from the 4 I 11/2 level. This process is not generally included in rate-equation models for resonantly pumped Er:YAG as it has minimal effect on the 1.6 µm lasing dynamics: the main effect on lasing is the removal of ions from 4I 13/2 due to the CR1 and ESA1 processes, and further excitation of these ions mostly impacts the laser through heating from multi-phonon decays [63]. Nevertheless, the readily visible green fluorescence of Er:YAG under resonant pumping is attributed to CR2 and ESA of 980 nm fluorescence, and so these processes are included for completeness. Since cross-relaxation is an energy transfer between neighbouring Er ions, it happens much more frequently in highly-doped crystals where the erbium ions are closer together. This increase in energy transfer leads to more highly-excited erbium ions, resulting in a greater proportion of upconversion and fluorescence. As a result, most 1.6 µm lasing crystals have low doping concentrations ( 0.5 %) to prevent excessive energy transfer which removes ions from the upper lasing state, though these effects cannot be eliminated. Another way that erbium ions can be excited to high energy levels is through ESA, where an ion in an excited state absorbs another pump or lasing photon and is excited further. In Er:YAG, ESA of nm wavelengths from the 4 I 13/2 state does not readily occur at room temperature [71, 72], though there is some ESA of lasing wavelengths which contributes to the green fluorescence observed in Er:YAG crystals when lasing [73]. Studies in erbium-doped materials [61, 65] indicate that the amount of upconversion fluorescence increases as the material is cryogenically cooled and increases with doping concentration. For a CW laser, the impact of upconversion is to increase the laser threshold and somewhat reduce the slope efficiency [63, 64, 74], as upconversion and associated fluorescence constitute a loss of lasing ions that must be compensated for with further pumping. 13

32 CHAPTER 2. ENERGY LEVELS AND SPECTROSCOPY OF ER:YAG Energy Structure at 300K and 77K While Figure 2.1 provides an overview of the energy levels of erbium ions, to determine lasing behaviour we need information about the sub-levels involved in the pump and lasing transitions. Figure 2.2 shows the energy levels in the ground state (4I 15/2 ) and first excited state (4I 13/2 ) manifolds of Er:YAG, as well as the thermal equilibrium occupation or Boltzmann occupation of each Stark-split level within the manifolds. The Boltzmann occupation of each state (labelled F (x) for each state x in a given manifold) is calculated according to: F (x) = E(x,T ) exp( ) kt y exp( E(y,T ) kt ) where E is the energy of the state x measured relative to the lowest energy state in the manifold, k is Boltzmann s constant, T is the temperature of the material. Figure 2.2: Ground state and first excited state manifolds for Er:YAG, with individual bands labelled L 1 to L 4 for clarity. Energy of levels at 300 K sourced from [33], originally from [66]. Energy of levels at 77 K sourced from [66]. Boltzmann distributions calculated. We observe that the upper part of the ground state manifold (L 2 in Figure 2.2) is partially occupied at room temperature, but nearly empty at 77 K. For the 16xx 14

33 2.2. ENERGY LEVELS IN ER:YAG lasing transitions, this reduces re-absorption at the lasing wavelength to almost zero, changing the inversion level required for laser gain from 9.2 % at room temperature to 0.1 % when the slab is cooled. Lasing at a 15xx nm transition is also possible, but unfavourable due to sub-optimal populations of the required states. For lasing at <1550 nm (from L 3 to L 1 ), there is strong re-absorption at the lasing wavelength as L 1 is highly populated. For lasing at >1550 nm (from L 4 to L 2 ), the low Boltzmann occupation of the L 4 state makes lasing difficult at room temperature, and nearimpossible at 77 K. These effects can also be explained in terms of the number of independent laser energy levels present in the pumping and lasing scheme. The key part of this designation is whether or not the levels are empty or full, and how quickly each can be depopulated. When pumping at 14xx nm and lasing at 16xx nm, we are utilising all four different bands of the 4I 15/2 and 4I 13/2 levels. At 77 K, the upper bands of each are empty, preventing stimulated emission at the pump wavelength and reabsorption at the laser wavelength. As we are using all four bands independently, this arrangement is often referred to as a four-level laser system. At 300 K, L 4 and L 2 are partially filled, resulting in different lasing dynamics. This is usually referred to as a quasi-three-level system, where the three levels in this case would be L 4, L 3, and the combined 4 I 15/2 ground state. The quasi designation is used because the 4 I 15/2 ground state could be separated into two levels but these levels do not operate sufficiently independently. To determine which lasing level in a given band has greater gain and is more likely to lase, we need to look at the relative occupation of the upper and lower lasing states in greater detail. The gain (G) at each wavelength is related to the intensity of light at the lasing wavelength by I out = I in exp(g) for I out and I in the laser intensity at the output and input faces of the slab under consideration, and where G is given by G = (σ e N U σ a N L )z βf U (1 β)f L for σ e and σ a the effective emission and absorption cross sections of the material at the wavelength of interest, N U and N L the numbers of ions in the upper and lower lasing manifolds, and z the length of the material. The second line results from breaking down the effective cross sections and removing common factors. The 15

34 CHAPTER 2. ENERGY LEVELS AND SPECTROSCOPY OF ER:YAG effective cross section is the absolute cross section σ, weighted by the occupation of the Stark sub-levels contributing to the transition: in this case, f U and f L the upper and lower lasing states respectively. Similarly, when we remove the common factor of the ion density of the material, N U and N L can be replaced with terms involving the inversion fraction β. By rearranging, we write G (f U + f L )β f L giving the slope of G proportional to (f U +f L ) with respect to inversion, and implying that the inversion fraction required for transparency is β t = f L f L + f U in simplest form. Table 2.1 presents the state occupation values, the threshold inversion, and f U + f L for the 1617 nm and 1645 nm lasing transitions. Table 2.1: Lasing state occupation properties 300K 77K Property 1617 nm 1645 nm 1618 nm 1645 nm Upper state occupation (%) Lower state occupation (%) Inversion required for transparency (%) Gain slope f U + f L At room temperature, 1645 nm has a lower transparency threshold, and is more likely to lase at low inversion levels. The transparency threshold for 1617 nm lasing is much higher due to the larger occupation fraction of the lower lasing level which leads to greater re-absorption. As the inversion increases, the larger gain slope of the 1617 nm laser line means that 1617 nm lasing will eventually have more gain and be favoured over 1645 nm lasing at high inversion levels. Excited State Absorption for 1617 nm and 1645 nm is similar [72] and does not contribute much to the choice of lasing wavelength at room temperature. At 77 K, both 1618 nm and 1645 nm lasing have very low thresholds and nearly absent ground state absorption, so the larger gain slope for 1618 nm means that this wavelength will have larger gain than the 1645 nm line above threshold. Examination of the Stark-split energy levels of Er:YAG predicts laser emission lines and pump absorption lines, but doesn t provide information about their absorption strength or spectral width. To determine the optimal pumping for a cryogenic 16

35 2.3. ABSORPTION SPECTROSCOPY TECHNIQUES Er:YAG laser, more information is required about the magnitude and spectral width of different absorption transitions. 2.3 Absorption Spectroscopy Techniques Description of Measurement System The absorption of each sample was measured as shown in Figure 2.3. A SuperK broadband supercontinuum source was used to illuminate the samples. The transmitted beam was focused onto a single-mode fibre and analysed using an Anritsu Optical Spectrum Analyser (OSA). Supercontinuum source Cryostat OSA Delivery fibre Single-mode fibre Collimation lens in fibre end Mounted sample Focusing lens (or microscope objective) Figure 2.3: Schematic of measurement apparatus for absorption spectroscopy While a computer-controlled grating spectrometer could have been used for this experiment (which would eliminate some sources of error related to focusing and aligning into the capture fibre), a fibre-coupled digital spectrum analyser was chosen due to the advantage that a fibre-coupled optical spectrum analyser allows us the freedom to move the spectrum analyser on the table to make room for additional equipment, without impacting the capture of the transmitted light from the sample. The light from the sample is focused into the core of a commercial single-mode fibre, the other end of which can be fitted to the standard connector in the OSA. With a grating spectrometer, light from the sample must be focused into the input slit directly, which may require significant free-space propagation. Due to the height of the cryostat, free-space propagation to a grating spectrometer would have required the use of a beam periscope. The Anritsu OSA has a stated wavelength accuracy of ± 0.3 nm. The actual wavelength accuracy of the OSA in the region of interest was not measured as part of this thesis, in part due to our major interest in tuning pump sources (which reduces the need for an absolute measurement of aborption wavelength), and in part because only one OSA was used for all experiments with a resolution better than 17

36 CHAPTER 2. ENERGY LEVELS AND SPECTROSCOPY OF ER:YAG 0.5 nm, reducing the need to calibrate multiple spectrum analysers to each other. As a result, this wavelength accuracy is neglected in all descriptions of measured wavelength unless specifically stated. Multiple measurements using the same measuring device in the same configuration also offers some additional opportunities. The sampling rate and resolution of the OSA can be decoupled, allowing fast scans to be made at high wavelength resolution. For example, a resolution of 0.1 nm can be combined with sampling every 0.5 nm to collect a fast scan which will not broaden narrow features or combine neighbouring features together, but may miss them when they lie between samples (this method captures power from the 0.1 nm bandwidth surrounding each sample point, but the sample points are separated by 0.5 nm, leaving regions of approximately 0.4 nm unmeasured). In contrast, sampling at 0.5 nm with a resolution of 0.5 nm will measure all the power in each 0.5 nm, summing the power from neighbouring absorptions, which may create the illusion of an absorption peak at a wavelength corresponding to a zero absorption between two narrow features. In practice, decoupling the sampling rate and resolution allows us to stitch low sample rate and high sample rate measurements together, as measurements made at the same wavelength will have measured the same power if the noise in the illuminating source is small. This is valuable for measuring thestrength of narrow absorption features and displaying them on the same graph as other broader features. Three spectroscopic samples of length 2 mm were used for this experiment, at both room temperature and at 77 K. The samples had nominal doping concentrations of 0.25, 0.5 and 1.0 atomic % and were sourced from Scientific Materials. Longer samples would have a greater absorption length and thus greater signal-tonoise on the measured absorptions, but longer samples are also more difficult to mount and align into inside the cryostat, as well as being more expensive. An additional 10 mm long, 0.25 % doped sample used as part of a previous Er:YAG study was also used for some measurements. While spectra for room temperature samples were recorded with the sample resting on a platform to eliminates losses due to the cryostat windows, spectra for cryogenic samples were recorded with the samples mounted to the cryostat cold finger as shown in Figure 2.4. The absorption of the samples was measured using the Beer-Lambert Law: I out = I in exp( αl), where I out and I in respectively are the intensities transmitted through and incident on the sample, α is the absorption per length (usually per cm), and L is the length 18

37 2.3. ABSORPTION SPECTROSCOPY TECHNIQUES Cold finger surface Thermal connection (copper) Aluminium plates Indium layers Er:YAG samples Figure 2.4: Schematic of sample mounting method used for cryogenic measurements of the sample in cm. Rearranging gives α = 1 L ln(i out I in ) = 1 L (ln(i out) ln(i in )) where ln is the natural logarithm. In a more realistic model, Fresnel reflections at the sample faces and cryostat window faces must be taken into account. By taking two sets of measurements - one with the sample, and one without - the absorption in the slab can be isolated. In this case, I out above would be the measured spectrum of the sample and windows together, while I in would be the measured spectrum without the sample in place. In this case, the Fresnel reflection at the sample faces still needs to be removed from our calculated α, but the Fresnel reflection is wavelength-independent over the range of interest and will appear as a zero-offset in the calculation of α. The magnitude of the Fresnel reflection can also be independently calculated using the refractive index of the material Characterisation of Measurements In Section it is suggested that two measurements of the cryostat system - one with the sample in place, and one without - is sufficient to characterise the absorption of Er:YAG using the Beer Lambert Law. Early spectra revealed a lack of reproducibility in the background shape of the curve between measurements of different samples (I out ) and the SuperK source (I in ) through the cryostat windows, as shown in Figure 2.5. The absorption spectra calculated naively using the Beer-Lambert Law show Er:YAG absorption features that are the same size relative to their respective back- 19

38 CHAPTER 2. ENERGY LEVELS AND SPECTROSCOPY OF ER:YAG -25 Measured spectrum Calculated Absorption Power (dbm) Wateruabsorption Er:YAGuabsorption Absorption (cm -1 ) -50 Largeuoffsetsufromuzero Wavelength (nm) Wavelength (nm) SuperKusource 0.5lu2mmusample 0.5lu3.2mmusample Figure 2.5: Measured spectra from samples mounted in the cryostat at room temperature, at 0.05 nm resolution. (Left) The absorption spectra of two samples (red,blue) were measured by scanning the SuperK beam (purple) from one sample to the next. (Right) The absorption α of these samples calculated by naively using the Beer-Lambert Law ground levels, and occur at the same wavelengths. The absorption features present in both the SuperK spectra and the sample spectra from approximately nm are believed to be water vapour absorptions inside the OSA, and are removed by are removed by the division I out /I in although a large background offset remains. As this was a preliminary experiment, additional experiments were required to determine if the large background offset was repeatable under constant conditions. Typical spectra for the 1 % doped sample mounted in the cryostat at room temperature also revealed repeatability issues, shown in Figure 2.6. Each measurement was recorded under consistent conditions for the cryostat, sample and SuperK, with only the alignment into the OSA capture fibre varied. In this case, only I out was recorded to avoid moving the SuperK between measurements, so the y-axis shows 1 ln(i L out) instead of α. Despite being measured under approximately constant conditions, these measured spectra varied significantly in background shape and background level. While it is possible that the variation in background shape could be a property of the SuperK supercontinuum source itself such that each of these measurements had a corresponding I in with the same background shape, the simpler conclusion is that the measured background shape is highly sensitive to fibre coupling. If the numerical aperture (NA) of the lens used to focus into the fibre doesn t match the NA of the fibre core, some of the light may couple into the fibre cladding instead of the fibre core. Additionally, spherical lenses will focus different wavelengths at slightly different positions, adding further wavelength-dependence to the coupling. A variety of different single-mode fibres were tested with the OSA, but in general the 20

39 2.3. ABSORPTION SPECTROSCOPY TECHNIQUES 1/L * ln (I out ), units cm Absorption of room temperature Er:YAG Wavelength (nm) Figure 2.6: Measurements of 1 L ln(i out) for the 1 % doped sample mounted in the cryostat at room temperature on different days. The SuperK path through the samples remains constant. background shape problem could not be eliminated. Light propagating in the fibre cladding could have been eliminated by stripping part of the fibre and applying an index-matching liquid to prevent total internal reflection in the cladding, but as this method irreversibly damages the fibers which were shared with other users of the OSA this method would not have been a good solution. Instead, we use a computational approach to remove the background variation in the measured spectra. Figure 2.6 indicates that there is still too much background shape variation to normalise the data by simply setting the smallest value to zero. We use a calibrated spectrophotometer to find multiple wavelengths where the Er:YAG absorption is zero (or close to zero), and use these wavelengths to determine the shape of the background curve. The room temperature absorption of the 2.4 mm 1 % doped Er:YAG sample was measured in the Cary UV-VIS grating spectrophotometer, and this spectrum is shown in Figure 2.7. This measurement reveals wavelengths where the Er:YAG sample does not absorb any of the incident light (or absorbs such a small fraction as to be negligible), which match with published absorption spectra of room-temperature Er:YAG [34]. Note that due to the large resolution bandwidth of the Cary, many of the points marked as being zero absorption appear non-zero in this image, a result of nearby strong absorption features. 21

