A Deep Ultraviolet Laser Light Source by Frequency Doubling of Gan Based External Cavity Diode Laser Radiation

A deep ultraviolet laser light source by frequency doubling of GaN based external cavity diode laser radiation

vorgelegt von M. Sc. Norman Ruhnke

an der Fakultät IV - Elektrotechnik und Informatik der Technischen Universität Berlin zur Erlangung des akademischen Grades

Doktor der Naturwissenschaften - Dr. rer. nat. -

genehmigte Dissertation

Promotionsausschuss:

Vorsitzender: Prof. Dr. Wolfgang Heinrich Gutachter: Prof. Dr. Günther Tränkle Gutachter: Priv.-Doz. Dr. Bernd Sumpf Gutachter: Prof. Dr. Paul Michael Petersen

Tag der wissenschaftlichen Aussprache: 24. August 2020

Berlin 2020

Zusammenfassung

Eine kompakte und portable Laserlichtquelle im Wellenlängenbereich zwischen 210 nm und 230 nm würde zahlreiche Anwendungen im alltäglichen Kontext ermöglichen, wie zum Beispiel die Sterilisation und Desinfektion von medizinischem Equipment und Wasserdesin- fektion, die Gasanalyse mittels Absorptionsspektroskopie, oder die Identiﬁkation und Quantiﬁzierung von Proteinen und Biomolekülen mittels laserinduzierter Fluoreszenz- oder Raman-Spektroskopie. Leuchtdioden sind zwar besonders kompakt, emittieren jedoch zu breitbandiges Licht für einige dieser Anwendungen. Auf der anderen Seite stehen etablierte Lasersysteme in diesem Wellenlängenbereich zur Verfügung, die zwar ausreichend schmalbandig emittieren, jedoch komplexe Laborsysteme mit hohem Stromverbrauch und großen Abmaßen darstellen, wodurch diese für Feldanwendungen außerhalb des Labors oftmals ungeeignet sind.

In dieser Arbeit wird daher ein neuartiges Konzept entwickelt und untersucht, um mittels Frequenzkonversion Diodenlaser-basierter Laserstrahlung eine besonders kompakte und portable im tiefen ultravioletten Spektralbereich emittierende Laserlichtquelle mit geringem Stromverbrauch zu realisieren, die diese Anwendungslücke schließen kann. Das Konzept basiert auf einer single-pass Frequenzverdopplung des blauen Lichtes einer kommerziell erhältlichen Hochleistungs-GaN-Laserdiode, die in dieser Form in dieser Arbeit nach bestem Wissen des Autors zum ersten Mal demonstriert wird.

Als Pumpquelle für die Frequenzverdopplung in einem BBO-Kristall wird aufgrund der geringen Konversionseﬃzienzen in diesem Wellenlängenbereich von ca. 10−4 W−1 eine Laserdiode mit hoher Ausgangsleistung über 1 W mit schmalbandiger Emission im Be- reich der Phasenanpassungs-Akzeptanzbandbreite des verwendeten BBO-Kristalls benötigt, um eine für die genannten Anwendungen ausreichende Ausgangsleistung von mindestens 100 µW zu erreichen. Da GaN-basierte Hochleistungslaserdioden für gewöhnlich ein brei- tes Emissionsspektrum mit einer vollen Halbwertsbreite von ∆λ = 1...2 nm aufweisen, wird in dieser Arbeit zum ersten Mal die Wellenlängenstabilisierung und Verringerung der spektralen Bandbreite einer solchen GaN-Hochleistungslaserdiode mithilfe externer wellenlängenselektiver optischer Elemente gezeigt und untersucht.

Um ein besseres Verständnis davon zu erlangen wie sich die verwendete Laserdiode unter optischer Rückkopplung verhält, wird zunächst in einer Machbarkeitsstudie ein makroskopischer Aufbau mit einem reflektierenden Oberflächengitter als externes optisches Element realisiert. Mithilfe des Aufbaus wird analysiert welche spektrale Bandbreite der optischen Rückkopplung und welche Rückkopplungsstärke zu einer für die nachfolgende Frequenzverdopplung ausreichend schmalen Emissionsbandbreite bei gleichzeitig hoher Aus- gansleistung führt. Es wird gezeigt, dass ein Rückkopplungsanteil von 15% ausreichend für die Wellenlängenstabilisierung der Laserdiode ist und die resultierende Emissionsbandbreite in etwa der spektralen Breite der optischen Rückkopplung des Oberflächengitters entspricht.

i ii

Mit diesem makroskopischen Aufbau wird eine schmale Emissionsbandbreite von ∆λ ≤ 20 pm bis zu einer optischen Ausgangsleistung von 470 mW, und von ∆λ ≤ 70 pm bis zu einer Ausgangsleistung von 680 mW mit einer maximalen Unterdrückung der longitudinalen Moden der Laserdiode von 42 dB (bei 545 mW Ausgansleistung) bei einer Emissionswellenlänge von 445 nm erreicht. Außerdem lässt sich die Emissionswellenlänge über einen Bereich von 4 nm schmalbandig durchstimmen.

Der makroskopische, wellenlängenstabilisierte Diodenlaser wird als Pumpquelle für die single-pass-Frequenzverdopplung zu 222.5 nm in einem BBO-Kristall getestet, um zu ver- stehen welchen Einﬂuss die Strahlformung auf die Konversionseﬃzienz und auch auf die Phasenanpassungstoleranzen hat. Es zeigt sich, dass die Phasenanpassungsbandbreiten im BBO-Kristall bei einem fokussierten Pumpstrahl aufgrund der verkürzten Wechsel- wirkungslänge im Kristall deutlich breiter sind als die mit der Näherung der ebenen Wellen kalkulierten Phasenanpassungsbandbreiten und darüber hinaus von der Stärke der Fokussierung abhängen. In diesem Setup wird mit einer Pumpleistung von 680 mW eine schmalbandige ultraviolette Ausgangsleistung von 16 µW erzeugt.

Basierend auf den erlangten Erkenntnissen wird ein miniaturisiertes Diodenlasermodul auf einem passiv gekühlten Kupferblock mit einer Grundﬂäche von 25 mm x 25 mm entwickelt, bei dem ein holographisches Volumen-Bragg-Gitter als wellenlängenselektives externes Element eingesetzt wird. Dieses miniaturisierte Lasermodul hat keine beweglichen Komponenten mehr und wird an einem Mikromontage-Messplatz mit hoher Präzision aufgebaut. Dadurch zeigt es verglichen mit dem makroskopischen Aufbau verbesserte spektrale Eigenschaften mit einer spektralen Halbwertsbreite von ∆λ ≤ 50 pm bis zu einer optischen Ausgangsleistung von 1.4 W bei einer Emissionswellenlänge von 445 nm und bei gleichzeitiger hoher Unterdrückung der longitudinalen Moden der Laserdiode von bis zu 53 dB über den gesamten Arbeitsbereich.

Das miniaturisierte Lasermodul wird dann als Pumpquelle für die Frequenzverdopplung eingesetzt, um eine ultraviolette Laserlichtquelle mit möglichst kleinen Abmaßen zu realisieren. Basierend auf den Erkenntnissen mit dem makroskopischen Aufbau wird die Strahl- fokussierung für die Frequenzverdopplung weiter optimiert. Aus der Analyse verschiedener Strahlformungen ergibt sich, dass für den asymmetrischen und nicht-beugungsbegrenzten Pumpstrahl der Laserdiode grössere Strahltaillenradien im Bereich von 20 µm bis 30 µm im BBO-Kristall zu höheren Konversionseﬃzienzen führen als der in der Theorie von Boyd und Kleinman für sphärische Gauss-Strahlen empfohlene optimale Strahltaillenradius von 15 µm. Mit der verbesserten Strahlformung und der höheren Pumpleistung des miniaturisierten Lasermoduls kann schließlich eine maximale optische Ausgangsleistung von 160 µW bei einer Wellenlänge von 222.5 nm generiert werden.

Durch das in dieser Arbeit gewonnene Verständnis wird mit dem entwickelten neuartigen Konzept schließlich eine ultraviolette Laserlichtquelle mit einer kompakten Grundﬂäche von ungefähr 5 cm x 30 cm realisiert, die aufgrund der unbeweglichen Komponenten äußerst robust ist und eine geringe Leistungsaufnahme von unter 10 W aufweist. Mit der demonstrierten Ausgangsleistung von über 100 µW eröﬀnen sich somit zahlreiche Anwen- dungsmöglichkeiten außerhalb von Laborumgebungen im alltäglichen Kontext und in der Industrie, für die bisherige Lasersysteme zu komplex und zu kosten- und energieinstensiv sind. Abstract

A compact and portable laser light source emitting in the wavelength range between 210 nm and 230 nm would enable numerous applications especially outside the laboratory environment such as sterilization and disinfection of medical equipment, water purification, gas and air analysis by means of absorption spectroscopy, or the identification and quan- tification of proteins and biomolecules by means of laser induced fluorescence or Raman spectroscopy. Light emitting diodes are especially compact but their spectrally broandband emission is not suitable for some of the mentioned applications. On the other hand, there are established laser systems in this wavelength region available that show sufficiently narrow emission bandwidths but are too complex laboratory systems with high power consumptions and large footprints making them not suitable for field applications outside of laboratory environments.

This thesis therefore developes and investigates a novel concept to realize an especially compact and portable light source with low power consumption emitting around 222 nm that is based on frequency conversion of laser diode emission and able to close this appli- cational gap. The concept is based on single-pass frequency doubling of a commercially available high-power GaN laser diode emitting in the blue spectral range and, to the best of the authors knowledge, will be presented for the ﬁrst time in this work.

Due to the low power conversion efficiencies of about 10−4 W−1 in this wavelength range, a laser diode with high optical output power above 1 W with narrowband emission in the range of the acceptance bandwidth of the applied nonlinear BBO crystal is required as pump source to achieve an ultraviolet output power sufficient for all of the mentioned applications of at least 100 µW. Since GaN based high-power laser diodes typically exhibit a broad emission spectrum of ∆λ = 1...2 nm, wavelength stabilization and narrowing of such GaN based high-power laser diode emission by the use of external wavelength selective elements will be presented and investigated for the first time in this work.

To gain a better understanding of the laser diodes behavior under optical feedback, an external cavity diode laser (ECDL) system with a surface diffraction grating as external element is realized as a proof-of-concept study. With this setup it is analyzed which spectral bandwidth of the grating feedback and which feedback strength leads to a sufficient narrowband emission with simultaneous high optical output power for the subsequent second harmonic generation. It will be shown that feeding as much as 15% of the laser diode radiation back into the laser is sufficient for the wavelength stabilization and that the resulting ECDL emission bandwidth is in the range of the spectral bandwidth of the optical feedback from the grating.

The proof-of-concept ECDL setup exhibits a narrow emission bandwidth of ∆λ ≤ 20 pm (FWHM) up to an output power of about 470 mW, and ∆λ ≤ 70 pm (FWHM) up to an output power of about 680 mW with a maximum suppression of the longitudinal laser

iii iv diode modes of 42 dB (at 545 mW output power), at an emission wavelength of 445 nm. Furthermore, the narrowband emission can be coarsely tuned over 4 nm.

The ECDL system is then tested as pump source for the single-pass second harmonic generation of laser light at 222.5 nm using a BBO crystal as nonlinear material to understand what inﬂuence the beam shaping has on the conversion eﬃciency and also on the phase matching tolerances. It is shown that due to the decreased interaction length inside the crystal the phase matching acceptance bandwidths of BBO are much broader for focused beams than the simulated phase matching acceptance bandwidths derived from the plane-wave approximation and also depend on the focusing strength. With the ECDL, narrowband DUV laser light with a continuous wave output power of 16 µW is generated with a pump power of 680 mW.

Based on the previous ﬁndings, a micro-integrated ECDL (µECDL) module assembled on a conduction cooled copper package with a footprint of 25 mm x 25 mm is developed. Here, a holographic volume Bragg grating serves as external wavelength selective element. This µECDL module has no moveable parts and is built using a mirco-assembly setup with high precision. Therefore, it shows an improved performance compared to the macroscopic EDCL setup having a narrow emission bandwidth of ∆λ ≤ 50 pm up to an output power of 1.4 W at an emission wavelength of 445 nm, and a mode suppression ratio as high as 53 dB over the whole operating range.

To realize an ultraviolet laser light source that is as compact as possible the µECDL module is then applied as pump source for the frequency doubling. Based on the findings with the macroscopic ECDL, the beam focusing for the frequency conversion is further optimized. From the analysis of different focusing conditions it is found that for the asymmetric and non-diffraction limited laser diode output, a larger beam waist radius in the range of 20 µm to 30 µm results in the highest conversion efficiency, in contrast to the Boyd-Kleinman theory for focused Gaussian beams that recommends an optimum beam waist radius inside the BBO crystal of 15 µm. With the improved beam shaping and the higher pump power, an optical output power of PDUV = 160 µW at a wavelength of λDUV = 222.5 nm is generated.

With the understanding for the novel concept gained in this work, a compact ultraviolet laser light source with a small footprint of approximately 5 cm x 30 cm is realized, that due to its immoveable components is exceptionally robust and has a low power consumption of less than 10 W. This light source with a demonstrated output power above 100 µW enables numerous applications outside of the laboratory environment in the everyday context and in the industry for which previous laser systems are too complex and too cost- and energy-intensive. List of publications

Parts of this work were published in peer-reviewed journals and presented on national and international conferences.

Peer-reviewed journal publications 1. N. Ruhnke, A. Müller, B. Eppich, M. Maiwald, B. Sumpf, G. Erbert, and G. Tränkle, 400 mW external cavity diode laser with narrowband emission at 445 nm, Optics Letters, 39(13), 3794-3797 (2014).

2. N. Ruhnke, A. Müller, B. Eppich, R. Güther, M. Maiwald, B. Sumpf, G. Erbert, and G. Tränkle, Single-pass UV generation at 222.5 nm based on high-power GaN external cavity diode laser, Optics Letters, 40(9), 2127-2129 (2015). 3. N. Ruhnke, A. Müller, B. Eppich, M. Maiwald, B. Sumpf, G. Erbert, and G. Tränkle, Micro-Integrated External Cavity Diode Laser With 1.4-W Narrowband Emission at 445 nm, IEEE Photonics Technology Letters, 28(24), 2791-2794 (2016). 4. N. Ruhnke, A. Müller, B. Eppich, M. Maiwald, B. Sumpf, G. Erbert, and G. Tränkle, Compact Deep UV System at 222.5 nm Based on Frequency Doubling of GaN Laser Diode Emission, IEEE Photonics Technology Letters, 30(3), 289-292 (2018).

Conference Contributions 1. N. Ruhnke, A. Müller, B. Eppich, M. Maiwald, B. Sumpf, G. Erbert, and G. Tränkle, 400 mW output power at 445 nm with narrowband emission from an external cavity diode laser system, Proc. SPIE 9382, Photonics West, San Francisco, USA, Feb. 07-12, 93820P (2015).

2. N. Ruhnke, A. Müller, B. Eppich, R. Güther, M. Maiwald, B. Sumpf, G. Erbert, and G. Tränkle, Narrowband GaN external cavity diode laser with 400 mW output power at 445 nm for deep ultraviolet frequency doubling, Conf. on Lasers and Electro- Optics/Europe and European Quantum Electronics Conf. (CLEO/Europe-EQEC 2015), Jun. 21-25, Munich, Germany, ISBN: 978-1-4673-7475-0, paper CB-P-6 (2015).

3. N. Ruhnke, A. Müller, B. Eppich, R. Güther, M. Maiwald, B. Sumpf, G. Erbert, G. Tränkle, Compact deep UV laser system at 222.5 nm by single-pass frequency doubling of high-power GaN diode laser emission, Proc. SPIE 9731, Photonics West, San Francisco, USA, Feb. 13-18, 97310A (2016).

4. N. Ruhnke, A. Müller, B. Eppich, M. Maiwald, B. Sumpf, G. Erbert, G. Tränkle, Compact deep UV laser system at 222.5 nm by frequency doubling wavelength-stabilized emission of a micro-integrated high-power GaN diode laser module, Proc. SPIE 10516, Photonics West, San Francisco, USA, Jan. 27 - Feb. 1, 10516-5 (2018).

v vi

5. B. Sumpf, N. Ruhnke, A. Müller, B. Eppich, M. Maiwald, G. Erbert, G. Tränkle, Deep UV light source at 222 nm based on second harmonic generation of GaN high power diode lasers, ICULTA (International Conference on UV LED Technologies & Applications), Berlin, Germany, April 22-25 (2018), (invited).

6. B. Sumpf, N. Ruhnke, A. Müller, B. Eppich, M. Maiwald, G. Erbert, G. Tränkle, Deep UV laser systems at 222.5 nm by single-pass frequency doubling of wavelength stabilized high-power GaN diode lasers, 2nd International UV WORKshop (hosted by LASER COMPONENTS), Olching, Germany, November 26-27 (2019), (invited). Danksagung

Zunächst gilt mein besonderer Dank Herrn Professor Günther Tränkle für die Möglichkeit, meine Doktorarbeit am Ferdinand-Braun-Institut, Leibniz-Institut für Höchstfrequenztech- nik (FBH), anfertigen zu dürfen. Ohne unsere regelmäßigen Besprechungen, in denen mir immer wieder neue Perspektiven auf meine Arbeit aufgezeigt wurden, und seine auch über inhaltliche Fragen hinausgehenden Ratschläge wäre diese Arbeit nicht möglich gewesen.

I would also like to express my sincere gratitude to Professor Paul Michael Petersen for his willingness to act as the external reviewer for my thesis. Einen besonderen Dank möchte ich auch an Herrn Götz Erbert aussprechen. Gerade in der Anfangszeit war er maßgeblich für die Ausrichtung und Zielsetzung meiner Arbeit mitverantwortlich und hat mit seiner langjährigen Erfahrung und seinem Wissen auf dem Gebiet der Laserdioden immer wieder für die richtigen Impulse gesorgt.

Bernd Sumpf hat als Gruppenleiter des Laser Sensors Lab und als mein direkter Be- treuer ebenso großen Anteil am Gelingen dieser Arbeit. Sein Enthusiasmus und seine wertvollen Tipps zu experimentellen Fragestellungen waren ein ständiger Antrieb. Und seine zahlreichen kleinen Anekdoten und Scherze haben die Arbeit am FBH ein ganzes Stück unterhaltsamer gemacht.

Auch meinen Kollegen Martin Maiwald, André Müller, Christof Zink und Bernd Eppich möchte ich für die angenehme Zusammenarbeit und die tagtägliche Unterstützung bei inhaltlichen Fragestellungen und im Labor danken.

Außerdem danke ich den Mitarbeitern der Werkstatt Sebastian Deutscher, Detlef Grimpe und Thomas Roos für die Anfertigung zahlreicher Bauteile und Ihre Unterstützung bei vielen kleinen und großen technischen Problemen. Neben allen anderen FBH-Mitarbeitern möchte ich speziell Manuela Münzelfeld danken, die ohne Übertreibung als die gute Seele der Optoelektronik-Abteilung des FBH bezeichnet werden kann.

Meinen Doktoranden-Kollegen Marcel Braune, Martin Winterfeldt, Mahmoud Tawﬁeq, Lara-Sophie Theurer, Jonathan Decker, Carlo Frevert, Matthias Karow, Thorben Kaul, Daniel Jedrzejczyk und Juliane Rieprich möchte ich ebenso meinen großen Dank aussprechen. Ihr habt die Zeit am FBH zu einem unvergesslichen Erlebnis gemacht und einige von Euch sind zu guten Freunden geworden.

Alex, Andreas, Eric, Florian, Jakob, Johannes, Josi, Karl, Linda, Lisa, Lukas, Matthias, Michelle, Ryan, Sascha und Tillmann. Danke für Eure aufmunternden Worte, gelegentliche Ablenkung, und Eure ständige Unterstützung. Gleiches und noch mehr gilt für meine Eltern und Oma Dora. Ohne Euch wäre ich nicht bis hierher gekommen, danke für alles.

vii

Contents

1 Introduction1

2 Fundamentals for frequency doubling with GaN based laser diodes7 2.1 Nonlinear optics...... 8 2.1.1 Second harmonic generation...... 9 2.2 Nonlinear materials for deep ultraviolet light generation...... 12 2.3 Birefringent phase matching in BBO...... 14 2.3.1 Angle tuning...... 16 2.3.2 Phase matching tolerances...... 18 2.4 Second harmonic generation with focused Gaussian beams...... 20

3 Characterization of the applied laser diode 23 3.1 Working principle of laser diodes and vertical layer structure...... 24 3.2 Longitudinal modes...... 26 3.3 Optical gain and threshold condition...... 27 3.4 Electro-optical characteristics...... 28 3.5 Spectral emission characteristics...... 29 3.6 Spatial emission characteristics...... 31 3.7 Implications for the development of a deep ultraviolet laser light source.. 35

4 External cavity diode lasers as pump sources for DUV generation 37 4.1 Wavelength stabilization by external optical feedback...... 37 4.1.1 ECDLs with surface diﬀraction gratings...... 40 4.2 Macroscopic external cavity diode laser in Littrow conﬁguration...... 44 4.2.1 Experimental Setup...... 45 4.2.2 Electro-optical characteristics...... 48 4.2.3 Spectral emission characteristics...... 51 4.2.4 Wavelength tuning...... 54 4.2.5 Spatial emission characteristics...... 55 4.3 Volume-Bragg-Grating stabilized external cavity diode laser module (µECDL) 56 4.3.1 Working principle of volume Bragg gratings...... 57 4.3.2 Concept and development of the µECDL module...... 60 4.3.3 Electro-optical characteristics...... 64 4.3.4 Spectral emission characteristics...... 65 4.3.5 Temporal stability of the µECDL emission...... 68 4.3.6 Spatial emission characteristics...... 69 4.4 Summary...... 70

5 Compact deep ultraviolet laser light source 71 5.1 Proof-of-concept setup with macroscopic ECDL as pump source...... 72 5.1.1 Detection of deep ultraviolet light...... 74 5.1.2 Investigation of conversion eﬃciency and phase matching tolerances 76

ix x Contents

5.2 Micro-integrated ECDL module as pump source...... 85 5.2.1 Compact deep ultraviolet laser light source...... 85 5.2.2 Optimization of the focusing conditions...... 87

6 Conclusion and Outlook 93

A Datasheet information for laser diode PL TB450B (Osram Opto Semi- conductors GmbH) 97

Bibliography 99

List of Figures 111

List of Tables 117 1 Introduction

Motivation The development of the laser diode with its high eﬃciency, compactness, robustness and long lifetime has enabled the construction of compact and non-sophisticated laser systems that have accessed many new applications outside of the laboratory environment. Nowa- days, many ﬁelds of application such as information and communication technology, data storage, consumer electronics, materials processing, spectroscopy, biophotonics, and life science are addressed with laser diodes.

A spectral region, that is particularly interesting for numerous applications, is the wavelength range below 300 nm, hereinafter called deep ultraviolet (DUV). DUV light between 200 nm and 300 nm can sterilise bacteria, viruses and other pathogens by disrupting the structure of their DNA and can therefore be used for chemical-free disinfection of medical equipment [1–3], air conditioning systems and for water puriﬁcation [4,5].

It also has a great potential for sensing applications as many gases, that are major atmospheric pollutants and are primarily produced by fossil fuel combustion such as NO, NO2, SO2, and NH3, exhibit strong electronic transitions in the wavelength region between 210 nm and 230 nm [6–8]. DUV absorption spectroscopy can be used to monitor the pollutant concentrations in ambient air. Typically, a spectral linewidth of about 5-10 cm−1 is usually sufficient for absorption spectroscopy of liquids and solids and in many cases also of gases [9]. Biomolecules and proteins such as tryptophan, NADH, tyrosine, DNA, RNA, and many others exhibit strong electronic transitions in the wavelength range below 250 nm as well and can be identified and quantified by DUV laser induced fluorescence (LIF) spectroscopy and related techniques [10–12]. In general, these techniques do not require a particularly narrow excitation linewidth and an excitation power of a few µW is sufficient [13].

Another prominent example for the wide range of application possibilities of DUV laser light is DUV Raman spectroscopy that can also be used for the identification of proteins and biomolecules [14, 15], explosives [16], and many other substances [17]. As the scattering cross section of the Raman signal scales with ω4 and due to the possible excitation of resonant effects, the Raman signal can be enhanced by many orders of magnitude under DUV excitation compared to excitation with visible or NIR laser light [9]. Addition- ally, for excitation wavelengths below 260 nm, the Raman signal is spectrally separated from the usually much stronger and therefore disturbing fluorescence spectrum of most molecules [18]. Figure 1.1 further illustrates this situation with an exemplary Raman spectrum of the so-called fingerprint region of polystyrene exhibiting a spectral width of ∆ν˜ = 1600 cm−1. The hatched curve with the maximum at 600 nm represents a typical fluorescence background. For simplicity, all spectra are normalized to 1. Depending on the excitation wavelength λex, the same Raman spectrum has a different spectral width on

1 2 1 Introduction

10 nm 50 nm 130 nm 250 nm

1.0 f l u o r e s c e n c e = 488 nm 488 = nm 785 = = 225 nm 225 = = 1064 nm 1064 = ex ex ex ex

0.5 N orm alizedintensity

0.0

200 400 600 800 1000 1200 1400

Wavelength nm

Figure 1.1: Illustration of the spectral separation of Raman signal and fluorescence background for DUV excitation using the example of the fingerprint region of the Raman spectrum of polystyrene. the wavelength scale. For 225 nm excitation, it has a spectral width of only 10 nm and is spectrally separated from the fluorescence background. The spectral separation of Raman and fluorescence signal can also be used to obtain both signals in a single measurement which might be advantageous in some cases. To resolve the investigated Raman spectra, the spectral emission bandwidth of the excitation laser light has to be in the range of the spectral width of individual Raman lines. For liquid and solid samples, these lines typically have a width of ∆ν˜ = 10 cm−1 [19]. For an excitation wavelength of λex = 225 nm, this translates into a required spectral width of ∆λ = 50 pm on the wavelength scale. Although higher optical output power is always desired, it was demonstrated that a DUV output power around 100 µW or even lower is already sufficient for absorption spectrocscopy of gases [20], for LIF detection [13], and DUV Raman spectroscopy [21, 22]. In a laboratory environment, an excitation power of a few nW was shown to be sufficient for absorption spectroscopy [7]. Especially when biological samples are under investigation too high output powers can lead to the destruction of the molecules as absorption in the DUV wavelength range is usually strong [22]. Table 1.1 summarizes the required wavelength range λ, average optical output power Popt, and spectral linewidth ∆ν˜ in wavenumbers for a laser light source targeting these applications, which all have a great potential to be used outside of laboratory environments. This requires a compact and robust, if possible handheld and battery driven DUV laser light source. Within the frame of this work, the development and characterization of such a DUV light source based on laser diode emission is pursued that meets the physical requirements summarized in table 1.1 and utilizes the advantages of laser diodes in terms of robustness, small footprint and low power consumption.

Established DUV laser light sources Established laser light sources emitting in the DUV wavelength range are gas lasers and frequency-quadrupled solid state lasers. Table 1.2 gives an overview of the emission wavelength λ, average optical output power Popt, and power consumption Pcon of these laser systems. 3

Application Technique λ Popt ∆ν˜

Sterilisation/disinfection DNA disruption by absorption 200 - 300 nm – – of med. equipment and water puriﬁcation

Gas/air analysis Absorption spectroscopy 210 - 230 nm ≥ nW < 10 cm−1

Identiﬁcation and Laser Induced Fluorescence 210 - 230 nm ≥ µW– quantiﬁcation of (LIF) spectroscopy proteins and biomolecules Raman spectroscopy < 260 nm ≥ µW ≤ 10 cm−1

Table 1.1: Possible applications for deep ultraviolet laser light, corresponding techniques, wavelength ranges, and required speciﬁcations.

Gas lasers like the KrF excimer laser (248 nm)[23] or the HeAg (224.3 nm) and NeCu (248.6 nm) hollow cathode lasers [24] directly emit in the wavelength range below 250 nm. The frequency doubled Ar+ laser emitting at 244 nm [25, 26] is a well established DUV laser light source, too. Despite offering high average output powers of more than 100 mW [23, 25, 26], KrF and Ar+ lasers have a large footprint, a high power consumption of more than 1 kW [23, 25, 26], require frequent maintenance, and entail considerable production costs. HeAg and NeCu hollow cathode lasers have a lower power consumption of less than 100 W, but deliver lower average output powers of less than 1 mW [24]. They also require frequent maintenance and have a relatively large footprint. Another commonly applied solution is the generation of the fourth harmonic of Nd:YLF or Nd:YAG solid state lasers, that directly emit at infrared wavelengths of 1047 nm or 1064 nm, respectively. These systems deliver output powers of more than 10 mW, have smaller footprints and a moderate power consumption of around 100 W [27, 28]. Unfor- tunately, they only emit on fixed wavelengths of 262 nm or 266 nm, whereas an emission wavelength well below 250 nm is more suitable for the aforementioned applications. DUV laser systems based on the generation of higher harmonics of the infrared emission from Ti:sapphire lasers offer tunable output in a wavelength range from 193 nm to 270 nm

Laser type λ mode average Popt Pcon

HeAg 224 nm CW < 1 mW < 100 W Ar+ (2nd harmonic) 244 nm pulsed > 100 mW > 1 kW KrF excimer 248 nm pulsed > 100 mW > 1 kW NeCu 249 nm CW < 1 mW < 100 W Nd:YLF (4th harmonic) 262 nm CW > 10 mW ≈ 100 W Nd:YAG (4th harmonic) 266 nm CW > 10 mW ≈ 100 W Ti:Sa (up to 4th harmonic) 193 - 270 nm pulsed 5 - 50 mW > 1 kW

Table 1.2: Established DUV laser light sources and their emission wavelength λ, typical average optical output power Popt, and power consumption Pcon. 4 1 Introduction

[29] together with excellent spectral and spatial beam properties. However, their high power consumption of more than 1 kW, their complexity and large footprint make them more suitable as scientiﬁc tools in the laboratory environment.

Each of the established light sources offers sufficient optical properties in terms of output power and emission linedwidth. However, they all have large footprints, high power consumptions and are rather complex, which usually restricts experiments to the laboratory environment. Furthermore, except for the Ti:sapphire based laser systems they all emit on fixed wavelengths only. To utilize DUV lasers in fields of application outside the laboratory environment as a portable in situ analysing and monitoring tool, a more compact and reliable DUV laser light source with minimal power consumption ideally suitable for battery operation is necessary. In this sense, a promising approach is frequency conversion of laser diode emission into the DUV wavelength range.

Diode laser based DUV light sources GaAs based laser diodes cover a wavelength range between 600 nm and 1200 nm with optical output powers in the watt range and reach wall-plug efficiencies of more than 70% [30] making them to the most efficient devices in converting electrical into optical energy [31]. Since the first demonstration of continuous wave emission from an InGaN based laser diode by Nakamura et al. in 1996 [32], also the green, blue, and parts of the ultraviolet spectral region can be addressed by direct emission from quantum well diode lasers. However, the shortest wavelength emitted by an electrically pumped AlGaN based laser diode so far is 336 nm [33] and direct DUV emission from laser diodes is still out of reach until now.

Therefore, a diode laser based DUV light source can only be realized by frequency conversion. A selection of relevant works on DUV frequency conversion of laser diode radiation is summarized in table 1.3. The concept of fourth harmonic generation of the infrared radiation of GaAs based laser diodes was already demonstrated in the 90s. It was realized by using either two successive single-pass (SP) [20, 34] or resonant cavity-enhanced (CE) [35, 36] frequency doubling (second harmonic generation, SHG) configurations. Goldberg [34] and Koplow [20] used very similar single-pass setups with a GaAlAs tapered amplifier (TA) laser diode emitting at 860 nm as pump source and a BBO crystal for DUV generation. Goldberg achieved a DUV optical output power of 15 µW at 215 nm. In the work of Kliner, the focus was on a minimized DUV emission bandwidth for absorption spectroscopy and an output power of 240 nW at 215 nm was generated. Zimmermann et al. demonstrated a scheme with a GaAs master oscillator power amplifier (MOPA) laser diode (LD) emitting at 972 nm as pump source and two successive cavity- enhanced frequency doubling stages [35]. In the second stage, a BBO crystal was again used for the DUV generation of 2.1 mW at a wavelength of 243 nm. Schwedes et al. used a seeded GaAs tapered amplifier laser diode emitting at 922 nm to generate 1 mW of laser light at 231 nm [36] with a similar setup. The latter concept leads to higher DUV output powers than the single-pass arrangements, but can also become increasingly complex. A commercially available system from TOPTICA Photonics AG based on successive cavity- enhanced frequency doubling nowadays offers DUV output powers of 10 mW at 213 nm and even 300 mW at 266 nm [37, 42]. This system has a footprint of 9 cm x 41 cm x 69 cm and a power consumption of typically 100 W[42]. 5

Author Laser type Method λDUV Popt

2x SHG Goldberg (1995) [34] GaAlAs TA SP 215 nm 15 µW Koplow (1998) [20] GaAlAs TA SP 215 nm 240 nW

Zimmermann (1995) [35] GaAs MOPA CE 243 nm 2.1 mW Schwedes (2003) [36] GaAs TA CE 231 nm 1 mW Toptica (2019) [37] GaAs TA CE 213 nm 10 mW 266 nm 300 mW

SFG Alnis (2000) [38] GaN ECDL + GaAs LD SP 254 nm 1 nW Carruthers (2005) [39] GaN ECDL + GaAs LD SP 254 nm 50 nW Anderson (2005) [40] GaN ECDL + GaAs LD SP 254 nm 4 nW

SHG Nishimura (2003) [41] GaN ECDL CE 209 nm 9 µW

Table 1.3: Overview of diode laser based DUV light sources. SHG: second harmonic generation, SFG: sum frequency generation, TA: tapered ampliﬁer, MOPA: master oscillator power ampliﬁer, SP: single-pass, CE: cavity-enhanced.