40 CHAPTER 2. ENERGY LEVELS AND SPECTROSCOPY OF ER:YAG Absorption (cm -1 ) Cary measurement at 1.6nm bandwidth Wavelength (nm) Figure 2.7: Measurement of the 1 % doped sample at room temperature, using the Cary spectrophotometer. Zero regions of the spectrum are circled in green. (Inset) Detail in nm region. The zero-absorption wavelengths (circled in Figure 2.7) can be used to remove the background variation on the OSA measurements via the following process: ˆ ˆ ˆ A= 1 ln(i L out) is calculated and plotted, and the values at each zero-absorption wavelength are saved. We linearly interpolate between the values at each zero-absorption wavelength, to produce a background curve composed of line segments. This background curve (representing 1 ln(i L in)) is subtracted from the data set A, producing a plot of absorption α. MATLAB code to automate this process is given in Appendix C.1. To demonstrate that this approach yields reproducible results for strong absorption features, the results of its application to the data in Figure 2.6 are shown in Figure 2.8. Additional studies revealed that the peak absorption of strong absorption features are consistent to within 1 % between repeated measurements under the same conditions, or to within 4 % when the fibre alignment is varied. This approach does have some weaknesses when measuring spectral features that are close to zero, or that are far from known zero wavelength points. Linear interpolation between zero wavelength points may represent these regions incorrectly if the spectral background curves too sharply. These effects are demonstrated in Figure 2.8 where the spectra > 1550 nm have much more variation in shape than the larger absorption features < 1550 nm (see inset Figure 2.8). To produce precise results, small absorption features require separate measurement with strongly absorbing samples, presented in 22

41 2.3. ABSORPTION SPECTROSCOPY TECHNIQUES Absorption (cm -1 ) Absorption of room temperature Er:YAG Wavelength (nm) Figure 2.8: Spectra from Figure 2.6 with background removed. (Inset) zoom of nm region showing overlap of measurements. 23

42 CHAPTER 2. ENERGY LEVELS AND SPECTROSCOPY OF ER:YAG Section The measured absorption of strong absorption features at 300 K and 77 K are presented in Section Summary of Measurement Techniques Methods for measuring the absorption spectrum of Er:YAG using an Optical Spectrum Analyser were described. Variations in fibre alignment result in large changes to the background shape, making naive application of the Beer-Lambert Law unreasonable. Instead, the variations in background shape were removed computationally by applying knowledge of wavelengths corresponding to zero Er:YAG absorption. These techniques result in less than 4 % variation in measured absorption strength and produce repeatable results for absorption peaks above the noise floor. We use this background removal method for all measurements of the Er:YAG absorption spectrum in the rest of this thesis. 2.4 Results Measured Absorption Cross-Section at 300 K and 77 K The absorption cross-section σ of an Er:YAG sample can be calculated using the absorption coefficient α and the number density N of Er ions in the sample: α = σn rearranges to σ = α/n. For 1.0 atomic % doping, the number density of Er 3+ ions is ions per cm 3 when Er ions replace Y ions in the YAG lattice, as given by Scientific Materials [75]. The number density can also be calculated from the density of YAG and its chemical composition Y 3 Al 5 O 12. The absorption cross-section of 1 % doped Er:YAG, measured at room temperature and at 77 K is presented in Figures 2.9 and 2.10, with a summary of the peak absorption and FWHM for each absorption peak given in Table 2.2. These are composite measurements, combining back-to-back measurements at different sample rates in order to capture the cross-section accurately. In particular, the 1532 nm peak at 77 K is measured separately at a finer resolution and higher sample rate and stitched into the final graph. The peak wavelengths reported at each temperature have a wavelength repeatability of ±0.1 nm compared to sample sets from other days and sample alignments at the same temperature. Absorption spectra for λ > 1550 nm are presented in Section 2.4.4, and the measured absorption spectra of samples of different concentrations are given in Section As the temperature of the sample decreases, the absorption peaks become taller and sharper and the regions between peaks flatten out toward zero absorption. For example, the peak near 1532 nm decreases in width from 0.7 nm full-width halfmaximum (FWHM) at 300 K to <0.1 nm FWHM at 77 K. Similar behaviour is 24

43 2.4. RESULTS Table 2.2: Peak absorption coefficient and FWHM from measured spectra of 1 % doped Er:YAG. Number of decimal places in measurement related to the sample rate/separation between measured points of spectral measurements used to determine the value. 300K 77K Measured Absorption Width Measured Absorption Width λ (nm) per cm (nm) λ (nm) per cm (nm)

44 CHAPTER 2. ENERGY LEVELS AND SPECTROSCOPY OF ER:YAG x10 Absorption Cross Section (cm 2 ) Absorption cross-section of 1% Er:YAG sample 300K 77K Wavelength (nm) Figure 2.9: Measured absorption cross-section for the 1 % doped Er:YAG sample ( nm), at room temperature (red) and at 77 K (blue) Absorption Cross Section (cm 2 ) x Absorption cross-section of 1% Er:YAG sample 300K 77K Wavelength (nm) Figure 2.10: Measured absorption cross-section for the 1 % doped Er:YAG sample ( nm), at room temperature (red) and at 77 K (blue) 26

45 2.4. RESULTS observed for most other peaks in the nm pump absorption band, and as a result, resonant pumping into this band at 77 K requires narrow pump sources (<1 nm). In the nm pump band, many absorption peaks remain relatively broad ( 1 nm or larger). Unlike the nm band, the absorption between peaks is often nonzero in the nm band at 77 K, for example between the peaks at 1453 and 1458 nm Temperature Dependence of Measured Spectra The absorption spectrum of the 1 % sample was also measured at selected additional temperatures between 300 K and 77 K. These measurements are snapshots without the multiple sweep averaging and varied sampling rate used for other measured spectra, as scans at high sampling rates take long enough that the variation in the temperature across the scan would would be high, possibly more than 10 degrees.with a faster OSA, higher sampling rates could be recorded, but none were available at the time of this experiment. The temperature-dependent spectra are shown in Figures 2.11 and In the 14xx nm absorption band, the strength of the 1453 nm and 1470 nm absorptions increases more between 123 K and 79 K than for other absorption lines in the band. This difference could be used as an indication of temperature in the Er:YAG samples if no other method of measuring sample temperature is available. Combined with the laser wavelength change discussed in Section 2.2.3, the temperature of a cryogenic Er:YAG laser slab during lasing could be tracked spectroscopically. This is examined further in Chapter 5. In this band, we also observe a slight decrease in the strength of the 1468 nm absorption. In the 15xx nm absorption band, the absorptions at 1514 and 1516 nm almost double in peak absorption between 120 K and 77 K, while the 1527 nm peak begins to split from the neighbouring 1528 nm absorption at approximately 140 K. Unfortunately due to the low sampling rate in these measurements, the behaviour of the other peaks in this band were not monitored accurately below approximately 140 K, but the creation of flat zero-absorption regions between the major absorption peaks can be observed. By examining the changes in the absorption profile at different temperatures, we can observe some changes in physics as the temperature (and thus the phonon energy) decreases. Further study would be required to make conclusions about the energy level dynamics of Er:YAG at intermediate temperatures, but this study does give an indication of absorption features that would be relevant to a cryogenic Er:YAG laser operating at a temperature above 77 K. 27

46 CHAPTER 2. ENERGY LEVELS AND SPECTROSCOPY OF ER:YAG Variation of absorption with temperature 7 Absorption of 1% sample (cm -1 ) Wavelength (nm) 296K 255K 195K 143K 123K 79K Figure 2.11: Variation of absorption of 1 % doped Er:YAG with temperature ( nm) Variation of absorption with temperature 20 Absorption of 1% sample (cm -1 ) Wavelength (nm) 296K 255K 195K 143K 123K 79K Figure 2.12: Variation of absorption of 1 % doped Er:YAG with temperature ( nm). Note that some peaks around 1530 nm appear to shrink at fixed width as the temperature decreases, but this is actually due to the absorptions becoming narrower than the measurement wavelength sampling rate 28

47 2.4. RESULTS Impact of Doping Concentration While changing doping concentrations of erbium effects the rates of energy transfer processes (see Section 2.2.2), the doping concentration can also influence lasing through changes to the absorption cross-section. Doped ions distort the YAG lattice, and as the concentration of doped ions increases, these distortions in the lattice alter the Stark splitting within states. As Er ions are similar in size to Y ions, distortions due to Er doping are expected to be less than for doping with Yb or Nd ions [76,77], but since we have already measured a variety of samples, it is prudent to compare the data. A comparison of the absorption coefficient and the calculated absorption crosssection of the 0.5 % and 1.0 % samples at 77 K measured under the same conditions as part of the same cooling cycle are shown in Figures 2.13 and Absorption of samples at 77K Absorption (cm -1 ) Wavelength (nm) 1.0% doped sample 0.5% doped sample Figure 2.13: Comparison of measured absorption coefficient α for 0.5 % (blue) and 1 % (red) doped samples at 77 K We find that no significant variations in the absorption cross-section were measured from different doping concentrations in the nm spectral region at 77 K. The comparison of the 0.5 % sample and 1.0 % sample here shows what might be slightly stronger absorption on the 0.5 % sample, but in this case this is more likely to be attributed to noise: the 0.5 % sample has a noise level of almost cm 2, comparable to the difference between the measurements. Measurements at room temperature from all the 2-3 mm samples and the 10 mm 29

48 CHAPTER 2. ENERGY LEVELS AND SPECTROSCOPY OF ER:YAG Absorption Cross Section of samples at 77K 7 x Absorption Cross Section (cm 2 ) Wavelength (nm) 1.0% doped sample 0.5% doped sample Figure 2.14: Comparison of measured absorption cross section σ for 0.5 % (blue) and 1 % (red) doped samples at 77 K 0.25 % sample have the same peak shape and structure within error margins given in Section For a more in-depth study of possible variations in absorption due to doping concentration, longer samples should be used to improve signal to noise. This would also require a more precise knowledge of the doping concentration of each sample. In this thesis, the nominal doping concentration is used for each of the samples, but this concentration is confirmed using comparisons between the measured absorption of different samples. As a result, we were able to identify a 0.5 % doped sample that had been marked as 1.0 % doped Measurements at Wavelengths >1535 nm As discussed in Section 2.3.2, small absorption features were often subject to measurement error. Since the effect of different doping concentrations was on absorption cross-section was minimal, however, this presents an opportunity to use a mixture of samples to find the lowest-noise measurements of small or noisy features. Figure 2.15 shows the 1542 and 1546 nm absorptions of cryogenic Er:YAG in high detail, measured using the 1 % doped sample. Pumping the 1546 nm peak minimises the quantum defect for lasing at 1618 nm, but would require a spectrally narrow pump source for high power lasing to minimise transmitted pump power and maximise pump power efficiency, as the absorption is only 0.24 nm FWHM. 30

49 2.4. RESULTS 0.8 Absorption of 1% doped Er:YAG at 77K Absorption (cm -1 ) Wavelength (nm) Figure 2.15: Measurement of 1546 nm absorption at 77 K Figure 2.16 shows the 300 K absorptions at wavelengths > 1550 nm, measured using longer Er:YAG samples and slabs. The 10 mm 0.25 % slab is used in some earlier measurements, but new measurements include a 22 mm 0.25 % doped slab from a previous Er:YAG laser, and the 20 mm 0.5 % Brewster-faced slab used for lasing later in this thesis (see Chapter 4). The spectra of the longer slabs were recorded by Miftar Ganija. All three samples have consistent peak absorption cross section values (within 1 %) for all absorption peaks in the nm band, including the 1546 nm peak shown in the left of Figure Unfortunately, the measured absorption at wavelengths >1550 nm is not consistent between samples, as the 10 mm sample measurement had greater variation across the background shape compared to the strength of the absorptions. The measured cross-sections for the 0.25 % and 0.5 % samples have similarly low noise levels due to their similar length, and because they were measured under the same conditions on the same day. As in the previous section, there is little difference between the absorptions of samples with different doping concentrations measured under the same conditions. We observe absorption peaks near 1571 nm, 1585 nm, 1601 nm, 1617 nm for both the long 0.5 and 0.25 % slabs, although further measurements would be required to characterise these >1550 nm absorptions in greater detail with greater precision and repeatability. 31

50 CHAPTER 2. ENERGY LEVELS AND SPECTROSCOPY OF ER:YAG Absorption Cross Section (cm 2 ) x Er:YAG absorption at 300K 10mm 0.25% sample 22mm 0.25% slab 20mm 0.5% Brewster slab Wavelength (nm) Figure 2.16: Measurement of absorption features >1540 nm at 300 K, using a selection of longer samples 2.5 Conclusion Er:YAG has complicated energy level dynamics, requiring an in-depth study of different pumping methods for a cryogenic Er:YAG laser. This chapter described methods for measuring the absorption cross section of Er:YAG, and presented absorption results at 300 K and 77 K for a variety of Er:YAG samples. The pumping wavelength for an optimal laser must also be determined. Resonant pumping into the 4 I 13/2 level - which includes the upper lasing state - will minimise the quantum defect and thus minimise thermal distortions in the laser slab. a result, the absorption spectrum of Er:YAG samples in this resonant band was measured and presented in Section 2.3. Many nonlinear effects are dependent on the erbium doping concentration, and as a result we examined whether doping concentration had an appreciable effect on absorption cross section. Spectroscopic measurements reveal no significant difference between the absorption cross-section of Er:YAG samples with low doping concentrations ( 1 %) in the resonant pumping band. This means that optimal doping concentration for a cryogenic CW Er:YAG laser is mostly restricted by pumping conditions and engineering constraints on the slab dimensions, although upconversion losses will still require some consideration. The minimum possible quantum defect occurs when pumping into the As

51 2.5. CONCLUSION 1550 nm wavelength band, though all of the absorptions in this band are measured less than 1 nm wide and would require an expensive spectrally narrow pump source for good pump efficiency. In contrast, the nm band has several absorptions (1453, 1458, and 1466 nm) which remain >1 nm at 77 K, with broad absorption regions between these peaks. Multiple peaks in this absorption band could also be pumped simultaneously with a spectrally wide pump source (>5 nm). 33

52 34

53 Chapter 3 Laser Diode Cooling 3.1 Introduction Laser diodes are an effective means of pumping room temperature Er:YAG lasers. Studies have determined the optimal spectral regions for pumping room temperature Er:YAG slabs [69], but to our knowledge no diode pumping study of a cryogenic Er:YAG laser exists. While there has been some investigation of using fibre laser sources to pump a cryogenic Er:YAG laser [48], and diode-pumped cryogenic Er:YAG lasers have been reported [47,49], there is little evidence of an optimal diode-pumping strategy for a cryogenic Er:YAG laser. In this chapter, we investigate the power and spectral properties of a diode laser source at room temperature and when cooled below 0 C, with the aim of using this source for diode-pumping studies to determine if it is advantageous to pump Er:YAG in the nm band. The power and focusing of the diode is described in Section 3.2, while the spectral properties are presented in Section 3.3. Finally, in Section 3.4 we describe the method used to cool the diode, with the results presented in Section Diode Properties at 300 K All lasing studies reported in this thesis used a three-bar fast axis collimated In- GaAsP diode from DILAS as a pump source. The characterisation of this diode at room temperature allowed us to confirm its specified performance, and to provide a base of comparison with cooled operation. In this section, the efficiency, threshold, focal spot size, and spectral properties of the diode are measured. The temperature of the diode is also tuned within a room temperature operating range to investigate the dependence of output power and wavelength on temperature. 35

54 CHAPTER 3. LASER DIODE COOLING Original Specifications This DILAS diode used was already somewhat aged at the beginning of this thesis, and as a result had somewhat reduced efficiency from original specifications. The original stated performance of the diode is presented in Table 3.1. Table 3.1: Specifications of DILAS E7B W module Parameter Centre wavelength Output power Polarisation Spectral width Threshold current Operating current Operating voltage Operating temperature Collimation DILAS module nominal value nm 40 W Linearly polarised <13 nm 4.2 A 36 A 3.15 V C Fast and Slow axis The output power curve for the diode supplied by DILAS is shown in Figure 3.1 (top). The emission spectrum at a current of 55 A was also supplied by DILAS, shown in Figure 3.1 (bottom). Note that this current is significantly above the diode s specified typical operating current of 36 A (40 W optical). The DILAS laser module is water cooled via internal micro-channels, and is mounted on a Delrin water manifold to connect to the rest of the water cooling system. Conductivity-stabilised de-ionised water is supplied by a Termotek DIwater chiller. Further information on these systems is given in Appendix B Measured Output Power Figure 3.2 shows the output power of the laser diode against supplied current, measured in March 2013 with the diode cooling at 20 C. This is the full output power of the diode, measured close to the diode with a large-aperture power meter to include components with larger beam divergence. This measured slope of W/A matches the diode slope efficiency from the original specifications, although Chang reports a diode slope efficiency of 1.45 W/A for the same diode in his thesis [78]. The measured threshold current of 4.5 A is a little higher than specifications, but matches Chang s 2010 result. Decay in the operating properties of the diode since its previous use is attributed to the diode s overall age, and improper storage in the 2-3 years since its previous use. Leaving the diode connected to the water supply without water flow degrades the humidity cartridge and may lead to condensation on the diode facets when the 36

55 3.2. DIODE PROPERTIES AT 300 K Figure 3.1: Output power and forward voltage (top) and spectrum (bottom) of DILAS E7B W module supplied with diode. Measured output power (W) Output power of DILAS diode y = 1.26x Current (A) Figure 3.2: Output power of DILAS diode at 20 C, measured on 7th March

56 CHAPTER 3. LASER DIODE COOLING diode is next cooled. Additionally, the diode was also left connected to the power supply instead of being shorted closer to the diode, risking electrical damage. The power plotted in Figure 3.2 is the total power from the diode as measured using a large aperture power meter placed close to the diode. For the results presented in this thesis, however, the output of the diode must be focused onto the gain medium within the cryostat, passing through a 25 mm aperture at the focusing lens, and at the cryostat window where applicable. As a result, we are careful to include either an aperture or the pump focusing lens in measurements of the diode in order to capture the main emission of the three diode bars relevant for pumping the slab, and exclude the low-power halo of additional emission with a larger divergence or incorrect pointing angle. As the lenses used are AR coated for the diode pump wavelength, these methods are equivalent within the error of the power meter s measurement. The measured power which would be incident on the slab is plotted in Figure 3.3 with 5 % error bars from the power meter measurement. The diode output power is measured here at 27 C, which is the temperature used for laser measurements in later chapters. The output power at different temperatures is examined later in Section If we compare the slope efficiency of the 16/06 measurement with the 05/01 measurement, we see a 6 % decrease in performance which may indicate decay of the diode attributed to operating outside of specifications as part of diode cooling experiments (see Section 3.4), but could also be attributed to the power meter error Beam Quality and Focusing The focusing distance and focal spot size of the diode was measured using a camera in order to determine the pump intensity profile inside the laser slab. The apparatus for this experiment is shown in Figure 3.4. To prevent damage to the InGaAs camera, all measurements were performed at low current (8 A with f=100 mm lens, 10 A with f=200 mm lens) with a number of Neutral Density (ND) filters between the diode and camera. Using a camera to measure the M 2 of a non-gaussian beam is a somewhat unusual way to measure its width, the standard method is to use a knife-edge measurement. In this case, since we were also interested in recording the intensity profile of the beam and estimating the divergence angle, a visual method was used. Fits to the measured profiles were performed using MATLAB code written by Ka Wu and Nick Chang, with some modifications. This code is included in Appendix C.2. Figure 3.5 shows both InGaAs camera images, and a plot of diode beam radius against camera position as M 2 fit for the diode. The minimum beam dimensions, diode M 2 and diode far-field divergence angle measured under different focusing 38

57 3.2. DIODE PROPERTIES AT 300 K Power incident on slab (W) DILAS diode pump performance 16/06/ /09/ /01/2015 y=1.13x y=1.06x Current (A) Figure 3.3: Power through 25 mm aperture for DILAS diode at 27 C, measured at different times over the course of this thesis. The 5 % error bars for the power meter are shown. Diode lens ND filters InGaAs camera (on moveable stage) 10-15cm Fixed ruler Figure 3.4: Schematic of apparatus for beam quality and focusing measurements of DILAS diode. The distance between the diode and lens is comparable to that used for lasing. 39

58 CHAPTER 3. LASER DIODE COOLING conditions are given in Table 3.2. Note that the width and height of the pump intensity profile were calculated at the 1/e 2 level, and that this is the full width of the diode rather than a half-width or radial width. The far-field divergence halfangle θ is calculated from the measured M 2 according to: θ = M 2 λ πw 0 for λ the diode wavelength and w 0 the minimum beam radius. Photograph of beam using InGaAs camera Fitting software calculates 1/e 2 width vertically and horizontally Beam Radius in y (mm) Vertical M 2 fit for f=100mm lens fitted curve M 2 = Beam propagation (mm) Figure 3.5: Image of DILAS diode near focus (top left), the same image with beam height and width marked by fitting program (bottom left), alongside M 2 y plot for data measured with f=100 mm lens (right) Table 3.2: Measured beam parameters of DILAS diode. Small-spot propagation distance labels the estimated distance for which the measured spot is <1.1 mm in diameter in y. Focusing Condition Min. Beam Diameter (mm) Small-Spot Propagation Distance (mm) M 2 Divergence Half-Angle ( ) x y x y x y f =100 mm f =250 mm

59 3.3. DIODE SPECTRAL PROPERTIES 3.3 Diode Spectral Properties Spectral Width The spectral properties of the diode were measured using a reflection from a glass wedge (or later, from a microscope slide) to limit the diode power incident on the capture fibre for the Anritsu OSA. A schematic of the system used for this experiment is shown in Figure 3.6. Diode Glass wedge Power meter Cooled using water-cooled chiller to OSA Figure 3.6: Schematic of apparatus for measuring diode spectrum A comparison of the diode spectral shape and centre wavelength at constant coolant temperature but different supplied current levels is given in Figure 3.7. Each spectrum is normalised to have area 1 to allow simpler visual comparison of different widths. The diode width at 14 A is approximately 5 nm, while the diode width at 33 A is approximately 7 nm. This indicates that the diode will be suitable for pumping two neighbouring Er:YAG absorption peaks simultaneously, as the typical separation of neighbouring absorption peaks in the nm absorption band is approximately 4-5 nm. These measured widths are significantly smaller than the 12 nm width of the diode at 55 A given in the diode s data sheet. A precise numerical value for the width of the diode is hard to characterise and compare between repeated measurements due to the presence of large spikes in the diode spectrum, which can cause errors in a naive FWHM measurement from a single measured spectrum. Instead, several measurements of the diode at constant temperature and current from different days are recorded and compared graphically, shown in Figure 3.8. Ignoring the large spikes in the dataset and instead taking the diode maximum at the centre of the noise envelope (see Figure), the width of the spectrum can be determined to within 5 %, or about 0.2 nm with multiple measurements. For completeness, the width of the spectral shape was also measured using only the top and bottom edges of the noise envelope, but these measurements were much less consistent than using the middle. The variation in 41

60 CHAPTER 3. LASER DIODE COOLING (Arbitrary Units) Diode spectra at 20 C 14A 22A 33A Wavelength (nm) Figure 3.7: Spectrum of DILAS diode for different supplied currents, at 20 C. Both the centre wavelength and width of the spectrum change with current. centre wavelength observed is a result of temperature drift on the water cooled chiller (see Appendix B.2). Measurements of the spectrum were also recorded using an aluminium scatter plate or a Teflon sheet instead of the glass wedge in order to ensure that the capture fibre samples all of the diode emitters. The entire diode spectrum differed in shape from measurements of particular groups of emitters, with different groups of emitters tending to have distinguishing features such as an extended low-power tail, or greater number of tall spikes. Nonetheless, the spectra of individual regions of the diode were found to be consistent and repeatable over time, and maintained a consistent relationship with the spectrum of the entire diode Effect of Diode Temperature on Wavelength The wavelength of a semiconductor diode is affected by its temperature. The system used to measure the temperature dependence was similar to that given in Figure 3.6. For this experiment, however, a fan-cooled DI chiller was used rather than the water cooled chiller, giving a more precise control over the coolant temperature. The results of this experiment are plotted in Figure 3.9. The measured diode centre wavelength varies by 0.26 nm/a, and 0.45 nm/ C. These measurements allow us to estimate the diode temperature required in order to 42

61 3.3. DIODE SPECTRAL PROPERTIES 0.35 Diode spectra for 33A current at 20 C (Arbitrary Units) Maximum of spectrum chosen in the middle of the noise shape (rather than top or bottom) Wavelength (nm) Figure 3.8: Repeatability of width measurement at 20 C Wavelength against current at 20 C Wavelength against temperature at 33A Diode Peak Wavelength (nm) y = 0.26x Current (A) Diode Peak Wavelength (nm) y = 0.45x Chiller temperature ( C) Figure 3.9: Plots of diode tuning behaviour against current for constant chiller temperature 20 C (left), or against temperature for a constant diode current of 33 A (right). 43

62 CHAPTER 3. LASER DIODE COOLING pump the 1453 nm Er:YAG absorption using this diode. At 33 A supplied current to the diode, the extrapolated temperature is -9.6 C, well outside the normal operating conditions of the diode Effect of Diode Temperature on Output Power The temperature dependence of diode output power at constant current was measured using the apparatus in Figure An aperture is included to block diode power that would not be focused onto the laser slab. For this experiment, the waterto-water DI chiller was used. This chiller has a lot of temperature hysteresis (see Appendix B.2), so an experimental method was developed to use the temperature changes in the chiller to slowly tune the diode centre wavelength. This tuning is most consistent when the chiller starts cold and then is set to a high temperature. As a result, the temperature (and thus wavelength) varies smoothly with time from close to 15 C up to 32 C and is stopped before the chiller reaches its a set-point of 35. Diode Microscope slide Aperture Power meter Cooled using water-cooled chiller to OSA Divergent low-power "halo" Figure 3.10: Schematic of apparatus for measuring diode spectrum and power simultaneously. Note that the aperture is used to block stray power that would not be focused onto the slab, as detailed in Section In this experiment, diode current is kept constant while the coolant temperature gradually increases. The diode spectrum and output power are recorded simultaneously at regular time intervals. While the diode centre wavelength is displayed on the OSA and could be recorded without saving the full diode spectrum, recording the spectrum also allowed us to check that the width and spectral shape of the diode remained consistent as the temperature changed. The diode tuning curves are presented in Figure The left-most point on each curve approximately corresponds to 15 C, and the right-most point to 32 C. We plot diode output power against diode centre wavelength, rather than against temperature: wavelength is the quantity we measure with greater precision in this experiment, and wavelength is more directly relevant to pumping a laser crystal than the diode temperature. 44

63 3.3. DIODE SPECTRAL PROPERTIES On the curve at 33 A, we can see a point which has fallen from the line of best fit: when the diode switches from heating to cooling the relationship between diode wavelength, temperature, and output power changes, and the points form a different curve (see Appendix B.2). Measured diode power (W) Diode tuning curves near room temperature A 30A 27A 25A 22A Diode Centre Wavelength (nm) Figure 3.11: Plot of diode output power against measured diode centre wavelength, for a variety of diode current levels. With this characterisation, the supplied current and measured diode wavelength are sufficient to determine the diode power as the diode is heating. If a measurement of diode temperature is required, these curves can be combined with the measured diode slope constant from Section to calculate an approximate diode temperature from diode current and output power alone Summary of Room-Temperature Results The power and spectral properties of the diode were examined in detail, including characterisations of the diode spectral width, centre wavelength, and temperature tuning. The diode has a measured spectral width of 5-7 nm FWHM as the supplied current varies from 15 A to 33 A, suitable for pumping two neighbouring Er:YAG absorption peaks simultaneously. A method to characterise the diode power that would be incident on the laser slab was determined, and this method was used in further experiments to characterise the diode output power and efficiency with respect to diode pump wavelength. We observe a maximum useable output power 45

64 CHAPTER 3. LASER DIODE COOLING of 29 W at nm when monitoring the centre wavelength, corresponding to 33 A supplied current at approximately 15 C. The wavelength tuning range of the diode for room temperature operating conditions is approximately 5 nm, spanning nm at 33 A. The key measurement from this early characterisation is that the diode wavelength shifts by 0.45 nm/ C, predicting a diode temperature of -9.6 C to access the 1453 nm pumping peak. In the next section, we describe a system developed to cool the diode to at least -10 C in order to access this wavelength region. 3.4 Cooling the Pump Diode Cooling Method In order to cool a laser diode without damaging it, care must be taken to avoid condensation on the diode facet surfaces. The best way to achieve this is to enclose the diode in a dry nitrogen environment, so that no water is present to condense onto the cold laser facets. A photograph and schematic of the aluminium cylinder constructed for this purpose is shown in Figure The canister is opaque with an Anti-Reflection coated (AR-coated) window mounted on the front to transmit the pump light. An earlier version of the diode canister used a microscope slide as the front window, but this resulted in an additional 8 % loss of pump light due to surface reflection. The canister has ports for the power cables and coolant tubes, and can be sealed with tape. The top has an inlet so that the box can be filled with nitrogen, used to flush water vapour from within the cylinder. Dry nitrogen supply Lid Power cables Window Dotted lines: slots cut in back for power/water Water supply Diode and canister bolted to same surface Figure 3.12: Schematic (left) and photograph (right) of diode cooling canister. This is the second iteration of the design, featuring an AR coated window. Cooling a water-cooled diode below 0 C requires some additional considerations. A conduction-cooled diode could be cooled directly with a Peltier cell to the appro- 46

65 3.4. COOLING THE PUMP DIODE priate temperature, with heat removed from the hot side of the Peltier via water cooling. With a water-cooled diode, however, the cooling liquid passes through the diode case directly and is used to set the diode temperature. In order to cool a water-cooled diode below zero degrees Celsius, an appropriate substitute coolant must be found. The coolant specifications for the DILAS diode are given in Table 3.3. Table 3.3: Coolant specifications for DILAS diode Coolant property Value Coolant flow rate 1 L/min Temperature of coolant C Conductivity < 2 µs/cm ph-value Suspended particle size < 5 µm For a goal temperature of -10 C, the DI coolant water for the diode must be replaced by a mixture of DI water and an appropriate liquid to prevent freezing. The initial options in this case are ethanol, ethylene glycol, and propylene glycol. The properties of each of these antifreeze liquids are given in Table 3.4. Table 3.4: Properties of different antifreeze mixtures (50/50 with DI water), sourced from [79] unless specified. Property Freezing point of mixture with water ( C) DI water ethanol ethylene glycol propylene glycol Dynamic viscosity (centipoise) at +25 C 22 at -20 C 42 at +25 C Specific Gravity at +20 C [80] 1.1 at -20 C at +15 C Specific Heat Capacity (kcal/(kg C)) at +20 C [81] 0.78 at -20 C 0.85 at +15 C Heat transfer capacity (specific gravity * specific heat capacity) Flammability at high concentrations at +20 C 0.86 at -20 C 0.88 at +15 C nil significant nil nil Initial trials with ethanol were unsuccessful as the heat transfer capacity of 47