Another approach is based on sum frequency generation (SFG) of the emission from a blue and a red emitting laser diode in a single-pass arrangement with BBO as nonlinear crystal [38–40]. In these works, only relatively low output powers in the nanowatt range were achieved and the concept has the inherent disadvantage of needing two laser diodes in operation. A concept that leads to a DUV laser light source with further reduced footprint, power consumption, complexity, and overall production cost is direct second harmonic generation of GaN based laser diode radiation in the blue spectral range. This was already demonstrated by Nishimura et al. [41] using a low-power GaN external cavity diode laser (ECDL) as pump source. To achieve suﬃcient DUV optical output power, a BBO nonlinear crystal was integrated in an enhancement cavity. With this setup, a continuous wave (CW) output power of 9 µW at 209 nm was generated from 26 mW pump power at 418 nm. Besides this work, no further studies of this concept have been published so far.

Goal of this work The goal of this work is to develop and characterize a novel diode laser based DUV light source with a smaller footprint and lower power consumption than previous light sources. The emission wavelength should be in the range between 210 nm and 230 nm with a 1 continuous wave output power of about 100 µW and an emission bandwidth < 10 cm− or < 50 pm. Its specifications are defined by the applications referred to above and are listed in table 1.4. The concept of direct second harmonic generation of GaN based laser diode radiation in the blue spectral range promises to result in the most compact and inexpensive DUV laser light source with minimal power consumption. Conversion efficiencies for DUV generation are quite low (typically 10−4 W−1) usually making cavity-enhanced frequency doubling stages necessary [41]. However, with the recent development of commercially available 6 1 Introduction

wavelength λ 210-230 nm output power P ≈ 100 µW linewidth ∆ν˜ < 10 cm−1 ∆λ < 50 pm

power consumption Pcon < 10 W

Table 1.4: Targeted speciﬁcations for the diode laser based DUV light source. high-power GaN laser diodes emitting around 450 nm with output powers beyond 1 W [43, 44], direct single-pass frequency doubling of such laser diode radiation with DUV output powers around 100 µW has become feasible. This work is intended to serve as proof-of-concept study, i.e. to demonstrate the feasibility of this concept and to investigate the physical challenges, that need to be considered.

The thesis is organized as follows: In chapter2, the theoretical background that needs to be considered for DUV frequency doubling using a high-power GaN based laser diode, is presented. This includes fundamentals of nonlinear optics with an emphasis on second harmonic generation (SHG), an overview of crystals suitable for DUV generation, and estimations regarding the expected phase matching tolerances and nonlinear conversion efficiencies. The applied laser diode (PL TB450B, OSRAM Opto Semiconductors) and its characteristics are presented in chapter3. At the time of this work, the laser diode provided the highest optical output power of P = 1.6 W from a commercially available GaN based laser diode. Its broad spectral emission of ∆λ ≈ 1...2 nm full width at half maximum (FWHM) limits its applicability for efficient frequency conversion and does not meet the application requirements on the DUV light source listed in table 1.4. Hence, the emission bandwidth is reduced by implementing an external cavity diode laser (ECDL) setup. As this is a challenging and also crucial step for the development of the DUV light source, the concept is tested in a proof-of-principle ECDL system using surface diffraction gratings in Littrow configuration. Chapter4 starts with a brief description of wavelength stabilization of laser diodes by external optical feedback and a short literature review on GaN ECDLs with surface gratings. In section 4.2 and the following, the proof-of-principle ECDL system in Littrow configuration and its performance is analyzed. A miniaturized ECDL module (µECDL) with a volume Bragg grating as wavelength selective optical element is presented and discussed in section 4.3. Both ECDL systems are demonstrated and evaluated in single-pass frequency doubling setups with BBO as nonlinear optical crystal in chapter5. In section 5.1, the whole concept is again tested in a proof-of-principle setup by using the macroscopic ECDL in Littrow configuration as pump source. Here, challenges and difficulties of the single-pass concept regarding optimal focusing conditions and phase matching tolerances are discussed. The more compact DUV laser light source with the micro-integrated µECDL as pump source is presented in section 5.2 and the optimization of the focusing conditions is discussed in section 5.2.2. Chapter6 concludes this work, outlines different ideas on possible improvements of the compact DUV system and summarizes remaining challenges of the examined concept. 2 Fundamentals for frequency doubling with GaN based laser diodes

This chapter gives an overview of the relevant physical background for the realization of a DUV laser light source based on second harmonic generation (SHG) of blue GaN laser diode emission. A brief theoretical insight into nonlinear optics with a special emphasis on second order phenomena and second harmonic generation is given in section 2.1. The derivations of the necessary equations are thereby taken from the textbooks of William P. Risk et al. [45], Robert W. Boyd [46], and Richard L. Sutherland [47]. A full description of nonlinear phenomena can also be found in standard textbooks on nonlinear optics [46–48].

Section 2.2 gives an overview of nonlinear crystals suitable for frequency conversion into the ultraviolet wavelength range in general and discusses the advantages and disadvantages of diﬀerent crystals with respect to second harmonic generation into the wavelength range below 250 nm. It will be shown that the nonlinear material of choice for the purpose of this work is β-BaB2O(β-barium borate, BBO). Using BBO for DUV generation by collinear second harmonic generation requires birefringent phase matching, which is explained in section 2.3. Here, the wavelength, temperature and angle tolerances for type I critical phase matching in BBO are calculated for the wavelength addressed in this work. A brief summary of the Boyd-Kleinman theory that predicts the optimum focusing conditions inside the nonlinear crystal for circular Gaussian beams is given in section 2.4.

7 8 2 Fundamentals for frequency doubling with GaN based laser diodes

2.1 Nonlinear optics

The interaction of dielectric materials with the electric field E(z,t) of an incident light wave propagating in z-direction is described by the polarization P (z,t). For moderate intensities of the incident light, the response of the material is linearly dependent upon the strength of the electric field. However, for high intensities of the optical field as provided by lasers, the response of the material can be nonlinearly dependent upon the optical field strength. The first discovery of a nonlinear phenomenon was demonstrated by Franken et al. in 1961 [49]. They observed the second harmonic at a wavelength of 347 nm generated by irradiating a quartz crystal with a ruby laser emitting at a wavelength of 694 nm.

In nonlinear optics, the response of the material is usually generalized by expressing P (z,t) as a power series in the ﬁeld strength E(z,t):

(1) (2) 2 (3) 3 P (z,t) = ε0 χ E(z,t) + χ E (z,t) + χ E (z,t) + ... , (2.1)

(1) (2) (3) where ε0 is the vacuum permittivity, and χ , χ , and χ are the first, second, and third-order susceptibilities, respectively. The first term describes linear phenomena like the index of refraction. The second term with the square of the electric field leads to second order phenomena like second harmonic generation (SHG), sum frequency generation (SFG), difference frequency generation (DFG), parametric fluorescence or optical rectification. The third term with the cube of the electric field treats third order phenomena like third- harmonic generation, the intensity-dependent refractive index, or Brillouin scattering ([45], p. 23) .

For the description of SHG, only the second-order polarization P (2)(z,t) is considered in the following: (2) (2) 2 P (z,t) = ε0χ E (z,t) (2.2) The strength of the induced polarization depends on the second-order susceptibility χ(2) which is taken to be constant in equation (2.2). This assumption is justified, if all the frequencies taking part in the interaction are lying far below the lowest resonance frequency of the nonlinear material. For SHG, this condition is usually fulfilled and the nonlinear susceptibility is independent of the frequency ([46], p. 37). In general, P~ (~r,t) and E~ (~r,t) are 3-dimensional vectors with components in x-, y-, and z-direction. The self-convolution of E~ (~r,t) in equation (2.2) leads to 3 x 3 possible components for each of the three components of the second-order polarization. This means, that (2) is actually a third-rank tensor written as (2) , where the indices , , can χ χijk i j k independently take on the values x, y, and z leading to 27 different components [46]. Some symmetries fortunately reduce the number of independent components [46]. For instance, in the case of SHG, the indices j and k become interchangeable and can therefore be replaced by a new index l according to the following scheme ([46], p. 38): jk xx yy zz yz,zy xz,zx xy,yx l 1 2 3 4 5 6

The second-order susceptibility is then expressed by the contracted notation of the nonlinear 2.1 Nonlinear optics 9

coeﬃcient dil, that is a 3 x 6 matrix with 18 components ([45], p. 28):

1 (2) d = χ (2.3) il 2 ijk Due to other symmetries including spatial crystal symmetries, the number of independent components in dil is further reduced for most nonlinear crystals. A detailed discussion on the symmetries influencing the nonlinear susceptibility can be found in reference [46] (p. 32 ff.). The second-order polarization can now be re-written in the following form ([46], p. 38):   ExEx  (2)     EyEy  Px d11 d12 d13 d14 d15 d16      (2)  = 2    EzEz  (2.4)  Py  ε0 d21 d22 d23 d24 d25 d26  2   (2)  EyEz d31 d32 d33 d34 d35 d36   Pz  2E E  | {z }  x z  dil 2ExEy For known polarization and propagation directions of the participating waves, an effective nonlinear coefficient deff can be calculated and equation (2.2) can be expressed with scalar values (see [46], p. 39): (2) 2 P (t) = 2ε0deffE (t) (2.5)

2.1.1 Second harmonic generation The electric ﬁeld of a plane wave propagating in z-direction and oscillating with frequeny ω is written as: E1(z,t) = A1 cos(k1z − ωt + φ1) (2.6) where A1 is the amplitude of the electric ﬁeld, k1 is the wave vector, and φ1 is the phase.

Inserting equation (2.6) into (2.5) gives the second-order polarization induced in a nonlinear material by the fundamental electric ﬁeld E1(z,t):

(2) 2 P (z,t) = 2ε0deﬀ [A1 cos(k1z − ωt + φ1)] (2.7)

After using a simple trigonometric identity, an expression for the second-order polarization consisting of two terms is obtained:

(2) 2 P (z,t) = ε0deffA1 [1 + cos(2k1z − 2ωt + 2φ1)] (2.8) The first term is at zero frequency giving rise to the process of optical rectification in which a static electric field is created within the nonlinear material. The second term in equation (2.8) oscillates at the second harmonic frequency 2ω. According to the inhomogeneous wave equation 2 2 2 1 2 ∂ ( ) n2 ∂ ( ) = ∂ (2)( ) (2.9) 2 E2 z,t − 2 2 E2 z,t 2 2 P z,t , ∂z c0 ∂t ε0c0 ∂t the second-order polarization can be a source of a second electric field E2(z,t) oscillating with the frequency 2ω: E2(z,t) = A2 cos(k2z − 2ωt + φ2) (2.10) 10 2 Fundamentals for frequency doubling with GaN based laser diodes

a) b) w w w (2) c 2w 2w w

Figure 2.1: a) Schematic sketch of a second harmonic generation process inside a nonlinear crystal. b) SHG process depicted in an energy level diagram.

In this process, called second harmonic generation (SHG), two photons oscillating at frequency ω are transformed into one photon oscillating at the second harmonic frequency 2ω. Figure 2.1 illustrates the SHG process geometrically (a) and in an energy level diagram (b). By substituting equation (2.10) into (2.9) and applying the slowly varying amplitude approximation, one obtains a coupled-amplitude equation for the amplitude of the generated second harmonic wave (see [47], p. 57):

d 2 A2 ωdeﬀ 2 i∆kz = i A1 · e (2.11) dz n2ωc

deff is the effective nonlinear coefficient, n2ω the refractive index for the generated wave, and ∆k is the wavevector mismatch:

∆k = 2k1 − k2. (2.12)

The amplitude of the generated second harmonic wave can be derived by integrating equation (2.11) over the length Lcr of a nonlinear crystal:

L 2 Z cr ωdeﬀ 2 i∆kz A2(Lcr) = i A1 e dz n2ωc 0 (2.13) i∆kL ! 2ωdeﬀ 2 e cr − 1 = i A1 n2ωc i∆k Substituting the amplitude for its intensity

2 I = 2ε0nc|A| , (2.14)

gives with the use of (see [46], p.75)

2 ei∆kLcr − 1 = 2 sinc2( 2) and (2.15) Lcr ∆kLcr/ ∆k

and ω = 2πc/λω, (2.16) an expression for the intensity of the generated second harmonic light: 8 2 2 = π deﬀ 2 2 sinc2 ∆kLcr (2.17) I2 2 2 I1 · Lcr · 2 ε0cn2ωnωλω 2.1 Nonlinear optics 11

This equation will be important for the calculation of the conversion efficiency in a nonlinear crystal in dependence of the phase matching angle and temperature presented in section 2.3. The second harmonic intensity strongly depends on the effective nonlinear coefficient deff, the intensity of the fundamental wave I1, and the crystal length Lcr. The intensity I2 has its maximum for a vanishing wavevector mismatch ∆k = 0. This so-called phase matching condition can be written as:

4π ∆k = 2k1 − k2 = (n(ω) − n(2ω)) = 0 (2.18) λω → n(ω) = n(2ω) (2.19) This means, that the refractive indices for fundamental and second harmonic wave have to be equal, i.e. the phase velocities of both waves have to be equal, so that constructive interference can occur and the SHG intensity can build up inside the nonlinear crystal. Due to dispersion, this condition is usually not fulﬁlled (n(ω) 6= n(2ω)).

However, there are methods to overcome phase-mismatch like quasi-phase matching in periodically poled nonlinear crystals or critical (angle tuning) and non-critical phase matching (temperature tuning) in birefringent nonlinear crystals. The phase matching technique applied in the course of this work is critical birefringent type I phase matching and will be explained in detail in section 2.3. 12 2 Fundamentals for frequency doubling with GaN based laser diodes

2.2 Nonlinear materials for deep ultraviolet light generation Established nonlinear crystals used for deep ultraviolet frequency conversion below 300 nm are summarized in table 2.1. A more comprehensive overview of nonlinear crystals including newly developed and rarely used crystal can be found in [50]. λcut-off is the UV wavelength at which the crystals exhibit close to "0" transmittance. Potassium fluoroboratoberyllate (KBe2BO3F2, KBBF) has the lowest cut-off wavelength of 147 nm, followed by lithium triborate (LiB3O5, LBO), potassium dihydrogen phosphate (KH2PO4, KDP), cesium lithium borate (CsLiB6O10, CLBO), and beta-barium borate (β-BaB2O4, BBO) with 155 nm, 176 nm, 180 nm, and 189 nm, respectively. Frequency conversion almost down to the respective transmission cut-off wavelength of each crystal has been demonstrated by means of sum frequency generation (KBBF: 163 nm [51], CLBO: 175 nm [52], LBO: 187.7 nm [53], KDP: 190 nm [54], BBO: 190.8 nm [55]).

The aim of this work however, is to convert GaN laser diode emission between 440 nm and 460 nm by direct SHG, which leads to the necessity of applying type I phase matching (see section 2.3). For LBO, KDP, and CLBO, the minimum SHG wavelength that can be phase matched with type I phase matching is 277 nm, 256 nm, and 240 nm, respectively, which is not small enough for the targeted wavelength range in this work of 210 nm to 230 nm. A relatively new crystal that was first reported in 1996 [56] and shows promising nonlinear optical properties is KBBF. Based on the Sellmeier equations from [57], phase matching down to 165 nm for type I SHG could in principle be achieved [58] with this crystal material. Togashi et al. experimentally demonstrated direct SHG down to 172.5 nm [51]. However, KBBF possesses a plate-like form and the growth of KBBF crystals thicker than a millimeter is challenging. This makes it difficult to cut the crystal along the phase matching direction for the generation of deep UV light. To circumvent this problem, a prism-coupling technique was suggested that is nonetheless inconvenient for deep UV SHG. Recently, KBBF crystals with a length of up to 8 mm were grown by a new growth technique. Unfortunately, due to their worse optical quality these longer crystals exhibit SHG conversion efficiences up to two orders of magnitude smaller than the plate-like crystals until now [64].

The most established and often used nonlinear crystal for frequency conversion below 300 nm, is beta-barium borate (β-BaB2O4, BBO), that was discovered by Chen et al.

Crystal λcut-off [50] λSHG-cut-off (Type I) deff

KBBF 147 nm 165 nm [58] 0.38 pm/V (λSHG = 223 nm) [59]

LBO 155 nm 277 nm [60] 0.16 pm/V (λSHG = 280 nm) [59]

KDP 176 nm 256 nm [61] 0.48 pm/V (λSHG = 260 nm) [59]

CLBO 180 nm 240 nm [62] 0.92 pm/V (λSHG = 240 nm) [59]

BBO 189 nm 205 nm [63] 1.26 pm/V (λSHG = 223 nm) [59]

Table 2.1: Transmission cut-off wavelength λcut-off, minimum SHG wavelength at which phase matching with type I SHG can be achieved at room temperature (T ≈ 293 K), and effective nonlinear coefficient deff for selected crystals. 2.2 Nonlinear materials for deep ultraviolet light generation 13

in 1985 [65]. It has a significantly higher effective nonlinear coefficient deff and also a higher laser induced damage threshold (10 GW/cm2 for 100 ps pulse-width at 1064 nm [50]) compared to the other crystals listed in table 2.1[ 59, 66]. The SHG conversion efficiency depends on the square of deff. A higher deff therefore means a higher intensity of the generated second harmonic light. Additionally, due to the strong birefringence in BBO, type I phase matching down to 205 nm is theoretically possible and was also demonstrated by Kato et al. in 1986 [63]. BBO is a negative uniaxial crystal that belongs to the point group symmetry class 3m and its nonlinear coefficient dil is [47]:   0 0 0 0 d15 −d22 =  0 0 0  (2.20) dil −d22 d22 d15  d31 d31 d33 0 0 0

A general derivation for calculating deff for each of the crystal classes was presented by Midwinter and Warner [67] (see [46], pp. 39). They showed, that the effective nonlinear coefficient deff for SHG with type I phase matching in a negative uniaxial crystal of crystal class 3m is given by the expression [67]:

deff = d31 sin(θ + ρ) − d22 cos(θ + ρ) sin(3φ) (2.21) where θ is the phase matching angle, ρ the walk-off angle, and φ the azimutal angle. In the literature, different absolute values for the coefficients d31 and d22 can be found. Shoji et al. recommend absolute values of d22 = 2.6 pm/V, and d31 = 0.04 pm/V [68]. However, for better comparison with the effective nonlinear coefficient of the other crystals, all deff from table 2.1 are derived using the software SNLO (Version 68) by Arlee Smith [59]. For BBO, the software uses the values d22 = 2.2 pm/V, and d31 = 0.08 pm/V and calcu- lates the effective nonlinear coefficient for type I SHG from 445 nm to 222.5 nm in BBO ◦ ◦ (θ = 65 , ρ = 4 , φ = π/2) to be deff ≈ 1.26 pm/V.

In summary, BBO is the best suitable nonlinear crystal for direct SHG to wavelengths between 210 nm and 230 nm and will therefore be applied in this work. The next section describes phase matching in BBO. As the refractive indices in BBO posses a relatively weak dependence on temperature [69], phase matching is usually achieved by angle tuning, which is described in the next section as well. Here, it should be noted that one drawback of BBO compared to other crystals is its high walk-oﬀ angle when critical phase matching is applied, which leads to a highly elliptical second harmonic beam. A way to circumvent the disadvantage of the heavy walk-oﬀ in BBO, would be the use of periodically poled nonlinear crystals. Conventional used nonlinear materials like PPLN and PPLT however show strong absorption in the wavelength range below 300 nm and the lowest generated second harmonic wavelength with these crystals is 325 nm generated in PPLT [70]. There is ongoing research towards periodically poled nonlinear crystals suitable for the DUV wavelength range. Recently, Hirohashi et al. demonstrated the generation of laser radiation at 266 nm in a periodically poled LaBGeO5 crystal [71]. In principle, this material can also be used to generate laser light in the wavelength range around 222 nm. However, this has not been achieved up to now. 14 2 Fundamentals for frequency doubling with GaN based laser diodes

2.3 Birefringent phase matching in BBO It was shown in section 2.1 that for eﬃcient second harmonic generation the phase matching condition ∆k = 0 needs to be fulﬁlled, which means that the refractive indices for the fundamental and the second harmonic wave have to be equal:

n(ω) = n(2ω) (2.22)

Unfortunately, this is prevented by the normal dispersion occuring in most materials for which the refractive index increases with increasing frequency. A phenomenon that can be used to compensate for the normal dispersion and to achieve phase matching is the birefringence of optically anisotropic materials. In such materials, the refractive index depends on the polarization and the direction of the light propagating through it.

The nonlinear crystal BBO which is applied throughout this work is an optically anisotropic material exhibiting birefringence, more precisely it is a negative uniaxial crystal. Uniaxial crystals are characterized by a particular direction, which is called the optical axis (c axis). A light wave polarized in the plane containing its own propagation vector k and the optical axis experiences the extraordinary refractive index ne(ω,θ), that depends on the angle θ between k and the optical axis. If the polarization of the light is polarized perpendicular to the plane containing k and the optical axis, the light wave experiences the ordinary refractive index no(ω). In a negative uniaxial crystal like BBO, the extraordinary refractive index is smaller than the ordinary refractive index (ne < no), which is illustrated in ﬁgure 2.2. One can see that there is no pair of refractive indices in BBO that fulﬁlls the phase matching condition for the targeted wavelength range so that no(ω) = ne(2ω). For example, for a fundamental wave at 445 nm, no(445 nm) = 1.68397 and ne(222.5 nm) = 1.65749. The birefringence in BBO can be used to compensate for the normal dispersion, if a nonlinear interaction is chosen that involves an ordinary polarized fundamental beam and an extraordinary polarized second harmonic beam. The combination for which two ordi-

Figure 2.2: Dispersion of the ordinary and extra-ordinary refractive index in BBO for a ◦ crystal temperature of TBBO = 50 C according to the Sellmeier equations from [72]. 2.3 Birefringent phase matching in BBO 15 nary polarized waves generate one extraordinary polarized wave is known as type I-phase matching (neω2 = noω1 + noω1, ω2 = 2ω1) and will be applied in this work. The phase matching can then be realized by either temperature or angle tuning. In order to include a temperature dependence for the following phase matching calculations ◦ the Sellmeier equations determined by Zhang et al. [72] at T0 = 293 K (19.85 C) are used and expanded by the known temperature derivatives of the refractive indices [69]. The temperature dependencies of the ordinary and the extraordinary refractive index of BBO are [69] dno = −16.6 · 10−6 K−1 (2.23) dT and dne = −9.3 · 10−6 K−1. (2.24) dT

The temperature dependent Sellmeier equations for BBO can then be written in the form = + ( ) [73]: nT nT0 T − T0 · dn/dT

0.01878 n2 = 2.7359 + − 0.01471λ2 + 0.0006081λ4 o λ2 − 0.01822 (2.25) − 0.0000674λ6 − 16.6 · 10−6(T − 19.85◦C) 0.01224 n2 = 2.3753 + − 0.01627λ2 + 0,00005716λ4 e λ2 − 0.01667 (2.26) − 0.00006305λ6 − 9.3 · 10−6(T − 19.85◦C)

With these equations one can also calculate the phase matching temperature acceptance of the BBO crystal as will be shown in section 2.3.2. 16 2 Fundamentals for frequency doubling with GaN based laser diodes

2.3.1 Angle tuning The method of angle tuning utilizes the strongly angle-dependent nature of the extraordinary refractive index ne(2ω, θ), which behaves according to the following equation: 1 sin2( ) cos2( ) = θ + θ (2.27) 2( ) 2 2 ne θ ne no By precisely adjusting the angle θ between the optical axis of the crystal and the wave vector k of the propagating light wave, the extraordinary refractive index that is experienced by the second harmonic wave can be matched with the ordinary refractive index no of the fundamental wave. The geometry of this method for SHG is illustrated in ﬁgure 2.3.

c w q 2w k ordinary extraordinary

nonlinear crystal

Figure 2.3: Schematic illustration of phase matching via angle tuning for second harmonic generation, top view (adapted from [46], p. 98).

Figure 2.4 shows the ordinary refractive index for a fundamental wavelength of 445 nm and the extraordinary refractive index for a second harmonic wavelength of 222.5 nm as ◦ ◦ a function of the angle θ in BBO at a crystal temperature of 50 C. For θ = 0 , ne(2ω,θ) ◦ conincides with the principal value no, and for θ = 90 it coincides with the principal value ◦ ne. For a phase matching angle of θPM = 64.97 , the refractive indices for the fundamental and second harmonic waves are equal and the phase matching condition (2.22) is satisﬁed. It can be rewritten as:

no(ω) = ne(2ω,θ) = 1.68397 (2.28)

Figure 2.4: Ordinary refractive index of the fundamental beam and extraordinary refractive index of the second harmonic beam as a function of the angle θ between the optical axis ◦ of the crystal and the wave vector k of the extraordinary beam in BBO (TBBO = 50 C). 2.3 Birefringent phase matching in BBO 17

The drawback of the angle tuning method is, that for a phase matching angle θ between angles of 0◦ and 90◦, the Poyting vector S for the extraordinary wave of the second harmonic is not parallel to the wave vector k, as illustrated in ﬁgure 2.5.

c k q

r S

nonlinear crystal

Figure 2.5: Schematic illustration of the spatial walk-oﬀ occuring with critical phase matching (top view).

Therefore, the energy flow of the second harmonic propagates with a walk-off angle ρ to the wave vector k of the fundamental, which significantly reduces the spatial overlap between both waves leading to a reduced conversion efficiency. The walk-off angle can be calculated according to [45]: 2 no tan(θ) 2 − 1 tan( ) = ne (2.29) ρ 2 no 2 1 + 2 tan (θ) ne

For instance, phase matching in BBO for an SHG process from 445 nm to 222.5 nm at a crystal temperature of 50◦C with the phase matching angle of θ = 64.97◦ leads to a walk-oﬀ angle of ρ = 4◦ for the extraordinary second harmonic wave.

A method to circumvent the drawback of walk-oﬀ is non-critical phase matching also referred to as temperature tuning, for which the angle θ is either exactly or close to 0◦ or 90◦. However, this method is only applicable for limited wavelength ranges and in the case of BBO this method cannot be applied for the DUV wavelength around 223 nm targeted in this work. The term non-critical refers to the fact, that the phase matching tolerances regarding the emission bandwidth of the fundamental, the phase matching angle θ, and the crystal temperature are much less critical compared to critical phase matching (angle tuning), which is applied in this work. 18 2 Fundamentals for frequency doubling with GaN based laser diodes

2.3.2 Phase matching tolerances The dependence of the SHG intensity upon the emission wavelength of the fundamental wave, the temperature of the BBO crystal, and the phase matching angle θ for a fixed fundamental wavelength and crystal temperature can all be simulated using equation (2.17) with the temperature dependent Sellmeier equations (2.25) and (2.26). It should be noted, that all calculations are carried out by assuming infinite plane waves. In real experiments, the propagating waves are not infinite plane waves but focused beams, which eventually results in some deviations from the theoretically calculated values especially for strongly focused beams. This circumstance will be discussed in the results section and further analyzed with respect to the Boyd-Kleinman theory for focused Gaussian beams in section 2.4.

The following tolerances are calculated for the BBO crystal used throughout this work with a length of 7.5 mm, and for a phase-matched fundamental wavelength of 445 nm. Figure 2.6 shows the normalized SHG intensity as a function of the fundamental wavelength for a BBO crystal temperature of 50◦C. For a spectral deviation of ±8 pm to the phase matching wavelength, the SHG efficiency already decreases to 90% of its maximum. The full width at half maximum of the wavelength acceptance in this particular case is ∆λPM = 42 pm (FWHM). The normalilzed SHG intensity as a function of the BBO crystal temperature is presented in figure 2.7. Here, the SHG intensity decreases to 90% of its maximum for a deviation of ±0.8 K from the phase matching temperature of 50◦C. The full-width of half maximum of the temperature acceptance curve is ∆TPM = 4.2 K. As a consequence, precise control of the crystal temperature and a fundamental emission bandwidth in the range of approximately ∆λ ≈ 40 pm (FWHM) is required for efficient SHG. The simulated normalized SHG intensity as a function of the phase matching angle θ ◦ in BBO (TBBO = 50 ) is shown in figure 2.8. It can be seen, that the estimated phase ◦ matching angle acceptance bandwidth is only ∆θPM = 0.013 (FWHM) requiring precise control of the phase matching angle.

Figure 2.6: Normalized SHG intensity as a function of the pump wavelength in BBO at a ◦ crystal temperature of T = 50 (LBBO = 7.5 mm). 2.3 Birefringent phase matching in BBO 19

Figure 2.7: Normalized SHG intensity as a function of the crystal temperature in BBO for ◦ a phase matching temperature of T = 50 (LBBO = 7.5 mm, λ = 445 nm).

Figure 2.8: Normalized SHG intensity as a function of the phase matching angle for a ﬁxed fundamental wavelength of 445 nm (LBBO = 7.5 mm).

As mentioned earlier, the phase matching tolerances are only valid for the approximation of infinite plane waves. In the single-pass SHG experiment presented in chapter5, a focused laser beam is used. It will be seen, that the strong focusing reduces the effective interaction length in the BBO crystal and thereby leads to wider phase matching tolerances. The plane wave approximation however can be seen as a limiting case. The weaker the focusing, the closer are the real phase matching tolerances to the tolerances in the plane wave approximation. It is worth noting, that for waveguide structured cystals, the plane wave approximation is always valid, because the light is confined to a narrow stripe over the whole crystal length and therefore behaves like a plane wave inside the crystal. 20 2 Fundamentals for frequency doubling with GaN based laser diodes

2.4 Second harmonic generation with focused Gaussian beams The intensity of the second harmonic wave in equation (2.17) increases with the square of the fundamental intensity and the square of the crystal length. This result is derived by assuming infinite plane waves taking part in the nonlinear interaction. With real optical beams, high intensities are achieved by realizing spatial confinement of the light through focusing. Stronger focusing results in higher intensities in the beam waist. However, the higher divergence of a strongly focused beam reduces its effective interaction length inside the nonlinear crystal, as the intensities away from the beam waist decrease more rapidly. On the other hand, too weak focusing results in relatively low intensities over the whole crystal length which leads to an unefficient nonlinear interaction as well. Both limiting cases are illustrated in figure 2.9. Another critcial parameter that has to be considered is the walk-off in the phase matching plane in case of critical phase matching which reduces the overlap between fundamental and second harmonic wave. Therefore, a trade-off between strong focusing and a maximized interaction length has to be found to optimize the nonlinear conversion efficiency for a given crystal length Lcr. Boyd and Kleinman [74] have treated this problem analytically for the case of non-depleted SHG with focused Gaussian beams in a negative uniaxial crystal in which critical type I- phase matching is applied. −r2/w2 Under the assumption of a circular Gaussian beam with a radial distribution e 0 that is characterized by its confocal parameter

2 b = 2πnωw0/λω (2.30)

(with the beam waist radius w0) and is focused into a crystal with negligible loss, they derived the generated SHG power according to

16 2 2 = π deﬀ 2 ( ) (2.31) P2 3 P1 · Lcr · h σ,β,κ,ξ,µ . ε0cnωn2ωλω

P1 is the fundamental power, and h(σ,β,κ,ξ,µ) is the Boyd-Kleinman (BK) function, that can be maximized by optimizing the focusing conditions. The parameters in the argument of the function are σ representing the wave mismatch, β representing the walk-oﬀ eﬀect, κ describing losses at the fundamental and the second

Figure 2.9: Geometry of a focused beam inside a nonlinear crystal of length Lcr. The solid line indicates a weakly focused and the dashed line a strongly focused beam. 2.4 Second harmonic generation with focused Gaussian beams 21 harmonic wavelength, ξ representing the focusing conditions, and µ the beam waist position in the crystal. Assuming a lossless material (κ = αb/2 = 0) with the focus at the center of the crystal (µ = 0) the BK function is reduced to three parameters in its argument h(σ, β, ξ) with

= b∆k (2.32) σ 2 , ρ β = (2.33) θ0 and Lcr ξ = . (2.34) b

Here, ∆k is the wave vector mismatch, ρ the walk-off angle, and θ0 the half beam divergence angle. To maximimize the Boyd-Kleinman function and thereby the nonlinear conversion efficiency ∆k needs to be optimized with regard to the focusing condition expressed by ξ. Figure 2.10a) shows the BK function as a function of the beam waist radius w0 on a logarithmic scale in the case without walk-off and with walk-off for a 7.5 mm long BBO crystal at a crystal temperature of 50◦C and for a pump wavelength of 445 nm. In case of no walk-off (ρ = 0), the maximum value of the Boyd-Kleinman function derived from the BK analysis is hmax = 1.07 at a beam waist radius of w0 = 10.5 µm, resulting in a confocal parameter b = Lcr/ξ = Lcr/2.84. In the case of critical phase matching and spatial walk-off additionally depicted on a linear scale in figure 2.10b), the Boyd-Kleinman function, i.e. the conversion efficiency, is significantly reduced. As explained in section 2.3.1, the second harmonic beam experiences a walk-off angle of about 4° in this case. Here, the maximum value of the BK function is hmax = 0.048 for a beam waist radius of w0 = 15 µm, leading to a confocal parameter of b = Lcr/ξ = Lcr/1.4. ξ is a measure for the focusing strength and therefore also takes the

Figure 2.10: a) Boyd-Kleinman function in dependence of the beam waist radius w0 for SHG in a BBO crystal without walk-oﬀ (dashed line) and with walk-oﬀ (solid red line) on a

logarithmic scale. b) Boyd-Kleinman function in dependence of the beam waist radius w0 in ◦ case of walk-oﬀ on a linear scale. (TBBO = 50 C, LBBO = 7.5 mm, λpump = 445 nm). 22 2 Fundamentals for frequency doubling with GaN based laser diodes

Figure 2.11: Simulated SHG power as function of the pump power according to the ◦ Boyd-Kleinman theory for a BBO crystal with TBBO = 50 C, LBBO = 7.5 mm and λpump = 445 nm. strength of the beam divergence generated by the focusing into account. b can be seen as the distance over which the beams cross sectional area is relatively constant. For a strongly focused beam, b is small and consequently ξ is large. The smaller value for ξ and the higher optimum beam waist radius w0 compared to the no walk-off case indicates, that weaker focusing is favorable in case of strong walk-off as observed in BBO. The strong decrease of the BK factor by a factor of 20 demonstrates the negative effect of the spatial walk-off on the nonlinear conversion efficiency. Figure 2.11 shows the expected SHG power PSHG as a function of the pump power Ppump. It follows the quadratic behavior for SHG without pump depletion according to equa- ◦ tion (2.31). For TBBO = 50 C, LBBO = 7.5 mm and λpump = 445 nm and with the focusing conditions optimized according to the BK analysis with w0 = 15 µm, the expected normalized conversion efficiency is η = 13.8 · 10−5 W−1, assuming an effective nonlinear coefficient of deff = 1.26 pm/V (see section 2.2). This means, that an SHG power of 138 µW at 222.5 nm can be expected for a pump power of 1 W at 445 nm from an optimally focused Gaussian beam.