66 CHAPTER 3. LASER DIODE COOLING ethanol is low due to low specific gravity. Instead, ethylene glycol was chosen as the coolant rather than propylene glycol, as it is compatible with more plastics and more types of PPE [82]. Commercial ethylene glycol coolants contain inhibitors that prevent corrosion and permit a long useable lifetime, but due to our low conductivity requirement we must use spectroscopic grade ethylene glycol in DI water. A 50/50 mixture of ethylene glycol in water has conductivity 5 µs/cm [83], though here the purity of the water is unspecified. We assumed that in a DI water mixture the conductivity would be sufficiently low that short bursts of operation should not damage the diode. A schematic and photograph of the final system used to cool the diode are given in Figure The ethylene glycol mixture is cooled to -20 C using a commercial chest freezer, which also housed the pump and manifold for the diode-cooling system. Using the freezer to house the chiller manifold and pump reduces the amount of heat transfer into the coolant from the chiller pipes, and also provides a simple way to return the coolant to the appropriate temperature between diode runs. The red insulated pipes, heat exchanger, reservoir and pump for this chiller were sourced from an inoperable Termotek DI water chiller. The chest freezer was not powerful enough to maintain cold temperatures while the pump and diode were running. As a result, we could not record any diode properties at a constant below-zero temperature, and instead the properties were recorded as the diode warmed slowly, similarly to the method used in Section Cooled Diode Results A schematic of the system used for measurements of the cooled diode is shown in Figure Many of the measurements reported in Section 3.2 are repeated in this section for the cooled diode. A notable exception is the diode focusing measurement; since the diode temperature could not be stabilised below 0 C, beam profiles could not be measured at constant temperature. An initial measurement of the diode output power during cooling is shown in Figure The output of the diode was measured in the same configuration at room temperature and when cooled. Cooling the diode results in moderate improvements to slope efficiency and output power. The slope efficiency of the diode laser improved by 3.5 % as a result of cooling and lasing threshold decreased from approximately 5 A to 2 or 3 A. Combined, these effects result in 10 % extra output power at 33 A current. As a result, the maximum current-limited output power of the diode at room temperature can be obtained with only 29 A supplied current when cooled sufficiently. The spectrum of the cooled diode is shown in Figure By overlaying the 48

67 3.4. COOLING THE PUMP DIODE To diode Particle filter Flow control Heat exchanger Pump Diode return Reservoir Flow meter/flow control Heat exchanger Diode return Particle filter To diode Reservoir Pump Figure 3.13: Schematic (top) and photograph (bottom) of the freezer manifold Diode Cooling canister Microscope slide Power meter to OSA Figure 3.14: Schematic of the apparatus used to measure the power and spectrum of the cooled diode. The canister window and the microscope slide are both uncoated 49

68 CHAPTER 3. LASER DIODE COOLING Measured power (W) Comparison of room temperature and cooled diode Room temperature diode Cooled diode y = 1.06x y = 1.03x Current (A) Figure 3.15: Output power of cooled diode at room temperature (red) and when cooled below 0 C (green) spectra for different temperatures but constant current, we find that no statistically significant change in spectral shape or width can be observed with respect to this temperature change. Narrowing is, however, observed for the diode at constant optical power, as shown in Figure Due to the efficiency improvements of operating at colder temperatures, the current can be reduced to obtain the same output power as at room temperature, which then results in a narrower spectral width due to reduced ohmic heating of the diode junctions. From room temperature (20 C) to 0 C approx, we observe a decrease in width from 6.5 nm ± 5 % at 33A, to 5.4 nm ± 5 % at 29 A. The efficiency and centre wavelength for the cooled diode are plotted in Figure The left-most point on both curves is measured at the coldest coolant temperature, and the right-most point at the warmest coolant temperature (0-5 C when the experiment was stopped). Unlike the measurement of diode power in Section 3.3.3, this is a measurement of the full output power of the diode. At 33 A, the centre wavelength of the diode can reach as low as 1450 nm when cooled using ethylene glycol coolant with an initial temperature of -20 C in the freezer. As a result, this method of diode cooling can be used to tune the diode through the nm absorption region of Er:YAG which would otherwise be inaccessible. 50

69 3.4. COOLING THE PUMP DIODE Diode Wavelength (nm) at -6 C A -6 C 33A 20 C (Arbitrary units) Diode Wavelength (nm) at 20 C Figure 3.16: Diode spectrum at room temperature (red) and when cooled (blue). Note that the cooled curve is shifted in wavelength by 13.5 nm to produce overlap. 0.3 Comparison of diode spectra 0.25 (Arbitrary Units) Wavelength (nm) 29A (29W) at -6C 33A (32.7W) at -6C 33A (29.1W) at 20C Figure 3.17: Diode spectrum at room temperature (blue) and when cooled (red,green). The blue and red curves have the same optical output power. 51

70 CHAPTER 3. LASER DIODE COOLING Output power (W) Diode tuning curves for the cooled diode 33A 20A Diode Centre Wavelength (nm) Figure 3.18: Plot of measured output power against diode centre wavelength for the cooled diode. Both 20 A (red) and 33 A (blue) recordings were recorded in the same warm-up experiment. 3.5 Conclusion This chapter described the power and spectral properties of a DILAS diode. The total diode output at 20 C matches specifications, with a slope efficiency of W/A, a threshold of 4.5 A, and an output power of just over 30 W at 33 A. The spectral width was found to vary with current, and measured approximately 7 nm at 33 A. Tuning curves for diode output power against diode centre wavelength were measured at both room temperature and below 0 C. When cooled, the diode has greater efficiency and a lower threshold. No statistically significant change in spectral width is detected, although the large shift in the measured diode centre wavelength allows the diode to be used to pump Er:YAG in the nm band. In Chapter 4, the diode will be used to a pump a room temperature Er:YAG slab laser. Techniques for spectrally monitoring and cooling the diode developed in Chapter 3 will be used in the rest of this thesis. 52

71 Chapter 4 Diode-Pumped Er:YAG laser at 300 K 4.1 Introduction In Chapter 3, temperature tuning of the spectrum of the diode was reported. When the diode is tuned and used to pump an Er:YAG slab, this method can be used to determine the optimal pump wavelength for Er:YAG at room temperature. In this chapter, construction of a room temperature Er:YAG laser is described as a preliminary to cryogenic laser development. In Section 4.2 and 4.3, techniques for mounting the Er:YAG slab with minimal stress and for checking the stress on the slab using an interferometer are investigated. In Section 4.4 the operation of the Er:YAG laser is initially characterised with the diode at approximately constant temperature. This is then extended to a temperature-tuned pumping study in Section 4.5, where the diode wavelength is scanned in both the nm band and the nm band to determine the optimal pumping wavelength for a room temperature Er:YAG laser. 4.2 Construction of the Laser Head The room-temperature laser head used parts designed by Chang for a Q-switched Er:YAG laser [78]. The dimensions of the 0.5 % doped Er:YAG slab are shown in Figure 4.1. A photograph of the 0.5 % doped Er:YAG laser slab and the copper mounting blocks when assembled is shown in Figure 4.2. The laser slab was mounted between two copper blocks with indium foil sheets to ensure good thermal contact, as shown in Figure 4.3. Good mounting technique requires a uniform crush on the indium, and so ceramic spacers were used to prevent over-tightening the screws and ensure that the final crush is uniform. 53

72 CHAPTER 4. DIODE-PUMPED ER:YAG LASER AT 300 K 20 mm 3.24 mm 3.22 mm Coated pump face: HR at 1645 nm AR at 1470 nm 28.8 Brewster angled face Figure 4.1: Diagram of Er:YAG laser slab with dimensions. The laser output face is cut to be at Brewster s angle for a laser mode propagating along the axis of the slab. Copper cooling blocks Steel spacer Slab Brewster face Ceramic spacers Water cooling connectors Aluminium mounting plate Figure 4.2: Photograph of the laser head assembly using steel spacers, with ceramic spacers shown alongside Screw holes Indium foil Er:YAG slab Ceramic spacers Copper block Figure 4.3: Schematic of the laser head. Not to scale. 54

73 4.3. INTERFEROMETRY AND THERMAL LENSING The spacers must be adjusted to the height of the slab and to the thickness of the available indium to produce the correct level of crush. Too much crush can stress the slab, while too little crush reduces thermal contact between the copper and slab. Great care was also taken to lap the copper faces flat to ensure uniform contact with the slab. It was found that the ceramic spacers available produced excessive crush when the upper copper block was screwed tight onto the spacers, so an alternate technique was developed to tighten the copper block. In this method, successful crush occurs when the screws in the copper are just tight and all three ceramic spacers may rotate when nudged lightly. A locked spacer that is unable to rotate indicates that the screw is too tight, and that the crush on the slab is non-uniform. Since only one successful assembly of the laser head was required, the laser head was assembled using fine motor control to feel the correct level of tension in the screws. Lasing was observed to be inefficient with a high threshold when the slab was mounted too tightly. While a high threshold can indicate poor slab mounting, using threshold or efficiency to characterise mounting is not recommended, as mounting effects can be difficult to distinguish from other factors such as laser or pump alignment. Instead, wavefront distortions corresponding to mounting stress in the slab may be measured directly using an interferometer. 4.3 Interferometry and Thermal Lensing Interferometry allows measurement of wavefront distortions in the slab due to mounting stress and due to thermal loading during pumping. The copper mounting blocks were maintained at 16 C during pumping to help remove heat from the slab, using an AquaCooler portable chiller to cycle distilled water through the closed loop. As the slab was end-pumped using a f=100 mm lens, there was little space for a transmitted interferometer beam and so the interferogram is measured in reflection as shown in Figure 4.4. The path length in each arm of the interferometer was approximately balanced. A Spiricon LBA-100A silicon CCD camera and analyser was used to record the interferograms. Interferograms from a well-mounted slab are shown in Figure 4.5. The dark spot on all interferograms is due to prior damage on the coated end-face of the slab, while the small interference spot to the right of this is from the ND filters in front of the camera. The slab is clamped on the left and right faces in each image. The left-most image corresponds to the unpumped slab, which shows a zero-fringe interference pattern, indicating stress-free mounting. Pumping of the slab resulted in thermal lensing due to the heating, also shown in Figure 4.5. The pumped slabs show approximately 3 rings (20 A) and 5 rings 55

74 CHAPTER 4. DIODE-PUMPED ER:YAG LASER AT 300 K Diode ND filters To camera Imaging lens 50/50 Splitter f=100mm lens Laser head 50/50 Splitter Incident HeNe beam Figure 4.4: Schematic of Mach-Zehnder interferometer used to measure stress and thermal lensing in the slab, showing slab arm (red) and reference arm (blue) Unpumped 20A Pumped 30A Without lasing With lasing Figure 4.5: Interferograms of Er:YAG slab under unpumped and pumped conditions, with and without lasing. 56

75 4.3. INTERFEROMETRY AND THERMAL LENSING (30 A) double pass. During lasing, especially at high power, we might expect to see a reduction in the amount of lensing when some absorbed pump light is re-emitted as lasing instead of heating the slab through non-radiative decays. In this case, there is no significant difference between the interferograms whether or not the slab is lasing. The approximate thermal lensing in the slab can be calculated from the phase pattern of the interferogram with some additional assumptions. First, the observed pattern of light and dark fringes can be converted to a pattern of optical path length differences in the slab. Bright fringes represent path differences of the form mλ and dark fringes (m + 1/2)λ, so the path length varies by λ/2 as we pass from dark to light fringes. This allows us to determine the total path length difference between the centre of the slab and each of the edges for the double-pass interferogram. As the total optical path length (OPL) of the double-passed HeNe beam in the slab is given by OPL = 2nL for n the refractive index and L the length of the slab, the pattern of optical path length differences in the double-pass interferogram is characterised by OPL = 2L n + 2n L The first term (2L n) corresponds to thermally-induced refractive index changes in the slab according to the relationship n = dn dt T + n stress while the second term (2n L) corresponds to thermal-induced mechanical stresses on the slab, such as bulging of the slab end-faces. For crystalline laser materials, the terms from n contribute the majority of the distortion ( 95 % in an Nd:YAG example in [84]), allowing us to neglect the mechanical effects when calculating the focal length of the induced thermal lens in the slab. Since n is maximum in the centre of the slab where the absorption (and thus temperature) are high, and decreases toward the cooled edges, to calculate the thermal lens in the slab we approximate n by a parabola in the x and y directions. This matches the refractive index pattern of a graded index (GRIN) lens, which has the form ( ) n(r) = n 0 1 Ar2 2 where r is the radial position from the optic axis, n 0 is the on-axis refractive index, and A is a positive constant. As a result, the formula for the focal length of a GRIN 57

76 CHAPTER 4. DIODE-PUMPED ER:YAG LASER AT 300 K lens can be used to calculate the focal length of the effective thermal lens of the slab. The pattern of refractive index change can also be used to approximate the temperature in the slab using dn/dt = K 1 for YAG [75], and neglecting the n stress term (which contributes 20 % in Nd:YAG [84]). The calculated thermal lens and estimated temperature difference between the edge and the centre of the slab are given in Table 4.1, and the MATLAB code used to calculate these properties is given in Appendix C.3. Table 4.1: Slab properties measured using interferometry Pump Current Effective Focal Length (mm) Temp. Difference ( C) x y x y 20A A A The focal length of the thermal lens is sufficiently long and the cavity length of the laser sufficiently short ( 10 cm) that stable laser operation is possible using a flat output coupler. The impact of different pump focusing geometries and output couplers on lasing is investigated in the next section. 4.4 Optimising Lasing A schematic of the laser configuration is shown in Figure 4.6. Pump light from the InGaAs diode characterised in Chapter 3 is focused into the slab to pump the laser, with unabsorbed pump light transmitted through the slab. The laser cavity is formed between the pump face of the slab (highly reflective at 1645 nm) and the output coupling mirror. InGaAs Diode Coated face: 1645nm 1470nm 0.5% Er:YAG Brewster face Output Coupler Pump focusing lens Copper heatsink Figure 4.6: Schematic of the Er:YAG laser 1645nm emission 58

77 4.4. OPTIMISING LASING For early characterisations of the laser, the pump diode was used at a constant temperature. Different diode temperatures - corresponding to different parts of the nm room temperature tuning band - are tested in Section 4.4.2, and the average pump absorption is also measured. In general, a diode temperature of 27 C is used, corresponding to a centre wavelength of 1469 nm at 33 A current Pump Focusing and Output Coupling The intensity and uniformity of pump light in the slab was varied by using different lenses to focus the diode power. f= 60 mm and 100 mm spherical lenses were tested. The ideal lens for focusing the pump light was the f=100 mm spherical lens, as this provided the most uniform distribution of pump light along the length of the slab and an intensity larger than the lasing threshold. The f=60 mm lens produced a higher intensity spot in the slab - which reduced diode current required for lasing - but also produced a stronger thermal lens, distorting the laser mode if an additional intracavity lens is not included. Due to its relative simplicity, the f =100 mm lens was used to focus the pump light into the slab for all measurements in this Chapter, unless otherwise stated. The laser performance was also compared for different output couplers. transmission of the output couplers was measured using the Cary UV-VIS Spectrophotometer, and the transmission of the two flat output couplers is shown in Table 4.2. Several curved output couplers were also tested, but none were as effective as the flat output couplers. Table 4.2: Output coupler transmission at pump and lasing wavelengths Wavelength Transmission (%) 90% OC 87% OC The The 87 % output coupler has similar transmission for both pump wavelength regions of interest, while the 90 % output coupler has a 10 % difference in transmission between the two pump wavelength regions under consideration. Both output couplers produce lasing at 27 % slope efficiency with threshold 14.3 W, although the laser emission from the 87 % output coupler rolled off at high pump power. Given that the performance of both flat output couplers is similar, the choice between the two is somewhat arbitrary. We suspect that the laser output power rolls off for both output couplers, but that the measurements of lasing with the 59

78 CHAPTER 4. DIODE-PUMPED ER:YAG LASER AT 300 K 90 % output coupler may also measure a significant amount of unabsorbed pump light, as approximately 45 % of the pump light is unabsorbed by the slab (see Section 4.4.2). Although it is also possible to reduce the fraction of unabsorbed pump light measured by propagating the laser mode further before measuring the output power to allow the pump light to diverge and diffuse, better results were obtained with the 87 % output coupler and by using an aperture in front of the power meter to block most of the transmitted pump light when measuring the laser. We choose the 87 % output coupler to compare 1453 nm and 1470 nm pumping for this reason, and because the reflectivity of the 87 % output coupler is more similar for the two possible pump wavelengths. Chang reports 6.1 W (from 32 W incident pump power) and 36 % slope with this laser slab using a 95 % flat output coupler [40,78], but this output coupler was not available for the above measurements. Using the 90 % output coupler, Chang reports 14.5 W threshold, 28 % slope and 4.3 W (from 30.5 W incident pump power), in line with results in this thesis despite Chang s choice of a different diode temperature, and thus different pump wavelength Pump Absorption The choice of chiller temperature (and thus pump wavelength) is important to the operation of the laser. Although the impact of tuning of the pump diode on lasing is investigated further in Section 4.5, a preliminary investigation of the laser behaviour with respect to diode temperature is described in this section. The pump absorption of the slab is also measured. Lasing for diode temperatures of 23, 27 and 32 C was investigated. These preliminary measurements indicate a chiller temperature of 27 C results in the best efficiency for a room temperature Er:YAG laser, as shown in Figure 4.7. The shape of the laser output power curve was not reproducible between repeated measurements, indicating some source of variation in the pump light. This is due to variations in chiller temperature, and thus diode temperature, changing the diode wavelength and efficiency. Characterisation of the chiller temperature variation is given in Appendix B.2. Even if the chiller temperature were constant, we note that it would not correspond to a constant diode wavelength because the diode wavelength also varies with current as described in Section This implies that the laser output above is not measured for consistent pump absorption. Given that the pump power incident on the slab is known from previous measurements, the fraction of power absorbed in the slab can be calculated using an additional measurement of the power transmitted by the slab. The absorbed frac- 60

79 4.4. OPTIMISING LASING Lasing for different chiller temperatures Laser output power (W) Diode at 27 C Diode at 32 C Diode at 23 C y = x y = x y = x Incident pump power (W) Figure 4.7: Laser performance for different pump diode temperatures, using 90 % output coupler tion for different pump currents and constant diode temperature is given in Figure 4.8. Fraction of pump power absorbed Diode at 27 C Diode at 32 C Pump absorption Diode current (A) Figure 4.8: Measured pump absorption against diode current, for different chiller set point temperatures These results show an average pump absorption fraction of 55 %. The changing pump absorption can explain the shape of the laser power curve when pumped by the 32 C diode: at low current, pump absorption is high, and the laser has a lower threshold. As the current increases from 20 to 25 A (around 22 W incident pump power), the pump wavelength changes and the pump absorption decreases, resulting 61

80 CHAPTER 4. DIODE-PUMPED ER:YAG LASER AT 300 K in a lower efficiency. An detailed characterisation of the laser behaviour as the pump wavelength (and pump absorption) changes in a more controlled manner is given in the next section. 4.5 Pump Wavelength Tuning Earlier measurements of the laser are affected by variations in diode wavelength, which result in changes to the pump absorption. These occur as the diode current is changed, or at constant current due to fluctuations in the chiller coolant temperature. In this Section, the temperature of the diode is tuned and the centre wavelength is measured during lasing, to monitor the effect of varying pump wavelength directly and produce a laser slope efficiency for constant pump wavelength. The diode is also cooled below 0 C to compare the behaviour of the laser when pumped in the nm pump band with the nm band Description of Measurement The schematic of the apparatus used for the wavelength-tuning measurement is shown in Figure 4.9. The diode was kept in the sealed canister for both room temperature and cooled operation to keep the experiment consistent for different diode temperatures. The f=100 mm lens used to focus pump light into the diode was mounted on a translation stage, so that the position of the lens could be optimised for cooled and room temperature operation independently. InGaAs Diode Microscope slide 1645nm 1470nm Brewster face Output Coupler Cooling canister (dry-nitrogen filled) to OSA f=100mm 1645nm emission Figure 4.9: Diagram of apparatus for temperature-tuned pumping experiment with room-temperature slab. In this experiment, the diode temperature was tuned while the diode remained at constant current, as described in Section The diode spectrum and laser output power were measured simultaneously. 62

81 4.5. PUMP WAVELENGTH TUNING Effect of Diode Wavelength on Laser Power The variation in laser output power as the central wavelength of the pump light is tuned is plotted in Figure Measurements for diode currents of 33 A and 27 A are shown, with the shape of the Er:YAG absorption cross section shown as a guide. Each of the points on the laser output curves represents a simultaneous measurement of the laser output power and diode spectrum K Er:YAG laser output power Diode spectrum Laser output power at 1645 nm (W) A diode current 27 A diode current Absorption spectrum of Er:YAG at 300K Diode centre wavelength (nm) Figure 4.10: Laser output power plotted against measured diode centre wavelength, for 27 A (pink) and 33 A (blue) supplied current. Er:YAG absorption cross section at 300 K also seen in Figure 2.9. Inset: diode spectrum corresponding to the maximum laser output power. The maximum laser output power of 4.5 W corresponds to a diode centre wavelength of approximately nm. Given that the spectral width of the diode is 7 nm at this current, this maximum output power corresponds to pumping a nm band which covers the 1466 and 1471 nm absorption peaks. There was insufficient high-temperature tuning available on the diode to centre on the 1475 nm absorption peak. For the cooled diode, the maximum laser output power 4 W occurred for a central pump wavelength of nm, which would pump both the 1454 and 1458 nm peaks in the nm band Effect of Diode Wavelength on Laser Efficiency The diode tuning curves in Section were used to convert from diode centre wavelength to diode power incident on the Er:YAG slab at each diode current. Linear interpolation was used to determine the diode power for currents where the tuning curve for the cooled diode was not directly measured. Figure 4.11 shows the laser output power plotted against incident diode power. The diode centre 63

82 CHAPTER 4. DIODE-PUMPED ER:YAG LASER AT 300 K wavelength corresponding to the maximum measured laser power on each curve is marked. Laser output power against incident pump power Laser Output Power (W) Incident Pump Power (W) Pump current: Pump band: 33A 30A 27A 25A 22A nm nm Figure 4.11: Laser output power curves for the room temperature diode (squares) and the cooled diode (circles) plotted against incident diode power. The wavelength corresponding to the maximum laser output power on each curve is shown. The maximum laser output power on each curve is used to determine the slope efficiency and threshold of the laser, shown in Figure This is the true slope efficiency of the laser: the laser efficiency measured in Section 4.4 is at a constant diode coolant temperature but not a constant pump wavelength supplied to the laser, producing reduced efficiency as the diode tunes off of the optimal pumping wavelength. The laser pumped at 1468 nm by the room temperature diode has 28 % slope efficiency and 12.5 W threshold, while the slope efficiency is only 20 % when pumped at 1456 nm by the cooled diode. This is to be expected, as the nm pump region has stronger absorptions than the nm region at 300 K. The threshold for the laser pumped by the cooled diode was approximately 8 % larger than the threshold of the laser pumped with the room temperature diode, again attributed to lower pump absorption at 1455 nm. 64

83 Laser output power (W) Performance of 300K Er:YAG laser Pumped at nm Pumped at nm 28% slope 4.6. CONCLUSION 20% slope Incident pump power (W) Figure 4.12: Laser efficiency when optimally pumped by cooled or room temperature diode 4.6 Conclusion The CW operation of the 0.5 % Er:YAG slab was optimised with respect to mounting conditions, pump focusing, and pump wavelength for available optics. The thermal lensing in the laser was measured using interferometry, and indicated that the temperature difference between the centre of the laser slab and the cooling interface was 10 C. Laser performance with respect to pump wavelength was measured by tuning the diode wavelength. The optimal pump wavelength for a room temperature Er:YAG laser is nm, resulting in a maximum output power of 4.5 W and slope efficiency of 28 In Chapter 5, the techniques used to optimise the room temperature Er:YAG laser are repeated for the same slab at cryogenic temperatures. 65

84 66

85 Chapter 5 Cryogenic Er:YAG lasers 5.1 Introduction The construction of a cryogenic laser requires greater attention to various engineering constraints than a room temperature laser of similar power. The largest engineering constraint is the design of the laser head and the quality of slab mounting, though the transparency of the cryostat windows, lasing cavity stability, and the alignment of the pump and laser mirrors also contribute. In this Chapter, possible stresses due to slab mounting are discussed in Section 5.2, and interferometry is used to observe the impact of these stresses to determine if lasing is viable. Several cryogenic Er:YAG lasers are also described in this Chapter. In Section 5.3, The alignment procedure of the laser and pump diodes is described as part of the construction of a preliminary cryogenic Er:YAG laser. This laser was pumped in the nm band and is described and characterised. Unfortunately, unexpected damage to the diode reduced its output power, so the diode is re-characterised in Section 5.4. Finally, in Section 5.5, a 77 K Er:YAG laser is presented and characterised and the limitations on lasing efficiency are discussed. 5.2 Slab Mounting and Interferometry In this section, the mounting of the slab in a laser head assembly for a cryogenic Er:YAG laser is described, and interferometry is used to check for mounting stress in the slab at room temperature and at 77 K to determine if lasing is viable under these conditions. We also describe a method for improving the contrast of measured interferograms. 67

86 CHAPTER 5. CRYOGENIC ER:YAG LASERS Mounting the Slab Mounting of the slab for a cryogenic laser is more difficult than for an equivalent laser at room temperature, as the large temperature change between room temperature assembly and the cryogenic operating conditions means that the thermal expansion of different parts of the assembly must be taken into account. The thermal expansion properties of aluminium, copper, molybdenum, steel and YAG are given in Table 5.1. The values for YAG and copper were calculated using where the curve for α T = 1 L L 77 L = L dl dt 1 dl L dt dt is given in the reference provided. Table 5.1: Integrated linear expansion coefficient for various materials Material Linear expansion ( L 77 L 293 L 293 ) ( 10 5 ) Calculated L 77 -L 293 (µm) for L 293 = 3 mm Aluminium [85] Copper [86] Molybdenum [85] Stainless Steel [85] YAG [87] As seen in the Table, molybdenum and YAG have similar thermal contraction across the 300 K to 77 K temperature range. Thus, a high power cryogenic Er:YAG laser should be designed with a molybdenum laser head and YAG or molybdenum spacers to minimise stress on the slab, as seen in Miftar Ganija s earlier work with cryogenic Yb:YAG lasers. [56] It was decided that we would obtain preliminary results on the effectiveness of 1455 nm pumping of cryogenic Er:YAG with the copper laser head used for the 300 K laser, rather than waiting for a molybdenum laser head to be produced. Copper is not an ideal mounting material for YAG as it has much greater thermal expansion coefficient and thus much greater contraction when cooled, increasing stress and risking damage to the slab as the copper contracts relative to YAG. Careful assembly and planning is crucial for ensuring that the indium on the slab has sufficient (but not excessive) crush at 77 K after being assembled at 300 K. A schematic of the laser head assembly is shown in Figure 5.1. The assembly uses stainless steel screws, which have a similar thermal contraction to copper. The ceramic spacers from the room temperature assembly were unsuitable for use with a cryogenic laser head, as the thermal expansion properties of the ceramic were unknown. Instead, steel spacers were used as part of this assembly. As in 68

87 5.2. SLAB MOUNTING AND INTERFEROMETRY Steel spacers Aluminium foil Er:YAG slab Indium foil Copper block Figure 5.1: Schematic of cryogenic laser head assembly. Miftar Ganija s thesis [88], spacers approximately the same size as the slab are used with aluminium foil. In Ganija s design for a cryogenic laser head, the spacers are made of YAG, and aluminium foil is used on the spacers (rather than indium foil as used on the slab) to provide a barrier that prevents excessive indium crush in the assembly as the laser head contracts. In this laser head, the spacers are made of steel 10 µm larger than the slab, which decreases to 4-5 µm larger than the slab at 77 K as the slab and spacers contract. In early assemblies, the aluminium and indium thicknesses were chosen so that the slab with indium and the spacers with aluminium were the same thickness at room temperature, but this left the slab loose when the laser head was assembled. Increasing thickness of indium relative to the aluminium was used in order to form a light crush and keep the slab stable at room temperature. With the spacers always slightly thicker than the slab, we prevent the slab from being excessively crushed due to the thermal contraction of the copper laser head at 77 K. The other possible shear stress on the slab is a result of the contraction of the copper and slab in the direction parallel to the indium surfaces. As seen in Table 5.1, the copper in contact with the indium contracts by an additional 7 µm compared to the YAG slab during cooling. We expect this to be somewhat mitigated by the ductility of the indium foil, which should allow some relative movement of the slab and mounting blocks. A photograph of the cryogenic laser head assembly mounted in the cryostat is shown in Figure 5.2. The copper laser head is attached to an aluminium mounting plate by a single screw, preventing stress due to the different thermal contraction of copper and aluminium and allowing the angle of the laser head and the aluminium mounting plate to be adjusted. This aluminium plate is then attached to 69