It should be noted, that the optimum BK factor in the Boyd-Kleinman theory is obtained at ∆k > 0, resulting from a phase shift in the focus of a Gaussian beam [75, 76].

In some expansions of the Boyd-Kleinman theory also elliptical focusing is considered. It was shown that weaker focusing in the phase matching (PM) plane reducing the walk-off effect, and strong focusing perpendicular to the PM plane can increase the efficiency in large walk-off crystals like BBO [77–79]. Furthermore, it was demonstrated in periodically poled crystals that for non-diffraction limited pump beams an increased spot size with respect to the BK theory is favorable [80]. 3 Characterization of the applied laser diode

In this chapter, the working principle of laser diodes is introduced briefly. Laser parameters are explained based on the electro-optical, spectral and spatial emission characteristics of the laser diode used in this work and only to an extent that is relevant for the following experiments. The laser diode applied throughout this work is a commercially available InGaN based laser diode from OSRAM Optosemiconductors GmbH (model PL TB450B). The laser diode is intentionally developed for high-quality projectors in the professional field or for stage and decoration illumination and even medical applications [81]. It is packaged in a standardized TO56 package, that can be seen in figure 3.1.

Figure 3.1: Photography of the applied laser diode packaged in a TO56 can [43].

One of the challenges of applying such a commercially available laser diode is the lack of information about the exact vertical quantum well structure and the strength of the front and back facet reflectivities. Only the experimentally accessible parameters and the information divulged by OSRAM Optosemiconductors GmbH in some publications and the datasheet (see appendixA) for the laser diode can be used for the design and simulation of the experiments. However, the electro-optical, spectral and spatial emission characteristics can be measured and are presented in this section. Wherever possible, measured parameters are compared with published results and the datasheet. Another challenge is the limited accessability of the laser diode due to its packaging in the TO56 package. This package has a window made of D 263® T eco glass (Schott AG, refractive index at 445 nm: n = 1.5342) with a thickness of 250 µm, through which the laser diode emission is coupled out. The window has a distance of about 0.8 mm to the front facet of the laser diode, which prevents the application of micro-lenses with shorter focal lengths. A first visual inspection of the laser diode is carried out with a simple microscopic measurement through the protective glass of the TO56 package. Figure 3.2 shows three microscope images with different magnification factors. In figure 3.2.a), the whole laser diode package is shown with fivefold magnification. On the left side of the image, an ESD protection diode protecting the laser diode from overvoltage can be seen. In the middle is the laser diode chip itself. Figure 3.2.b) is an image of the laser diode chip with tenfold

23 24 3 Characterization of the applied laser diode

a) b) c)

15m m 50x

200m m 5x100m m 10x

Figure 3.2: Optical microscope images of the applied laser diode with different magnification factors (5x (a), 10x (b), 50x (c)). magnification. The chip width is measured to be 200 µm and its height to be 100 µm. The image is taken with a minimal applied voltage so that blue spontaneous emission from the laser diode is identifiable. The noticeable brighter spot originates from the laser diodes active region, that can be seen magnified fifty times in figure 3.2.c). The lateral width of the active region is measured to be 15 µm. Considering the time of the publication and comparing the optical characteristics, one can assume that the laser diode used throughout this work is essentially comparable with the R&D laser device presented in a paper by Vierheilig et al. in 2012 [82]. In this paper, a ridge waveguide laser diode with a ridgewidth of 15 µm and a resonator length of 1.2 mm is shown. The laser diode is reported to have a peak conversion efficiency of 29% and a thermal roll-over at 2.5 W at room temperature. Its slope efficiency is specified to be about 1.7 W/A. These values are also in accordance with the characterization measurements presented in this chapter.

3.1 Working principle of laser diodes and vertical layer structure In principle, laser operation is achieved by amplification of stimulated emission inside of an active medium that provides optical gain. This active medium is placed into a resonator, that creates optical feedback into the active medium. Figure 3.3 shows a simple schematic illustration of a laser diode, in which the optical feedback is provided by the cleaved rear and front facets, that serve as plane mirrors with reflectivities R1 and R2, and hereby form a Fabry-Perot (FP) resonator of length L. In a simple semiconductor quantum well-based diode laser, the active medium is based on a heterostructure consisting of an undoped semiconductor layer with a direct band gap embedded between p- and n-doped semiconductor layers with a larger band gap energy than the undoped layer. By applying a forward bias, holes from the p-side and electrons from n-side are injected and recombine in the undoped active region, which leads to the emission of photons with energies slightly larger than the energy band gap. In figure 3.3, the emission direction is the z direction, in which a high photon density in the active region required for constant lasing is realized by the optical feedback. For efficient operation, photons and charge carriers need to be confined in the active region in both transverse directions (x and y direction) as well. In the laser diode applied in this work, the lateral confinement is realized by the already mentioned ridge waveguide. For the vertical direction, charge carrier and photon confine- 3.1 Working principle of laser diodes and vertical layer structure 25

p-doped active region z n-doped

L R1 R2

Figure 3.3: Schematic illustration of a Fabry-Perot laser diode resonator of length L with mirror reflectivities R1 and R2 of the rear and front facet, respectively. ment is realized by the design of the vertical layer structure. As presented in the paper by Vierheilig, the diode is grown on a c-plane GaN substrate by metal organic vapor phase epitaxy (MOVPE). The active region consists of InGaN multiple quantum well (MQW) structures instead of a simple intrinsic undoped semiconductor layer. It is embedded between n-type and p-type AlGaN cladding and GaN waveguide layers [82]. From another paper by Hempel et al., that investigates the kinetics of catastrophic optical damage in these GaN based diode lasers, one can find that the chip is mounted by hard solder in a p-side up configuration [83]. Figure 3.4.a) shows the vertical layer structure without knowledge of the number of quamtum wells and the individual layer thicknesses. Such a structure is called separate confinement heterostructure (SCH) as the optical confinement of the propagating wave is separated from the confinement of the charge carriers in the vertical (x) direction. The energy band structure for such a laser diode is exemplary illustrated in figure 3.4.b). The MQWs confine the charge carriers in the active region, where they recombine and emit photons with energies sligtly larger than the band gap energy Eg. The AlGaN layers with the higher band gap energies also exhibit a lower refractive index than the GaN waveguide layers, which leads to an optical confinement of the generated wave. The advantage of the separate confinement is that both charge carrier and optical confinement can be optimized separately.

a) b) MQW p-AlGaN EC p-GaN x y GaN/InGaN E MQW g p-AlGaN n-GaN y n-AlGaN x n-AlGaN EV GaN substrate optical waveguide

Figure 3.4: a) Exemplary transverse layer structure of a multiple quantum well separate conﬁnement heterostructure GaN based laser diode. b) Exemplary transverse energy band

structure for a multiple quantum well separate conﬁnement heterostructure laser diode. Eg is the band gap energy of the quantum wells. 26 3 Characterization of the applied laser diode

3.2 Longitudinal modes In an FP resonator, only longitudinal modes fulﬁlling the following relation can oscillate:

λ0 L = m (3.1) 2ng,eff m is the order number of allowed modes (m = 1,2,3,..), λ0 is the vacuum wavelength, and ng,eff is the effective group refractive index of the active region. The spectral distance between the allowed modes is called free spectral range (FSR) and is defined as 2 λ0 ∆λFSR = , (3.2) 2ng,effL where 2ng,effL is the distance travelled by the light in one roundtrip around the cavity. For the laser diode applied in this work, the FSR can be determined by measuring an emission spectrum below the threshold current, where many longitudinal modes can be observed. Figure 3.5 shows such a spectrum at an injection current of I = 140 mA and T = 20◦C heatsink temperature for the laser diode applied in this work. It is measured with a double echelle monochromator (Demon, LTB Lasertechnik Berlin GmbH) with a resolution of 6 pm around 445 nm. A spectral distance of ∆λFSR = 27 pm between the longitudinal modes occuring around 442 nm is observed. By knowing the cavity length of L = 1.2 mm [82], an effective group refractive index of ng,eff = 3.02 can be calculated from equation (3.2) using λ0 = 442 nm. This value is in good agreement with the value of ng = 3.0297 derived for the group index of GaN at λ0 = 442 nm using the dispersion formula given on the webpage www.refractiveindex.info [84], that is taken from the paper by Barker and Ilegems [85]:

1.75λ2 4.1λ2 n2 − 1 = 2.60 + + (3.3) λ2 − 0.2562 λ2 − 17.862

dn and solving ng = n − λ0 . λ0

Figure 3.5: Emission spectrum of the laser diode applied in this work below threshold at a current of I = 140 mA and at T = 20◦C heatsink temperature. 3.3 Optical gain and threshold condition 27

3.3 Optical gain and threshold condition

This part follows the description in the textbook of R. Diehl ([86] p. 10 ﬀ.). When a planar optical wave propagates through an absorbing material in z direction, its intensity I exponentially decreases:

−αz I = I0e , (3.4) where I0 is the initial intensity and α the intensity absorption coefficient. In a laser active material, the wave is amplified by stimulated emission leading to an exponential intensity increase, that can be described with a negative value of α, which is called the optical gain g of the semiconductor material itself. In a laser resonator, only a part of the intensity pattern of the optical mode overlaps with the active region of the optical waveguide. The ratio of this overlap is described by the confinement factor Γ and the gain of the optical mode is then described by the modal gain gmodal = Γ g, that is significantly lower than the material gain g. Lasing can only occur when the gain provided by the optical mode Γ g compensates all the losses in the laser resonator, which leads to the threshold condition [86]:

1 1 Γ gth = αi + αmirror = αi + ln (3.5) 2L R1R2 The modal gain depends on the density of the injected charge carriers. The minimum gain at which laser operation starts and the losses are compensated is called threshold gain gth. αi describes the intrinsic absorption losses in the active material, and αmirror represents the combined mirror losses expressed through the reﬂectivities of the rear and front facet after one roundtrip around the cavity. In ﬁgure 3.6, the allowed longitudinal modes in a FP resonator with the spectral distance ∆λFSR to each other are illustrated together with the modal gain Γ g, that is represented by a Gaussian distribution for the sake of simplicity. When the peak of the modal gain at λp reaches the threshold value, the longitudinal mode in the closest vicinity to λp starts to oscillate. As the modal gain is usually much broader than the free spectral range of the FP resonator, FP diode lasers exhibit several simultaneously lasing modes referred to as longitudinal multi-mode emission. To achieve longitudinal single-mode emission, further techniques have to be applied. One of them is to use wavelength selective optical feedback from external gratings, which will be discussed in chapter4.

Figure 3.6: Spectrum of the modal gain and the longitudinal FP modes of a laser diode at threshold. The modal gain takes its maximum Γ gth at λp. 28 3 Characterization of the applied laser diode

3.4 Electro-optical characteristics By applying the rate equations for steady-state operation together with the threshold condition (3.5), an expression for the optical output power of a FP laser diode can be derived ([86], p. 41 ﬀ.):

αi hν hν Popt = ηi (I − Ith) = ηd (I − Ith) (3.6) αi + αmirror q q

The internal efficiency ηi represents the fraction of the current injected current into the laser diode that generates carriers in the active region ([86], p. 38). Above the threshold current Ith, the optical output power linearly increases with the injection current I. ηd is the differential efficieny and is defined as ([86], p. 44) :

q dPopt ηd = (3.7) hν dI It describes the differential increase in emitted photons divided by the differential increase in injected electrons per time above the lasing threshold. dPopt/dI is the slope efficiency S in W/A from the linear part of the power-current characteristics. By simultaneously measuring the applied voltage U, one can derive the electro-optical conversion efficieny ηc (also referred to as the wall-plug efficiency) from the power-current characteristic of a laser diode as the ratio of the total electrical power Pel injected into the laser diode to the emitted optical output power Popt:

Pel U · I ηc = = (3.8) Popt Popt Figure 3.7 shows the electro-optical characteristics of the laser diode applied in this work for a heatsink temperature of T = 20◦C.

5 2.0 1.0

4 1.6 0.8 c / W / opt P

3 1.2 0.6 / V /

U S = d P / d I

opt

2 0.8 0.4 Voltage C onversionefficiency

1 0.4 0.2 O pticaloutput pow er

T = 20 °C

0 0.0 0.0

0.0 0.2 0.4 0.6 0.8 1.0 1.2

Injection current I / A

Figure 3.7: Voltage U, optical output power P and electro-optical conversion eﬃciency ηc as a function of the injection current I for the laser diode applied in this work at a heatsink temperature of T = 20◦C. 3.5 Spectral emission characteristics 29

The laser diode reaches a maximum optical output of P = 1.6 W at an injection current of I = 1.2 A. The power current characteristic exhibits a slope eﬃcieny of S = 1.55 W/A. The threshold current is Ith = 145 mA. The maximum operating voltage is U = 4.5 V. For the maximum injection current of 1.2 A, an electro-optical conversion eﬃcieny of ηc = 30% is measured. These values are in good accordance with the paper from Vierheilig et al. [82] and the datasheet published by OSRAM Opto Semiconductors GmbH (see table A.1 in the appendix).

3.5 Spectral emission characteristics The laser diode applied in this work exhibits longitudinal multi-mode emission with a spectral emission bandwidth of ∆λ = 1...2 nm (FWHM) over the entire operating range. Figure 3.8 shows an exemplary emission spectrum at an injection current of I = 1.2 A and T = 20◦C heatsink temperature, measured with a double echelle monochromator (Demon, LTB Lasertechnik Berlin GmbH) with a resolution of 6 pm around 445 nm. The emission bandwidth is ∆λ = 1.4 nm (FWHM). The emission wavelength of a diode laser is temperature dependent. Here, one has to distinguish between the shift of the longitudinal modes and the shift of the spectral gain curve with increasing temperature. The further is mainly caused by an increase of the refractive index with increasing temperature, which leads according to equation (3.1) to an increase of the longitudinal wavelength. Additionally, the length of the FP cavity increases with temperature, which also leads to a higher emission wavelength. The stronger eﬀect is the shift of the gain curve, which is mainly caused by the decrease of the band-gap energy with temperature.

Figure 3.9.a) shows the wavelength shift with increasing temperature for the laser diode applied in this work at a constant injection current of I = 1.2 A. A linear increase of the central emission wavelength of ∆λC/∆T = 52 pm/K is measured for a temperature increase from 15◦C to 45◦C heatsink temperature. This value is also in good accordance with

I = 1.2 A 1.0

T = 20°C

0.5 = 1.4 nm

FW HM Normalized intensity intensity Normalized

0.0

440 442 444 446 448 450

Wavelength / nm

Figure 3.8: Emission spectrum for an injection current of I = 1.2 A at a heatsink temperature of T = 20◦C. 30 3 Characterization of the applied laser diode

a) b)

448 446

I = 1.2 A T = 20°C

445 / n / m n / m

447 C C

444

446

443

/ I = 3.5 nm/A / T= 52 pm/K

C C 445

442 C enterw avelength C enterw avelength

444 441

15 20 25 30 35 40 45 200 400 600 800 1000 1200

Heatsink temperature T / °C Injection current I / mA

Figure 3.9: a) Shift of the central emission wavelength λC versus the heatsink temperature at an injection current of I = 1.2 A. b) Shift of the central emission wavelength λC versus the injection current at T = 20◦C heatsink temperature.

the reported value of ∆λC/∆T = 50 pm/K for this laser diode [82]. Figure 3.9.b) shows the shift of the central emission wavelength of the laser diode with increasing injection ◦ current of ∆λC/∆I = 3.5 nm/A for T = 20 C heatsink temperature. This eﬀect is caused by Joule heating with increasing injection current leading to an increased temperature in the active region, even though the heatsink temperature is set to 20◦C. Considering ∆λC/∆T = 52 pm/K, means that the temperature increases with increasing injection current with ∆T/∆I = 67 K/A. 3.6 Spatial emission characteristics 31

3.6 Spatial emission characteristics To couple the emission of a laser diode into an optical system or a lens it is necessary to have knowledge of the spatial emission characteristics of the laser diode. When the laser diode emission is coupled out through the front facet it experiences diffraction to a different extent in the two transverse directions. The smaller vertical dimension of the optical waveguide leads to a quickly diverging beam in the vertical direction, which is therefore referred to as the fast axis. The larger dimension of the optical waveguide in the lateral direction leads to a slowly diverging beam and this direction is referred to as the slow axis. Consequently, the resulting spatial beam profile has an elliptical shape. The beam divergence angles, also referred to as the far-field angles, for the laser diode applied ◦ ◦ throughout this work are measured to be θlat = 8 (FHWM) and θvert = 23 (FWHM) in the lateral and vertical direction, respectively [87]. These values are in accordance with the values from the datasheet provided by OSRAM (see table A.1 in the appendix).

Theoretically, a diﬀraction-limited Gaussian beam is ideal for practical applications due to its minimal divergence. Real laser beams however, often exhibit deviations from the ideal Gaussian proﬁle, which results in a higher divergence and a larger beam waist diameter after focusing by a lens.

When a Gaussian beam propagating in z direction, as illustrated in figure 3.10, is focused by a lens, the beam diameter d(z) increases with increasing distance to the beam waist, where the beam has its smallest diameter d0 = 2w0 at the position z = 0, according to [88]: s z 2 d(z) = d0 1 + (3.9) zR The Rayleigh length = 2 4 is defined as the distance to , at which the beam √ zR πd0/ λ z0 diameter is 2 times larger than its minimum value d0. In the far-field for z zR, the diameter increases linearly with z and the full beam divergence angle θ is defined to [88]:

d0 4λ θ = = (3.10) zR πd0 An important parameter, that describes the spatial beam quality and thereby the focusing capability of a laser beam, is the beam parameter product.

2 w w 0

Figure 3.10: Geometry of a Gaussian beam propagating in z direction. 32 3 Characterization of the applied laser diode

For a diﬀraction-limited Gaussian beam, the beam parameter product is deﬁned to [88]:

d0θ = λ (3.11) 4 π

If higher order transverse modes oscillate, the beam waist diameter d0 and the beam divergence angle θ are both increased by a factor M, referred to as the beam propagation factor, and the beam parameter product is written as [88]:

d0θ λ = M 2 (3.12) 4 π As a result, an M 2 > 1 in each transverse direction indicates to what extent a real beam deviates from an idealized diﬀraction-limited Gaussian beam (M 2 = 1) and can be seen as a measure of the beam quality of a laser beam with reference to a Gaussian beam.

For the spatial characterization of the laser diode applied in this work, the beam propagation factor M 2 in both transverse directions is determined according to the ISO 11146 standard [88]: The collimated beam is focused by a spherical lens and thereby a beam waist is created. The beam proﬁle is then measured with a CCD camera at at least ten positions along the propagation direction. Half of the measurement points should lay in between the Rayleigh length zR and the other half should lay at positions with a distance to the beam waist greater than 2zR.

A CCD image of the intensity profile of the collimated beam, that can be interpreted as the far-field of the laser diode applied in this work, is presented in figure 3.11. The beam is collimated by an aspheric lens with a focal length of f = 4.02 mm (NA = 0.6, Thorlabs C671TME-A) with an AR-coating from 350 nm to 700 nm. It can be seen that in both

2 Vertical position / mm / position Vertical

0 1 2 3 4 5 6 7

Lateral position / mm

Figure 3.11: Intensity profile of the collimated beam from the laser diode applied in this work. 3.6 Spatial emission characteristics 33 transverse directions the intensity profiles are asymmetric. In the lateral axis, a multi-lobed lateral far-field pattern can be observed, that is charactertic for broad ridge (Al,In)GaN laser diodes [89]. And on one side of the beam, low peripheral intensities appear. This lateral far-field pattern probably occurs from an interplay of the photon density with the lateral refractive index profile, the local carrier density and thermal effects, that lead to self focusing of the optical laser mode, which is known as filamentation [89].

For such a laser beam clearly exhibiting a beam propagation factor M 2 > 1, it is anything but trivial to choose an appropriate criterion for the determination of the beam diameter from a measured intensity profile, as different criteria might result in significantly different beam diameters. Furthermore, the appropriate criterion may also depend upon the intended application for the laser beam under investigation. The definition for the beam diameter that is maybe the most accurate with respect to the real beam is the variance definition, that weights low peripheral intensities quadratically. It uses the second moment of the beam intensity profile I(x,y) across the transverse coordinate x (or alternatively across the y coordinate) according to [90]:

R ∞ 2 (x − x0) I(x,y)dxdy 2 = −∞ (3.13) σx R ∞ ( ) , −∞ I x,y dxdy where x0 is the focal point of the beam. For this method, the peripheral intensities have to be included accurately as they are quadratically weighted. The square root of the variance is known as the standard deviation σ and the beam waist radius is deﬁned to be twice the standard deviation. The beam waist diameter in both transverse directions is then four times the standard deviation:

dx,y = 4σx,y (3.14) The hypberbolic ﬁt from equation (3.9) can then be rewritten for the variance deﬁnition of the beam diameter as:

v 2 u 4 ( )! ( ) = u1 + 2 λ z − z0 (3.15) d4σ z d0,4σt M4σ 2 . πd0,4σ

Another possibility to define the beam diameter in both transverse directions is to use the knife-edge method. In this method, the total intensity of the beam is used as the reference. An imaginary knife-edge is then translated across the beam profile and the transmitted power is measured as a function of the knife-edge position. The criterion, that is used in this work, is to define the diameter as the spatial distance between the points at which the transmitted power is 95% and 5% of the total power in each transverse direction, respectively. This means, that on both sides of the intensity profile, 5% of the peripheral intensities are neglected and the beam diameter in each transverse direction contains 90% of the total intensity. The hypberbolic fit from equation (3.15) can then be expressed for the 90%-knife-edge definition of the beam diameter as:

v u !2 u 2 4λ(z − z0) d90%(z) = d0,90%t1 + M90% 2 . (3.16) πd90% 34 3 Characterization of the applied laser diode

The knife-edge method results in smaller beam diameters and beam divergences and therefore in smaller M 2 values than the variance method. However, the 90%-knife-edge method seems to be an appropriate criterion considering second harmonic generation as the intended application of the beam. Since the generated SHG intensity depends on the square of the fundamental intensity, very low intensities have a small contribution to the conversion process and can be neglected. In this sense, the peripheral intensities of the laser beam shown in ﬁgure 3.11 would be over-interpreted by the variance method.

Figure 3.12 shows the caustic measurements according to ISO 11146 for the variance and the knife-edge method in the fast (left) and slow (right) axis of the laser diode emission for the maximum injection current of 1.2 A. In the fast axis, a second moment beam propagation factor of 2 = 2 6 and a 90%-knife M4σ . 2 edge value of M90% = 1.1 is measured. This means, that the beam quality in vertical direction is relatively good and as can be seen from figure 3.11 the beam shape in vertical direction is close to a Gaussian shape. In the slow axis, the second moment beam propagation factor is determined to be 2 = 5 9 and the 90%-knife edge value is 2 = 4 5, M4σ . M90% . which correponds to the lateral multi-lobed far-field pattern. As explained above, the 90%-knife edge value seems to be more appropriate considering this particular beam profile and second harmonic generation as application. Therefore, these values will be considered when the focusing conditions for efficient second harmonic generation are discussed in chapter5.

2.0 2.0

Variance (4 ) Variance (4 )

Knife edge (90%) Knife edge (90%) / mm / d caustic fit caustic fit 1.5 1.5

M = 5.9

2 4

M = 2.6

4 1.0 1.0

M = 4.5

90% 2

M = 1.1 0.5 0.5

90%

fast axis I = 1.2 A slow axis I = 1.2 A Beam waist diameter diameter waist Beam

0.0 0.0

0 100 200 300 400 500 0 100 200 300 400 500

Position z / mm

Figure 3.12: Caustic measurements of the laser diode in fast (left) and slow (right) axis according to the variance and the 90%-knife-edge diameters at T = 20◦C heatsink temperature and I = 1.2 A. 3.7 Implications for the development of a deep ultraviolet laser light source 35

3.7 Implications for the development of a deep ultraviolet laser light source After having discussed several applications, that can be addressed with deep ultraviolet laser light, it has become clear, that single-pass frequency doubling of blue GaN based laser diode radiation is a novel and promising concept for a compact and portable DUV laser light source. A desired emission wavelength below 250 nm is most suitable, and although higher optical output power is always desired, it has been shown, that an output power around 100 µW or even lower is already sufficient. In chapter2, an introduction to nonlinear optics related to second harmonic generation was given. It was shown, that BBO is the most suitable nonlinear crystal material with the highest expected conversion efficiency. The simulation of the birefringent phase matching in BBO and the simulation of the expected conversion efficiency using the Boyd-Kleinman theory revealed, that a pump source with high optical output power in the watt-range together with narrowband emission in the range of the acceptance bandwidth of BBO of about 40 pm is needed to achieve a DUV output power around 100 µW with the single-pass concept. Therefore, a commercial GaN based laser diode from OSRAM Opto Semiconductors providing the - at the time of this work - highest available optical output power of 1.6 W around 450 nm was chosen to serve as pump source and its electro-optical, spectral and spatial characteristics were presented in this chapter. It was shown, that the laser diode has a spectral width ∆λ ≈ 1 nm, which is too broad for efficient second harmonic generation as well as for the intended applications. Therefore, the spectral emission needs to be reduced by means of wavelength selective optical feedback, which will be presented in the following chapter4.

4 External cavity diode lasers as pump sources for DUV generation

Due to their broad gain spectrum diode lasers usually exhibit an emission bandwidth of a few nanometers making them unsuitable for nonlinear frequency conversion. Two main approaches to reduce the bandwidth of laser diodes are most commonly applied. Either wavelength selective sections like distributed feedback (DFB) gratings or distributed Bragg reﬂectors (DBR) are integrated on-chip or external cavities are realized by external wavelength selective elements in an external cavity diode laser (ECDL) setup. The implementation of DFB [91] and DBR [92] structures has been realized for GaAs based laser diodes [93–95] and DFB and DBR diode lasers are already commercially available for this material system [42]. In contrast, the manufacturing of such structures in GaN based laser diodes faces some technological challenges and remains an objective of intensive research until now [96–99]. Therefore, the realization of an ECDL system is currently a more feasible approach to achieve narrowband emission from a GaN based laser diode.

4.1 Wavelength stabilization by external optical feedback The basic principle of an ECDL system consists in feeding a narrow bandwidth portion of the light back into the laser diode cavity via an external optical element. If a dispersive, i.e. wavelength selective element is used, a reduced loss over a narrow spectral range is induced forcing the laser diode to emit in this desired wavelength range with a significantly reduced bandwidth. In general, optical feedback can cause variations in the lasing threshold, the output power, and the emission bandwidth and wavelength. These manifold effects have been studied extensively [100–109]. Hereby, the variations of the spectral characteristics under different strenghts of optical feedback also from dispersive external cavity elements have been investigated with special emphasis. The interaction between the internal field from the laser diode and the external field is most importantly dependent on the length of the external cavity, i.e. the distance between the reflective element and the coupling facet, the strength of the optical feedback and the reflectivity of the coupling facet. Depending on these parameters, the description of ECDL properties can become rather complex, which is beyond the scope of this work. Instead, the reader may be referred to the works already referenced above and to some more detailed reviews on that topic [110, 111].

Instead, a three-mirror-cavity model [112] as depicted in figure 4.1 helps to illustrate the situation in an ECDL setup with a wavelength selective element and to explain the most important parameters that have an influence on the spectral response of the ECDL in a rather simple way. The internal laser diode cavity is formed by the rear and the front facet of the laser diode interpreted as mirrors with intensity reflectivities R1 and R2, respectively. The laser diode emission is coupled out via the front facet and collimated by a lens. In the case described here, a wavelength selective and reflective filter with reflectivity R3 is

37 38 4 External cavity diode lasers as pump sources for DUV generation

R1 R2 R3

laser output laser diode cavity lens

d LECDL Figure 4.1: Schematic illustration of an external cavity diode laser as a three-mirror cavity laser. R1 and R2 are the reﬂectivities of the back and front facet and R3 is the reﬂectivity of the external dispersive element.

placed in front of the laser diode in a distance LECDL to the front facet. Depending on the dispersion strength and the reflectivity of the dispersive element, a defined portion within a defined spectral bandwidth of the emitted light is fed back into the laser diode via its front facet. With stray or absorption losses at the dispersive element being neglected, the main portion of the light T = 1 − R3 is coupled out via the dispersive element serving as the laser output of the ECDL. In principle, three different Fabry-Perot resonators - between the front and rear facet, the rear facet and the dispersive element and between the front facet and the dispersive element - are formed and influence the emission spectrum of an ECDL. Saliba et al. give an intuitive and didactic description of the main wavelength dependent gain and loss factors, that influence the emission wavelength of an ECDL with a surface grating as dispersive element [113]. These factors are the transmission functions of the three resonators, the semiconductor gain profile g and the dispersion of the surface grating D. The product of these factors gives the spectral response of the ECDL:

TECDL = T12 · T13 · T23 · g · D (4.1)

T12, T13 and T23 are the transmission functions of the resonators between R1 and R2, R1 and R3, and R2 and R3, respectively, that are described by the Airy-function [114]: 1 T = . (4.2) 1 + F sin2(∆φ/2)

√ √ 2 F = 4 R1R2/(1 − R1R2) is the finesse coefficient with, in case of the laser diode cavity, the intensity reflectivities of the rear and front facets R1 and R2, respectively. ∆φ = 4πng,effLc/λ is the phase difference after one round trip with ng,eff being the effective group refractive index and Lc the respective cavity length. The dispersion of the surface grating can be described as a diffracted intensity D, approximated by assuming a square-slit profile with the slit width equal to half of the grating period dG of the surface grating [114]:

2 sin(kNdG sin αL) kdG sin αL D = sinc2 , (4.3) N sin(kdG sin αL) 2 where k = 2π/λ is the wave vector, N = 2r/dG the number of illuminated grooves for a given beam radius r and grating constant dG (distance between grating grooves), and αL is the Littrow angle, that is introduced in section 4.1.1. 4.1 Wavelength stabilization by external optical feedback 39

Figure 4.2.a) illustrates a simulation of the relative dispersive factors on the basis of a practical example of an ECDL setup, that is very similar to the ECDL systems realized in this work. The black curve shows the transmission function of the laser diode cavity, whereas the green curve shows the combined transmission functions of the cavities formed between mirror R1 and R3 and between R2 and R3. The assumed values for the intensity reflectivities in this example are R1 = 0.96, R2 = 0.05, and R3 = 0.15. Even though the exact laser diode facet reflectivities are unknown, the assumed values for R1 and R2 represent typical values for a high-power InGaN based laser diode (see [115], p.2). The value for R3 corresponds to the reflectivity of the surface grating and the volume Bragg grating used in this work. The cavity length of the laser diode is L = 1.2 mm. The group refractive index was calculated to be ng,eff = 3.02 at 442 nm. This gives a free spectral range of ∆λFSR = 27 pm. As can be seen, these values correspond to the characterization of the applied laser diode in the previous chapter. The distance between the front facet and the dispersive element is set to be L = 2 cm, which is approximately the distance between the volume Bragg grating and the front facet of the laser diode in the micro-integrated ECDL setup realized in this work (see section 4.3). The resulting free spectral range of the combined external cavities is about ∆λFSR = 4 pm. The red line illustrates the grating feedback dispersion for a beam radius r = 1.4 mm, assuming a grating with a groove density of 3600 lines/mm resulting in a grating constant of dG = 278 nm which gives N = 10080. The center wavelength of the grating feedback ◦ is at λ = 445 nm for a Littrow angle of αL = 53.23 . These values correspond to the experimental setup of the macroscopic ECDL in Littrow configuration, that will be presented in section 4.2. In this case, the optical feedback from the grating has a spectral bandwidth of ∆λ = 38 pm (FWHM). The gain profile g (blue line) is very broad compared to the other dispersive factors and therefore appears to be constant in figure 4.2.a), even though it is approximated with a Gaussian distribution. The resulting spectral response of the ECDL system as the product of all dispersive factors is depicted in figure 4.2.b) (black line). The grating dispersion

Figure 4.2: a) Calculated transmission functions in a three-mirror ECDL: The green line is the combined transmission function of the the cavities formed between mirror R1 and R3 and between R2 and R3. The black line is the transmission function of the laser diode cavity. The red line indicates the dispersion D of the surface grating and the blue line the semiconductor gain. b) Spectral response of the ECDL system as the product of all dispersive factors (black line) and grating dispersion D (grey line) for comparison. 40 4 External cavity diode lasers as pump sources for DUV generation indicated as grey line is the dominant mechanism here and one can see several external cavity modes within the spectral window of the grating feedback. Even though these modes have a very similar combined gain, it is possible to achieve single-mode operation under such conditions [116–118]. However, ﬁgure 4.2.b) shows, that without further active control of the setup, several external cavity modes will oscillate at the same time and the ECDL emission bandwidth is expected to be similar to or slightly smaller than the spectral bandwidth of the external feedback.

4.1.1 ECDLs with surface diffraction gratings ECDLs can be realized with different optical elements as wavelength selective filter. In the most common approach, diffraction gratings are used to direct a narrow bandwidth portion of the incident light back into the laser diode cavity. These gratings can be surface diffraction gratings or volume Bragg gratings. The former are typically aligned in either the Littrow [116, 117, 119] or the Littman-Metcalf [120, 121] configuration as illustrated in figure 4.3. In the Littrow configuration (see figure 4.3.a)), the first diffraction order from the grating is directly sent back into the laser diode through the front facet of the laser diode chip and the light reflected from the grating, e.g. the zeroth order diffraction, is used as output beam. This is realized by aligning the grating in the so-called Littrow angle for which the incident beam and the diffracted first order beam are anti-parallel to each other. The Littrow angle can be calculated from the grating equation mλ = dG(sin α + sin β), where m is the diffraction order, λ the wavelength of the incident light, and α and β are the angles between the incident beam and the grating surface normal and between the diffracted beam and the grating surface normal, respectively. As α equals β for the first diffraction order, the Littrow angle αL is calculated by: λ αL = arcsin . (4.4) 2dG Wavelength tuning is realized by tilting the grating and thereby changing the angle of the incident light with respect to the grating normal, e.g. the Littrow angle. The drawback of the Littrow configuration is the change of the output beam direction during wavelength tuning. This lateral beam displacement can be compensated by using an additional mirror parallel to the grating [122] or a triangular prism [123]. In the Littman-Metcalf, also called grazing-incidence, configuration (see figure 4.3.b)), the first diffraction order from the grating is directed onto a mirror, which reflects the light

b) a) mirror 0th order 0th order

st a 1 order L st 1 order

laser lens grating laser lens grating diode diode Figure 4.3: Illustration of an external cavity diode laser in a) Littrow and b) Littman- Metcalf configuration. 4.1 Wavelength stabilization by external optical feedback 41 back onto the grating from where it is diffracted back to the front facet of the laser diode chip. The zeroth order reflection serves as output beam. Wavelength tuning in the Littman-Metcalf configuration is achieved by a rotation of the mirror, that reflects the first order diffraction back onto the grating, that is kept in a fixed position. The advantage of the Littman-Metcalf configuration is a fixed output beam during wavelength tuning, which is more practical in experimental setups. Additionally, the bandwidth of the optical feedback is smaller than in the Littrow configuration, as the light is diffracted twice on the surface grating before it is directed back into the laser diode cavity. The drawback of this configuration is a lower output power than in the Littrow configuration, because the reflection from the first diffraction order beam on the grating cannot be used (dashed arrow in figure 4.3.b)). In practice, the emission bandwidth of an ECDL system is influenced by electronic and acoustic noise, and mechanical vibrations and therefore the simpler and more robust Littrow configuration often shows smaller emission bandwidths.

To achieve a high suppression of the laser diode cavity modes, the outcoupling facet of the laser diode is often anti-reﬂection (AR) coated resulting in a very low reﬂectivity R2 in the order of 10−4. When a high AR-coating is applied, the laser diode only serves as the active medium and does not show laser operation itself. In this case, laser operation is only possible with the additional external optical feedback and the external cavity modes originating from the cavity formed by the rear facet (R1) and the external dispersive element (R3) dominate the mode spectrum relatively independent of the operating conditions. With such ECDL systems, a high stability of the lasing frequency and extremely narrow linewidths down to the Hz regime have been achieved at low output powers using narrow stripe GaAs based laser diodes [124].

In the following, publications on GaN based ECDL setups with surface diffraction gratings are briefly presented. Since the Littrow configuration promises to result in higher optical output powers and also in a more compact and robust laser system, only setups in this configuration will be presented. 42 4 External cavity diode lasers as pump sources for DUV generation

Low-power GaN based ECDLs in Littrow configuration Low-power lateral single-mode GaN based laser diodes emitting in the violet and blue spectral range have already been integrated into external cavity setups with surface gratings in Littrow configuration [125–130]. The electro-optical ECDL parameters from a few selected publications are summarized in table 4.1. In some works, the laser diodes were treated with an AR coating: Lonsdale et al. demonstrated a coarse tuning range of 6.3 nm around a wavelength of 398 nm with an emission bandwidth of 11 MHz (6 fm) using a GaN based laser diode with a maximum continuous wave (CW) output power of 5 mW [126]. They showed, that an AR coating significantly increases the tuning range compared to the same device without AR coating. Hildebrandt et al. demonstrated an ECDL output power of 20 mW (CW) at 410 nm with an emission bandwidth of 0.8 MHz (0.45 fm) using an AR coated GaN based laser diode [127]. They also showed the tuning enhancement effect of an AR coating and evaluated the effect of optical feedback on the lasing threshold and the photon density of the FP laser diode. Using a laser diode emitting around 405 nm, that had a front facet reflecitivity below 1%, Tanaka et al. presented a Littrow-ECDL with a single-mode output power of 80 mW (CW) and a linewidth of ∆ν = 21 MHz (12 fm) [130].

Ref. AR λ PLD PECDL ∆ν ∆λtun,contin ∆λtun,coarse

[126] (2002) y 398 nm 5 mW N/A 11 MHz 0.4 pm 6.3 nm [127] (2003) y 410 nm 22 mW 20 mW 0.8 MHz 28 pm 4 nm [130] (2007) y 405 nm N/A 80 mW 21 MHz N/A N/A

[125] (2000) n 392 nm 5 mW 3.5 mW 5 MHz 3 pm 2.7 nm [128] (2004) n 450 nm 5 mW 1 mW 8 MHz 71 pm N/A [129] (2005) n 410 nm 30 mW 1 mW N/A 50 pm N/A

Table 4.1: ECDL parameters from a selection of publications on low-power GaN based ECDLs in Littrow configuration. PLD: nominal maximum laser diode output power, PECDL: maximum ECDL output power, ∆ν: ECDL emission bandwidth, ∆λtun,contin: mode- hop-free or continuous tuning range, ∆λtun,coarse: manual or coarse tuning range. An easier and more inexpensive approach is to use a commercially available diode laser, that may be housed in a standardized TO package and that may not have an AR coated facet. Conroy et al. demonstrated an output power of 3.5 mW (CW) at 392 nm with an emission bandwidth of ∆ν = 5 MHz and a coarse tuning range of 2.7 nm using a laser diode without AR-coating [125]. Burns et al. and Hult et al. used GaN based laser diodes without AR coatings emitting at 410 nm and 450 nm in ECDL setups delivering a single-mode CW output power around 1 mW with emission bandwidths of 8 MHz (5 fm)[128, 129]. By modulating the diode injection current, and thus modulating the effective laser diode cavity length, in synchro- nization with the external cavity tuning, they matched the diode FP and external cavity modes and thereby achieved a mode-hop-free tuning range of 50 pm (90 GHz) and 71 pm (105 GHz) at 410 nm and 450 nm, respectively. In order to obtain higher output powers, a GaN diode laser-based master-oscillator power- amplifier was presented by Shimada et al. and 110 mW (CW) of single-mode output power at 461 nm was achieved [131]. 4.1 Wavelength stabilization by external optical feedback 43

High-Power ECDLs Wavelength stabilization of the emission of broad-area [132] and tapered diode lasers [133, 134] and even diode-array bars [135, 136], that are based on GaAs, was realized with surface diﬀraction gratings in Littrow conﬁguration. In these works, output powers in excess of 1 W with emission bandwidths in the GHz [132, 134, 135] and even MHz [133, 136] range were achieved. At the time of this work, no publications on high-power GaN based ECDLs with surface gratings could be found in the literature.

The works cited above used lateral single-mode laser diodes in optical feedback setups with a focus on a very narrow linewidth in the kHz and MHz regime and high tunability of the emission wavelength rather than achieving high optical output powers. As stated earlier, the object of this work is to realize a pump source for frequency conversion with an output power in the watt range and narrowband emission in the order of the wavelength acceptance bandwidth of the BBO crystal of ∆λ = 42 pm (∆ν = 64 GHz at 445 nm) (FWHM). Only recently, GaN based laser diodes have reached output powers beyond 1 W and wavelength stabilization of high-power GaN based laser diodes in an external cavity setup was not realized until now. Therefore, the first objective of this work was to test the feasibility of using a commercially available high-power GaN based laser diode as the active medium in a non-sophisticated external cavity diode laser setup, that uses a surface diffraction grating in Littrow configuration as wavelength selective element. The laser diode used throughout this work and that was characterized in the previous chapter (PL TB450B, OSRAM Opto Semiconductors) was chosen, because it provided the highest optical output power of P = 1.6 W from a commercially available GaN based laser diode at the time of this work. In the next section, the macroscopic ECDL in Littrow configuration will be presented and, for the first time, wavelength stabilization of a high-power GaN based laser diode will be demonstrated. 44 4 External cavity diode lasers as pump sources for DUV generation

4.2 Macroscopic external cavity diode laser in Littrow configuration To understand the performance of the external cavity system, it is necessary to evaluate a few basic properties of the free-running laser diode, such as the peak position of its gain curve below threshold, the center wavelength of the emission above threshold, and its wavelength shift with increasing injection current. These values are important, because the ECDL performance strongly depends on the spectral position of the optical feedback in comparison to the center wavelength of the laser diode emission, e.g. the peak position of the gain curve. For the macroscopic ECDL, another laser diode of the same batch with the same specifications as the one presented in section3 is used. The electro-optical parameters might therefore slightly differ. Figure 4.4 shows a measurement of the amplified spontaneous emission (ASE) following the gain curve of the laser diode far below threshold at an injection current of I = 0.1 A. The spectrum is measured with a high dynamic range optical spectrum analyzer (OSA) (Yokogawa, AQ6373) with a dynamic range of 60 dB and a spectral resolution of 50 pm. The ASE takes its maximum at a wavelength of λC = 444.96 nm and shows an emission bandwidth of around 8 nm (FWHM, - 3 dB). Three emission spectra of the free-running laser diode at an injection current of I = 0.2 A, I = 0.7 A, and I = 1.2 A, again measured with the optical spectrum analyzer, are depicted in figure 4.5. The center wavelength of the emission shifts with ∆λ/∆I = 3.9 nm/A from λC = 444.01 nm at I = 0.2 A to λC = 447.87 nm at I = 1.2 A. This shift is mainly caused by Joule heating of the laser diode chip for higher injection currents and is sligtly higher here than for the laser diode presented in section3 due to differences in the packaging or mounting that influence the thermal management of the chip.

I = 0.1 A

T = 20°C 0

= 444.96 nm

C / dB / I

-5 Intensity Intensity

-10

430 435 440 445 450 455 460

Wavelength / nm

Figure 4.4: ASE emission spectrum of Figure 4.5: Emission spectra of the laser the laser diode at an injection current of diode at an injection current of I = 0.2 A, I = 0.1 A. 0.7 A, and 1.2 A. 4.2 Macroscopic external cavity diode laser in Littrow conﬁguration 45

4.2.1 Experimental Setup The setup of the external cavity diode laser system, that is built up as a proof-of-concept of using a commercially available high-power GaN laser diode in an ECDL system, is shown in figure 4.6. The FP laser diode (1), that was characterized in section3, is used as gain medium. Based on the measured spatial emission characteristics (section 3.6) of the laser diode and optical beam shaping simulations according to ISO 11146 using the program WinABCD1, an aspheric lens with a focal length of f = 4.02 mm (NA = 0.6, Thorlabs C671TME-A) with an AR-coating from 350 nm to 700 nm is used for collimation (2). The collimated beam has an elliptic beam shape with beam diameters of dlat = 1.1 mm and dvert = 2.8 mm in lateral and vertical direction, respectively. In the course of this work, holographic surface diffraction gratings (3) with different groove densities of 1800, 2400, and 3600 grooves/mm were tested in the same setup. The gratings on a 12.5 mm x 12.5 mm x 6 mm substrate (Richardson Gratings) are mounted on a four-axis diffraction grating mount (Newport, DGM-1) with the Pivot point located on the surface of the grating. The distance between the laser diode and the surface grating is approximately 15 cm. For the characterization measurements of the ECDL system, the output beam is redirected by a mirror (4).

1 2 a L 3

L Pivot point FP L inner

L outer

Figure 4.6: Schematic view of the ECDL system in Littrow conﬁguration: (1) FP laser diode, (2) collimating lens, (3) surface grating, (4) mirror.

The diffraction efficiency of the gratings in the first diffraction order depends on the orientation of the laser diode polarization with respect to the grooves of the grating. Additionally, the spectral bandwidth of the grating dispersion depends on the number of illuminated grating grooves. Figure 4.7 illustrates two possible configurations of the ECDL setup in Littrow configuration. In configuration a), the laser diodes fast axis is aligned parallel to the grating grooves, and, as the laser diode is TE-polarized (TE:TM = 100:1), the polarization direction E is perpendicular to the grating grooves. Here, the lateral plane (slow axis) of the laser diode undergoes dispersion. The number of illuminated grating grooves for a grating with 3600 grooves/mm (groove density dG = 278 nm) is N = 2 · rlat./dG = 2 · 0.55 mm/278 nm = 3960 in this configuration. In configuration b), the laser diode is aligned with its fast axis perpendicular to the grating grooves and therefore its polarization is in the p-plane of the grating.

1 WinABCD is a ray transfer matrix analysis software developed at the Ferdinand-Braun-Institut by Dr. Bernd Eppich 46 4 External cavity diode lasers as pump sources for DUV generation

a) fast b) axis E

LD E LD fast axis grating grating

Figure 4.7: Illustration of two ECDL conﬁgurations: a) The laser diode (LD) polarization is perpendicular to the grating grooves, the grating dispersion occurs in the lateral LD plane. b) The LD polarization is parallel to the grating grooves, the grating dispersion occurs in the vertical LD plane.

Here, the vertical plane of the laser diode undergoes the grating dispersion. Due to the larger beam diameter in vertical direction of 2.8 mm, more grating grooves (N = 10080) are illuminated in this configuration, which leads to a higher grating resolution than in configuration a), i.e. a smaller spectral bandwidth of the grating feedback. On the other hand, the vertical aperture of a laser diode is much smaller than the lateral aperture. Therefore, configuration b) requires a higher precision of the optical alignment to efficiently couple the optical feedback back into the laser diode cavity and the sensitivity of the setup to mechanical vibrations is higher.

The grating constants dG and the Littrow angles αL for a wavelength of 445 nm are listed in table 4.2 for different groove densitites of 1800, 2400, and 3600 grooves/mm. Depending on the groove density and the respective beam diameter, the number of illuminated grooves varies for configuration a) and b) and is listed in table 4.2 as well. Inserting these values into equation (4.3) gives the expected spectral bandwidth ∆λ (FWHM) of the grating feedback for each grating and configuration as listed in the table. The diffraction efficiency DE given by the supplier (Richardson Gratings) at a wavelength of λ = 445 nm [137] is also listed in table 4.2 for each grating and configuration. Only for a grating with 3600 grooves/mm aligned in configuration b), a sufficient spectral resolution of ∆λk = 38 pm in the range of the wavelength acceptance bandwidth of the BBO crystal can be achieved. Additionally, the diffraction efficiency in this case is approximately

DEk = (15 ± 3) %, given by the supplier [137]. This amount of optical feedback should be suﬃcient to stabilize the FP laser diode whereas it also assures that as much as 85 % of the

conﬁguration a) conﬁguration b)

grooves/mm dG αL N⊥ ∆λ⊥ DE⊥ Nk ∆λk DEk

1800 556 nm 23.61◦ 1980 184 pm 45 % 5040 78 pm 72 % 2400 417 nm 32.28◦ 2640 138 pm 94 % 6720 58 pm 48 % 3600 278 nm 53.23◦ 3960 92 pm 78 % 10080 38 pm 15 %

Table 4.2: Spectral bandwidth ∆λ and diffraction efficiency DE for the three gratings with different groove densities G with the laser diode polarization E perpendicular to the grating grooves (∆λ⊥, DE⊥), or parallel (∆λk, DEk). The measurement uncertainty for the diffraction efficiency values is specified by Richardson Gratings to be ±3 % [137]. 4.2 Macroscopic external cavity diode laser in Littrow configuration 47

Figure 4.8: Simulated normalized diffracted intensity D in the first diffraction order for a surface diffraction grating with 3600 grooves/mm in configuration b). optical output power of the free-running laser diode can be coupled out through the zeroth order reflection - assuming the ideal case of no absorption or stray light losses. In all other configurations, the expected spectral bandwidth of the grating feedback is broader and the diffraction efficiency in the first diffraction order is much higher, which would result in a broader emission spectrum and significantly less optical output power of the ECDL system.

As expected from these values, the ECDL system with a holographic surface grating with 3600 grooves/mm in configuration b) showed the best results with respect to a narrow spectral emission bandwidth and high optical output power. Therefore, only the results for this ECDL system will be presented in section 4.2.2. For convenience, the simulated normalized diffracted intensity D in the first diffraction order for a grating with 3600 grooves/mm and at a wavelength of λ = 445 nm set up in configuration b) is presented in figure 4.8. It is calculated according to equation (4.3) by using the experimental parameters of the setup from figure 4.6, that are summarized again in table 4.3:

groove distance dG 278 nm

wavelength λ 445 nm

◦ Littrow angle αL 53.23

beam radius r 1.4 mm

illuminated grooves N 10080

Table 4.3: Experimental parameters used for the calculation of the grating dispersion (ﬁgure 4.8) for a grating with 3600 grooves/mm. 48 4 External cavity diode lasers as pump sources for DUV generation

4.2.2 Electro-optical characteristics In the following, the performance of the ECDL system with a surface grating with 3600 grooves/mm in configuration b) (see figure 4.7) is presented. By tilting the surface grating around the Pivot point, the Littrow angle and therefore the center wavelength of the grating feedback is changed so that the ECDL emission wavelength can be tuned over a certain wavelength range. The grating mount has a rotation sensitivity of ∆θ = 0.003◦ (10 arc sec). According to equation (4.4) this correspondents to a wavelength tuning sensitivity of ∆λsens ≈ 17 pm. The lower graph in figure 4.9 shows the lasing threshold current of the ECDL system for different wavelengths. For each measurement point the grating is tilted by 0.09◦ corresponding to wavelength steps of 0.5 nm. The threshold current is determined by slowly reducing the injection current and observing the emission spectrum of the ECDL system with a double-echelle monochromator (LTB Lasertechnik Berlin GmbH) with a resolution of approximately 6 pm. For each measurement point, the ECDL alignment is slightly readjusted and the lowest injection current at which stable, single longitudinal mode emission is observed, is taken as the threshold current. The curve clearly follows the gain curve in figure 4.4, that is again shown in the upper graph of figure 4.9 for better comparison. At the maximum of the gain curve at a wavelength of λ = 445 nm, the lowest threshold current of Ith,ECDL = 115.4 mA is measured. This is due to the fact, that the threshold carrier density of the free-running laser diode is the lowest at the maximum of the gain curve [86, 138]. The dashed red line in figure 4.9 indicates the lasing threshold for the free-running laser diode, which is Ith,LD = 164 mA. It can be seen that the ECDL lasing threshold is smaller than the LD threshold for all ECDL wavelengths, which is in agreement with the theory [101] and other experiments [125–127]. The highest suppression of the amplified spontaneous emission of the laser diode and therewith the best ECDL performance can therefore be expected for an ECDL emission wavelength of λECDL = 445 nm.

Figure 4.9: Upper part: ASE emission spectrum of the laser diode at an injection current of I = 0.1 A. Lower part: Lasing threshold current versus emission wavelength for the ECDL system at a heatsink temperature of 20◦C. 4.2 Macroscopic external cavity diode laser in Littrow conﬁguration 49

Figure 4.10 shows the optical output power of the ECDL system emitting at 445 nm and of the free-running FP laser diode as a function of the injection current, both at a heatsink temperature of T = 20◦C. The curve for the ECDL system was only measured up to an injection current of 600 mA, because of concerns in the beginning of the experiments, that the laser diode could be damaged for high levels of feedback at higher injection currents.

ECDL at �ECDL = 445 nm FP laser diode

Figure 4.10: Optical output power of the ECDL system emitting at 445 nm and the FP laser diode for a heatsink temperature of 20◦C.

The FP laser diode has a lasing threshold of Ith,LD = 164 mA and a slope efficiency of SLD = 1.6 W/A. At the maximum studied injection current of 600 mA, the ECDL has an optical output power of 545 mW and the FP laser diodes output power at this current is 690 mW. The ECDL system exhibits a reduced lasing threshold of Ith,ECDL = 130 mA, and a reduced slope efficiency of SECDL = 1.2 W/A. The reduced slope efficiency is also observed in other ECDL systems and can be understood by reminding the relation between the differential efficiency ηd and the slope efficiency S given in equation (3.7), which says that S ∝ ηd. From equation (3.6) for the power characteristics of a laser diode above threshold the following equation for ηd can be derived (see [86], p.45): 1 1 2 = + αi (4.5) . ηd ηi ηi ln 1 R1R2

If the reflectivity of the front facet R2 is now replaced by an effective reflectivity R2,eff that is the composite reflectivity of the back facet reflectivity R2 and the reflectivity R3 of the external element (in a three mirror cavity model as described for instance in [101]), one can see that 1 ! S ∝ ηd ∝ ln . (4.6) R1R2,eff

In ECDL systems, R2,eff is typically larger than R2 (R2,eff > R2), and this should also be the case for the ECDL systems presented in this work. From the proportionality given in equation (4.6) then directly follows that the slope efficiency S is reduced. The deviation in the threshold current to the previous measurement (in figure 4.9) is caused by the high sensitivity to even minor adjustments. Due to thermal effects, the optimum adjustment with respect to a narrow emission bandwidth depends on the operating point 50 4 External cavity diode lasers as pump sources for DUV generation of the injection current and may therefore be different for low and high injection currents. For the power current characteristics, the ECDL system is adjusted for a minimal emission bandwidth at the maximum injection current of 600 mA. This adjustment then leads to a slightly higher threshold current compared to the previous adjustment for a the minimized threshold current. 4.2 Macroscopic external cavity diode laser in Littrow configuration 51

4.2.3 Spectral emission characteristics The spectral emission characteristics of the ECDL system are depicted in figure 4.11. It shows emission spectra of the ECDL system for an ECDL wavelength of λECDL = 445 nm for different injection currents of I = 0.2 A, 0.4 A, 0.6 A, 0.8 A, 1.0 A, and 1.2 A. The spectra of figure 4.11.a) are measured with a spectral resolution of 6 pm at 445 nm. Figure 4.11.b) shows corresponding emission spectra measured with the OSA with a spectral resolution of 50 pm. The emission has a spectral bandwidth of ∆λ ≤ 15 pm (FWHM) up to an injection current of 0.6 A, corresponding to an optical output power of PECDL = 545 mW. In this operating range, the emssion spectrum is dominated by a single longitudinal mode of the laser diode cavity. However, an increasing number of side modes starts to oscillate around the main peak with increasing injection current. These side-modes have a spectral distance to the main peak and to each other, respectively, of ≈ 27 pm and can therefore be identified as longitudinal modes of the laser diode cavity. The emission spectra for higher injection currents are dominated by more than one longitudinal mode from the laser diode cavity, which is exemplary shown for an injection current of 0.8 A, and 1.0 A.

a) b)

1.0 I = 0.2 A 0 I = 0.2 A

24 dB

15 pm

0.5 -30

-60 0.0

1.0 I = 0.4 A 0 I = 0.4 A

32 dB

14 pm

0.5 -30

0.0 -60

1.0 I = 0.6 A 0

I = 0.6 A

14 pm 42 dB

0.5 -30

0.0 -60

1.0 I = 0.8 A 0 I = 0.8 A

19 dB

33 pm Intensity / dB / Intensity 0.5 -30

0.0

Normalized intensity Normalized -60

1.0 I = 1.0 A 0 I = 1.0 A

14 dB

40 pm

0.5 -30

0.0 -60

1.0 I = 1.2 A 0 I = 1.2 A 4 dB

15 pm

0.5 -30

0.0 -60

444.8 445.2 435 440 445 450 455

Wavelength / nm Wavelength / nm

Figure 4.11: Emission spectra of the ECDL system at injection currents of I = 0.2 A, 0.4 A, 0.6 A, 0.8 A, 1.0 A, and 1.2 A measured with a) a double-echelle monochromator with a spetral resolution of 6 pm and b) an optical spectrum analyzer with a spectral resolution of 50 pm, for 20◦C heatsink temperature. 52 4 External cavity diode lasers as pump sources for DUV generation

The spectral bandwidth however does not exceed 40 pm (FWHM), which is almost identical to the simulated spectral bandwidth of the diffracted intensity from the surface grating of 38 pm (FWHM, see figure 4.8). At I = 1.2 A, the ECDL bandwidth is smaller than at I = 1.0 A, which can be explained by the wavelength shift of the gain with respect to the ECDL wavelength. At I = 1.2 A, the wavelength difference between gain maximum and ECDL wavelength is larger than at 0.8 A or 1.0 A as can be seen in figure 4.11.b). Thus, the laser diode cavity modes around 445 nm experience less gain and the mode discrimination induced by the grating dispersion is therefore strong enough to favor only one of the longitudinal laser diode cavity modes. However, the complete emission spectrum at 1.2 A, which can be seen in figure 4.11.b), is dominated by the laser diode cavity modes around the gain maximum as their suppression is only 4 dB. Figure 4.11.b) shows how the suppression of the amplified spontaneous emission and laser diode cavity modes depends on the spectral position of the gain maximum with respect to the ECDL emission wavelength. Here and in the following, this suppression ratio will be denoted as SR value in decibel.

As shown in figure 4.5, the center wavelength of the free-running laser diode emission exhibits a shift of 3.9 nm/A. For lower injection currents, the gain takes its maximum at shorter wavelengths than the center wavelength of the feedback at 445 nm. With the shift of the gain maximum with increasing current, the SR increases and reaches its highest value of 42 dB at I = 0.6 A, where the wavelength of the gain maximum coincides with the ECDL emission wavelength. The onset of the side-modes seen in figure 4.11.a) is due to the increasing gain of the laser diode cavity modes, that are not sufficiently suppressed by the mode discrimination from the grating feedback for higher injection currents. At the operating point of 0.6 A, corresponding to an ECDL output power of 545 mW, the ECDL system shows the best performance in terms of emission bandwidth reduction together with a high SR value. Figure 4.12 shows a spectrum of the ECDL and the laser diode (LD) at 0.6 A for comparison. An emission bandwidth reduction by a factor of 73

Figure 4.12: Emission spectra of the FP laser diode (grey) and the ECDL system (red) at I = 0.6 A and T = 20◦C. 4.2 Macroscopic external cavity diode laser in Littrow conﬁguration 53

from ∆λFP = 1.1 nm to ∆λECDL = 15 pm (FWHM) is achieved. Above I = 0.6 A, the gain shifts towards longer wavelengths than the feedback wavelength and the suppression of the laser diode cavity modes around the gain maximum decreases in comparison to the modes selected by the optical feedback. Therefore, a majority of the ECDL output power comes from the laser diode cavity modes around the gain maximum and does not lie in the spectral window of the grating feedback anymore.

This is evaluated by linearizing the normalized spectra from figure 4.11.b) and comparing the integral over the ECDL peak wavelength range from 445.01 nm to 445.29 nm (ECDL peak wavelength is 445.15 nm) with the integral over the whole laser diode emission from 420 nm to 470 nm. This way, the ratio of the optical power PRECDL, that is contained in the ECDL peak, can be calculated and is summarized in table 4.4. The higher the SR value, the higher is the ratio PRECDL of the ECDL peak power with a maximum of 99.7 % at 0.6 A. This ratio significantly decreases for higher injection currents. At 1.2 A, less than 10 % of the total laser diode power is included in the ECDL peak wavelength range. It will be seen in section 5.1.2 that this behavior has a negative effect on the SHG conversion efficieny.

I ∆λ (FHWM) SR PRECDL

0.2 A 15 pm 24 dB 93.4 % 0.4 A 14 pm 32 dB 99.0 % 0.6 A 14 pm 42 dB 99.7 % 0.8 A 33 pm 19 dB 88.9 % 1.0 A 40 pm 14 dB 73.3 % 1.2 A 15 pm 4 dB 9.9 %

Table 4.4: Summary of the measured spectral parameters of the macroscopic ECDL system emitting at 445 nm. 54 4 External cavity diode lasers as pump sources for DUV generation

4.2.4 Wavelength tuning By tilting the surface grating and thereby changing the Littrow angle the emission wavelength of the ECDL in Littrow configuration is tuned. The wavelength tuning behavior of the ECDL system at an injection current of I = 0.6 A can be seen in figure 4.13. It shows emission spectra of the ECDL system for different Littrow angles corresponding to emission wavelengths of 443 nm, 444 nm, 445 nm, 446 nm, and 447 nm, indicating a coarse tuning range of about 4 nm. While the emission bandwidth of the ECDL peak is well below 20 pm (FWHM) over the whole tuning range, the suppression of the longitudinal modes of the laser diode cavity (SR) depends on the position of the ECDL wavelength with respect to the gain maximum around 445 nm.

I = 0.6 A

0 = 443 nm

ECDL 13 dB

-20

-40

-60

= 444 nm

ECDL

32 dB

-20

-40

-60

= 445 nm

ECDL

-20 42 dB

-40

-60 Intensity / dB / Intensity

= 446 nm

ECDL

34 dB

-20

-40

-60

0 = 447 nm 12 dB ECDL

-20

-40

-60

435 440 445 450 455

Wavelength / nm

Figure 4.13: Emission spectra of the ECDL system at an injection current of I = 0.6 A for diﬀerent Littrow angles measured with the optical spectrum analyzer with a resolution of 50 pm at 20◦C heatsink temperature.

This again illustrates why the laser diode coupling facet is usually AR-coated in ECDL systems. Without AR-coating, the oscillating laser diode cavity modes can only be su- pressed efficiently when the feedback is close to the gain maximum, as it is the case at λ = 445 nm for this ECDL system. In general, the tuning range depends on the feedback strength and the reflection of the facet through which the optical feedback is coupled back into the laser diode cavity. An AR-coating would likely enhance the wavelength tuning range of the ECDL system compared to the ECDL system without AR-coated coupling facet [127]. Larger feedback also enhances the tuning range, but leads to a lower ECDL output power [138]. 4.2 Macroscopic external cavity diode laser in Littrow configuration 55

4.2.5 Spatial emission characteristics The beam quality of the ECDL system is determined by a caustic measurement according to ISO 11146 using the variance and the 90%-knife edge method, respectively. Figure 4.14 shows the caustic in the fast (left) and slow axis (right) at an injection current of I = 0.6 A, corresponding to an output power of 545 mW. The variance beam propagation factors are M2 = 4 0 and M2 = 7 7 in the fast and slow axis, respectively. The 90%-knife 4σ,fast . 4σ,slow . edge beam diameter values are M2 = 2 1 and M2 = 4 7 in the fast and slow 90%,fast . 90%,slow . axis, respectively.

The values in both directions are slighlty larger than the ones for the free-running laser diode. Though it must be noted that the laser diode used in the macroscopic ECDL is not the same that was presented in chapter3. However, the slightly larger values could be attributed to an increase of peripheral intensities around the central beam profile that might be caused by random stray light from the surface diffraction grating. As explained in section 3.6, the precise determination of the beam quality for this laser diode is difficult due to the high amount of peripheral intensities and must therefore interpreted with caution. In conclusion, it is believed that the optical feedback does not have a relevant effect on the beam quality of the ECDL system.

2.0

Variance (4 ) Variance (4 )

Knife edge (90%) Knife edge (90%) / mm / d caustic fit caustic fit 1.5

2 2

M = 7.7 1.0 M = 4.0 4 4

M = 4.7 0.5 M = 2.1 90%

90%

fast axis I = 0.6 A slow axis I = 0.6 A Beam waist diameter diameter waist Beam

0.0

100 200 300 400 500 50 100 150 200 250 300 350

Position z / mm

Figure 4.14: Measured caustic in the fast (left) and slow axis (right) of the ECDL system at an injection current of I = 0.6 A using variance and the 90%-knife edge method. 56 4 External cavity diode lasers as pump sources for DUV generation

4.3 Volume-Bragg-Grating stabilized external cavity diode laser module (µECDL) The proof-of-concept study presented in section 4.2 has proven the feasibility of a spectral bandwidth reduction of the emission from a high-power GaN based laser diode by using a surface diﬀraction grating as wavelength selective optical element in a simple ECDL system without an additional AR-coating of the laser diode front facet. It has also been shown that the spectral bandwidth of the grating dispersion needs to be in the range of the desired spectral emission bandwidth of the external cavity system.