88 CHAPTER 5. CRYOGENIC ER:YAG LASERS the aluminium cryostat cold finger via another single screw. Location of copper attachment to aluminium Steel spacer Attachment screw to cold plate Slab: coated end face Water-cooling fittings Figure 5.2: Cryogenic laser head assembly in the cryostat Aluminium spacer plates are used between the aluminium mounting plate and the cold finger to align the slab vertically with the centre of the cryostat windows. The stack of alignment plates combined with the single screw connection between the mounting plate and cold finger can result in poor thermal contact or slab misalignment during cooling, however. In an ideal design, the size of the mounting plate would be calculated taking into account the height of the slab in the copper assembly to eliminate the need for spacer plates, but as previously discussed, it was most important to collect preliminary cryogenic Er:YAG lasing results than to construct an ideal system Interferometry A Mach-Zehnder interferometer was used to look for stress on the slab following the technique used in Section 4.3. The particular measurement configuration used for the slab mounted in the cryostat is shown in Figure 5.3. According to [89] and [90], the precise stresses and strains on the slab could be determined by combining the Mach-Zehnder interferometer with polarising optics to isolate the components of stress-induced birefringence in the laser slab. This would be important for the optimal design of a high-power laser head, but in this case, the copper laser head was already known to be sub-optimal and confirming the presence or absence of stresses is more important than quantifying them. Polarisation techniques were not used in this experiment. Unlike at room temperature, in the cryogenic case the cryostat windows resulted in some loss of the probe beam (the windows are AR coated nm but not at the HeNe wavelength 633 nm). This unbalanced the intensity in the two arms of the interferometer, reducing contrast in the interferogram. While the intensity in the other interferometer arm can be reduced using ND filters, this is an unsatisfactory 70

89 5.2. SLAB MOUNTING AND INTERFEROMETRY Cryostat Mounted slab Collimating lenses Incident HeNe beam 50/50 splitter To camera ND filters Imaging lens 50/50 splitter Figure 5.3: Schematic of Mach-Zehnder interferometer used for measurement of cryogenic slab. Slab arm (red) and reference arm (blue) shown. solution when filters are only available in a limited range of values, and sometimes the contrast could not be sufficiently improved with the values available. The splitter for the incident beam was replaced by a variable splitter in an attempt to improve contrast, but this was still not sufficient to produce clear interferograms with this interferometer, and produced additional alignment problems due to its large size. A different way to increase the contrast in the interferogram is to separate the phase pattern from the images of the two arms of the interferometer. If the reference arm of the interferometer has an electric field of the form E 1 exp(iωt) at the camera, this implies that the slab arm has an electric field of the form E 2 exp (i(ωt + k x x)) at the camera if the path lengths of the two interferometer arms are the same and we assume vertical straight fringes for simplicity. Then the intensity at the camera is given by: I = ( E 1 exp(iωt) + E 2 exp(i(ωt + k x x) ) ( E 1 exp( iωt) + E 2 exp( i(ωt + k x x)) ) = ( E 1 + E 2 exp(ikx) )( E 1 + E 2 exp( ik x x) ) = E E E 1 E 2 cos(k x x) where E 1 2 and E 2 2 are the intensity of the reference arm and slab arm of the interferometer respectively, and 2E 1 E 2 cos(kx) is the fringe pattern. 71

90 CHAPTER 5. CRYOGENIC ER:YAG LASERS As a result, while the visibility of the fringe pattern can be enhanced by ensuring that E 1 and E 2 are equal in magnitude, we can also enhance the fringe visibility in post-processing by subtracting the background image in the reference arm ( E 1 2 ) and the slab arm ( E 2 2 ) from the measured interferogram. A pictographic representation of this process showing the interferogram, images of each arm, and the final phase pattern is shown in Figure 5.4. Measured interferogram (low contrast) Block slab arm and reference arm in turn to measure each of the arms of the interferometer Subtract these images from the measured interferogram to isolate the phase pattern Figure 5.4: Process used to increase contrast on interferograms, including images in sequence. This post-processed image shows the interference pattern for the mounted slab more clearly, which was barely visible in the original image. Some additional fringes are still visible on this image, and we notice that they are also present on the images of the slab and reference arm. These fringes are due to reflections between the two uncoated lenses used to collimate the HeNe, and as a result they are present in both the slab and reference arm images. As all the measured fringes drift slowly over time, the time delay between measurements of the interferogram, slab, and reference means that that this stripe pattern is difficult to completely eliminate without taking multiple measurements. The position of the centre of the lens pattern can be adjusted relative to the slab by adjusting the relative angle of the lenses, so we ensure that the centre of this pattern does not appear on our interferogram of the slab to prevent confusion between the lens pattern and the slab interferogram. This post-processing method nonetheless allows us to examine the phase pattern 72

91 5.3. PRELIMINARY CRYOGENIC LASER PUMPED AT 1470 NM produced by the mounted slab and look for mounting stress. Figure 5.5 shows the interferograms of the Er:YAG slab from the same mounting at 300K and at 77K. 300K 77K Raw Improved contrast Figure 5.5: Interferograms for the mounted slab at room temperature before cooling (left) and after cryogenic cooling (right). Both the raw images and the processed interferograms are shown. These interferograms show 1 full fringe at room temperature, increasing to 2-3 fringes when the slab is cooled to 77 K. This indicates that the slab was under some stress when initially mounted, and the stress increased when the laser head assembly was cooled. Despite the low quality of the interferograms, we were still able to extract sufficient information about the slab stress to decide whether or not to lase with the slab or to re-assemble. As the phase variation across the centre part of the slab is minimal, we concluded that the stress was still sufficiently low to attempt preliminary lasing. 5.3 Preliminary Cryogenic Laser Pumped at 1470 nm While low wavefront distortion is important, the alignment and pump focusing of the cryogenic end-pumped laser also needed to be characterised early in the laser development to prevent delays if modifications to the system were required. A preliminary cryogenic Er:YAG laser was constructed, aligned, and tested to provide a performance baseline for comparison with later results. 73

92 CHAPTER 5. CRYOGENIC ER:YAG LASERS System Configuration A schematic of the cryogenic Er:YAG laser configuration is presented in Figure 5.6. Cryostat "pump" window Cryostat Output Coupler InGaAs Diode Microscope slide to OSA f=100mm Laser head Cryostat "lasing" window Figure 5.6: Schematic of cryogenic Er:YAG laser. Similar to the room temperature configuration, the f=100 mm lens was used to focus the pump light for cryogenic lasing. As a result, the Er:YAG slab was mounted close to the cryostat pump window. This resulted in a longer distance between the slab and the lasing window, and thus also between the slab and the output coupling mirror. Due to the long cavity length of this configuration, a flat output coupler was no longer appropriate and a variety of curved output couplers were tested during preliminary operation. The original 2-inch-diameter cryostat windows which were not AR coated for 1.6 µm were replaced by 1-inch AR coated windows, requiring new mounting plates to hold the smaller windows over the large window apertures. The new mounting plates were designed such that the laser and pump beams could pass through the windows at a near 0 angle of incidence. This required a 4 angle on the lasing window and an 8 angle on the pump window. A schematic and photograph of one pair of window mounts is shown in Figure 5.7. Window) opening Cryostat Slot)for)1-inch) O-ring (Behind))flat)face) to)seal)with)2-inch) O-ring O-rings AR)window Angular)offset Copper mounting blocks Figure 5.7: Schematic (left) and photograph (right) of angled window mounts for AR windows 74

93 5.3. PRELIMINARY CRYOGENIC LASER PUMPED AT 1470 NM While the modified windows reduce loss, they also introduce new problems. The change from 2-inch windows to 1-inch windows reduced visibility inside the cryostat, and made pump and laser alignment much more difficult due to the reduced aperture size. While we used the 1-inch windows and adapted for the reduced visibility, we conclude that a high power cryogenic Er:YAG laser should use 2-inch AR coated windows to improve alignment visibility Aligning the Slab and Pump Beam The large distance between the Brewster face of the slab and the lasing window also resulted in complications during the alignment process, as the mounting point (and thus the centre of rotation) for the laser head assembly in the cryostat was significantly displaced from the optic axis of the slab. A schematic of the laser head alignment scheme is shown in Figure 5.8. The laser head is aligned using a HeNe alignment beam at 633 nm. The difference in refractive index of YAG at 633 nm and at 1618 nm corresponds to a 1 change in the angle of incidence on the Brewster face for propagation parallel to the slab sides. This difference in angle is important for aligning the laser output coupler, but does not contribute much to the slab alignment in the cryostat. A 1 angular change of the angle into the slab corresponds to approximately a 2 mm shift of the beam on the window, which is difficult to observe with the 3-4 mm diameter HeNe spot used to fill the slab during the alignment. Cryostat )pump) window Cryostat HeNe(alignment(beam Transmitted HeNe(beam Aperture representing window Laser(head Back-reflection from(pump(face (overlaps(input(hene beam(when(slab(is positioned(correctly) Figure 5.8: Schematic of the laser head alignment system. The back-reflection from the flat pump face of the slab is shown. The alignment procedure for the laser head is presented below. 75

94 CHAPTER 5. CRYOGENIC ER:YAG LASERS Laser Head Alignment Procedure: 1. With the bottom can of the cryostat in place, the HeNe beam is aligned to the centre of the laser window at normal incidence to the window. An aperture is mounted in front of the pump window to mark its position. 2. The bottom can is removed, and the slab assembly is mounted to the cold finger such that the HeNe passes along the centre of the slab. The alignment and position of the laser head is adjusted until the back-reflection from the pump face is aligned with the input HeNe beam, and the beam transmitted through the slab passes through the centre of the aperture representing the pump window. 3. The can is replaced. The beams through each window of the slab are checked, and then vacuum is applied to the cryostat chamber. If the beam height and window alignments still look acceptable, the cryostat is cooled. Alignment of the pump beam with the laser mode is difficult as the slab cannot be viewed directly through the pump or lasing windows. A schematic of the pump diode and cryostat during pump alignment is shown in Figure 5.9, indicating the windows where the slab may be observed. The pump alignment procedure is presented below. Cryostat "pump" window Cryostat "lasing"b window HeNeB alignmentb beam InGaAsBdiode Unabsorbed pumpblight LaserBhead ViewBofB pumpbface ViewBof Brewster face Figure 5.9: Schematic for the pump diode alignment procedure, including the position of cryostat windows used to view the pump alignment. Pump Alignment Procedure: 1. The pump diode is placed in front of the pumping window leaving space for the f=100 mm lens, but without the lens in place. The pump diode is turned 76

95 5.3. PRELIMINARY CRYOGENIC LASER PUMPED AT 1470 NM on to just above threshold, and adjusted so that the transmitted HeNe spot is in the centre of the diode, and the pump spot is in the centre of the pumping window. The unabsorbed pump light passing through the slab should be visible from the lasing window if the diode is positioned and angled approximately correctly. 2. The pump focusing lens is inserted, and the position of the unabsorbed pump light relative to the HeNe beam can be used to estimate the pump beam position and angle in the slab, and thus estimate the orientation required for the lens. 3. The pump power is increased until the green upconversion fluorescence is visible in the slab. This allows fine-tuning of the pump position in the slab (using the view of the slab pump face) and the pump focus point and angle (using the view of the slab Brewster face). This must be combined with visual examination of the unabsorbed pump spot transmitted through the window, as the fluorescence alone does not provide sufficient information about the pump angle. While the pump beam could be aligned to the slab at room temperature before the cryostat is cooled, there is no advantage to doing so. The diode and focusing lens are close enough to the cryostat window when properly aligned that they need to be moved out of the way to take the can on and off. As a result, aligning the diode must be completed with the can on the cryostat, so alignment follows the same procedure at 300 K and 77 K, except that the green pump upconversion fluorescence should not be used to align the pump beam in the uncooled slab due to excessive heating. The laser output coupler is also aligned using the HeNe beam. An aperture is used to reduce the spot diameter of the HeNe to approximately 1 mm, and the mirror is coarsely aligned using the ingoing HeNe beam. The output coupler is aligned more finely while the slab is pumped. Early experiments using a 30 cm radius of curvature output coupling mirror yielded some success, but the short radius of curvature meant the mirror needed to be located as close as possible to the cryostat window. This mirror had a low reflectivity (80 %) preventing direct comparison with the 300 K laser, and was 12.7 cm in diameter (0.5 inches) making alignment difficult. A 25.4 cm (1-inch) diameter, 80 cm radius of curvature mirror with 90 % nominal reflectivity was much easier to align, and matched the 90 % output coupling used for the room temperature laser for comparison of the two lasers under similar operating conditions. The preliminary laser did not have stable output using a flat output coupler. 77

96 CHAPTER 5. CRYOGENIC ER:YAG LASERS Lasing Wavelength The expected laser wavelength for an Er:YAG laser at 77 K is 1618 nm (see Section 2.2.3). The preliminary cryogenic Er:YAG laser, however, was observed to lase at nm and nm simultaneously, with similar output power contribution from both laser lines. Cryogenic lasing on both wavelengths was also observed by Merkle et al. without wavelength selective optics across a K range, with both wavelengths lasing at similar power levels around 110 K [91]. If we assume that our cryogenic laser works similarly despite being CW instead of pulsed, this indicates that the slab is lasing at a temperature well above 77 K, probably close to 110 K in the centre of the slab. This high temperature was suspected to be a result of poor thermal contact between the slab and the cryostat cold finger int he preliminary laser. This was addressed in later iterations of the laser by using thicker indium to improve crush and thus thermal contact Laser Performance The performance of the preliminary cryogenic Er:YAG slab laser is compared to a similar 300 K Er:YAG slab laser in Figure Both lasers were pumped at approximately 1470 nm with the same pump intensity profile. The cryogenic slab laser uses an 89 % reflective output coupling mirror with an 80 cm radius of curvature, while the room temperature Er:YAG laser uses an 87 % reflective flat output coupler as described in Chapter 4. Laser output power (W) Comparison of Er:YAG lasers Cryogenic laser (89% OC 80cm curvature) Room temperature laser (87% flat OC) y=0.27x y=0.29x Incident pump power (W) Figure 5.10: Graph comparing the cryogenic Er:YAG laser (green) with an earlier room temperature laser (purple) The cryogenic laser had a lower threshold than the room temperature laser, which 78

97 5.3. PRELIMINARY CRYOGENIC LASER PUMPED AT 1470 NM is qualitatively as expected given the much lower inversion required for transparency in a cryogenic Er:YAG laser (see Chapter 2). Since the threshold inversion for transparency in the slab is 10 % for the 300 K laser and 0.1 % for a 77 K laser, we might expect two orders of magnitude difference in the lasing threshold, but this does not take into account other loss mechanisms such as scatter within the gain medium, present in both lasers. The cryogenic laser also has some additional loss compared to the room temperature laser, as the coefficient of upconversion is four times stronger at 77 K than at 300 K [65], which will marginally increase threshold. Additionally, while the lasing window is AR coated with less than 0.1 % variation in reflectivity between 1618 nm and 1645 nm, the reflectivity of the coated end-face of the slab at 1618 nm was not measured Pump Wavelength Tuning Temperature tuning of the pump diode was used to investigate the optimal pumping wavelength for the preliminary cryogenic Er:YAG laser, using the same methodology as described in Section 4.5. Due to space constraints, however, the cooling canister could not be used on the diode with the f=100 mm pump lens, so only the effect of tuning the diode near room temperature was measured. The laser configuration is unchanged from the schematic included in Figure 5.6. The measured output power curves for temperature-tuned pumping of the preliminary cryogenic laser are presented in Figure 5.11, with the absorption spectrum of Er:YAG at 120 K shown for comparison. No other absorption spectra between 120 K and 77 K had been recorded. The maximum measured output power for this laser is 5 W. Unlike the room temperature laser, there is a significant measured shift between the optimal wavelength at 27 A and the optimal wavelength at 33 A. This is be related to the spectral width of the diode. At low current, the diode output spectrum is narrower and it is more effective to centre on the strong and narrow 1470 nm absorption. At higher current, the diode spectrum is broad enough to cover both the 1466 nm and 1470 nm absorptions, and so a wavelength between the two absorptions is favoured. Due to damage to the diode, cryogenic lasing pumped in the nm band could not be investigated Summary of Preliminary Experiments The room temperature Er:YAG laser head was mounted in the cryostat to create a cryogenic Er:YAG laser. The cryostat windows were modified to enable lasing with this slab, and an alignment procedure was developed to align the slab and pump diode despite low visibility into the cryostat. 79