In this section, spectral narrowing and wavelength stabilization of the same type of laser diode is demonstrated in a micro-integrated external cavity diode laser module (µECDL). In this module, the wavelength selective optical feedback is provided by a holographic reflecting volume Bragg grating (VBG). The use of a VBG opens up the possibility of a very compact and robust laser diode module without moveable parts. Reflecting VBGs can have reflectivities up to 99% in a narrow wavelength range below 100 pm. For GaAs laser diodes, reflecting VBGs were used with ridge-waveguide diode lasers in a rear external cavity [139]. Mainly, they are used with reflectivities of 20%-60% for wavelength narrowing and stabilization of broad-area diode lasers and arrays by placing them directly behind the collimation optics [140, 141], as it will also be presented in this section. First, the working principle of a VBG will be explained. A detailed description of the concept and the development of the µECDL module follows before the electro-optical, spectral and spatial emission characteristics of the module are analyzed. 4.3 Volume-Bragg-Grating stabilized external cavity diode laser module (µECDL) 57

4.3.1 Working principle of volume Bragg gratings A holographic VBG is a structure with a uniformly spaced refractive index variation in the volume of a photosensitive medium. The refractive index modulation is recorded in so-called photo-thermo-refractive (PTR) glass by exposure with the interference pattern of UV laser light, typically with a wavelength between 280 nm and 350 nm, and a consecutive thermal development at a temperature of around 520◦C. The first recording of such a hologram in PTR glass was realized in the late 80s [142] and resulted in a further development of the technology and the production of highly efficient VBGs in the following years [143–145]. A comprehensive review of the photo-thermo mechanism in PTR glass is given by Lumeau and Zanotto [146]. PTR glass is Na2O-ZnO-Al2O3-SiO2 glass doped with silver (Ag), cerium (Ce), and fluorine (F). It is transparent in a wavelength range of 350 nm - 2500 nm and characterized by very stable thermo-mechanical properties up to a temperature of 400◦C. With dn/dT = 5 · 10−8 K−1 the thermal variation of the refractive index is very low and results in a small thermal shift of the Bragg wavelength of ∆λB/∆T = 7 pm/K [147].

The working prinicple of a VBG is based on the diffraction of light in a Bragg grating structure and was described in detail with a coupled wave theory formulated by Kogelnik in 1969 [148]. Figure 4.15 shows a model of a reflecting volume Bragg grating with parallel Bragg planes of varying refractive index indicated by dashed lines, that are slanted at an angle φ with respect to the surface of the VBG medium. The grating vector K has a length of K = 2π/Λ (with Λ being the grating period) and is oriented perpendicular to the parallel Bragg planes. The incident beam Rin hits the grating with the Bragg angle θ to the grating surface normal, experiences diffraction inside the grating and is coupled out via the output beam Sout.

S out K f z q Rin

Figure 4.15: Model of a holographic reﬂecting VBG with thickness d. Rin: incident beam, Sout: output beam, K: grating vector, θ: Bragg angle, φ: slanting angle, Λ: grating period, z: optical axis.

Based on the results obtained with the surface diffraction grating in the macroscopic ECDL system, a reflecting Bragg grating with comparable diffraction efficiency and spectral selectivity, produced by the company Optigrate Corp., is applied in the µECDL module. 58 4 External cavity diode lasers as pump sources for DUV generation

The diffraction efficiency DE for a lossless reflecting volume Bragg grating is

2 2 !−1 1 − ξ /ν DE = 1 + (4.7) sinh2 pν2 − ξ2 and also includes a description of the angular and wavelength sensitivity of the grating. In case of an unslanted (φ = 0◦) reﬂecting VBG and an incident beam perpendicular to the grating surface (θ = 0◦), the parameters ξ and ν become

ϑ · d ξ = − , (4.8) 2(1 − λ/(nav · Λ))

= π · δn · d (4.9) ν i p λ · 1 − λ/(nav · Λ) with 2 = π π · λ (4.10) ϑ − 2 . Λ nav · Λ The other parameters in the equations are the grating thickness d, the average refractive index of the grating nav, the refractive index modulation δn, and the wavelength λ.

The chosen grating has an aperture of 10 mm x 10 mm and a thickness of 2.7 mm. It is AR-coated on both sides with R < 0.5% at the resonant wavelength of 445 nm. As specified by Optigrate Corp, the VBGs maximum diffraction efficiency is DEmax = 17.9% at its resonant Bragg wavelength λB = 445 nm. By using the parameters summarized in table 4.5, the diffraction efficiency of the grating as a function of the wavelength is calculated using equation (4.7) and is depicted in figure 4.16. The grating is unslanted (φ = 0) and has a Bragg angle of θ = 0 for 445 nm. The spectral selectivity of the grating, i.e. the spectral bandwidth of the diffracted light, is specified to be

Parameter Symbol Value

Bragg wavelength λB 445 nm Grating thickness d 2.7 mm

Average refractive index nav 1.4933 Grating period Λ 149 nm Refractive index modulation δn 22 ppm

Simulated diﬀraction eﬃciency DEmax 15.7%

Simulated spectral selectivity (FWHM) ∆λsim 24 pm

Manufacturer speciﬁcations

Diﬀraction eﬃciency DEmax 17.9% Spectral selectivity (FWHM) ∆λ (30 ± 10) pm

Table 4.5: Parameters used to simulate the diffraction efficiency of the reflecting VBG applied in the µECDL module and the obtained values for DEmax at 445 nm and the spectral selectivity ∆λ. Lower part: manufacturer (Optigrate Corp.) specifications for DEmax and ∆λ. 4.3 Volume-Bragg-Grating stabilized external cavity diode laser module (µECDL) 59

∆λ = (30 ± 10) pm (FWHM). The simulated diffraction efficiency curve shows a maximum DEmax = 15.7% at 445 nm and a spectral bandwidth of ∆λsim = 24 pm (FWHM). The simulated spectral bandwidth is well within the uncertainty of the specified spectral bandwidth and the deviation in the diffraction efficiency can be explained by usual deviations in the manufacturing process.

The simulated normalized diffraction efficiency as a function of the deviation from the Bragg angle is shown in figure 4.17. For a deviation of ±0.48◦ of the incident light beam from the Bragg angle, the diffraction efficiency decreases to 50% of its maximum. This means, that the light incident on the VBG needs to be precisely collimated to make use of the maximum diffraction efficiency.

Figure 4.16: Simulation of the wavelength dependence of the diffraction efficiency DE according to equation (4.7) for a reflecting VBG with the parameters from table 4.5.

Figure 4.17: Simulation of the angle dependence of the diffraction efficiency DE according to equation (4.7) for a reflecting VBG with the parameters from table 4.5. 60 4 External cavity diode lasers as pump sources for DUV generation

4.3.2 Concept and development of the µECDL module The concept of the µECDL module is illustrated in ﬁgure 4.18. For the µECDL, the laser diode characterized in section 3.4 is used as gain medium. The protective glass of the TO56 can (W) has a thickness of 250 µm and the laser diode cavity length is 1.2 mm. Based on the lateral and vertical far-ﬁeld angles of the laser diode emission and optical simulations, an aspheric lens (L) with a focal length of f = 2.54 mm and a numeric aperture of NA = 0.66 is used to collimate the laser light, which is then directed onto the VBG. The lens (R < 0.5%) and the VBG (R < 0.5%) are AR-coated for 445 nm. Figure 4.18 also shows the dimensions of the µECDL elements and their distances to each other along the optical axis z in millimeters. L outer L L inner LD y W x LD z L

VBG 1.6 1.2 1.8 4.0 2.7 Figure 4.18: Concept of the µECDL module. Also shown are the distances between the different elements of the module and the three competing resonators of length LLD, Linner, and Louter . As explained in section 4.1, without AR-coating of the laser diodes front facet, three competing resonators can in principle influence the frequency response of the µECDL module: the laser diode resonator CLD between the front and end facet, the inner resonator Cinner between the front facet and the VBG, and the outer resonator Couter between the back facet and the VBG. Each resonator has a different optical path length resulting in a different free spectral range (FSR). The optical path length Lopt = P ni · si of each resonator is the sum of the optical path lengths si · ni of each contributing section, with ni being the respective refractive index. For the dimensions indicated in figure 4.18, Lopt and the resulting free spectral range for each resonator are summarized in table 4.6.

Resonator Lopt FSR

CLD 3.7 mm 27 pm

Cinner 12.8 mm 8 pm

Couter 16.5 mm 6 pm

Table 4.6: Optical path length and free spectral range for the three resonators CLD, Cinner, and Couter formed in the µECDL module. 4.3 Volume-Bragg-Grating stabilized external cavity diode laser module (µECDL) 61

The required precision for the alignment of the optical elements is determined with optical beam shaping simulations according to ISO 11146 using the program WinABCD, that was developed at the Ferdinand-Braun-Institut by Dr. Bernd Eppich. The simulated residual full divergence angle of the beam in the lateral and vertical plane in dependence of the collimating lens (L) position along the optical axis z is shown in figure 4.19. The lens needs to be aligned at a distance z = (1.575 ± 0.035)mm to the front facet to achieve a residual divergence in both planes that is smaller than the 95%-angular selectivity δθ95% of the VBG, in order to make use of at least 95% of the VBGs maximum diffraction efficiency.

1.2

lateral divergence

vertical divergence

1.0

0.8

VBG-

95%

0.6

0.4

0.2 Full div. angle / degree / angle div. Full

0.0

1.50 1.52 1.54 1.56 1.58 1.60 1.62 1.64

Lens position z / mm

Figure 4.19: Simulated residual full divergence angle in lateral and vertical direction as a function of the lens position.

The most critical parameters in the alignment of the µECDL module are the VBG tilting angles around the x- and y-axis indicated by the coordinate system in ﬁgure 4.18. Tilting of the VBG around the y-axis changes the position of the back-coupled light along the vertical plane of the laser diodes front facet. Tilting around the x-axis inﬂuences the coupling in the lateral plane of the laser diode.

vertical axis lateral axis

1.0

95% 95%

0.002° = 35 rad 0.028° = 0.5 mrad

0.5 FW HM FW HM

0.1° = 1.7 mrad 0.008° = 140 rad Coupling efficiency Coupling

0.0

-0.025 0.000 0.025 -0.2 -0.1 0.0 0.1 0.2

VBG tilt / degree VBG tilt / degree

Figure 4.20: Simulated coupling eﬃciency of the back-coupled light in the vertical and lateral axis as a function of the VBG tilt around the x- and y-axis, respectively. 62 4 External cavity diode lasers as pump sources for DUV generation

The laser diodes entrance aperture in the simulation is estimated from the measurements of the far-field angles to be 0.8 µm and 15 µm in the vertical and the lateral direction, respectively. Figure 4.20 shows the simulated coupling efficiency as a function of the VBG tilt in both directions. The smaller vertical aperture leads to a higher sensitivity in this direction. For a tilt larger than ±0.001◦ or ±17.5 µrad the coupling efficiency already sinks below 95% of its maximum. In the lateral direction, the coupling efficiency is above 95% for tilting angles smaller than ±0.014◦ or ±250 µrad.

Mounting of the µECDL module A schematic top and side view of the µECDL module is shown in figure 4.21. All elements are assembled on a conduction cooled package (CCP) (5) with a footprint of 25 mm x 25 mm. The CCP is custom-made out of one piece of copper (Cu)1, and electroplated with a 250 µm thick gold (Au) layer to protect the surface from oxidation and increase its thermal conductivity. The laser diode in the TO56 can (1) is positioned inside a tube in the CCP, that has a slightly larger diameter than the base plate of the TO56 can. The base plate is clamped onto the CCP with a screw nut with an external thread (4), which is then fixed to the CCP body by UV-curable adhesive. To protect the optical elements the CCP is covered with a black anodized aluminum plate (6). The beam height of the module is h = 8 mm, so that it is compatible with the fixed beam height of the micro-assembly and laser diode characterization station that is described in detail in the PhD thesis of Martin Maiwald [149]. The lens (2) and the VBG (3) are both aligned during laser operation. For this, they are hold by a vacuum tweezer, that is mounted on a six-axis positioning stage (H-206, Physik Instrumente (PI) GmbH & Co. KG) with high precision in the translational and

25 mm top view

2 1 3

25 mm 4

6 side view

15 mm 8 mm

Figure 4.21: Schematic top and side view of the micro-integrated ECDL module: (1) laser diode in TO56 can, (2) collimating lens, (3) reﬂecting volume Bragg grating, (4) thread for attachment of the laser diode, (5) conduction cooled package, (6) cover plate.

1 The manufacturing of the CCP was done by the FBH colleagues Thomas Roos, Detlef Grimpe, and Bastian Deutscher. 4.3 Volume-Bragg-Grating stabilized external cavity diode laser module (µECDL) 63 rotational axes. The stage provides a minimal translational step size of 0.1 µm and a minimal rotational step size of 2 µrad, both being sufficient for the required alignment precision shown in figure 4.19 and 4.20. For the alignment of the collimating lens (L), the laser diode is operated far below threshold and the amplified spontaneous emission is focused onto a CCD camera, that is positioned with its chip in the focal plane of a lens with a focal length of 150 mm. The collimating lens and the lens with 150 mm form a telescope, and the optimal collimation is achieved, when the spot size on the CCD chip becomes minimal. The lens is then mounted onto the CCP with UV-curable adhesive, that is treated with UV light for approximately 10 minutes.

The VBG alignment and mounting follows a similar procedure. However, alignment optimization is carried out by monitoring the emission spectrum of the module with the double-echelle monochromator with a resolution of 6 pm at 445 nm. Initial coarse alignment is carried out at a high operating current until an effect of the optical feedback like fluctuations in the spectral intensity or even a narrowing of the emission bandwidth is observed in the spectrum. Then, the injection current is decreased and the VBG is re-adjusted. This is done until the smallest injection current is reached, at which an effect on the emission spectrum can be observed.

It must be noted, that minimizing the threshold current may be the optimal alignment for low injection currents. However, as explained above, the tolerances for the VBG alignment in this system are extremely small. Therefore, the optimal alignment at high operating currents in terms of a preferably narrow emission bandwidth and a high ASE suppression may slightly differ from the optimum at low injection currents. This is due to eventual minimal changes in the position of the optical elements or the laser diode cavity caused by thermal expansion, which can have a direct effect on the coupling efficiency. For this reason, the injection current is set to the maximum value of 1.2 A and the VBG is again slightly adjusted until the emission spectrum is stable and as narrow as possible. Finally, the VBG is then mounted onto the CCP with UV-curable adhesive, that is treated with UV light for approximately 10 minutes. 64 4 External cavity diode lasers as pump sources for DUV generation

4.3.3 Electro-optical characteristics For all measurements presented below the heatsink temperature is set to T = 20◦C. The optical output power versus the injection current for the free-running Fabry-Perot laser diode and the µECDL module is shown in figure 4.22. The laser diode characterized in chapter3 is used as gain medium. It has a lasing threshold of Ith,LD = 145 mA and a slope efficiency of SLD = 1.55 W/A. Laser operation for the µECDL starts at Ith,µECDL = 115 mA and a reduced slope efficiency of SµECDL = 1.3 W/A is observed. At the maximum injection current of I = 1.2 A, the FP laser diode has an output power of PLD = 1.57 W. Due to the much better spectral characteristics (see section 4.3.4) compared to the macroscopic ECDL system, the injection current could be extended until the maximum injection current of 1.2 A. Here, the µECDL module reaches a maximum optical output power of PµECDL = 1.43 W. The sawtooth-like power characteristic of the µECDL emission for higher injection currents can be attributed to mode-hops.

In general, a comparison between the electro-optical parameteres of the macroscopic ECDL and the µECDL emission is difficult as different laser diodes are applied. However, the higher relative feedback from the VBG compared to the surface grating should lead to a lower threshold current and lower slope efficiency. Whereas the µECDL threshold current is indeed lower, the slope efficiency is higher than that for the macroscopic ECDL. This effect seems counterintuitive and could not be conclusively explained in the course of this work. The adjustment of the optical elements of the µECDL module is carried out with higher precision than the manual adjustment of the macroscopic ECDL system. Furthermore, the external cavity length of the µECDL module is much shorter than in the macroscopic ECDL. These aspects probably lead to a higher coupling efficiency of the optical feedback and to a lower sensitivity to mechanical vibrations for the µECDL module, which consequently results in a better performance of the µECDL module.

Figure 4.22: Optical output power versus injection current for the µECDL module (solid line) and the FP laser diode (dashed line) at a heatsink temperature of T = 20◦C. 4.3 Volume-Bragg-Grating stabilized external cavity diode laser module (µECDL) 65

4.3.4 Spectral emission characteristics Without external optical feedback the FP laser diode emits in longitudinal and lateral multi-mode with a spectral emission width of about 1 nm (FWHM) (see section 3.5 and 3.6). Figure 4.23 and figure 4.24 show optical spectra of the laser diode emission and of the µECDL emission as a function of the injection current I, respectively. The spectra in figure 4.23 are measured with a spectrometer providing a spectral resolution of 0.6 nm. The measuring interval is ∆I = 20 mA. The spectra in figure 4.24 are measured with the double-echelle monochromator providing a spectral resolution of 6 pm at 445 nm, and the measuring interval is ∆I = 5 mA. Each spectrum is normalized in intensity to 1 and all spectra are presented in a false color contour diagram.

The spectral emission of the free-running laser diode shows a shift of the center wavelength of ∆λC = 3.5 nm/A (see section 3.5). At I = 1.2 A, it has a center wavelength of λC = 445.3 nm. The spectral emission bandwidth is around 1 nm (FWHM). The spectral emission of the µECDL module is wavelength stabilized for the whole operating range. At P = 0.1 W (I = 0.225 A), the peak wavelength is λ = 445.06 nm and at I = 1.4 W (I = 1.2 A) the peak wavelength is λ = 445.12 nm. With a typical wavelength shift of the Bragg wavelength of the VBG of ∆λ/∆T = 7 pm/K [147], the peak wavelength shift would correspond to an increase of the temperature of the µECDL of about ∆T ≈ 9 K from P = 0.1 W to P = 1.4 W. The entire µECDL emission stays within a spectral window of 250 pm, which is indicated by the white dashed lines in both ﬁgures.

455

Normalized intensity

0 0.5 1

450

250 pm / nm /

445

440 Wavelength Wavelength

T = 20°C

435

0.0 0.2 0.4 0.6 0.8 1.0 1.2

Injection current I / A

Figure 4.23: Contour plot of multiple Figure 4.24: Contour plot of multiple emission spectra of the FP laser diode as a µECDL emission spectra as a function of function of the injection current with mea- the injection current with measurement surement steps of 20 mA. Each spectrum is steps of 5 mA. Each spectrum is individu- individually normalized in intensity to 1. ally normalized in intensity to 1.

For a better illustration of the wavelength stabilization, ﬁgure 4.25.A shows the full-width at half maximum emission bandwidth for the free-running LD and the µECDL. The µECDL emission bandwidth is in the range of the spectral selectivity of the VBG of 24 pm (FWHM) up to an output power of about 1 W, corresponding to an injection current of I = 0.9 A. For higher output powers, more longitudinal modes of the laser diode start to oscillate. 66 4 External cavity diode lasers as pump sources for DUV generation

450

free-running LD free-running LD A B

ECDL ECDL / nm / 448 / nm / peak

446 FWHM

0.1

444

0.01

442 Bandwidth Bandwidth Peak wavelength wavelength Peak

0.001 440

0.2 0.4 0.6 0.8 1.0 1.2

Injection current I / A Injection current I / A

Figure 4.25: A: Peak wavelength of the free-running laser diode and of the µECDL emission as a function of the injection current. B: Half-logarithmic plot of the FWHM bandwidth of the free-running laser diode and the µECDL emission as a function of the injection current.

However, the FWHM emission bandwidth does not exceed 50 pm up to I = 1.2 A. On average, the emission bandwidth of the free-running laser diode is reduced by almost two orders of magnitude with the µECDL module. The peak wavelength λpeak of the free-running laser diode and the µECDL emission as a function of the injection current are shown in figure 4.25.B. For high operating currents, the emission wavelength of the free-running laser diode shifts towards the µECDL emission wavelength (∆λC = 3.5 nm/A). This influences the strength of the ASE and longitudinal laser diode mode suppression (SR) for the µECDL, which becomes noticeable in figure 4.26.A. It shows spectra of the free-running LD and the µECDL emission at selected injection currents of 0.3 A, 0.6 A, 0.9 A, and 1.2 A. The spectra are again measured with the optical spectrum analyzer.

The spectral behavior is comparable to that of the macroscopic ECDL described in section 4.2.3 with the diﬀerence that the central emission wavelength λC, i.e. the gain spectrum, of the laser diode used in the µECDL is approximately 2 nm blue-shifted. Ac- cordingly, the gain maximum and the µECDL emission wavelength at 445 nm coincide for higher injection currents and the suppression of the ASE and the longitudinal laser diode modes (SR) reaches its maximum with 53 dB at an injection current of I = 0.9 A. Compared to the macroscopic ECDL the SR value is higher over a wider operating range. Above I = 0.9 A, the gain shifts towards longer wavelengths but is still close to the µECDL emission and the SR value only slightly decreases to 48 dB at I = 1.2 A.

Figure 4.26.B shows the spectra of the µECDL emission at the same operating points with a higher resolution of 6 pm. As already mentioned, the spectrum is dominated by only one longitudinal FP laser diode mode up to I = 0.9 A, where the emission has a spectral width of ∆λFWHM = 22 pm. And a spectral width of only ∆λ95% = 46 pm even contains 95% of the intensity. At I = 1.2 A (P = 1.43 W) several other modes occur. However, the emission still has a spectral width of ∆λFWHM = 14 pm and ∆λ95% = 120 pm. The other modes have a spectral distance of about 27 pm to each other and can therefore be 4.3 Volume-Bragg-Grating stabilized external cavity diode laser module (µECDL) 67

A B

444.5 445.0 445.5 435 440 445 450 455

1.0 ECDL

0 I = 0.3 A I = 0.3 A

-20 38 dB

11 pm FW HM

0.5

-40

25 pm 95%

-60

0.0

1.0

0 I = 0.6 A I = 0.6 A

-20

51 dB 21 pm FW HM

0.5

-40

39 pm 95%

-60

0.0

1.0

0 I = 0.9 A I = 0.9 A

-20

22 pm FW HM 53 dB

0.5

-40 Relative intensity / dB / intensity Relative

46 pm 95% -60

0.0

1.0 units) (arb. intensity Normalized

0 I = 1.2 A I = 1.2 A

-20

48 dB 14 pm FW HM

0.5

-40

120 pm 95%

-60

0.0

435 440 445 450 455 444.5 445.0 445.5

Wavelength / nm Wavelength / nm

Figure 4.26: A: Emission spectra of the free-running LD and the µECDL system at injection currents of I = 0.3 A, 0.6 A, 0.9 A, and 1.2 A measured with a high-dynamic range optical spectrum analyzer with a spectral resolution of 50 pm. B: µECDL emission spectra at I = 0.3 A, 0.6 A, 0.9 A, and 1.2 A measured with a double-echelle monochromator with a spetral resolution of 6 pm. indentified as longitudinal modes from the FP laser diode. In conclusion, narrowband emission together with a high ASE and longitudinal laser diode mode suppression is achieved for much higher output powers up to 1.4 W with the µECDL module. The reasons for this improved behavior were already mentioned, namely the higher diffraction efficiency of the VBG of 17.9%, the more precise adjustment of the optical elements, the reduced external cavity length of the µECDL module, and the coninciding gain peak and VBG feedback wavelengths for higher injection currents. 68 4 External cavity diode lasers as pump sources for DUV generation

4.3.5 Temporal stability of the µECDL emission To evaluate the temporal stability of the µECDL emission wavelength and optical output power, both are measured over an operation time of one hour with a measurement step size of 10 s. The injection current is set constant to I = 575 mA and is controlled using a current source with a long-term stability of < 15 µA, as speciﬁed by the supplier (Newport, model 560B). Figures 4.27.a) and c) show the peak wavelength and the optical output power as a function of the operation time, respectively. Over the whole operation time the mean peak wavelength of λpeak,mean = 445.10 nm is stable within a spectral window of 6 pm, corresponding to the spectral resolution of the spectrometer. The mean optical output power is Pmean = 528 mW with a peak-to-peak variation of ±3 mW (< ±1%). Figure 4.27.b) shows an exemplary spectrum after an operating time of top = 30 min. It has a spectral width of 14 pm FWHM.

a) b)

Opera tion time t / m in

0 10 20 30 40 50 60

445.12

1.0

I = 575 mA

t = 30 min

445.11 op

445.10 0.5

14 pm / nm In ten sity peak 445.09

6 pm

0.0

445.08

444.5 445.0 445.5

W a velength / nm

560

I = 575 mA

550

540

530 / mW

520 opt P

510

500

0 10 20 30 40 50 60

Opera tion time t / m in

Figure 4.27: Temporal stability of the µECDL emission for an injection current of I = 575 mA. a) Peak wavelength of the µECDL emission measured over a time period of 1 hour. b) Exemplary emission spectrum at top = 30 min. c) Optical output power of the µECDL over a time period of 1 hour. 4.3 Volume-Bragg-Grating stabilized external cavity diode laser module (µECDL) 69

4.3.6 Spatial emission characteristics The beam quality of the µECDL module is determined by a caustic measurement according to ISO 11146 using the variance and the 90%-knife edge method, respectively. Figure 4.14 shows the caustic in the fast (left) and slow axis (right) at an injection current of I = 0.8 A, corresponding to an output power of 860 mW. The variance beam propagation factors are M2 = 2 7 and M2 = 7 5 in the fast and slow axis, respectively. The 90%-knife 4σ,fast . 4σ,slow . edge beam diameter values are M2 = 1 1 and M2 = 4 5 in the fast and slow 90%,fast . 90%,slow . axis, respectively.

2.0 2.0

Variance (4 ) Variance (4 )

Knife edge (90%) Knife edge (90%) / mm / d caustic fit caustic fit 1.5 1.5

M = 7.5

1.0 1.0

M = 2.7

2 2

M = 4.5 M = 1.1

90% 0.5 90% 0.5

fast axis I = 0.8 A slow axis I = 0.8 A Beam waist diameter diameter waist Beam

0.0 0.0

0 100 200 300 400 500 0 100 200 300 400 500

Position z / mm

Figure 4.28: Variance and 90%-knife edge caustic measurement of the µECDL emission in fast and slow axis for an injection current of I = 0.8 A. The values in both directions are within the measurement uncertainty in good agreement with the values for the free-running laser diode. In particular, the 90%-knife edge values are precisely the same, indicating that the VBG does not have an eﬀect on the beam quality of the laser diode and therewith also no eﬀect on the subsequent SHG process. 70 4 External cavity diode lasers as pump sources for DUV generation

4.4 Summary Table 4.7 summarizes the electro-optical, spectral and spatial parameters of the free-running laser diode (LD), the macroscopic ECDL and the µECDL at the respective maximum studied injection current.

The macroscopic ECDL only shows narrowband emission < 40 pm (FWHM) together with a high suppression ratio of the ASE and laser diode cavity modes (SR) up to an injection current of 0.6 A, corresponding to an output power of 0.55 W. The µECDL maintains narrowband emission < 40 pm (FWHM) with an SR ≥ 38 dB for all operating points up to the maximum injection current of 1.2 A, corresponding to an output power of 1.43 W. The maximum SR is 53 db at 0.9 A.

The big advantage of the macroscopic ECDL system is its principle ability to tune the emission wavelength. As was shown, the tunability is limited to about 4 nm, because the longitudinal laser diode cavity modes are not suppressed suﬃciently for ECDL wavelengths too far away from the laser diode gain maximum.

A poorer beam quality of the macroscopic ECDL in the fast axis is indicated by somewhat higher M2 values. As the laser diodes fast axis is in the plane in which the beam is diﬀracted by the surface grating, the slightly worse beam quality might originate from aberrations caused by the diﬀraction. In the slow axis, the caustic measurements resulted in similar M2 values. Considering all 90%-M2 values, which are almost identical (except for the value of the macroscopic ECDLs fast axis), the ECDL setups seem to have no impact on the beam quality. In general, the relatively poor lateral (slow axis) beam quality of the laser diode is a drawback for the subsequent SHG process as it limits the focusability of the beam.

However, both the macroscopic proof-of-concept ECDL system in Littrow configuration and the micro-integrated µECDL module have proven to fulfill the phasematching tolerances for efficicient frequency conversion in a BBO crystal. In the next chapter, both systems are applied as pump sources for single-pass SHG down to 222.5 nm.

parameter symbol LD macro ECDL µECDL

maximum injection current Imax 1.2 A 0.6 A 1.2 A maximum output power P 1.6 W 0.55 W 1.43 W emission wavelength λ ≈ 445 nm 443-447 nm 445 nm emission bandwidth (FWHM) ∆λ 1...2 nm < 40 pm < 40 pm

coarse tuning range ∆λcoarse - 4 nm -

suppression ratio at Imax SR - 42 dB 48 dB 2 2 beam quality (fast axis) M4σ /M90% 2.8 / 1.1 4.0 / 2.1 2.7 / 1.1 2 2 beam quality (slow axis) M4σ /M90% 5.9 / 4.5 7.7 / 4.7 7.5 / 4.5

Table 4.7: Summary of the electro-optical, spectral and spatial parameters of the free- running laser diode (LD), the macroscopic ECDL and the µECDL module. 5 Compact deep ultraviolet laser light source

This chapter presents the results obtained with single-pass second harmonic generation using the macroscopic ECDL in Littrow configuation and the micro-integrated µECDL module as pump sources. The setup with the macroscopic ECDL system again serves as a proof-of-principle to evaluate the feasibility of the concept and to develop an idea of the critical aspects, that need to be considered for efficient frequency conversion. The development of the single-pass SHG setup with the macroscopic ECDL as pump source is presented in section 5.1. The first challenge of the experiments arose from the need to develop a setup, with which deep ultraviolet output powers even in the lower µW range can be reproducibly detected and measured. This aspect is discussed in section 5.1.1. First SHG results and the influence of different focusing conditions on the conversion efficiency are discussed in section 5.1.2. Here, it is also examined how far the simulated curves for the phase matching tolerances can be reproduced and the differences between theory and experiment will be discussed.

As the the technical goal of this work is the development of a compact DUV laser light source, the more compact µECDL is used as pump source in section 5.2. With a better understanding of the setup, the focusing conditions are now optimized, and the achieved results are presented in section 5.2.2. It is discussed if and how far the results can be compared with the Boyd-Kleinman analysis for focused Gaussian beams given the fact of the elliptical beam proﬁle and the relatively poor beam quality of the laser diode emission especially in the lateral axis.

71 72 5 Compact deep ultraviolet laser light source

5.1 Proof-of-concept setup with macroscopic ECDL as pump source A schematic view of the experimental arrangement is given in figure 5.1. The macroscopic ECDL in Littrow configuration emitting at 445 nm described in section 4.2 serves as the pump source for the single-pass frequency doubling setup. The laser diode polarization is oriented parallel to the grating grooves along the y-axis indicated by the coordinate system in figure 5.1. This is also the correct alignment for the type-I phase matching applied in this setup where the fundamentals polarization has to be perpendicular to the phase matching plane, which, as indicated in figure 4.2, is the xz-plane in this setup. The collimated beam from the ECDL has an elliptic beam shape of 1.1 mm x 2.8 mm in lateral (y-axis) and vertical (x-axis) direction, respectively.

All lenses applied in front of the crystal are AR-coated with R ≈ 10−3 for 450 nm. Lens L1 and L2 are two cylindrical lenses with a focal length of f = 22 mm (L1) and f = 76 mm (L2), respectively. They are used as a telescope arrangement to expand the beam in lateral direction, in the exemplary case shown here to a lateral beam width of 4.6 mm. This is 2 2 done for several reasons: First, the relatively high lateral M value of M90% = 4.7 makes the lateral expansion necessary to enable a smaller beam waist diameter inside of the 2 crystal. Here and in the following the 90%-knife edge beam diameter M90% values are used, because they seem to be more appropriate in the context of the SHG process (see section 3.6). Secondly, by changing the magnification factor of the lateral telescope with different lenses L1 and L2, different focusing conditions can be tried out to optimize the SHG conversion efficiency. Moreover, after the lateral beam expansion the beam is focused into the BBO crystal by a spherical achromatic lens (L3). The lateral expansion minimizes the relatively large residual divergence in lateral direction of the ECDLs output beam from 0.18◦ to 0.04◦ (simulated with WinABCD). Thereby it is assured that the focal points in lateral and vertical direction after focusing with lens L3 are at the same position along the z-axis inside the BBO crystal.

For lens L3, different focal lengths of 50 mm, 75 mm, 100 mm, and 150 mm are applied to gain an idea of the effect of different focusing conditions on the SHG conversion efficiency and compare the experimentally generated beam waist radii with the expected optimum beam waist radius and focusing condition predicted by the Boyd-Kleinman analysis.

Photodiode, Spectrometer

Focusing Lens ECDL Beam Focusing 222.5 nm Shaping Lens

BBO Beam 445 nm Dump

CaF Prism L1 L2 L3 2 x 445 nm Collimation y,E Lens z

Figure 5.1: Schematic top view of the single-pass frequency doubling setup with the macroscopic ECDL emitting at 445 nm as pump source. 5.1 Proof-of-concept setup with macroscopic ECDL as pump source 73

The BBO crystal applied here has a length of 7.5 mm and an entrance aperture of 4 mm x 4 mm. The front and back facets are AR-coated for the fundamental and the second harmonic wavelength, respectively. As specified by the supplier (Cryslaser Inc.), the ◦ crystal was cut at a phase matching angle of θPM = 63.6 for a phase matching wavelength of λ = 449 nm at a temperature T = 25◦C. To assure stable phase matching conditions, the crystal is put into a crystal oven (Eksma Optics), and the temperature is stabilized at T = 50◦C. The oven is mounted on a manual 3-axis alignment stage (Newport Corporation, model: 562F-XYZ + model: SM-25) with a specified sensitivity of 1µm. An additional manual rotation stage (Newport Corporation, model: M-UTR80S) specified with a sensitivity of 4 arc sec or 0.001◦ is used for the alignment of the phase matching angle. The crystal center is positioned in the beam focus. To achieve phase matching for the fundamental wavelength of 445 nm at a crystal temperature of 50◦C, the crystal is ◦ ◦ angle-tuned by 1.4 to a phase matching angle of θPM = 65 .