98 CHAPTER 5. CRYOGENIC ER:YAG LASERS Laser output power (W) K Er:YAG Laser Output Power 33A 27A 22A Absorption spectrum of Er:YAG at 120K Diode centre wavelength (nm) Figure 5.11: Temperature-tuned pumping of a preliminary cryogenic laser. The measured absorption spectrum at 120 K (see Section 2.4.2) is shown for comparison. A cryogenic Er:YAG laser pumped at 1470 nm with 5.4 W output power, 27 % slope and a threshold of 6.2 W is presented. This threshold is less than half that of the 300 K Er:YAG laser presented in Chapter 4. The preliminary cryogenic laser had an estimated operating temperature of approximately 110 K, with emission on both the 1645 nm and 1618 nm lines. When the pump wavelength was temperature tuned, the optimal pumping wavelength in the nm band shifted by -2 nm as the diode current increased from 27 A to 33 A. Although lasing results at a controlled slab temperature are preferred, these results form a preliminary characterisation of the cryogenic laser which can be compared to the performance of the 300 K Er:YAG laser, and can be extended for further work at 77 K using the cooled diode. While the cooled diode couldn t be used for this laser, the absorption spectrum of Er:YAG at 77 K suggests that the nm pump band will be more strongly absorbed by the slab, resulting in a lower laser threshold relative to incident pump power and a greater efficiency when pumped in the nm band. 80

99 5.4. CHARACTERISING A LOWER-POWERED DIODE 5.4 Characterising a Lower-Powered Diode Further investigation of the cryogenically cooled Er:YAG slab was interrupted by damage to the diode when one of the components in the supply broke, which resulted in electrical damage to the diode due to insufficient surge protection. The damage affected the output power, slope efficiency and output beam shape of the diode, and required re-characterisation of the diode behaviour before it could be used to pump the 77 K Er:YAG laser. The diode s power output after damage is compared to its previous operation in Figure The damaged diode has a higher threshold (approximately 10 A) and reduced efficiency. Measured Diode Power (W) Diode performance Before damage After damage y = 1.04x -6.5 y = 0.84x Current (A) Figure 5.12: Comparison of diode output power before (red) and after damage (purple) The temperature-tuning curves are shown in Figure 5.13, for both room temperature and cooled operation. Both sets of measurements are recorded with the aluminium canister in place. Compared with Figure 3.11, the damaged diode now covers a spectral region nm at maximum current near room temperature, rather than the previous nm tuning region. The diode spectrum is compared with the previously measured spectrum in Figure Both spectra are normalised to have area 1, so that the width is easily comparable. We observe that the diode spectral width and shape are similar before and after damage, when measured at the same current. A comparison of the diode beam shapes before and after the damage are shown in Figure Prior to damage, the output consisted of three bar shapes stacked vertically, and at low current only the top and bottom diode bars would emit. After damage, bar shapes were less uniform, some emitters do not turn on, and the top bar 81

100 CHAPTER 5. CRYOGENIC ER:YAG LASERS 25 Diode tuning curves after damage Measured Output Power (W) Diode centre wavelength (nm) 36A 33A 30A 26A 23A 20A Figure 5.13: Output power of damaged diode against diode centre wavelength, measured near room temperature (right) and when the diode is cooled (left). 0.3 Diode wavelength after damage (nm) (Arbitrary units) Diode wavelength before damage (nm) Figure 5.14: Comparison of diode spectral shape before (blue) and after damage (red), both spectra are normalised to have area 1. The spectrum for the damaged diode is shifted to overlap the two curves. 82

101 5.5. ER:YAG LASING AT 77 K is now the last to turn on as the current is increased. These images are recorded close to the far field, after the diode beam has propagated through a lens and through the focus to show the shape of the diode without imaging individual emitters. Before damage After damage Figure 5.15: Comparison of diode emission shape before and after damage. The before image (left) is at 8 A current supplied, while the after image (right) is measured at 22 A supplied to show the third diode bar. Despite the difference in appearance of the diode pump spot, the diode still has similar focusing behaviour. The diode focused-spot-sizes before and after damage are presented in Table 5.2. For the f=100 mm lens, no significant change in the focal spot size is observed. Table 5.2: Focused 1/e 2 spot size before and after diode damage Before damage After damage Focusing condition x diameter (mm) y diameter (mm) x diameter (mm) y diameter (mm) f =100 mm f =250 mm In summary, while the diode has reduced output power and a different wavelength tuning range, the focusing properties and spectral width of the diode are similar. Thus, the diode could still be used for further investigation of cryogenic Er:YAG lasers. 5.5 Er:YAG Lasing at 77 K By re-mounting the slab with thicker indium and being careful to tighten the screws into the cold finger sufficiently, the Er:YAG slab can be maintained at 77 K. Since the method for pumping the slab and establishing lasing is already well-developed, 83

102 CHAPTER 5. CRYOGENIC ER:YAG LASERS the temperature in the slab can be verified experimentally by checking the lasing wavelength. For all laser measurements in this section, the Er:YAG slab is confirmed to lase at 1618 nm and not at 1645 nm, indicating that the slab temperature is below 90 K System Design The f=100 mm lens configuration could not be used with the diode cooling canister due to space constraints. Instead, the pump diode is placed further from the cryostat and a second lens is used to limit the divergence of the diode pump spot, so that the spot does not clip on the f=100 mm lens. This new focusing configuration for the laser is shown in Figure 5.16, and the measured focal spot size of the diode in this configuration is compared to the spot size in the f=100 mm configuration in Table cm 26cm 3cm Cryostat Cooling canister To OSA f=500mm f=100mm Laser head Figure 5.16: Diagram of the pump configuration used for the temperature-tuned laser Table 5.3: Focused spot size for one lens and two lens focusing Focusing condition x diameter (mm) y diameter (mm) f =100 mm Two lens (100 mm and 500 mm) Unfortunately, this configuration produces a less intense pump spot than the f=100 mm lens alone, preventing direct comparison with the 300 K Er:YAG laser and with the cryogenic Er:YAG laser described in Section 5.3. Due to the damage to the diode, however, a comparison between 77 K and 300 K lasing under consistent pumping conditions was already impossible without repeating all of the room temperature results. Instead, we continue the experiment to learn more about operating an Er:YAG laser at 77 K and compare pumping in the nm and nm bands. 84

103 5.5. ER:YAG LASING AT 77 K Temperature-Tuned Pumping The pump configuration of the laser is changed to the two-lens configuration shown in Figure 5.16, but the rest of the lasing configuration is still the same as in Figure 5.6. The laser output power is plotted against pump wavelength for the pump diode at room temperature and when cooled in Figure All four sets of measurements were recorded for the same pumping configuration without adjustment or fine-tuning of alignment between measurements. 77K Er:YAG Laser Output Power Laser output power (W) A 30A 26A Absorption spectrum of Er:YAG at 77K Diode centre wavelength (nm) Figure 5.17: Laser output power against diode centre wavelength. The Er:YAG absorption spectrum at 77 K is shown for comparison. In this Figure, we observe that at 30 A pump current, the cooled diode results in greater laser output power (1.75 W) than when the room temperature diode is used (1.2 W). Naively, using the cooled diode seems to be more efficient for lasing than the room temperature diode relative to supplied current, but with only one measured curve we have insufficient information to comment on the slope. As an additional concern, this method of displaying the data does not take into account the increase in diode efficiency as the diode cools, which will be addressed in the next section. This graph also indicates that there may be either wavelength calibration problems or additional factors contributing to the optimal wavelength. In Chapter 4, the optimal pump tuning for the cooled diode lay between the two absorption peaks in the band, but in this graph, the largest laser output power occurs when the pump is tuned to centre on the small 1458 nm peak. We also observe similar laser output power when the diode is centred on the 1453 nm absorption peak as when it is centred on 1460 nm where no absorption peak is present. In contrast, the wavelength 85

104 CHAPTER 5. CRYOGENIC ER:YAG LASERS tuning curves for the room temperature diode match with expectations from the room temperature experiment and the earlier experiment shown in Figure The factors contributing to the optimal lasing wavelength for cooled diode pumping are investigated further and discussed in the next sections Measuring the Slope Efficiency As before, the diode characterisation curve was used to determine the pump power at each centre wavelength. Figure 5.18 shows a comparison of the temperature-tuned 120 K laser from Section and the 77 K laser from Figure 5.17 against pump power, rather than against pump wavelength. The single measurement of the 77 K laser pumped by the cooled diode has also been included for completeness. After these initial measurements, the cryostat returned to room temperature unexpectedly. The slab was thus remounted and the pump and laser cavity was re-aligned before the next set of measurements. The new measurements are presented in Figure 5.19 alongside the previous measurements. 6 Comparison of cryogenic Er:YAG lasers Laser Output Power (W) y = 0.22x y = 0.20x Incident Pump Power (W) 120K laser, Pumped at 1470nm 77K laser, Pumped at 1469nm 77K laser, Pumped at 1457nm Figure 5.18: Laser output power against incident diode power for various cryogenic Er:YAG lasers We observe similar behaviour for the temperature-tuned 120 K laser (blue diamonds) and the temperature-tuned 77 K laser (green triangles). Both lasers exhibit a roll-off in laser output power as the diode current increases and the spectral width 86

105 5.5. ER:YAG LASING AT 77 K of the diode increases. If we were pumping with the same intensity for both of these lasers, they might also have a similar slope and threshold. Comparing the operation of the laser when pumped by the cooled diode (orange circle) with the room temperature diode (green triangles), we don t observe any major improvement from using the cooled diode, but we already knew that more measurements are required to determine the slope efficiency and threshold. We note again that the orange circle and the middle green triangle occur for the same pump current. 6 Comparison of cryogenic Er:YAG lasers Laser Output Power (W) y = 0.19x -1.5 y = 0.22x y = 0.20x -1.9 y = 0.09x Incident Pump Power (W) 120K laser, Pumped at 1470nm 77K laser, Pumped at 1469nm 77K laser, Pumped at 1457nm 77K laser, Pumped at 1470nm 77K laser, Pumped at 1458nm Figure 5.19: Laser output power against incident diode power for various cryogenic Er:YAG lasers. This graph matches Figure 5.18 with two additional laser curves. The new 77 K laser pumped at room temperature (red squares) behaved similarly to the previous 77 K laser (green triangles). In contrast, the new measurements for the laser pumped using the cooled diode (blue circles) produce a completely different curve from the previous cooled-diode measurement. The repeatable results when pumping with the room temperature diode lead us to believe that the problem is related to a change when cooling the diode, rather than a slab mounting problem. We hypothesise that while the pump mode in the slab was similarly aligned with the laser mode during room temperature pumping for both of these lasers, when the diode was cooled the pump mode and laser mode became misaligned. This misalignment effect must be further investigated in order to make any conclusions about the effectiveness of pumping the 77 K laser with a cooled diode. 87

106 CHAPTER 5. CRYOGENIC ER:YAG LASERS Effectiveness of Two-Lens Focusing A quantitative way to measure the pump power incident on the slab lasing mode was needed in order to measure the suspected misalignment effect when the diode was cooled. We chose a 1 mm diameter pinhole to simulate the aperture of the lasing mode in the slab, and used the pump power transmitted by the pinhole to represent the pump power mode-matched to the laser. The configuration used for this experiment is shown in Figure The pinhole is aligned to maximise the power on the power meter when the diode is operating with room temperature coolant (27 C). The power transmitted through the pinhole and the diode centre wavelength are simultaneously recorded while the temperature of the diode is tuned. 9cm 26cm Cooling canister To OSA f=500mm f=100mm 1mm pinhole Power meter Figure 5.20: Schematic of pinhole experiment The pumping configuration for the cryogenic laser used tilt on the two lenses to align the pump beam into the centre of the slab, as the slab is fixed and it is difficult to make fine adjustments to the position or angle of the diode. The first set of measurements with the pinhole used the lenses in the same tilted positions, and nothing was moved during the diode cooling process. For the second set of measurements we aligned the pump beam through the centre of each lens with the lenses perpendicular to the diode beam, and aligned the pinhole to the pump beam at room temperature. Unlike in the previous experiment, the position of the pinhole was adjusted to maximise the power transmitted while the diode was cooled and warmed back up to room temperature. The results of these measurement are plotted in Figure The fraction of power transmitted by the fixed pinhole varies significantly when the diode is cooled. We observe a change from 43 % transmission to 57 % transmission as the diode centre wavelength increases by 7 nm (approximately 18 C temperature change) when the pinhole is fixed. During the second experiment, the pinhole was adjusted to maximise transmission, and we discovered that the majority of the beam misalignment reducing the transmitted power was in the vertical direction. This is unsurprising, as the pump spot is slightly smaller than the pinhole in the y-direction (see Table 5.3), making the transmitted power more sensitive to 88

107 5.6. CONCLUSION Fraction transmitted by pinhole Power transmitted by pinhole First experiment: pinhole fixed Second experiment: pinhole adjusted Diode Centre Wavelength (nm) Figure 5.21: Plot of fraction of diode power transmitted by the pinhole against diode wavelength. The pinhole was either stationary (red) or adjusted to maximise transmission (blue) the position of the spot in y. In contrast, the spot size in the x direction is approximately twice the size of the pinhole, and as a result the transmitted power will vary more slowly in the x direction. Additionally, the optimal position of the pinhole was 2 mm further from the f=100 mm lens when using the cooled diode at 1453 nm than for the room temperature diode. Thus it appears that significant beam pointing errors can occur when the diode is cooled. If we consider the >10 % change in pump transmission across the nm tuning range, we can qualitatively explain the shift in the optimal pumping wavelength described in Section To quantitatively measure the effectiveness of pumping cryogenic Er:YAG in the nm band and determine the optimal pumping wavelength, we would need a way to eliminate this focusing problem. One way to achieve this is to modify the freezer system to keep the coolant and diode colder for much longer and allow adjustment of the pump diode during measurements at approximately constant temperature, or alternately to adjust the cryostat stand to allow the cooled diode to fit closer to the cryostat window and use the f=100 mm lens. Unfortunately, there was insufficient time to modify either the freezer cooling system for the diode or the cryostat holder in order to present further results as part of this thesis. 5.6 Conclusion An initial investigation of a cryogenic Er:YAG laser has been presented, including refinement of the laser head, alignment in the cryostat, and pumping of the Er:YAG 89

108 CHAPTER 5. CRYOGENIC ER:YAG LASERS laser. For early mounting, interferometry was used to measure stress on the slab at room temperature and at 77 K. In later experiments involving a preliminary laser, the quality of the thermal contact between the slab and the cryostat cold finger was determined by measuring the laser output wavelength, with the presence of the 1645 nm emission wavelength indicating insufficient thermal contact in the system. This preliminary laser has a 6.2 W threshold, 27 % slope efficiency, and 5.4 W maximum output power with a dual-wavelength output that suggests an operating temperature of approx 110 K according to literature. The optimal pump wavelength was observed to change as the diode current (and thus diode spectral bandwidth) increased, when pump wavelength was tuned in the nm band. A cryogenic laser with only 1618 nm emission operating at 77 K is also described. Unfortunately due to damage to the diode, the available pump power for this laser was limited, but a maximum output power of 1.65 W with a 9.5 W threshold and 20 % slope was achieved using a lower-intensity pumping geometry than for the 110 K laser presented above. The output power of the laser is limited by available pump intensity for both pump bands, but for the nm band, lasing efficiency is also reduced by a measured decrease in mode overlap between the pump and lasing modes when the diode is cooled. This suggests that if the mode overlap could be improved, the nm band may be more effective for pumping the laser relative to supplied pump power, when compared to the nm band. This hypothesis could be confirmed with additional equipment in a further experiment, outside the scope of this project. 90

109 Chapter 6 Conclusion 6.1 Thesis Summary This thesis has described the measurement of the absorption cross section of cryogenic Er:YAG and the investigation of cryogenic Er:YAG lasers at under a variety of pumping conditions Spectroscopy High resolution measurements of the absorption cross section of Er:YAG at 300 K and 77 K were presented. In the nm absorption band, many absorption features have widths >1 nm at 77 K suitable for diode pumping, most notably the 1453 nm absorption peak. Measurements of the small 1542 and 1546 nm absorptions at 77 K are also presented at high resolution: the 1546 nm peak has a peak absorption cross section of cm 2 and spectral width 0.7 nm, suitable for pumping with a spectrally narrow fibre laser. The absorptions in the nm band which correspond to re-absorption at lasing wavelengths at 300 K were also measured. The absorption cross section of Er:YAG was also measured for intermediate temperatures between 300 K and 77 K at a lower sampling rate. To our knowledge, no measurement of the absorption cross section between 77 K and 300 K is presented in literature, making these results unique. The measurement of the absorption cross section at 120 K was also useful in the characterisation of cryogenic Er:YAG lasers Er:YAG Lasers Diode-pumped Er:YAG lasers were constructed and characterised at room temperature and at cryogenic temperatures. Ethylene glycol was used as a coolant to allow sufficient temperature control for the 1470 nm nominal-wavelength InGaAs diode to 91

110 CHAPTER 6. CONCLUSION be tuned to pump the 1453 nm Er:YAG absorption. A room temperature Er:YAG laser was constructed and analysed. The diode wavelength was tuned to determine the optimal pumping wavelength for the laser. The optimal laser was pumped at nm, with a threshold power of 12.7 W, a slope efficiency of 28 % relative to incident pump power, and maximum measured output power of 4.5 W. When pumped by the cooled diode, the optimal pump wavelength is nm, producing a threshold power of 13.8 W, slope efficiency 20 %, and maximum measured output power 4.0 W. The effect of mounting conditions on the operation of cryogenic Er:YAG lasers was investigated in detail as part of the construction and characterisation of a cryogenic Er:YAG laser. An alignment procedure for the placing the slab in the cryostat consistently and repeatedly was also developed. Initial mounting conditions resulted in a cryogenic Er:YAG laser lasing at 1618 and 1645 nm simultaneously, corresponding to an operating temperature of approximately 120 K with a threshold of 6.2 W, a slope efficiency of 27 %, and a maximum recorded output power of 5.4 W. This laser was also pump-wavelength-tuned, with an optimal pump wavelength that shifted as the pump power (and thus the width of the pump diode spectrum) increased. Unfortunately, subsequent investigation of the slab laser at 77 K was not directly comparable to the 300K Er:YAG laser or the 120 K cryogenic laser due to electrical damage to the diode that significantly reduced diode power and changed pumping conditions. Nevertheless, an Er:YAG slab laser at 77 K was constructed and characterised in both the nm and nm pump bands. In the nm band, we recorded an output power of 1.65 W from 17 W incident pump power at 1469 nm, while in the nm pump band we recorded 1.7 W output from 18 W incident pump power at 1458 nm. The actual optimal laser power for 77 K Er:YAG pumped in the nm band is likely higher than the 1.7 W measured, as changes in the pump focusing as the diode was cooled resulted in low overlap between the pump and laser modes. 6.2 Future Work Characterisation of Optimal Pumping Due to diode damage and mode overlap problems, a satisfactory comparison between pumping a 77 K Er:YAG laser at 1455 nm and 1468 nm could not be obtained under the time constraints of this thesis. To complete this investigation, a method of aligning the pump mode with the lasing mode when the diode is emitting in the nm range is required. This could be achieved with a new cooling system maintaining cold temperatures on the diode for longer periods of time, or by 92

111 6.2. FUTURE WORK modifying the cryostat stand to fit the diode cooling canister closer to the cryostat window to pump with the f=100 mm lens Development of a Higher-Power Cryogenic Er:YAG laser A cylindrical molybdenum laser head was designed for use with a 30 mm 0.5 % doped Er:YAG slab available. The production of this laser head was underway at the submission of this thesis. An obvious future work is to continue the development and testing of this end-pumped cryogenic Er:YAG laser. A high power single-frequency single-mode cryogenic Er:YAG laser suitable for use in third-generation GWI could also be developed in the long-term using knowledge gained from the end-pumped cryogenic Er:YAG laser. This could feature a zigzag slab as used in Miftar Ganija s work producing a 200 W cryogenic Yb:YAG laser. [56] 93

112 94

113 Appendix A Publications This appendix contains publications associated with this work. A.1 Conference Publications A.1.1 Room temperature and cryogenic operation of an Er:YAG laser using a cooled InGaAsP diode Sophie Hollitt, Miftar Ganija, Peter Veitch, and Jesper Munch, abstract presented to the Conference on Optics, Atoms, and Laser Applications (KOALA), Adelaide 2014 Room temperature and cryogenic operation of an Er:YAG laser using a cooled InGaAsP diode Sophie Hollitt 1, Miftar Ganija 1, Peter Veitch 1, Jesper Munch 1 sophie.hollitt@adelaide.edu.au 1 School of Chemistry and Physics, The University of Adelaide, Australia Laser diodes are an inexpensive way of pumping high-power solid state lasers with broad absorption bands. In Er:YAG, a series of peaks in the nm region are typically used for diode pumping. To pump a narrower absorption peak efficiently, either the spectrum of the diode must be narrowed using expensive and complex grating components on each emitter, or a different pump source must be used. We investigate improving the properties of standard InGaAsP diodes by cooling them below 0 C, in order to produce an efficient cryogenic Er:YAG laser. Characterisation of a cooled InGaAsP diode has already been completed, along with lasing results for a room temperature Er:YAG slab. The cooled InGaAsP diode has higher efficiency, lower threshold, and smaller beam divergence than the same diode operated at room temperature. Spectroscopic studies indicate that the cooled InGaAsP pump will be superior to a room temperature InGaAsP pump for the cryogenic Er:YAG slab due to greater pump absorption combined with the higher efficiency of the diode, with laser results forthcoming. 95

114 APPENDIX A. PUBLICATIONS A.1.2 Comparison of diode pumping efficiency of an Er: YAG laser at 300 K and 77 K for Gravitational Wave Interferometry Sophie Hollitt, Miftar Ganija, Peter Veitch, and Jesper Munch, abstract presented to the AIP Congress The Art of Physics, Canberra 2014 Comparison of diode pumping efficiency of an Er:YAG laser at 300 K and 77 K for Gravitational Wave Interferometry Sophie Hollitt 1, Miftar Ganija 1, Peter Veitch 1 and Jesper Munch 1 1 Dept of Physics,The University of Adelaide, SA 5005 High power, high stability single frequency laser sources are required for gravitational wave interferometry conducted by the LIGO and VIRGO collaborations.[1] The next generation of LIGO may use cryogenic silicon test masses to reduce thermal noise, and as a result a new laser system operating in the m transmission band of silicon is needed.[2] A cryogenic Yb:YAG laser with excellent beam quality and high stability at high output power has already been developed here at the University of Adelaide.[3] By adapting this technology to a YAG crystal doped with erbium rather than ytterbium, a high power laser operating at µm may be developed for LIGOs next generation requirements. Laser diodes are an inexpensive way of pumping high-power solid state lasers with broad absorption bands. In Er:YAG, a series of peaks in the nm region are typically used for diode pumping, shown in Figure 1. To pump a narrower absorption peak efficiently, either the spectrum of the diode must be narrowed using expensive and complex grating components on each emitter, or a different pump source must be used. We investigate improving the properties of standard InGaAsP diodes just by cooling them. Measured absorption cross section of 1% doped Er:YAG x Absorption cross section (cm 2 ) K 77 K Wavelength (nm) Figure 1: Measured absorption cross-section of 1 % doped Er:YAG. Absorption at 300 K in red, absorption at 77 K in blue We have found that the cooled laser diode has superior pumping properties compared to the same diode operated at room temperature. The divergence of the diode is lower, allowing better mode-matching and thus better lasing efficiency. The laser diode itself is also more efficient when operated below 0 C, with a lower threshold and additional 10 % output power at maximum current when operated at 8 C. This temperature corresponds to the 1454 nm pump peak. A 0.5 % doped Er:YAG slab is used with a flat-flat resonator is used to test the InGaAsP pump diode. In the first round of experiments, the Er:YAG slab is maintained at room temperature. Initial results with both room temperature slab and room temperature operation of the diode are shown in Figure 2. A slope efficiency of 39 % incident pump power is demonstrated. This corresponds to a maximum recorded output power of 4.6 W at 1645 nm when the diode spectrum is centred on the 1465 nm absorption peak. Laser studies with the cooled diode operating at 1454 nm are continuing, and will be presented. Laser output at 1645 nm (W) Er:YAG lasing with room temperature pumping y = x Incident pump power (W) Figure 2: Performance of Er:YAG laser pumped at 1465 nm The efficiency of room-temperature Er:YAG lasers is limited by their quasi-three-level behaviour, with 9 % inversion required for lasing. By operating at cryogenic temperatures, Er:YAG becomes a four level laser and reabsorption at the laser wavelength is removed. Additionally, many of the thermal properties of YAG are improved at cryogenic temperatures, including dn/dt. In the next stage of experiments, we will be cryogenically cooling the Er:YAG slab to compare laser operation for both the room temperature and cooled diode configurations. Broad-band InGaAsP diodes are typically designed to operate near 1470 nm such that their spectral output overlaps with multiple absorption peaks in this band. Similarly, broad-band InP diodes can be used to overlap with and pump the Er:YAG absorptions at 1454 and 1458 nm. By cooling an InGaAsP diode below 0 C we can reach the 1454 nm absorption and directly compare lasing outcomes for different pump peaks using the same laser diode stack at different temperatures. References [1] Gregory M Harry, Advanced LIGO: the Next Generation of Gravitational Wave Detectors. Classical and Quantum Gravity, 27(8):084006, April 2010 [2] LIGO Scientific Collaboration, Instrument Science White Paper, 2012 [3] Miftar Ganija, David Ottaway, Peter Veitch and Jesper Munch, Cryogenic, high power, near diffraction limited Yb:YAG slab laser, Optics Express, 21(6):6973-8, Mar

115 A.1. CONFERENCE PUBLICATIONS A.1.3 Development of a cryogenic Er:YAG slab laser for Gravitational Wave Interferometry Sophie Hollitt, Miftar Ganija, Peter Veitch, and Jesper Munch, poster presented to the AIP Congress The Art of Physics, Canberra 2014 Development of a cryogenic Er:YAG slab laser for Gravitational Wave Interferometry Sophie Hollitt, Miftar Ganija, Peter Veitch and Jesper Munch sophie.hollitt@adelaide.edu.au Dept of Physics, School of Chemistry and Physics, the University of Adelaide, 5005 Institute for Photonics and Advanced Sensing, the University of Adelaide, 5005 Pumping study: Er:YAG laser at 300K: Motivation: High power, stable single frequency laser sources are required for gravitational wave interferomety. The sensitivity of current gravitational detectors will be limited by thermal noise in the test masses when the stored optical power is increased. A solution to this problem is to change the material for the test masses from fused silica to single crystal silicon. [1] Such a change will require stable lasers with wavelengths longer than the bandgap of silicon, while still short enough to make use of InGaAs detectors. This choice leads to a preferred laser wavelength in the µm band. We propose a high power cryogenic Er:YAG laser operating at µm, adapting knowledge from a cryogenic Yb:YAG laser already developed at the University of Adelaide.[2] Aim: Assess if a cryogenic Er:YAG laser operating at µm is scalable to the 600W single frequency, TEM00 mode required. This requires an analysis of spectroscopy and pumping approach: Determine the optimal doping concentration and pump wavelength using spectroscopy. Test efficiency of pumping with broadband laser diodes, to avoid expensive Volume Bragg Gratings if possible. Examine the change in efficiency when cooling the diode below 0 C. Preliminary Er:YAG laser design: This design is based on our previous 200 W Yb:YAG laser experience. Cryostat (with AR coated windows) Focusing optics HR@1617nm AR@1470nm AR@1617nm MaxR@1470nm Output Coupler 1617 nm Pump source 0.5% Er:YAG Pumped using a 1470 nm-nominal-wavelength InGaAsP diode 20mm Brewster-cut 0.5% Er:YAG slab mounted between two copper watercooled heatsinks. InGaAs Diode The InGaAsP diode is cooled to access the nm pump band. As the diode warms, the wavelength increases, allowing us to scan the diode centre wavelength. For the cooled diode, we observe: Up to 9% increase in slope efficiency Up to 25% increase in output power at maximum current. This allows us to use a lower current (and thus lower spectral bandwidth) to receive the same output power as at room temperature. Reduced beam divergence 0.5% Er:YAG HR@1645nm AR@1470nm AR@1470nm Brewster face R=87% f=10cm Sealed box (dry nitrogen filled) To OSA Copper heatsink 1645 nm The 300K Er:YAG laser demonstrates CW output power of 4.5 W when pumped with just under 30 W of diode power at 1468 nm, or just under 4 W of output power when pumped with 34 W of diode power at 1456 nm. At 77K, we anticipate the cooled diode to outperform the room temperature diode due to the large, broad absorptions in the nm band. Molybdenum laser head (two semi-cylindrical pieces) Interferometry of Er:YAG gain medium at 77K: Molybdenum mount has similar thermal expansion to YAG to reduce mounting stress. Final high-power design to incorporate zig-zag slab gain medium. Design verification experiments include: 1. Spectroscopy 2. Pumping studies at 300K and 77K 3. Initial laser design verification using existing hardware while final design is in production. This includes benchmarking room temperature behaviour and comparing with lasing at 77K. 300K 77K Raw Increased Contrast Spectroscopy: Measured broader absorption features in nm band suitable for diode pumping. Peaks at 1453, 1458 and 1466 nm remain broad even at 77K - most suitable for diode pumping. Strong, narrow absorptions (especially 1532 nm) only suitable for narrow fibre-laser pumping. Steel spacers Water-cooling fittings Slab: coated end face Uses the same copper mounting block as for the 300K pumping study laser Interferograms are recorded in a double-pass configuration. Copper head imparts additional stress on the slab when cryogenically cooled (one double-pass fringe). This stress will be eliminated using the molybdenum laser head design. Continuing work: Full characterisation of cooled and room temperature diode pumping for the cryogenic slab. Lasing with molybdenum laser head assembly. Refinement to higher power, TEM00 cryogenic Er:YAG laser. References: [1] LIGO Scientific Collaboration, Instrument Science White Paper, 2012 [2] Miftar Ganija, David Ottaway, Peter Veitch and Jesper Munch, Cryogenic, high power, near diffraction limited Yb:YAG slab laser, Optics Express, 21(6):6973-8, Mar