The radiation behind the crystal is collimated by an uncoated UV fused silica lens with a focal length of f = 75 mm.A 10 mm thick UV fused silica plate is speciﬁed by the supplier (Thorlabs Inc.) to have a transmission of 90.5% at 222 nm.A CaF2 prism with a speciﬁed transmission > 90% at 222 nm (Thorlabs Inc.) is used to spatially separate the fundamental from the generated second harmonic light. The fundamental is blocked by a beam dump and the second harmonic UV beam is focused onto a SiC photodiode by an UV fused silica lens with a focal length of f = 50 mm to determine the generated DUV output power. The DUV light generates a photocurrent in the photodiode, that depends on its spectral sensitivity and is measured with a picoammeter (Keithley Instruments).

To measure spectra of the generated DUV light, the SiC photodiode is replaced by a spectrometer (Newport MiniSpec 78355) having a spectral resolution of ≈ 1.3 nm over its entire wavelength range. Unfortunately, there was no spectrometer with a higher resolution in the DUV wavelength range available in the laboratory at the time of this work, so that the actual emission bandwidth of the generated DUV light could not be resolved. 74 5 Compact deep ultraviolet laser light source

5.1.1 Detection of deep ultraviolet light As estimated from the Boyd-Kleinman theory, the expected deep ultraviolet output power, that can be achieved in a single-pass second harmonic generation setup with a BBO crystal, lays between tens and a few hundreds of microwatts. Measuring optical output powers in such low power regimes is challenging. Additionally, the deep ultraviolet light is invisible to the naked eye impeding an easy laser beam adjustment to an optical power detector. For the initial coarse alignment of the crystal position and phase matching angle in the frequency doubling stage even lower optical output powers in the nW range need to be detected.

For the initial alignment of the phase matching angle of the BBO crystal and for the adjustment of the DUV laser beam, a fluorescence glas filter, that is doped with the highly fluorescent Rare Earth terbion ion Tb3+ (Lumilass G9, Sumita Optical Glass Europe GmbH), is used. This material can be excited in a wavelength range from 200 nm to 390 nm and has a fluorescent emission in the green spectral range with an intensity peak at λ = 542 nm. As specified by the supplier, the filter has a minimum sensitivity of < 1 µW/cm2 for an excitation wavelength around the absorption maximum of about 250 nm. This means that for a spot size of 1 mm x 1 mm an optical power of 10 nW could be detected. During the experiments it was possible to see green fluorescence on the filter with the naked eye for an optical output power at 222.5 nm as low as approximately 100 nW.

After a rough adjustment of the phase matching angle the initial DUV light is then used to adjust the DUV laser beam and direct it onto the photodiode. Fine adjustment of the phase matching angle can then be done by monitoring and maximizing the photocurrent that is generated by the DUV laser light.

Besides the low DUV output powers, that need to be measured, another obstacle to the reproducible and reliable power measurement occured with the experimental arrangement.

Figure 5.2: a) Spectral sensitivity of the used SiC photodiode (sglux SolGel Technologies, model SG01XL) and for comparison of a conventional UV extended Si photodiode (Thorlabs Inc., model S130VC). b) Spectral sensitivity of the applied SiC photodiode in logarithmic scale. Data taken from the suppliers calibrations. 5.1 Proof-of-concept setup with macroscopic ECDL as pump source 75

Even though the separation of fundamental and second harmonic light by the CaF2 prism works suﬃciently good, there is still random stray light from the fundamental even in the optical path of the DUV laser beam. This random stray light is measured with a silicon (Si) photodiode positioned in the DUV beam to be in the lower mW range. This is still higher than the expected amount of generated DUV laser light and therefore impedes a clear and reliable DUV power measurement with a conventional Si photodetector.

For this reason, a solar-blind photodiode based on silicon carbide (SiC) provided by the company sglux SolGel Technologies is used for the determination of the generated DUV output power. The advantage of the SiC photodiode over conventional Si based photodiodes for our measurements can be seen in ﬁgure 5.2.a). It shows the spectral sensitivity of the used SiC photodiode (sglux, model SG01XL) and for comparison the spectral sensitivity of an exemplary conventional UV extended Si photodiode (Thorlabs Inc., model S130VC) between 200 nm and 600 nm. The data for the SiC photodiode is taken from the manufacturer calibration and for the Si photodiode from the website of the provider [150].

The spectral sensitivity of the Si photodiode is higher over the whole wavelength range. However, the SiC photodiode has its highest sensitivity in the DUV wavelength range with a peak sensitivity at 286 nm and has an extremely low sensitivity at the fundamental wavelength of 445 nm. This can be seen in detail on a logarithmic scale in figure 5.2.b). At the fundamental wavelength λ = 445 nm, the sensitivity is 7.2 · 10−8A/W, compared to 6.7 · 10−3A/W at the second harmonic wavelength λ = 222.5 nm, which means a difference of five orders of magnitude. At 222.5 nm, 100 µW result in a photocurrent of 670 nA. Even an optical power of 100 mW of the fundamental beam would generate a photocurrent of only 7.2 nA, which is about 1% of the photocurrent resulting from 100 µW of DUV light and could therefore be neglected.

In conclusion, even though the absolute spectral sensitivity of the SiC photodiode is lower than that of conventional Si photodiodes, its strong wavelength discrimination or "solar-blindness" enable a reliable DUV power measurement and the fundamental stray light can be neglected. 76 5 Compact deep ultraviolet laser light source

5.1.2 Investigation of conversion eﬃciency and phase matching tolerances

Figure 5.3 shows the optical output power, here named pump power Ppump, versus the injection current of the ECDL system emitting at 445 nm. The fundamental pump power is measured behind the mirror, that redirects the output beam of the ECDL. The difference to the power characteristics curve in figure 4.10, is that the curve shown here is measured manually in steps of 20 mA during the SHG experiment. Therefore the values differ somewhat to the earlier power characteristics as the ECDL is adjusted before each DUV power measurement with the goal of a maximized DUV output power.

At an injection current of I = 800 mA, the ECDL system has an output power of 680 mW. The lasing threshold is IECDL-th = 130 mA and the slope efficiency is measured to be SECDL = 1.0 W/A. Inset A and B in figure 5.3 show optical spectra (see also fig. 4.11) of the ECDL emission measured with the optical spectrum analyzer for an injection current of 0.6 A and 0.8 A, corresponding to an output power of 0.47 W and 0.68 W, respectively. At 0.47 W, the suppression of the ASE and longitudinal laser diode modes (SR) reaches a maximum of 42 dB and decreases from thereon to 19 dB at 0.68 W (see figure 4.11 in section 4.2.3).

Figure 5.3: Pump power Ppump of the macroscopic ECDL system emitting at 445 nm as a function on the injection current. Inset A: ECDL emission spectrum at 0.6 A (0.47 W). Inset B: ECDL emission spectrum at 0.8 A(0.68 W).

The spectrum of the ECDL emission is also observed during the SHG experiment using the double-echelle monochromator. For output powers up to 0.4 W corresponding to an injection current of 0.53 A, the ECDL emission shows a spectral width of ∆λ ≤ 20 pm (FWHM). For higher output powers, the spectral emission of the ECDL exhibits a slight broadening. Figure 5.4 shows two emission spectra of the ECDL at an output power of a) 0.47 W (I = 0.6 A) and b) 0.68 W (I = 0.8 A). The spectral width of the ECDL emission increases from 25 pm to 64 pm (FWHM) between the two working points. These spectra 5.1 Proof-of-concept setup with macroscopic ECDL as pump source 77

Figure 5.4: Emission spectra of the ECDL system at a) I = 0.6 A and b) I = 0.8 A corresponding to an output power of I = 0.47 W and I = 0.68 W, respectively at a heatsink temperature of T = 20◦C. also represent working points at which the SHG conversion efficiency was maximized by adjusting the ECDL. The comparison between the earlier presented ECDL spectra (see figure 4.11) leads to the conclusion that a slightly broader pump spectrum with several longitudinal modes oscillating is favorable regarding a higher conversion efficiency. This finding is also in agreement with observations in other publications, where several longitudinal modes within the acceptance bandwidth of the nonlinear crystal caused an enhancement of the frequency doubling efficiency compared to a longitudinal single-mode pump source [151, 152]. 78 5 Compact deep ultraviolet laser light source

Conversion efficiency and focusing conditions After the lateral beam expansion, the beam diameters are 4.6 mm x 2.8 mm in lateral and vertical direction, respectively. In this setup, the vertical plane of the laser diode is the phase matching (PM) plane considered by the Boyd-Kleinman analysis to find the optimum focusing conditions, assuming a non-diffraction limited spherical beam. The lateral beam expansion effects the conversion efficiency insofar as it leads to a higher power density in the focal region inside the BBO crystal due to the enabled stronger focusing in this direction. The drawback however is a high beam divergence after focusing, which restricts the high power density to a relatively small region inside the BBO crystal.

Due to the different beam propagation factors of the ECDL emission in the vertical and lateral plane it is impossible to generate symmetric spherical focusing inside the BBO crystal, which limits the comparibility of the experimental results with the predictions from the Boyd-Kleinman analysis. The values for the optimum beam waist radius and the predicted SHG output power obtained from the Boyd-Kleinman analysis can therefore only serve as an initial reference point. In general, there is no analytical solution to exactly calculate the optimum focusing conditions and expected SHG output power for a non-diffraction limited and highly elliptical beam as applied in this experiment. The analysis of the optimum focusing parameters for this experiment is therefore restricted to a qualitative discussion of the achieved SHG output powers with different beam shaping optics in comparison to the Boyd-Kleinman theory (see section 2.4).

For a first assessment of the influence of different focusing conditions on the conversion efficiency, different focal lenghts for lens L3 (see figure 5.1) of 50 mm, 75 mm, 100 mm, and 150 mm are applied. The resulting beam waist radii are simulated with WinABCD and listed in table 5.1. For the simulation, the beam propagation factors are assumed to be M2 = 2 1 in the PM plane and M2 = 4 7 in the plane perpendicular to the PM,90% . ⊥,90% . 2 PM plane. These values correspond to the M90% values from the caustic measurements of the ECDL emission using the 90%-knife-edge beam diameter method (see figure 4.14). The resulting beam waist radii in the PM plane are in the vicinity of the Boyd-Kleinman predicted optimum of 15 µm for a BBO crystal with a length of 7.5 mm.

In the perpendicular plane, the resulting diameters are larger. A stronger lateral beam expansion would result in smaller beam waist radii but also in a larger beam divergence. As explained in section 2.4, a trade-oﬀ between strong focusing and a maximized interaction

fL3 w0,PM w0,⊥

50 mm 11 µm 15 µm 75 mm 16 µm 22 µm 100 mm 21 µm 29 µm 150 mm 32 µm 43 µm

Boyd-Kleinman 15 µm 15 µm

Table 5.1: Simulated beam waist radii in the phase matching plane (PM) and perpendicular to the phase matching plane for diﬀerent focal lengths of lens L3. 5.1 Proof-of-concept setup with macroscopic ECDL as pump source 79

Figure 5.5: Generated SHG power at 222.5 nm versus pump power at 445 nm for a focal length of lens L3 of 50 mm, 75 mm, 100 mm, and 150 mm. Dashed curves are the quadratic ﬁts of the data points. The solid black curve shows the expected SHG power for a focused Gaussian beam according to the Boyd-Kleinman analysis. length has to be found. The lenses applied here can be seen as a ﬁrst approach to the optimum focusing condition.

Figure 5.5 shows the second harmonic output power PSHG at 222.5 nm versus the fundamental pump power Ppump at 445 nm for lens L3 having a focal length fL3 of 50 mm, 75 mm, 100 mm, and 150 mm. The experimental data points for each measurement are 2 fitted with the function PSHG = η · Ppump according to the quadratic conversion efficiency behavior without pump depletion [74]. The quadratc fits give values for the normalized conversion efficiency, that are summarized in table 5.2.

fL3 η

50 mm 2.4 · 10−5 W−1 75 mm 3.4 · 10−5 W−1 100 mm 4.2 · 10−5 W−1 150 mm 4.1 · 10−5 W−1

Boyd-Kleinman 13.8 · 10−5 W−1

Table 5.2: Normalized SHG conversion eﬃciency η for the diﬀerent focal lengths of lens L3.

For a pump power of 680 mW, a maximum UV power of 16 µW is generated with lens L3 having a focal length of f = 100 mm. The corresponding normalized conversion efficiency −5 −1 is η100mm = 4.2 · 10 W . However, a focal length of 150 mm leads to almost the same −5 −1 second harmonic output power with a conversion efficiency of η150mm = 4.1 · 10 W . The theoretically expected conversion coefficient according to the Boyd-Kleinman analysis −5 −1 is ηBK = 13.8 · 10 W with an expected optimum beam waist radius of w0,BK = 15 µm. 80 5 Compact deep ultraviolet laser light source

The simulated beam waist radii for fL3 = 100 mm and fL3 = 150 mm in the vertical phase matching plane are w0,PM = 21 µm and w0,PM = 32 µm, respectively. It can be estimated that the optimum beam waist radius in the PM plane for this focusing condition is somewhere between 21 µm and 32 µm and therefore larger than the theoretical optimum of 15 µm. For fL3 = 50 mm and fL3 = 75 mm, the conversion eﬃciency decreases as the focusing in the PM plane is too strong.

The achieved conversion efficiency is a factor of 3.3 smaller than the Boyd-Kleinman prediction. This is attributed to the fact, that the ECDL beam is highly elliptical after the lateral beam expansion (aspect ratio of d0,PM : d0,⊥ = 1.6 : 1) and additionally the beam is not diffraction limited with different beam propagation factors in both transverse directions. A comparison to the Boyd-Kleinman analysis is therefore somewhat difficult and can only be made in a qualitative manner. In section 5.2, when the µECDL is used as pump source, the optimum focusing conditions for this setup will be discussed in more detail.

In the SHG versus pump power curves in figure 5.5, an increasing deviation of the 2 experimental data points from the quadratic fit PSHG = η · Ppump can be observed for a pump power higher than 0.47 W, indicated by the first dashed line (A) in figure 5.5. This behaviour can be explained with the ratio of the ASE and FP mode suppression (SR) of the ECDL emission, that decreases significantly between a pump power of 0.47 W (A) and 0.68 W as was illustrated in the insets of figure 5.3. As examined in section 4.2.3 the ratio of the optical power PRECDL that is contained in the ECDL peak is therefore reduced from 99.7% at a pump power of 0.47 W (0.6 A) to 88.9% at a pump power of 0.68 W (0.8 A). Hence, the actual power within the ECDL peak at the maximum applied current is only about 0.61 W. Using this value with the obtained quadratic fits gives SHG powers that are close to the experimentally measured DUV output powers at 0.68 W. This indicates, that the reduced SR value is the main reason for the reduced SHG conversion efficiency for higher pump powers. On the other hand, the slight broadening of the spectral emission bandwidth of the ECDL for higher pump powers seems to have a comparatively small effect on the SHG conversion efficiency.

Figure 5.6: Spectrum of the generated SHG light at λSHG = 222.5 nm and of the residual stray light from the fundamental beam at λpump = 445 nm. 5.1 Proof-of-concept setup with macroscopic ECDL as pump source 81

Figure 5.6 shows a spectrum of the generated second harmonic emission at λ = 222.5 nm and the residual stray light from the fundamental at 445 nm at a pump power of 0.4 W, corresponding to a generated DUV power of 7 µW. The intensities at both wavelengths are normalized to 1. The spectrometer (Newport MiniSpec 78355) has a spectral resolution of approximately 1.3 nm over the entire wavelength range. Therefore, the actual spectral width of the UV emission could not be determined. However, it is known from the analysis of the ECDL emission characteristics, that its spectral bandwidth does not exceed 20 pm for output powers up to 0.4 W. It is therefore reasonable to assume, that the spectral width of the generated DUV light is in the range of the spectral width of the ECDL emission of about ∆λDUV < 20 pm up to 7 µW and about ∆λDUV < 64 pm up to 16 µW DUV output power.

Temperature acceptance bandwidth of the BBO crystal for a focused pump beam

For the setup with lens L3 having a focal length of fL3 = 150 mm, the generated SHG intensity as a function of the BBO crystal temperature TBBO is exemplary measured at a moderate pump power of Ppump = 240 mW (I = 380 mA). With the focusing conditons applied here, a beam waist in the phase matching plane having a simulated radius of 32 µm is generated. The crystal oven temperature is calibrated using a pre-calibrated PT1000 temperature sensor. The generated SHG power is maximized for a crystal temperature of about 50◦C. For the measurement, the crystal temperature is then tuned from 30◦C to 70◦C in steps of about 1.3 K.

Figure 5.7.a) shows the normalized SHG intensity as a function of the crystal temperature TBBO. The red dots are the measured data points. As can be seen, the phase matching temperature acceptance bandwidth is ∆T = 26 K (FWHM), which is much wider than the theoretically expected temperature acceptance bandwidth of ∆T = 4.2 K for a BBO

Figure 5.7: a) Measured normalized SHG intensity at 222.5 nm as a function of the crystal temperature TBBO for a 7.5 mm long BBO crystal. Experimental data points are represented by the red squares. b) sinc2 fit (solid line) of the experimental data from a) using an effective crystal length of 1.2 mm. 82 5 Compact deep ultraviolet laser light source crystal with a length of 7.5 mm using the known Sellmeier equations for BBO [72] (see figure 2.7 in section 2.3). This deviation between the experimentally observed and the theoretically calculated temperature acceptance bandwidth is thought to originate from the spatial walk-off of the second harmonic and the fact that the sinc2 calculations are carried out in the plane-wave approximation whereas the beam in the experiment is focused. This results in a significantly decreased effective interaction length for focused beams which leads to wider phase matching temperature, angle, and wavelength acceptance bandwidths.

Boyd and Kleinman have defined an effective aperture length [74] to: √ πw0 La,eff = , (5.1) ρ where w0 is the beam waist radius and ρ the walk-off angle. For the focusing condition discussed here with w0 = 32 µm and ρ = 4°, this effective aperture length is calculated to be La,eff = 0.8 mm. The red line figure 5.7.b) shows a sinc2 fit of the data according to equation (2.17) having a FWHM of ∆T = 26 K with the crystal length set to a value of La,eff = 1.2 mm.

The deviation between the calculation using equation (5.1) and the sinc2 function are attributed to uncertainties in the simulation of the relevant beam waist radius, i.e. the relevant M2 value in the PM plane. For example a beam waist radius of 47 µm (implying 2 MPM = 3) applied in equation 5.1 would lead to an effective aperture length of 1.2 mm. Given the mentioned difficulties with the precise measurement of the M2 values, it seems justified to use M2 = 3 and thereby have a good agreement between theory and experiment. This result is also qualitatively in good agreement with the temperature acceptance bandwidth measured by Kumar et al. of ∆T = 33 K for a beam waist radius of 55 µm inside a 10 mm long BBO crystal [153].

Figure 5.8: Phase-matching wavelength acceptance bandwidth (a) and phase matching angle acceptance bandwidth (b) calculated in the plane-wave approximation for a crystal ◦ length of La,eﬀ = 1.2 mm (TBBO = 50 C). 5.1 Proof-of-concept setup with macroscopic ECDL as pump source 83

As a consequence, one must note that the eﬀective interaction length inside the BBO crystal and thereby the temperature acceptance strongly depends on the chosen focusing conditions. This not only eﬀects the temperature acceptance bandwidth but also the phase matching angle and the wavelength acceptance bandwidth in BBO. Figure 5.7 shows the calculated phase matching and wavelength acceptance bandwidths in the plane-wave approximation for a crystal length of 1.2 mm. The phase matching wavelength acceptance bandwidth in this case would be ∆λ = 250 pm (FWHM), compared to the earlier calculated ∆λ = 42 pm (FWHM). And the FWHM phase matching angle acceptance bandwidth would be ∆θ = 0.08◦ (1.4 mrad), compared to the earlier calculated ∆θ = 0.013◦ (0.2 mrad). 84 5 Compact deep ultraviolet laser light source

Temperature acceptance for a parallel beam in the PM plane It has become clear on the previous pages that the phase matching temperature acceptance bandwidth of the BBO crystal in the plane-wave approximation only holds true for a parallelized beam in the phase matching plane. This is evaluated using the setup illustrated in ﬁgure 5.9.a). The beam from the ECDL system is compressed in the phase matching plane by a telescope arrangement consisting of two cylindrical lenses L1 and L2 with focal length of fL1 = 73 mm and fL2 = 40 mm, respectively, so that the beam diameter is smaller than the BBO crystals entrance aperture of 4 mm. Another cylindrical lens L3 with a focal length of fL3 = 40 mm is used to focus the beam into the crystal in the plane perpendicular to the phase matching plane to achieve a higher power density inside the crystal.

Figure 5.9.b) shows the normalized SHG intensity as a function of the crystal temperature. The red dots represent the experimental data. The dashed curve represents the calculated temperature acceptance bandwidth according to the plane-wave approximation for a BBO crystal with a length of LBBO = 7.5 mm, having a FHWM of ∆T = 4.2 K. It can indeed be seen that the experimental data shows good agreement with the theoretically expected acceptance bandwidth for a parallelized beam. The origin of the higher intensities on the low temperature tail is not clear yet and still under investigation. a) b)

lateral plane photodiode ECDL

L1 L2 L3 crystal oven vertical, PM plane ΔT ECDL BBO

Figure 5.9: a) Setup for the measurement of the temperature acceptance with a parallelized beam in the phase matching plane. b) Resulting phase matching temperature acceptance curve. The dots are the experimental data and the solid curve is the simulated

temperature acceptance in BBO for a crystal length of LBBO = 7.5 mm.

In conclusion, it can be stated that the stronger the focusing in the phase matching plane the larger are the phase matching tolerances for the pump wavelength, the crystal temperature, and the phase matching angle. The increased wavelength phase matching tolerance for a focused beam also implies that the decreasing conversion eﬃciency for output powers above 0.47 W observed in ﬁgure 5.5 is more likely attributed to the decrease of the SR value than to the slight increase of the emission bandwidth of the ECDL. However, in regard to the targeted applications it is still favorable to achieve an emission bandwidth as narrow as possible. 5.2 Micro-integrated ECDL module as pump source 85

5.2 Micro-integrated ECDL module as pump source The single-pass frequency doubling setup with the macroscopic ECDL as pump source has shown, that further optimization of the focusing conditions is necessary to increase the conversion efficiency of the SHG process. However, these first experiments also give an indication, that the Boyd-Kleinman theory can not be used to precisely calculate the optimum beam waist diameter inside the BBO crystal and to predict the generated UV power. This is due to the strong elliptical beam shape of the ECDL emission as well as its non-diffraction limited nature in both transverse directions.

5.2.1 Compact deep ultraviolet laser light source The experimental arrangement for the single-pass frequency doubling with the µECDL as pump source is illustrated in figure 5.10 in a side and top view. Except for the different pump source and different beam shaping optics, i.e. the lenses have different focal lengths, the setup is the same as the previous one. For simplicity, the detection path with the separation of fundamental and second harmonic is not included here but is also identical to the arrangement with the macroscopic ECDL as pump source presented in figure 5.1.A detailed description of the µECDL and its electro-optical, spectral, and spatial properties were presented in section 4.3.

The collimated beam from the µECDL module has an elliptic shape with a beam diameter of dvert = 1.8 mm and dlat = 0.7 mm in the vertical and the lateral plane, respectively. For the critical type-I phase matching applied here, the laser diode polarization has to be perpendicular to the phase matching plane of the BBO crystal. Therefore, the TE-polarized (TE:TM = 100:1) laser diode is positioned with its fast axis (vertical plane) parallel to the phase matching plane, and the vertical plane is denoted phase matching (PM) plane in the following. The pump beam is again expanded by two cylindrical lenses L1 and L2 forming a telescope in the lateral plane, which is perpendicular to the phasemathing plane of the BBO crystal. The beam is then focused into the BBO crystal with a spherical lens (L3). For the longest focal length applied for lens L3 of 150 mm, the optical path length between the laser diode and the output facet of the BBO crystal is approximately Lopt = 30 cm.

mECDL 445 nm side view 222.5 nm (lateral plane) L1 BBO L2 L3

Lopt

LD VBG 445 nm top view 222.5 nm (vertical / pm plane) L1

Figure 5.10: Schematic side and top view of the single-pass SHG arrangement with the µECDL module as pump source. 86 5 Compact deep ultraviolet laser light source

For the experiments with the µECDL, a new crystal oven was manufactured at the FBH1. The BBO crystal is placed into the oven and is this time stabilized to a temperature of T = 40◦C to assure stable phase matching conditions.

For the convenience of the reader, figure 5.11 again shows the optical pump power of the µECDL emission as a function of the injection current up to I = 1.2 A. The inset in figure 5.11 shows an emission spectrum of the µECDL module at an injection current of 1.2 A, corresponding to an output power of 1.4 W with a central emission wavelength of 445 nm (measured with an optical spectrum analyzer, Yokogawa AQ6373). The suppression of the ASE and the longitudinal laser diode modes (SR) is measured to be 48 dB. This exemplary emission spectrum illustrates that in contrast to the spectral behavior of the macroscopic ECDL system in Littrow configuration, the SR for the µECDL module stays almost constant up to the maximum output power of Ppump = 1.4 W (I = 1.2 A), (see also figure 4.26.A in section 4.3).

Figure 5.11: Optical pump power of the µECDL module as a function of the injection current for T = 20◦C heatsink temperature. Inset: Emission spectrum of the µECDL module at an injection current of 1.2 A, corresponding to an output power of 1.4 W at the emission wavelength of 445 nm (measured with a high dynamic range optical spectrum analyzer, Yokogawa AQ6373).

1 The manufacturing of the crystal oven was done by the FBH colleagues Thomas Roos, Detlef Grimpe, and Bastian Deutscher. 5.2 Micro-integrated ECDL module as pump source 87

5.2.2 Optimization of the focusing conditions As already mentioned, the asymmetry of the µECDL beam in terms of beam quality and beam diameter makes it difficult to easily find the optimum focusing condition by simply realizing the beam waist radius recommended by the Boyd-Kleinman (BK) analysis using a spherical focusing lens. The optimum beam waist radius in the vertical, i.e. phase matching plane (PM), may be different to the optimum beam waist radius in the lateral plane in this experiment. For this reason, three different lateral telescopes are applied to expand the beam in the lateral plane. As mentioned, the expansion is necessary to reduce the residual divergence and to enable smaller beam waist radii in the lateral plane inside the BBO crystal. The aim of this approach is to find the best pair of lateral and PM plane focusing conditions and thereby maximize the SHG output power.

Table 5.3 gives an overview of the focal lengths of the cylindrical lenses L1 and L2 applied for the three lateral telescopes. Beam expansion (1) leads to a lateral magniﬁcation factor of Xlat = 4.5, expanding the lateral beam diameter from 0.7 mm to 3.1 mm. For beam expansion (2), the magniﬁcation factor is Xlat ≈ 10, leading to a lateral beam diameter of 6.6 mm. And expansion (3) results in Xlat ≈ 13 and dlat = 8.8 mm. The resulting aspect ratios dPM : dlat are listed in the table as well.

beam expansion fL1 fL2 Xlat dPM dlat dPM : dlat

w/o 1.8 mm 0.7 mm 1 : 0.4 (1) 22 mm 98 mm 4.5 1.8 mm 3.1 mm 1 : 1.7 (2) 7.55 mm 74 mm 10 1.8 mm 6.6 mm 1 : 3.7 (3) 7.55 mm 98 mm 13 1.8 mm 8.8 mm 1 : 4.9

Table 5.3: Overview of the diﬀerent applied lateral beam expansions. fL1, fL2: focal length of lens L1 and L2. Mlat: lateral magniﬁcation factor. d0,PM, d0,PM: beam diameter in phase matching and lateral plane. d0,PM : d0,lat: aspect ratio. After the lateral expansion, the beam is then focused into the BBO crystal by a spherical lens (L3) for which focal lengths of 50 mm, 75 mm, 100 mm, and 150 mm are applied in each case. For the SHG measurement, the pump power Ppump is increased in steps of 100 mW up to the maximum output power of the µECDL of 1.4 W and for each step the the DUV output power at 222.5 nm is measured with the SiC photodiode.

Figure 5.12 shows the generated SHG power PSHG at 222.5 nm for the three applied lateral beam expansions and the different focal lengths of lens L3 as a function of the pump power Ppump at 445 nm. The black solid curve shows the SHG output power predicted by the Boyd-Kleinman analysis for a beam waist radius of w0 = 15 µm leading to a conversion efficiency of η = 13.8 · 10−5 W−1. The experimental data points for each measurement 2 are again fitted with the function PSHG = η · Ppump according to the quadratic conversion efficiency behavior without pump depletion [74], and are illustrated by the dashed curves. The numerical fits give the values for the normalized conversion efficiencies η summarized in table 5.4. The highest DUV output of PSHG = 160 µW for a pump power of Ppump = 1.4 W is achieved with the lateral beam expansion (3) and lens L3 having a focal length of f = 150 mm. The normalized conversion efficiency for this configuration 88 5 Compact deep ultraviolet laser light source

Figure 5.12: Generated SHG power at 222.5 nm versus the pump power of the fundamen- ◦ tal at 445 nm for all applied focusing conditions (TBBO = 40 C). The dashed lines are the quadratic ﬁts according to = 2 . The black solid curves show the SHG output PSHG η · Ppump power predicted by the BK analysis.

−5 −1 is η(3),150mm = 7.9 · 10 W . This is a factor of 1.9 higher than the maximum achieved −5 −1 conversion efficiency with the macroscopic ECDL as pump source (ηmacro = 4.2 · 10 W ) and only a factor of 0.57 below the conversion efficiency predicted by the Boyd-Kleinman analysis. In contrast to the macroscopic ECDL, the µECDL exhibits no decrease of the suppression ratio of the ASE and longitudinal laser diode modes for higher pump powers (see figure 5.11) and therefore no deviation between the quadratic fits and the measured data points at higher pump powers can be observed here.

The beam waist radii in the PM and the lateral plane w0,PM and w0,lat, the Rayleigh length in the PM and lateral plane zR,PM and zR,lat, and the consequential focusing parameters in the PM and lateral plane ξPM and ξlat, respectively, are also summarized in table 5.4 on the following page for all applied conﬁgurations. For comparison, the parameters resulting from the BK analysis are listed as well.

The geometrical beam parameters are determined by a caustic measurement according to ISO 11146 at the position of the BBO crystal and by applying the previously described 90%-knife-edge beam diameter criterion. The Rayleigh lengths inside the BBO 5.2 Micro-integrated ECDL module as pump source 89

−5 −1 beam expansion fL3 w0,PM w0,lat zR,PM zR,lat ξPM ξlat η / 10 W

(1) 50 mm 12 µm 25 µm 1.0 mm 1.5 mm 3.8 2.5 7.2

(Mlat = 4.5) 75 mm 17 µm 42 µm 2.2 mm 3.4 mm 1.7 1.1 6.4 100 mm 20 µm 55 µm 3.9 mm 6.0 mm 1.0 0.6 5.3 150 mm 30 µm 85 µm 8.9 mm 13.5 mm 0.4 0.3 3.8

(2) 50 mm 11 µm 13 µm 1.0 mm 0.3 mm 3.8 12.5 4.4

(Mlat = 10) 75 mm 18 µm 19 µm 2.2 mm 0.7 mm 1.7 5.4 5.6 100 mm 23 µm 27 µm 3.9 mm 1.3 mm 1.0 2.9 6.5 150 mm 35 µm 42 µm 8.9 mm 2.9 mm 0.4 1.3 7

(3) 50 mm 11 µm 10 µm 1.0 mm 0.2 mm 3.8 18.8 3.4

(Mlat = 13) 75 mm 15 µm 13 µm 2.2 mm 0.4 mm 1.7 9.4 4.8 100 mm 21 µm 17 µm 3.9 mm 0.7 mm 1.0 5.4 5.9 150 mm 30 µm 29 µm 8.9 mm 1.7 mm 0.4 2.8 7.9

Boyd-Kleinman 15 µm 15 µm 2.7 mm 2.7 mm 1.4 1.4 13.8

Table 5.4: Summary of the focus parameter for the diﬀerent applied focusing conditions. w0,PM: beam waist radius in the PM plane. w0,lat: beam waist radius in the lateral plane. zR,PM: Rayleigh length in the PM plane. zR,lat: Rayleigh length in the lateral plane. ◦ η: SHG conversion eﬃciency (LBBO = 7.5 mm, TBBO = 40 C). crystal are simulated using the parameters from the caustic measurements and assuming a refractive index of BBO at the pump wavelength of 445 nm of nBBO = 1.685. The 90%-knife-edge beam propagation factors from the caustic measurements are determined to be approximately M90%,PM ≈ 1.3 and M90%,lat ≈ 6 in the PM and the lateral plane, respectively. These values are also used for the optical simulations and in general a good agreement between experimentally measured and simulated parameters is observed. Here, it is worth noting that the beam propagation factors in the lateral plane have increased compared to the previous µECDL caustic measurements (µECDL at 1.4 W: M90%,lat = 4.5). This is attributed to the lateral telescope in which the small focal length of lens L1 leads to a large beam diameter at lens L2, that probably causes spherical aberrations.

The configuration with the lateral telescope (3) creates nearly the same dimensions in the beam waist in both planes. For fL3 = 150 mm, that yields the highest conversion efficiency, the beam waist radii are measured to be w0,PM = 30 µm in the PM plane, and w0,lat = 29 µm in the lateral plane. This indicates that compared to the BK analysis somewhat higher radii than w0,BK = 15 µm are beneficial in this experimental arrangement. This is also in accordance to earlier results by Übernickel et al., that made the same ob- servation for non-diffraction limited pump beams for SHG in periodically poled crystals [80].

To further elucidate the optimum focusing conditions, the conversion eﬃciencies for all applied focusing conﬁgurations as the function of the average beam waist radius

(w0,PM + w0,lat)/2 are plotted in figure 5.13. The dashed curve represents the predicted conversion efficiency by the BK analysis for a spherical Gaussian pump beam. A clear tendency for higher conversion efficiencies towards beam waist radii in the range between 90 5 Compact deep ultraviolet laser light source

Figure 5.13: SHG conversion eﬃciency η as a function of the average beam waist radius 0.5 · (w0,PM + w0,lat) for the three diﬀerent applied lateral beam expansions. For comparison, the dashed line represents the simulated Boyd-Kleinman function h for an idealized Gaussian beam.

20 µm and 30 µm can be observed, whereas the efficiency significantly decreases for smaller and larger beam waist radii. This is also in accordance with the results with the macroscopic ECDL as pump source. However, the plot as a function of the average beam waist radius does not account for the highly asymmetric nature of the µECDL pump beam and can therefore only be seen as a rough orientation. Additionally, due to the different lateral beam expansions, similar beam waist radii generated by different lateral beam expansion configurations lead to very different focusing parameters ξ.

ξ is deﬁned as ξ = Lcr/b with the crystal length Lcr and the confocal parameter b = 2zr and is a measure for the focusing strength and therefore also takes the strength of the beam divergence generated by the focusing into account. b can be seen as the distance over which the beams cross sectional area is relatively constant. For a strongly focused beam, b is small and consequently ξ is large. Accordingly, b is large and ξ is small for a weakly focused beam.

Nonetheless, due to the high lateral beam propagation factor, i.e. the poor lateral beam quality, a high ξ value is necessary to create a sufficiently small lateral beam waist radius. As mentioned earlier, optimum focusing conditions are achieved, if a good trade-off between a small beam waist radius with high power densities in the crystal center and a sufficiently long interaction length, i.e. a not to high ξ value or not to high beam divergence, is found. It can be seen that for beam expansion (1) with a moderate lateral magnification factor, the highest conversion efficiency is achieved with lens L3 having a focal length of 50 mm, as in this configuration a beam waist radius in the favorable range between 20 µm and 30 µm is achieved for smaller focal lengths. In the PM plane however, the beam propagation factor is almost 1, indicating that the BK analysis is more justified in this plane. Thus, the generated beam waist of w0 = 12 µm with beam expansion (1) and fL3 = 50 mm seems to be smaller than the optimum. 5.2 Micro-integrated ECDL module as pump source 91

Considering all applied focusing conditions but concentrating on the focusing condition with the highest conversion effiency (beam expansion (3) and fL3 = 150 mm), a ξ value slightly larger than the BK prediction but smaller than three (1.4 < ξlat < 3) seems to be favorable in the lateral plane, whereas in the PM plane, a beam waist radius only slightly larger than the BK prediction of 15 µm but a ξ value smaller than the BK prediction (ξPM < 1.4) seems to be beneficial. This finding is good agreement with previous works that studied the optimum focusing for highly elliptical but diffraction limited beams [77–79]. It was shown, that weaker focusing in the (PM) plane, reducing the negative walk-off effect, and strong focusing perpendicular to the PM plane can increase the efficiency in large walk-off crystals like BBO. Especially in the paper by Steinbach et al. it was found that a ξ value of ξPM = 0.25 in the PM plane and of ξlat = 3.3 perpendicular to the PM plane for the case of strong walk-off (B = 16) resulted in a conversion efficiency even slightly larger than predicted by the Boyd-Kleinman analysis for focused spherical Gaussian beams [78]. Taking into account that the beam in the experiment presented in this work is additionally not diffraction limited, this is in suprisingly good agreement with the ξ values with the highest conversion efficiency derived in this experiment of ξPM = 0.4 and ξlat = 2.8 (B = 14.7).

The inherent downside of this setup is that it is difficult to optimize the focusing conditions in both planes independently of each other. And as there is no straightforward analytical solution to this problem, the only feasible approach is to test many different beam geometry configurations. Nonetheless, given the fact of the asymmetry of the beam and the relatively poor beam quality it is believed that the achieved conversion efficiency of η = 7.9·10−5 W−1, which is a factor of 0.57 smaller than the Boyd-Kleinman prediction for a focused Gaussian beam, is already close to what is achievable with this pump source.

With regard to a preferably compact DUV laser light source however, the conﬁguration with beam expansion (1) and fL3 = 50 mm results in only a slightly lower maximum optical output power of 142 µW, but a 10 cm shorter optical path length and therefore more

Figure 5.14: a) Emission spectrum of the generated SHG radiation with a DUV power of 161 µW at 222.5 nm for a pump power at 445 nm of 1.4 W measured with a spectral resolution of ∆λ ≈ 1.3 nm. b) Emission spectrum of the µECDL at a pump power of 1.4 W measured with a spectral resolution of 6 pm. 92 5 Compact deep ultraviolet laser light source compact light source compared to the configuration with the highest conversion efficiency. For the sake of completeness, figure 5.14.a) shows a spectrum of the generated DUV light at λ = 222.5 nm with an output power of 161 µW for a pump power of 1.4 W. It has an emission bandwidth of 1.3 nm again limited by the resolution of the spectrometer (Newport MiniSpec 78355). However, it should be at least in the range of the spectral width of the µECDL emission of ∆λ = 32 pm (FWHM) and ∆λ95% = 152 pm at a pump power of 1.4 W shown in figure 5.14.b). 6 Conclusion and Outlook

A compact laser light source emitting in the wavelength range between 210 nm and 230 nm is of great interest for numerous applications, especially outside the laboratory environment. In this thesis, a compact laser light source emitting around 222 nm was developed based on single-pass frequency doubling of a commercially available high-power GaN laser diode emitting in the blue spectral range. As a pump source for the frequency doubling stage, a laser diode with high optical output power above 1 W with narrowband emission in the range of the acceptance bandwidth of the applied nonlinear BBO crystal was required. Since GaN based high-power laser diodes exhibit a broad emission spectrum of ∆λ = 1...2 nm, wavelength stabilization and narrowing by the use of an external wavelength selective element was necessary.

First, an external cavity diode laser system with a surface diffraction grating as external element was realized as a proof-of-concept study. It was found that feeding as much as 15% of the laser diode radiation back into the laser is sufficient for wavelength stabilization even though the laser diodes facet was not treated with an anti-reflection coating. And the suppression of the ASE and longitudinal laser diode modes worked best when the wavelength of the optical feedback coincided with the maximum of the laser diodes gain bandwidth. The ECDL setup exhibited a narrow emission bandwidth of ∆λ ≤ 20 pm up to an output power of about 470 mW, and ∆λ ≤ 70 pm (FWHM) up to an output power of about 680 mW, at an emission wavelength of 445 nm.

The ECDL system was then used as pump source in a proof-of-concept setup for single- pass second harmonic generation of laser light at 222.5 nm in a BBO crystal. Thereby, narrowband DUV laser light with a continuous wave output power of 16 µW was generated with a pump power of 680 mW. The measurement of the phase matching temperature acceptance bandwidth of BBO revealed, that the phase matching acceptance bandwidths derived from the plane-wave approximation are not valid, when focused beams are used. Rather is the eﬀective interaction length inside the BBO crystal reduced by the focusing, which leads to a broadening of the temperature, wavelength and angle phase matching acceptance bandwidths in comparison to the simulations in the plane-wave approximation.

To further reduce the footprint of the complete setup, a micro-integrated ECDL module (µECDL) assembled on a conduction cooled package with a footprint of 25 mm x 25 mm was developed. Here, a holographic volume Bragg grating served as external wavelength selective element. Compared to the macroscopic ECDL, the µECDL module showed an improved performance having a narrow emission bandwidth of ∆λ ≤ 50 pm up to an output power of 1.4 W at an emission wavelength of 445 nm, and a mode suppression ratio as high as 53 dB over the whole operating range.

With the µECDL module as pump source, the focusing conditions were optimized and it was found that for the asymmetric and non-diﬀraction limited µECDL output a larger beam

93 94 6 Conclusion and Outlook waist radius in the range of 20 µm to 30 µm resulted in the highest conversion eﬃciency compared to the Boyd-Kleinman analysis for focused Gaussian beams, that recommends an optimum beam waist radius inside the BBO crystal of 15 µm.

In conclusion, a novel compact DUV laser light source with a footprint of approximately 5 cm x 30 cm delivering a continuous wave output power of PDUV = 160 µW at a wavelength of λDUV = 222.5 nm was developed. The presented concept enables compact, reliable and portable DUV laser light sources with low power consumption, that are potentially suitable for new applications outside of laboratory environments.

Outlook Even though the achieved DUV output power of 160 µW is already sufficient for some applications, a higher output power is always desirable. The most obvious approach is power scaling by means of a pump source with higher optical output power. At the time of this work, the laser diode (OSRAM Opto Semiconductors, model: PLTB 450B) applied throughout this work delivered the highest output power (P = 1.6 W) from a commercially available laser diode. Currently, the highest commercially available output power from a laser diode in the blue spectral range is specified with P = 3.5 W [44, 154] (Nichia Corporation, model: NDB7K75, Osram Opto Semiconductors: model PLPT9 450DE A01). Taking the achieved normalized conversion efficiency of η = 7.9 · 10−5 W−1 as a basis, a pump power of 3.5 W in the blue spectral range, could generate a DUV output power of approximately PDUV ≈ 968 µW. In the next years, even higher output powers from blue GaN based laser diodes can be expected making the generation of more than 1 mW of DUV laser light by single-pass frequency doubling in a BBO crystal easily feasible.

Another approach to increase the DUV output power would be the enhancement of the effective nonlinear coefficient deff by the use of newly developed crystals. Unfortunately, to the best of the authors knowledge, an alternative nonlinear bulk crystal material, that exhibits a higher effective nonlinear coefficient than BBO and can be phase matched to SHG wavelengths below 230 nm has not yet been developed. However, some efforts have been made to increase the conversion efficiency in BBO itself by implementing waveguide structures into the crystal [155, 156] or using walk-off compensating arrangements with multiple successive BBO plates [157, 158]. Hara et al. for example, realized a walk-off compensating structure consisting of room-temperature bonded BBO plates and achieved a 1.8 times increase of the nonlinear conversion efficiency compared to a bulk BBO crystal with the same overall length [158]. A further advantage of such a walk-off compensating structure is a nearly circular DUV output beam. In contrast, the heavy walk-off in a bulk BBO crystal leads to a strongly elliptical shape of the generated DUV beam, which usually requires additional beam shaping optics in order to generate a more practical circular beam profile. However, the manufacturing procedures of such structures are very complex and have not yet reached the maturity for the realization of commercially available structured crystals.

Another way to circumvent the disadvantage of the heavy walk-oﬀ in BBO, would be the use of periodically poled nonlinear crystals. Conventional used nonlinear materials like PPLN and PPLT however show strong absorption in the wavelength range below 300 nm and the lowest generated second harmonic wavelength with these crystals is 325 nm 95 generated in PPLT [70]. There is ongoing research towards periodically poled nonlinear crystals suitable for the DUV wavelength range. Recently, Hirohashi et al. demonstrated the generation of laser radiation at 266 nm in a periodically poled LaBGeO5 crystal [71]. In principle, this material can also be used to generate laser light in the wavelength range around 222 nm. However, this has not been achieved up to now.

Although cavity enhanced frequency doubling setups are thought to be too sensitive and complex for the realization of a compact and robust DUV laser light source, the concept of a multi-pass SHG in a semi-monolithic concave-plano resonator presented by Klappauf et al. represents an interesting alternative to single-pass frequency doubling [159]. Implementing a BBO crystal in such a resonator could lead to a compact and robust DUV laser light source in the mW range with pump powers far below 1 W.

Besides the power scaling possibilities, there are also some other challenges to solve in the presented setup in order to realize an even more compact and practical DUV laser light source: First, the footprint of the device can be reduced by using an anamorphic prism pair to shape the pump beam in front of the nonlinear crystal instead of using a telescope arrangement with a rather long optical path length. Furthermore, it would be desirable to filter the intensity of the fundamental by using a band-pass filter, that only transmits the generated DUV light instead of using the prism to spatially separate fundamental and second harmonic beam. This is especially challenging as such filters show relatively low transmission rates of less than 50% in the DUV, resulting in a high loss of the useable DUV output power.

With respect to the GaN based external cavity diode laser as pump source, it would be desirable to implement a laser diode with an AR coated facet into the external cavity setup. This is believed to result in a further reduced emission bandwidth of the laser diode, an increase in the wavelength tuning range, and a higher spectral stabilitity of the ECDL emission. Consequently, these improvements would also translate into the DUV wavelength range. Here, especially a wavelength tuneable DUV laser light source would be of great interest for many applications.

A Datasheet information for laser diode PL TB450B (Osram Opto Semiconductors GmbH)

Features • Typ. emission wavelength 450 nm

• Eﬃcient radiation source for cw and pulsed operation

• TO56 package

• ESD protection diode

• Laser diode isolated against package

Applications • Projection

• Metrology

• Stage lighting

Electro-optical parameters

Parameter Symbol

Emission wavelength λ 450 nm

Threshold current Ith 0.2 A

Output power (I = 1.2 A) Popt 1.6 W

Operating current (Popt = 1.6 W) I 1.2 A

Operating voltage (Popt = 1.6 W) U 4.8 V ◦ Beam divergence (FWHM) θk 7 ◦ Beam divergence (FWHM) θ⊥ 23 Polarization ratio (TE:TM) PR 100 : 1

Thermal resistance (junction to case) Rk 15 K/W

Table A.1: Electro-optical parameters of the laser diode applied in this work for a case ◦ temperature of Tcase = 25 C. Taken from the datasheet published by OSRAM Opto Semi- conductors GmbH.

Bibliography

[1] Nerandzic, Michelle M. ; Cadnum, Jennifer L. ; Pultz, Michael J. ; Donskey, Curtis J.: Evaluation of an automated ultraviolet radiation device for decontamination of Clostridium difficile and other healthcare-associated pathogens in hospital rooms. In: BMC infectious diseases 10 (2010), Nr. 1, S. 197 [2] Nerandzic, Michelle M. ; Cadnum, Jennifer L. ; Eckart, Kevin E. ; Donskey, Curtis J.: Evaluation of a hand-held far-ultraviolet radiation device for decontamination of Clostridium difficile and other healthcare-associated pathogens. In: BMC infectious diseases 12 (2012), Nr. 1, S. 120 [3] Memarzadeh, Farhad ; Olmsted, Russell N. ; Bartley, Judene M.: Applications of ultraviolet germicidal irradiation disinfection in health care facilities: effective adjunct, but not stand-alone technology. In: American journal of infection control 38 (2010), Nr. 5, S. S13–S24

[4] Crawford, Mary H. ; Banas, Michael A. ; Ross, Michael P. ; Ruby, Douglas S. ; Nelson, Jeﬀrey S. ; Boucher, Ray ; Allerman, Andrew A.: Final LDRD report: ultraviolet water puriﬁcation systems for rural environments and mobile applications. In: Sandia Report 1 (2005), S. 352005

[5] Kowalski, Wladyslaw: Ultraviolet germicidal irradiation handbook: UVGI for air and surface disinfection. Springer Science & Business Media, 2010 [6] Mellqvist, Johan ; Rosén, Arne: DOAS for flue gas monitoring - I. Temperature effects in the UV/visible absorption spectra of NO, NO2, SO2 and NH3. In: Journal of Quantitative Spectroscopy and Radiative Transfer 56 (1996), Nr. 2, S. 187–208 [7] Kliner, Dahv A. ; Koplow, Jeffrey P. ; Goldberg, Lew: Narrow-band, tunable, semiconductor-laser-based source for deep-UV absorption spectroscopy. In: Optics letters 22 (1997), Nr. 18, S. 1418–1420

[8] Hodgkinson, Jane ; Tatam, Ralph P.: Optical gas sensing: a review. In: Measure- ment Science and Technology 24 (2012), Nr. 1, S. 012004 [9] Long, Derek A.: The Raman effect: a unified treatment of the theory of Raman scattering by molecules. In: West Sussex (2002) [10] Hargis, Philip J. ; Sobering, TJ ; Tisone, Gary C. ; Wagner, John S. ; Young, Steve A. ; Radloff, RJ: Ultraviolet fluorescence identification of protein, DNA, and bacteria. In: Optical Instrumentation for Gas Emissions Monitoring and Atmospheric Measurements Bd. 2366 International Society for Optics and Photonics, 1995, S. 147–154

[11] Bhartia, Rohit ; Hug, Willam F. ; Salas, Everett C. ; Reid, Ray D. ; Sijapati, Kripa K. ; Tsapin, Alexandre ; Abbey, William ; Nealson, Kenneth H. ; Lane,

99 100 Bibliography

Arthur L. ; Conrad, Pamela G.: Classiﬁcation of organic and biological materials with deep ultraviolet excitation. In: Applied spectroscopy 62 (2008), Nr. 10, S. 1070–1077

[12] Bhartia, Rohit ; Salas, Everett C. ; Hug, William F. ; Reid, Ray D. ; Lane, Arthur L. ; Edwards, Katrina J. ; Nealson, Kenneth H.: Label-free bacterial imaging with deep-UV-laser-induced native fluorescence. In: Applied and environmental microbiology 76 (2010), Nr. 21, S. 7231–7237 [13] Sivaprakasam, Vasanthi ; Killinger, Dennis K.: Tunable ultraviolet laser-induced fluorescence detection of trace plastics and dissolved organic compounds in water. In: Applied optics 42 (2003), Nr. 33, S. 6739–6746 [14] Fodor, Stephen P. ; Spiro, Thomas G.: Ultraviolet resonance Raman spectroscopy of DNA with 200-266-nm laser excitation. In: Journal of the American Chemical Society 108 (1986), Nr. 12, S. 3198–3205 [15] Nelson, WH ; Manoharan,R; Sperry, JF: UV resonance Raman studies of bacteria. In: Applied Spectroscopy Reviews 27 (1992), Nr. 1, S. 67–124 [16] Tuschel, David D. ; Mikhonin, Aleksandr V. ; Lemoff, Brian E. ; Asher, Sanford A.: Deep ultraviolet resonance Raman excitation enables explosives detection. In: Applied spectroscopy 64 (2010), Nr. 4, S. 425–432 [17] Efremov, Evtim V. ; Ariese, Freek ; Gooijer, Cees: Achievements in resonance Raman spectroscopy: Review of a technique with a distinct analytical chemistry potential. In: Analytica chimica acta 606 (2008), Nr. 2, S. 119–134 [18] Asher, Sanford A. ; Johnson, Craig R.: Raman spectroscopy of a coal liquid shows that fluorescence interference is minimized with ultraviolet excitation. In: Science 225 (1984), S. 311–314

[19] McCreery, Richard L.: Raman spectroscopy for chemical analysis. Bd. 225. John Wiley & Sons, 2005

[20] Koplow, Jeffrey P. ; Kliner, Dahv A. ; Goldberg, Lew: Development of a narrowband, tunable, frequency-quadrupled diode laser for UV absorption spectroscopy. In: Applied optics 37 (1998), Nr. 18, S. 3954–3960 [21] Storrie-Lombardi, MC ; Hug, WF ; McDonald, GD ; Tsapin, AI ; Neal- son, KH: Hollow cathode ion lasers for deep ultraviolet Raman spectroscopy and fluorescence imaging. In: Review of scientific instruments 72 (2001), Nr. 12, S. 4452–4459

[22] Kumamoto, Yasuaki ; Taguchi, Atsushi ; Smith, Nicholas I. ; Kawata, Satoshi: Deep UV resonant Raman spectroscopy for photodamage characterization in cells. In: Biomedical optics express 2 (2011), Nr. 4, S. 927–936

[23] LightMachinery: -. https://lightmachinery.com/lasers/excimer-lasers/, Abruf: 30.04.2019

[24] PhotonSystems: -. https://photonsystems.com/products/deep-uv-sources/ lasers/, Abruf: 29.04.2019 Bibliography 101

[25] Inc., Coherent: Data sheet for model INNOVA 90C FreD (Frequency doubled Ar ion laser, 244 nm). https://edge.coherent.com/assets/pdf/ Innova-90C-FRED-Data-Sheet.PDF, Abruf: 29.04.2019

[26] AG, AMS T.: Data sheet LEXEL 95-SHG Ar ion laser at 244 nm. http://www.amstechnologies.com/fileadmin/amsmedia/downloads/4835_ lexel8595shglasers.pdf, Abruf: 29.04.2019

[27] Crylas: -. http://www.crylas.de/products/cw_laser.html, Abruf: 29.04.2019

[28] Toptica: -. http://www.toptica.com/products/single-frequency-lasers/ topwave-266/, Abruf: 29.04.2019 [29] Bykov, Sergei ; Lednev, Igor ; Ianoul, Anatoli ; Mikhonin, Aleksandr ; Munro, Calum ; Asher, Sanford A.: Steady-state and transient ultraviolet resonance Raman spectrometer for the 193–270 nm spectral region. In: Applied spectroscopy 59 (2005), Nr. 12, S. 1541–1552

[30] Knigge,A; Erbert,G; Jonsson,J; Pittroff,W; Staske,R; Sumpf,B; Weyers,M; Trankle, G: Passively cooled 940 nm laser bars with 73% wall-plug eﬃciency at 70 W and 25/spl deg/C. In: Electronics Letters 41 (2005), Nr. 5, S. 250–251

[31] Frevert, Carlo: Optimization of broad-area GaAs diode lasers for high powers and high eﬃciencies in the temperature range 200-220 K, Technische Universität Berlin, Diss., 2018

[32] Nakamura, Shuji ; Senoh, Masayuki ; Nagahama, Shin-ichi ; Iwasa, Naruhito ; Yamada, Takao ; Matsushita, Toshio ; Sugimoto, Yasunobu ; Kiyoku, Hi- royuki: Room-temperature continuous-wave operation of InGaN multi-quantum-well structure laser diodes. In: Applied Physics Letters 69 (1996), Nr. 26, S. 4056–4058 [33] Yoshida, Harumasa ; Yamashita, Yoji ; Kuwabara, Masakazu ; Kan, Hirofumi: Demonstration of an ultraviolet 336 nm AlGaN multiple-quantum-well laser diode. In: Applied Physics Letters 93 (2008), Nr. 24, S. 241106 [34] Goldberg, Lew ; Kliner, Dahv A.: Deep-UV generation by frequency quadrupling of a high-power GaAlAs semiconductor laser. In: Optics letters 20 (1995), Nr. 10, S. 1145–1147

[35] Zimmermann,C; Vuletic,V; Hemmerich,A; Hänsch, TW: All solid state laser source for tunable blue and ultraviolet radiation. In: Applied physics letters 66 (1995), Nr. 18, S. 2318–2320

[36] Schwedes, Ch ; Peik, Ekkehard ; Zanthier, Joachim von ; Nevsky, A Y. ; Walther, H: Narrow-bandwidth diode-laser-based blue and ultraviolet light source. In: Applied Physics B: Lasers and Optics 76 (2003), Nr. 2, S. 143–147 [37] Eismann, Ulrich ; Paasch-Colberg, Tim: Diode Lasers: Direct and frequency- converted diode lasers provide all wavelengths for holography. In: Laser Focus World 54 (2018), Nr. 5, S. 39–42 102 Bibliography

[38] Alnis,J; Gustafsson,U; Somesfalean, Gabriel ; Svanberg, Sune: Sum- frequency generation with a blue diode laser for mercury spectroscopy at 254 nm. In: Applied Physics Letters 76 (2000), Nr. 10, S. 1234–1236 [39] Carruthers, Antonia E. ; Lake, Tanya K. ; Shah, Anjali ; Allen, John W. ; Sibbett, Wilson ; Dholakia, Kishan: Single-scan spectroscopy of mercury at 253.7 nm by sum frequency mixing of violet and red microlensed diode lasers. In: Optics communications 255 (2005), Nr. 4-6, S. 261–266 [40] Anderson, TN ; Magnuson, JK ; Lucht, RP: Diode-laser-based sensor for ultraviolet absorption measurements of atomic mercury. In: Applied Physics B 87 (2007), Nr. 2, S. 341–353

[41] Nishimura, Tomoo ; Toyoda, Kenji ; Watanabe, Masayoshi ; Urabe, Shinji: A compact 209-nm deep UV cw light source for spectroscopy using frequency doubling of a diode laser. In: Japanese journal of applied physics 42 (2003), Nr. 8R, S. 5079

[42] AG, Toptica P.: High-power, tunable, frequency-quadrupled diode laser, TA-FHG pro. http://www.toptica.com/products/tunable-diode-lasers/ frequency-converted-lasers/ta-fhg-pro/, Abruf: 03.05.2019

[43] GmbH, OSRAM Opto S.: Visible Laser Diode, model: PL TB450B. https://www. osram.com/os/products/index.jsp, Abruf: 05.03.2018

[44] Corporation, Nichia: LASER Diode, model: NDB7K75. http://www.nichia.co. jp/en/product/laser.html, Abruf: 08.12.2017

[45] Risk, William P. ; Gosnell, Timothy R. ; Nurmikko, Arto V.: Compact blue-green lasers. Cambridge University Press, 2003

[46] Boyd, Robert W.: Nonlinear Optics. 2nd Edition. Academic Press, 2003

[47] Sutherland, Richard L.: Handbook of nonlinear optics. CRC press, 2003

[48] Shen, Yuen-Ron: The principles of nonlinear optics. In: New York, Wiley- Interscience, 1984, 575 p. (1984) [49] Franken, eg P. ; Hill, Alan E. ; Peters, CW e. ; Weinreich, G: Generation of optical harmonics. In: Physical Review Letters 7 (1961), Nr. 4, S. 118

[50] Nikogosyan, David N.: Nonlinear optical crystals: a complete survey. Springer Science & Business Media, 2006

[51] Togashi, Tadashi ; Kanai, Teruto ; Sekikawa, Taro ; Watanabe, Shuntaro ; Chen, Chuangtian ; Zhang, Chengqian ; Xu, Zuyan ; Wang, Jiyang: Generation of vacuum-ultraviolet light by an optically contacted, prism-coupled KBe 2 BO 3 F 2 crystal. In: Optics letters 28 (2003), Nr. 4, S. 254–256 [52] Petrov, Valentin ; Noack, Frank ; Rotermund, Fabian ; Tanaka, Mitsuhiro ; Okada, Yukikatsu: Sum-frequency generation of femtosecond pulses in CsLiB 6 O 10 down to 175 nm. In: Applied optics 39 (2000), Nr. 27, S. 5076–5079 Bibliography 103

[53] Wu, Baichang ; Xie, Fali ; Chen, Chuangtian ; Deng, Daoqun ; Xu, Zuyan: Generation of tunable coherent vacuum ultraviolet radiation in LiB3O5 crystal. In: Optics communications 88 (1992), Nr. 4-6, S. 451–454 [54] Takagi,Y; Sumitani,M; Nakashima,N; Yoshihara, K: Eﬃcient generation of picosecond coherent tunable radiation between 190 and 212 nm by sum-frequency mixing from Raman and optical parametric radiations. In: IEEE journal of quantum electronics 21 (1985), Nr. 3, S. 193–195 [55] Watanabe,M; Hayasaka,K; Imajo,H; Umezu,J; Urabe, S: Generation of continuous-wave coherent radiation tunable down to 190.8 nm in β-BaB2O4. In: Applied Physics B 53 (1991), Nr. 1, S. 11–13 [56] Wu, Baichang ; Tang, Dingyuen ; Ye, Ning ; Chen, Chuangtian: Linear and nonlinear optical properties of the KBe2BO3F2 (KBBF) crystal. In: Optical Materials 5 (1996), Nr. 1-2, S. 105–109 [57] Chen, Chuangtian: Recent advances in deep and vacuum-UV harmonic generation with KBBF crystal. In: Optical Materials 26 (2004), Nr. 4, S. 425–429 [58] Zhang,H; Wang,G; Guo,L; Geng,A; Bo,Y; Cui,D; Xu,Z; Li,R; Zhu, Y; Wang, X u. a.: 175 to 210 nm widely tunable deep-ultraviolet light generation based on KBBF crystal. In: Applied Physics B: Lasers and Optics 93 (2008), Nr. 2, S. 323–326

[59] Smith, Arlee: free software SNLO (oﬀers several functions to assist in selecting a nonlinear crystal and modeling its performance.), Version 68 (August 9, 2017). http://www.as-photonics.com/snlo. Version: August 2017, Abruf: December 20, 2017 [60] Borsutzky,A; Brünger,R; Huang, Ch ; Wallenstein, R: Harmonic and sum-frequency generation of pulsed laser radiation in BBO, LBO, and KD* P. In: Applied Physics B: Lasers and Optics 52 (1991), Nr. 1, S. 55–62 [61] Barnes, Norman P. ; Gettemy, Donald J. ; Adhav, Ratan S.: Variation of the refractive index with temperature and the tuning rate for KDP isomorphs. In: JOSA 72 (1982), Nr. 7, S. 895–898 [62] Ryu, Gihan ; Yoon, Choon S. ; Han, Thomas P. ; Gallagher, Hugh G.: Growth and characterisation of CsLiB 6 O 10 (CLBO) crystals. In: Journal of crystal growth 191 (1998), Nr. 3, S. 492–500

[63] Kato, K: Second-harmonic generation to 2048 Å in B-Ba 2 O 4. In: IEEE journal of quantum electronics 22 (1986), Nr. 7, S. 1013–1014 [64] Chen, CT ; Wang, GL ; Wang, XY ; Xu, ZY: Deep-UV nonlinear optical crystal KBe2 BO3 F2 - discovery, growth, optical properties and applications. In: Applied Physics B: Lasers and Optics 97 (2009), Nr. 1, S. 9–25 [65] Chuangtian, Chen ; Bochang, Wu ; Aidong, Jiang ; Guiming, You: A NEW- TYPE ULTRAVIOLET SHG CRYSTAL - β-BaB 2 O 4. In: Science in China Series B-Chemistry, Biological, Agricultural, Medical & Earth Sciences 28 (1985), Nr. 3, S. 235–243 104 Bibliography

[66] Nikogosyan, DN: Beta barium borate (BBO). In: Applied Physics A: Materials Science & Processing 52 (1991), Nr. 6, S. 359–368 [67] Midwinter, JE ; Warner, J: The eﬀects of phase matching method and of uniaxial crystal symmetry on the polar distribution of second-order non-linear optical polarization. In: British Journal of Applied Physics 16 (1965), Nr. 8, S. 1135 [68] Shoji, Ichiro ; Nakamura, Hirotaka ; Ohdaira, Keisuke ; Kondo, Takashi ; Ito, Ryoichi ; Okamoto, Tsutomu ; Tatsuki, Koichi ; Kubota, Shigeo: Absolute measurement of second-order nonlinear-optical coeﬃcients of β-BaB 2 O 4 for visible to ultraviolet second-harmonic wavelengths. In: JOSA B 16 (1999), Nr. 4, S. 620–624 [69] Eimerl, David ; Davis,L; Velsko,S; Graham, EK ; Zalkin, A: Optical, mechanical, and thermal properties of barium borate. In: Journal of applied physics 62 (1987), Nr. 5, S. 1968–1983

[70] Meyn, J-P ; Laue,C; Knappe,R; Wallenstein,R; Fejer, MM: Fabrication of periodically poled lithium tantalate for UV generation with diode lasers. In: Applied Physics B: Lasers and Optics 73 (2001), Nr. 2, S. 111–114 [71] Hirohashi, Junji ; Taniuchi, Tetsuo ; Imai, Koichi ; Furukawa, Yasunori: PP- LBGO device with 2nd-order QPM structure for 266nm generation. In: CLEO: Science and Innovations Optical Society of America, 2015, S. STh3H–5 [72] Zhang, Dongxiang ; Kong, Yufei ; Zhang, Jing-yuan: Optical parametric properties of 532-nm-pumped beta-barium-borate near the infrared absorption edge. In: Optics communications 184 (2000), Nr. 5-6, S. 485–491

[73] Ghosh, Gorachand: Temperature dispersion of refractive indices in β-BaB2O4 and LiB3O5 crystals for nonlinear optical devices. In: Journal of applied physics 78 (1995), Nr. 11, S. 6752–6760

[74] Boyd, GD ; Kleinman, DA: Parametric interaction of focused Gaussian light beams. In: Journal of Applied Physics 39 (1968), Nr. 8, S. 3597–3639

[75] Gouy, Léon G.: Sur une propriété nouvelle des ondes lumineuses. Gauthier-Villars, 1890

[76] Feng, Simin ; Winful, Herbert G.: Physical origin of the Gouy phase shift. In: Optics letters 26 (2001), Nr. 8, S. 485–487 [77] Librecht,F; Simons, J: Second-harmonic generation in birefringent crystals by a Gaussian beam with an elliptical cross section. In: IEEE Journal of Quantum Electronics 11 (1975), Nr. 10, S. 850–852 [78] Steinbach,A; Rauner,M; Cruz, FC ; Bergquist, JC: CW second harmonic generation with elliptical Gaussian beams. In: Optics communications 123 (1996), Nr. 1, S. 207–214

[79] Freegarde, Tim ; Coutts, Julian ; Walz, Jochen ; Leibfried, Dietrich ; Hänsch, TW: General analysis of type I second-harmonic generation with elliptical Gaussian beams. In: JOSA B 14 (1997), Nr. 8, S. 2010–2016 Bibliography 105

[80] Uebernickel,M; Güther,R; Blume,G; Fiebig,C; Paschke,K; Erbert, G: Study of the properties of the SHG with diode lasers. In: Applied Physics B: Lasers and Optics 99 (2010), Nr. 3, S. 457–464

[81] Strauβ, Uwe ; Hager, Thomas ; Brüderl, Georg ; Wurm, Teresa ; Somers, André ; Eichler, Christoph ; Vierheilig, Clemens ; Löffler, Andreas ; Ristic, Jelena ; Avramescu, Adrian: Recent advances in c-plane GaN visible lasers. In: Gallium Nitride Materials and Devices IX Bd. 8986 International Society for Optics and Photonics, 2014, S. 89861L

[82] Vierheilig, Clemens ; Eichler, Christoph ; Tautz, Sönke ; Lell, Alfred ; Müller, Jens ; Kopp, Fabian ; Stojetz, Bernhard ; Hager, Thomas ; Brüderl, Georg ; Avramescu, Adrian u. a.: Beyond blue pico laser: development of high power blue and low power direct green. In: Proc. SPIE Bd. 8277, 2012, S. 82770K [83] Hempel, Martin ; Tomm, Jens W. ; Stojetz, Bernhard ; König, Harald ; Strauss, Uwe ; Elsaesser, Thomas: Kinetics of catastrophic optical damage in GaN-based diode lasers. In: Semiconductor Science and Technology 30 (2015), Nr. 7, S. 072001

[84] Group refractive Index of GaN. https://refractiveindex.info/?shelf=main& book=GaN&page=Barker-o, Abruf: 21.08.2019 [85] Barker Jr, AS ; Ilegems, M: Infrared lattice vibrations and free-electron dispersion in GaN. In: Physical Review B 7 (1973), Nr. 2, S. 743

[86] Diehl, Roland: High-power diode lasers: fundamentals, technology, applications. Bd. 78. Springer Science & Business Media, 2003

[87] Ruhnke,N; Müller,A; Eppich,B; Maiwald,M; Sumpf,B; Erbert,G; Tränkle, G: 400 mW external cavity diode laser with narrowband emission at 445 nm. In: Optics letters 39 (2014), Nr. 13, S. 3794–3797 [88] Eichler, Jürgen ; Dünkel, Lothar ; Eppich, Bernd: Die Strahlqualitaet von Lasern–Wie bestimmt man Beugungsmasszahl und Strahldurchmesser in der Praxis? In: Laser Technik Journal 1 (2004), Nr. 2, S. 63–66 [89] Braun, Harald ; Rogowsky, Stephan ; Schmidtke, Bernd ; Schwarz, Ulrich T. ; Brüninghoff, Stefanie ; Lell, Alfred ; Strauß, Uwe: High power broad ridge (Al, In) GaN laser diodes: Spatial and spectral stability. In: physica status solidi (a) 206 (2009), Nr. 6, S. 1211–1214

[90] Siegman, Anthony E.: How to (maybe) measure laser beam quality. In: Diode Pumped Solid State Lasers: Applications and Issues Optical Society of America, 1998, S. MQ1

[91] Nakamura,M; Yariv,A; Yen, HW ; Somekh,S; Garvin, HL: Optically pumped GaAs surface laser with corrugation feedback. In: Applied Physics Letters 22 (1973), Nr. 10, S. 515–516

[92] Reinhart, FK ; Logan, RA ; Shank, CV: GaAs-Al x Ga1-x As injection lasers with distributed Bragg reﬂectors. In: Applied Physics Letters 27 (1975), Nr. 1, S. 45–48 106 Bibliography

[93] Wenzel,H; Klehr,A; Braun,M; Bugge,F; Erbert,G; Fricke,J; Knauer, A; Ressel,P; Sumpf,B; Weyers, M u. a.: Design and realization of high-power DFB lasers. In: Proc. of SPIE Vol Bd. 5594, 2004, S. 111 [94] Wenzel,H; Fricke,J; Klehr,A; Knauer,A; Erbert, G: High-power 980-nm DFB RW lasers with a narrow vertical far field. In: IEEE photonics technology letters 18 (2006), Nr. 6, S. 737–739 [95] Müller, André ; Fricke, Jörg ; Bugge, Frank ; Brox, Olaf ; Erbert, Götz ; Sumpf, Bernd: DBR tapered diode laser with 12.7 W output power and nearly diffraction-limited, narrowband emission at 1030 nm. In: Applied Physics B 122 (2016), Nr. 4, S. 87 [96] Hofstetter, Daniel ; Thornton, Robert L. ; Romano, Linda T. ; Bour, David P. ; Kneissl, Michael ; Donaldson, Rose M.: Room-temperature pulsed operation of an electrically injected InGaN/GaN multi-quantum well distributed feedback laser. In: Applied physics letters 73 (1998), Nr. 15, S. 2158–2160 [97] Masui, Shingo ; Tsukayama, Kazutaka ; Yanamoto, Tomoya ; Kozaki, Tokuya ; Nagahama, Shin-ichi ; Mukai, Takashi: CW operation of the first-order AlInGaN 405 nm distributed feedback laser diodes. In: Japanese journal of applied physics 45 (2006), Nr. 12L, S. L1223 [98] Slight, Thomas J. ; Odedina, Opeoluwa ; Meredith, Wyn ; Docherty, Kevin E. ; Kelly, Anthony E.: InGaN/GaN distributed feedback laser diodes with deeply etched sidewall gratings. In: IEEE Photonics Technology Letters 28 (2016), Nr. 24, S. 2886–2888 [99] Kang, Ji H. ; Wenzel, Hans ; Hoffmann, Veit ; Freier, Erik ; Sulmoni, Luca ; Unger, Ralph-Stephan ; Einfeldt, Sven ; Wernicke, Tim ; Kneissl, Michael: DFB Laser Diodes Based on GaN Using 10th Order Laterally Coupled Surface Gratings. In: IEEE Photonics Technology Letters 30 (2017), Nr. 3, S. 231–234 [100] Eliseev, PG ; Ismailov,I; Man’ko, MA ; Strakhov, VP: Injection semiconductor laser with compound resonator. In: Soviet Journal of Experimental and Theoretical Physics Letters 9 (1969), S. 362 [101] Salathe, RP: Diode lasers coupled to external resonators. In: Applied Physics A: Materials Science & Processing 20 (1979), Nr. 1, S. 1–18 [102] Hirota,O; Suematsu, Y: Noise properties of injection lasers due to reflected waves. In: IEEE Journal of Quantum Electronics 15 (1979), Nr. 3, S. 142–149 [103] Glasser, L: A linearized theory for the diode laser in an external cavity. In: IEEE Journal of Quantum Electronics 16 (1980), Nr. 5, S. 525–531 [104] Olsson,A; Tang, C: Coherent optical interference effects in external-cavity semiconductor lasers. In: IEEE Journal of Quantum Electronics 17 (1981), Nr. 8, S. 1320–1323 [105] Lang, Roy ; Kobayashi, Kohroh: External optical feedback effects on semiconductor injection laser properties. In: IEEE journal of Quantum Electronics 16 (1980), Nr. 3, S. 347–355 Bibliography 107

[106] Osmundsen,J; Gade, Niels: Inﬂuence of optical feedback on laser frequency spectrum and threshold conditions. In: IEEE Journal of Quantum Electronics 19 (1983), Nr. 3, S. 465–469

[107] Tkach,R; Chraplyvy, AR: Regimes of feedback eﬀects in 1.5-µm distributed feedback lasers. In: Journal of Lightwave technology 4 (1986), Nr. 11, S. 1655–1661 [108] Rong-Qing, Hui ; Shang-Ping, Tao: Improved rate equations for external cavity semiconductor lasers. In: IEEE Journal of Quantum Electronics 25 (1989), Nr. 6, S. 1580–1584

[109] Zorabedian, Paul: Axial-mode instability in tunable external-cavity semiconductor lasers. In: IEEE journal of quantum electronics 30 (1994), Nr. 7, S. 1542–1552

[110] Ye, Cunyun: Tunable external cavity diode lasers. World Scientiﬁc, 2004 [111] Mroziewicz, B: External cavity wavelength tunable semiconductor lasers-a review. In: Opto-Electronics Review 16 (2008), Nr. 4, S. 347 [112] Petermann, Klaus: External optical feedback phenomena in semiconductor lasers. In: IEEE Journal of Selected Topics in Quantum Electronics 1 (1995), Nr. 2, S. 480–489

[113] Saliba, Sebastian D. ; Junker, Mark ; Turner, Lincoln D. ; Scholten, Robert E.: Mode stability of external cavity diode lasers. In: Applied optics 48 (2009), Nr. 35, S. 6692–6700

[114] Born, Max u. a.: E. Wolf Principles of optics. In: Pergamon Press 6 (1980), S. 188–189

[115] Piprek, Joachim: Analysis of eﬃciency limitations in high-power InGaN/GaN laser diodes. In: Optical and Quantum Electronics 48 (2016), Nr. 10, S. 471 [116] Wieman, Carl E. ; Hollberg, Leo: Using diode lasers for atomic physics. In: Review of scientiﬁc instruments 62 (1991), Nr. 1, S. 1–20 [117] Ricci,L; Weidemüller,M; Esslinger,T; Hemmerich,A; Zimmermann,C ; Vuletic,V; König,W; Hänsch, Th W.: A compact grating-stabilized diode laser system for atomic physics. In: Optics Communications 117 (1995), Nr. 5, S. 541–549

[118] Arnold, AS ; Wilson, JS ; Boshier, MG: A simple extended-cavity diode laser. In: Review of Scientiﬁc Instruments 69 (1998), Nr. 3, S. 1236–1239 [119] MacAdam, KB ; Steinbach,A; Wieman, Carl: A narrow-band tunable diode laser system with grating feedback, and a saturated absorption spectrometer for Cs and Rb. In: American Journal of Physics 60 (1992), Nr. 12, S. 1098–1111 [120] Littman, Michael G. ; Metcalf, Harold J.: Spectrally narrow pulsed dye laser without beam expander. In: Applied optics 17 (1978), Nr. 14, S. 2224–2227 [121] Liu, Karen ; Littman, Michael G.: Novel geometry for single-mode scanning of tunable lasers. In: Optics Letters 6 (1981), Nr. 3, S. 117–118 108 Bibliography

[122] Hawthorn, CJ ; Weber, KP ; Scholten, RE: Littrow configuration tunable external cavity diode laser with fixed direction output beam. In: Review of scientific instruments 72 (2001), Nr. 12, S. 4477–4479 [123] Takamizawa, Akifumi ; Yonezawa, Gen ; Kosaka, Hideo ; Edamatsu, Keiichi: Littrow-type external-cavity diode laser with a triangular prism for suppression of the lateral shift of output beam. In: Review of scientific instruments 77 (2006), Nr. 4, S. 046102

[124] Stoehr,H; Mensing,F; Helmcke,J; Sterr, U: Diode laser with 1 Hz linewidth. In: Optics letters 31 (2006), Nr. 6, S. 736–738 [125] Conroy, RS ; Hewett, JJ ; Lancaster, GPT ; Sibbett,W; Allen, JW ; Dholakia, K: Characterisation of an extended cavity violet diode laser. In: Optics communications 175 (2000), Nr. 1, S. 185–188 [126] Lonsdale, DJ ; Willis, AP ; King, TA: Extended tuning and single-mode operation of an anti-reﬂection-coated InGaN violet laser diode in a Littrow cavity. In: Measurement science and technology 13 (2002), Nr. 4, S. 488 [127] Hildebrandt, Lars ; Knispel, Richard ; Stry, Sandra ; Sacher, Joachim R. ; Schael, Frank: Antireﬂection-coated blue GaN laser diodes in an external cavity and Doppler-free indium absorption spectroscopy. In: Applied optics 42 (2003), Nr. 12, S. 2110–2118

[128] Burns, IS ; Hult,J; Kaminski, CF: Spectroscopic use of a novel blue diode laser in a wavelength region around 450 nm. In: Applied Physics B: Lasers and Optics 79 (2004), Nr. 4, S. 491–495

[129] Hult, Johan ; Burns, Iain S. ; Kaminski, Clemens F.: Wide-bandwidth mode-hop- free tuning of extended-cavity GaN diode lasers. In: Applied optics 44 (2005), Nr. 18, S. 3675–3685

[130] Tanaka, Tomiji ; Takahashi, Kazuo ; Sako, Kageyasu ; Kasegawa, Ryo ; Toishi, Mitsuru ; Watanabe, Kenjiro ; Samuels, David ; Takeya, Motonobu: Littrow-type external-cavity blue laser for holographic data storage. In: Applied optics 46 (2007), Nr. 17, S. 3583–3592

[131] Shimada, Yosuke ; Chida, Yuko ; Ohtsubo, Nozomi ; Aoki, Takatoshi ; Takeuchi, Makoto ; Kuga, Takahiro ; Torii, Yoshio: A simplified 461-nm laser system using blue laser diodes and a hollow cathode lamp for laser cooling of Sr. In: Review of Scientific Instruments 84 (2013), Nr. 6, S. 063101 [132] Jechow, Andreas ; Raab, Volker ; Menzel, Ralf: Tunable 6.8 W narrow bandwidth emission from a single-stripe continuous-wave broad-area laser diode in a simple external cavity. In: Applied optics 47 (2008), Nr. 10, S. 1447–1450 [133] Stry, Sandra ; Hildebrandt, Lars ; Sacher, Joachim ; Buggle, Christian ; Kemmann, Mark ; Klitzing, Wolf von: Compact tunable diode laser with diffraction-limited 1 Watt for atom cooling and trapping. In: Proceedings of SPIE Bd. 4973, 2004, S. 17–25 Bibliography 109

[134] Friedmann, Patrick ; Gilly, Jürgen ; Moritz, Stefan ; Ostendorf, Ralf ; Kelemen, Márc T: 5 W frequency stabilized 976 nm tapered diode lasers. In: Proc. of SPIE Vol Bd. 6876, 2008, S. 68761J–1 [135] Chann, Bet a. ; Nelson,I; Walker, TG: Frequency-narrowed external-cavity diode-laser-array bar. In: Optics letters 25 (2000), Nr. 18, S. 1352–1354 [136] Jechow, Andreas ; Raab, Volker ; Menzel, Ralf ; Cenkier, Michael ; Stry, Sandra ; Sacher, Joachim: 1W tunable near diﬀraction limited light from a broad area laser diode in an external cavity with a line width of 1.7 MHz. In: Optics Communications 277 (2007), Nr. 1, S. 161–165

[137] Gratings, Richardson: Manual for surface diﬀraction grating with 3600 lines/mm. www.gratinglab.com. Version: August 2017, Abruf: 10.08.2017 [138] Chi, Mingjun ; Jensen, Ole B. ; Erbert, Götz ; Sumpf, Bernd ; Petersen, Paul M.: Tunable high-power narrow-spectrum external-cavity diode laser at 675 nm as a pump source for UV generation. In: Applied optics 50 (2011), Nr. 1, S. 90–94 [139] Maiwald, Martin ; Ginolas, Arnim ; Muller, AndrÉ ; Sahm, Alexander ; Sumpf, Bernd ; Erbert, GÖtz ; Trankle, GÜnther: Wavelength-stabilized compact diode laser system on a microoptical bench with 1.5-W optical output power at 671 nm. In: IEEE Photonics Technology Letters 20 (2008), Nr. 19, S. 1627–1629 [140] Volodin, BL ; Dolgy, SV ; Melnik, ED ; Downs,E; Shaw,J; Ban, VS: Wavelength stabilization and spectrum narrowing of high-power multimode laser diodes and arrays by use of volume Bragg gratings. In: Optics letters 29 (2004), Nr. 16, S. 1891–1893

[141] Venus, George B. ; Sevian, Armen ; Smirnov, Vadim I. ; Glebov, Leonid B.: High-brightness narrow-line laser diode source with volume Bragg-grating feedback. In: Proc. of SPIE Bd. 5711, 2005, S. 167 [142] Borgman, VA ; Glebov, LB ; Nikonorov, NV ; Petrovskii, GT ; Savvin, VV ; Tsvetkov, AD: Photothermal refractive eﬀect in silicate glasses. In: Soviet Physics Doklady Bd. 34, 1989, S. 1011 [143] Glebov, Leonid B. ; Smirnov, Vadim I. ; Stickley, C M. ; Ciapurin, Igor V.: New approach to robust optics for HEL systems. In: Proc. SPIE Bd. 4724, 2002, S. 101–109

[144] Efimov, Oleg M. ; Glebov, Leonid B. ; Smirnov, Vadim I. ; Glebova, Larissa: Process for production of high eﬃciency volume diﬀractive elements in photo-thermo- refractive glass. Juli 1 2003. – US Patent 6,586,141

[145] Efimov, Oleg M. ; Glebov, Leonid B. ; Smirnov, Vadim I.: High eﬃciency volume diﬀractive elements in photo-thermo-refractive glass. Januar 6 2004. – US Patent 6,673,497

[146] Lumeau, Julien ; Zanotto, Edgar D.: A review of the photo-thermal mechanism and crystallization of photo-thermo-refractive (PTR) glass. In: International Materials Reviews 62 (2017), Nr. 6, S. 348–366 110 Bibliography

[147] Glebov, Leonid: Fluorinated silicate glass for conventional and holographic optical elements. In: Proc. SPIE Bd. 6545, 2007, S. 654507

[148] Kogelnik, Herwig: Coupled wave theory for thick hologram gratings. In: Bell Labs Technical Journal 48 (1969), Nr. 9, S. 2909–2947

[149] Maiwald, Martin: Halbleiterlaser basierte Mikrosystemlichtquellen für die Raman- Spektroskopie, Technische Universität Berlin, Diss., 2009

[150] Inc., Thorlabs: Raw data for model S130VC. https://www.thorlabs.com/ thorproduct.cfm?partnumber=S130VC, Abruf: 25.07.2019 [151] Helmfrid, Sten ; Arvidsson, Gunnar: Second-harmonic generation in quasi-phasematching waveguides with a multimode pump. In: JOSA B 8 (1991), Nr. 11, S. 2326–2330

[152] Qu, Yujiang ; Singh, Surendra: Second-harmonic generation and photon bunching in multimode laser beams. In: Physical Review A 47 (1993), Nr. 4, S. 3259 [153] Kumar, S C. ; Casals, J C. ; Wei, Junxiong ; Ebrahim-Zadeh, Majid: High- power, high-repetition-rate performance characteristics of β-BaB 2 O 4 for single-pass picosecond ultraviolet generation at 266 nm. In: Optics express 23 (2015), Nr. 21, S. 28091–28103

[154] GmbH, OSRAM Opto S.: Visible Laser Diode, model: PLPT9 450DE A01. https: //www.osram.com/os/products/index.jsp, Abruf: 17.08.2019 [155] Degl’Innocenti,R; Guarino,A; Poberaj,G; Günter, P: Second harmonic generation of continuous wave ultraviolet light and production of β-Ba B 2 O 4 optical waveguides. In: Applied physics letters 89 (2006), Nr. 4, S. 041103 [156] Degl’Innocenti,R; Majkic,A; Sulser,F; Mutter,L; Poberaj,G; Günter, P: UV Second Harmonic Generation at 266 nm in He+ Implanted β-BaB 2 O 4 Optical Waveguides. In: Optics express 16 (2008), Nr. 15, S. 11660–11669 [157] Friebe,J; Moldenhauer,K; Rasel, EM ; Ertmer,W; Isaenko,L; Yelis- seyev,A; Zondy, J-J: β-BaB 2 O 4 deep UV monolithic walk-off compensating tandem. In: Optics communications 261 (2006), Nr. 2, S. 300–309 [158] Hara, Kenjiro ; Matsumoto, Shinnosuke ; Onda, Tomomi ; Nagashima, Wataru ; Shoji, Ichiro: Efficient ultraviolet second-harmonic generation from a walk-off- compensating β-BaB2O4 device with a new structure fabricated by room-temperature bonding. In: Applied Physics Express 5 (2012), Nr. 5, S. 052201 [159] Klappauf, Bruce G. ; Bidel, Yannick ; Wilkowski, David ; Chaneliere, Thierry ; Kaiser, Robin: Detailed study of an efficient blue laser source by second-harmonic generation in a semimonolithic cavity for the cooling of strontium atoms. In: Applied optics 43 (2004), Nr. 12, S. 2510–2527 List of Figures

1.1 Illustration of the spectral separation of Raman signal and ﬂuorescence background for DUV excitation using the example of the ﬁngerprint region of the Raman spectrum of polystyrene...... 2

2.1 a) Schematic sketch of a second harmonic generation process inside a nonlinear crystal. b) SHG process depicted in an energy level diagram...... 10 2.2 Dispersion of the ordinary and extra-ordinary refractive index in BBO for a ◦ crystal temperature of TBBO = 50 C according to the Sellmeier equations from [72]...... 14 2.3 Schematic illustration of phase matching via angle tuning for second harmonic generation, top view (adapted from [46], p. 98)...... 16 2.4 Ordinary refractive index of the fundamental beam and extraordinary refractive index of the second harmonic beam as a function of the angle θ between the optical axis of the crystal and the wave vector k of the extraordinary ◦ beam in BBO (TBBO = 50 C)...... 16 2.5 Schematic illustration of the spatial walk-off occuring with critical phase matching (top view)...... 17 2.6 Normalized SHG intensity as a function of the pump wavelength in BBO at ◦ a crystal temperature of T = 50 (LBBO = 7.5 mm)...... 18 2.7 Normalized SHG intensity as a function of the crystal temperature in BBO ◦ for a phase matching temperature of T = 50 (LBBO = 7.5 mm, λ = 445 nm). 19 2.8 Normalized SHG intensity as a function of the phase matching angle for a fixed fundamental wavelength of 445 nm (LBBO = 7.5 mm)...... 19 2.9 Geometry of a focused beam inside a nonlinear crystal of length Lcr. The solid line indicates a weakly focused and the dashed line a strongly focused beam...... 20 2.10 a) Boyd-Kleinman function in dependence of the beam waist radius w0 for SHG in a BBO crystal without walk-off (dashed line) and with walk-off (solid red line) on a logarithmic scale. b) Boyd-Kleinman function in dependence of ◦ the beam waist radius w0 in case of walk-off on a linear scale. (TBBO = 50 C, LBBO = 7.5 mm, λpump = 445 nm)...... 21 2.11 Simulated SHG power as function of the pump power according to the Boyd- ◦ Kleinman theory for a BBO crystal with TBBO = 50 C, LBBO = 7.5 mm and λpump = 445 nm...... 22 3.1 Photography of the applied laser diode packaged in a TO56 can [43]..... 23 3.2 Optical microscope images of the applied laser diode with different magnification factors (5x (a), 10x (b), 50x (c))...... 24 3.3 Schematic illustration of a Fabry-Perot laser diode resonator of length L with mirror reflectivities R1 and R2 of the rear and front facet, respectively. 25

111 112 List of Figures

3.4 a) Exemplary transverse layer structure of a multiple quantum well separate confinement heterostructure GaN based laser diode. b) Exemplary transverse energy band structure for a multiple quantum well separate confinement heterostructure laser diode. Eg is the band gap energy of the quantum wells. 25 3.5 Emission spectrum of the laser diode applied in this work below threshold at a current of I = 140 mA and at T = 20◦C heatsink temperature...... 26 3.6 Spectrum of the modal gain and the longitudinal FP modes of a laser diode at threshold. The modal gain takes its maximum Γ gth at λp...... 27 3.7 Voltage U, optical output power P and electro-optical conversion efficiency ηc as a function of the injection current I for the laser diode applied in this work at a heatsink temperature of T = 20◦C...... 28 3.8 Emission spectrum for an injection current of I = 1.2 A at a heatsink temperature of T = 20◦C...... 29 3.9 a) Shift of the central emission wavelength λC versus the heatsink temperature at an injection current of I = 1.2 A. b) Shift of the central emission ◦ wavelength λC versus the injection current at T = 20 C heatsink temperature. 30 3.10 Geometry of a Gaussian beam propagating in z direction...... 31 3.11 Intensity profile of the collimated beam from the laser diode applied in this work...... 32 3.12 Caustic measurements of the laser diode in fast (left) and slow (right) axis according to the variance and the 90%-knife-edge diameters at T = 20◦C heatsink temperature and I = 1.2 A...... 34 4.1 Schematic illustration of an external cavity diode laser as a three-mirror cavity laser. R1 and R2 are the reflectivities of the back and front facet and R3 is the reflectivity of the external dispersive element...... 38 4.2 a) Calculated transmission functions in a three-mirror ECDL: The green line is the combined transmission function of the the cavities formed between mirror R1 and R3 and between R2 and R3. The black line is the transmission function of the laser diode cavity. The red line indicates the dispersion D of the surface grating and the blue line the semiconductor gain. b) Spectral response of the ECDL system as the product of all dispersive factors (black line) and grating dispersion D (grey line) for comparison...... 39 4.3 Illustration of an external cavity diode laser in a) Littrow and b) Littman- Metcalf configuration...... 40 4.4 ASE emission spectrum of the laser diode at an injection current of I = 0.1 A. 44 4.5 Emission spectra of the laser diode at an injection current of I = 0.2 A, 0.7 A, and 1.2 A...... 44 4.6 Schematic view of the ECDL system in Littrow configuration: (1) FP laser diode, (2) collimating lens, (3) surface grating, (4) mirror...... 45 4.7 Illustration of two ECDL configurations: a) The laser diode (LD) polarization is perpendicular to the grating grooves, the grating dispersion occurs in the lateral LD plane. b) The LD polarization is parallel to the grating grooves, the grating dispersion occurs in the vertical LD plane...... 46 4.8 Simulated normalized diffracted intensity D in the first diffraction order for a surface diffraction grating with 3600 grooves/mm in configuration b)... 47 List of Figures 113

4.9 Upper part: ASE emission spectrum of the laser diode at an injection current of I = 0.1 A. Lower part: Lasing threshold current versus emission wavelength for the ECDL system at a heatsink temperature of 20◦C..... 48 4.10 Optical output power of the ECDL system emitting at 445 nm and the FP laser diode for a heatsink temperature of 20◦C...... 49 4.11 Emission spectra of the ECDL system at injection currents of I = 0.2 A, 0.4 A, 0.6 A, 0.8 A, 1.0 A, and 1.2 A measured with a) a double-echelle monochromator with a spetral resolution of 6 pm and b) an optical spectrum analyzer with a spectral resolution of 50 pm, for 20◦C heatsink temperature. 51 4.12 Emission spectra of the FP laser diode (grey) and the ECDL system (red) at I = 0.6 A and T = 20◦C...... 52 4.13 Emission spectra of the ECDL system at an injection current of I = 0.6 A for different Littrow angles measured with the optical spectrum analyzer with a resolution of 50 pm at 20◦C heatsink temperature...... 54 4.14 Measured caustic in the fast (left) and slow axis (right) of the ECDL system at an injection current of I = 0.6 A using variance and the 90%-knife edge method...... 55 4.15 Model of a holographic reflecting VBG with thickness d. Rin: incident beam, Sout: output beam, K: grating vector, θ: Bragg angle, φ: slanting angle, Λ: grating period, z: optical axis...... 57 4.16 Simulation of the wavelength dependence of the diffraction efficiency DE according to equation (4.7) for a reflecting VBG with the parameters from table 4.5...... 59 4.17 Simulation of the angle dependence of the diffraction efficiency DE according to equation (4.7) for a reflecting VBG with the parameters from table 4.5.. 59 4.18 Concept of the µECDL module. Also shown are the distances between the different elements of the module and the three competing resonators of length LLD, Linner, and Louter ...... 60 4.19 Simulated residual full divergence angle in lateral and vertical direction as a function of the lens position...... 61 4.20 Simulated coupling efficiency of the back-coupled light in the vertical and lateral axis as a function of the VBG tilt around the x- and y-axis, respectively. 61 4.21 Schematic top and side view of the micro-integrated ECDL module: (1) laser diode in TO56 can, (2) collimating lens, (3) reflecting volume Bragg grating, (4) thread for attachment of the laser diode, (5) conduction cooled package, (6) cover plate...... 62 4.22 Optical output power versus injection current for the µECDL module (solid line) and the FP laser diode (dashed line) at a heatsink temperature of T = 20◦C...... 64 4.23 Contour plot of multiple emission spectra of the FP laser diode as a function of the injection current with measurement steps of 20 mA. Each spectrum is individually normalized in intensity to 1...... 65 4.24 Contour plot of multiple µECDL emission spectra as a function of the injection current with measurement steps of 5 mA. Each spectrum is individually normalized in intensity to 1...... 65 114 List of Figures

4.25 A: Peak wavelength of the free-running laser diode and of the µECDL emission as a function of the injection current. B: Half-logarithmic plot of the FWHM bandwidth of the free-running laser diode and the µECDL emission as a function of the injection current...... 66 4.26 A: Emission spectra of the free-running LD and the µECDL system at injection currents of I = 0.3 A, 0.6 A, 0.9 A, and 1.2 A measured with a high-dynamic range optical spectrum analyzer with a spectral resolution of 50 pm. B: µECDL emission spectra at I = 0.3 A, 0.6 A, 0.9 A, and 1.2 A measured with a double-echelle monochromator with a spetral resolution of 6 pm...... 67 4.27 Temporal stability of the µECDL emission for an injection current of I = 575 mA. a) Peak wavelength of the µECDL emission measured over a time period of 1 hour. b) Exemplary emission spectrum at top = 30 min. c) Optical output power of the µECDL over a time period of 1 hour...... 68 4.28 Variance and 90%-knife edge caustic measurement of the µECDL emission in fast and slow axis for an injection current of I = 0.8 A...... 69 5.1 Schematic top view of the single-pass frequency doubling setup with the macroscopic ECDL emitting at 445 nm as pump source...... 72 5.2 a) Spectral sensitivity of the used SiC photodiode (sglux SolGel Technologies, model SG01XL) and for comparison of a conventional UV extended Si photodiode (Thorlabs Inc., model S130VC). b) Spectral sensitivity of the applied SiC photodiode in logarithmic scale. Data taken from the suppliers calibrations...... 74 5.3 Pump power Ppump of the macroscopic ECDL system emitting at 445 nm as a function on the injection current. Inset A: ECDL emission spectrum at 0.6 A(0.47 W). Inset B: ECDL emission spectrum at 0.8 A(0.68 W).... 76 5.4 Emission spectra of the ECDL system at a) I = 0.6 A and b) I = 0.8 A corresponding to an output power of I = 0.47 W and I = 0.68 W, respectively at a heatsink temperature of T = 20◦C...... 77 5.5 Generated SHG power at 222.5 nm versus pump power at 445 nm for a focal length of lens L3 of 50 mm, 75 mm, 100 mm, and 150 mm. Dashed curves are the quadratic fits of the data points. The solid black curve shows the expected SHG power for a focused Gaussian beam according to the Boyd-Kleinman analysis...... 79 5.6 Spectrum of the generated SHG light at λSHG = 222.5 nm and of the residual stray light from the fundamental beam at λpump = 445 nm...... 80 5.7 a) Measured normalized SHG intensity at 222.5 nm as a function of the crystal temperature TBBO for a 7.5 mm long BBO crystal. Experimental data points are represented by the red squares. b) sinc2 fit (solid line) of the experimental data from a) using an effective crystal length of 1.2 mm.. 81 5.8 Phase-matching wavelength acceptance bandwidth (a) and phase matching angle acceptance bandwidth (b) calculated in the plane-wave approximation ◦ for a crystal length of La,eff = 1.2 mm (TBBO = 50 C)...... 82 List of Figures 115

5.9 a) Setup for the measurement of the temperature acceptance with a parallelized beam in the phase matching plane. b) Resulting phase matching temperature acceptance curve. The dots are the experimental data and the solid curve is the simulated temperature acceptance in BBO for a crystal length of LBBO = 7.5 mm...... 84 5.10 Schematic side and top view of the single-pass SHG arrangement with the µECDL module as pump source...... 85 5.11 Optical pump power of the µECDL module as a function of the injection current for T = 20◦C heatsink temperature. Inset: Emission spectrum of the µECDL module at an injection current of 1.2 A, corresponding to an output power of 1.4 W at the emission wavelength of 445 nm (measured with a high dynamic range optical spectrum analyzer, Yokogawa AQ6373). 86 5.12 Generated SHG power at 222.5 nm versus the pump power of the funda- ◦ mental at 445 nm for all applied focusing conditions (TBBO = 40 C). The 2 dashed lines are the quadratic fits according to PSHG = η · Ppump. The black solid curves show the SHG output power predicted by the BK analysis... 88 5.13 SHG conversion efficiency η as a function of the average beam waist radius 0.5 · (w0,PM + w0,lat) for the three different applied lateral beam expansions. For comparison, the dashed line represents the simulated Boyd-Kleinman function h for an idealized Gaussian beam...... 90 5.14 a) Emission spectrum of the generated SHG radiation with a DUV power of 161 µW at 222.5 nm for a pump power at 445 nm of 1.4 W measured with a spectral resolution of ∆λ ≈ 1.3 nm. b) Emission spectrum of the µECDL at a pump power of 1.4 W measured with a spectral resolution of 6 pm... 91

List of Tables

1.1 Possible applications for deep ultraviolet laser light, corresponding techniques, wavelength ranges, and required specifications...... 3 1.2 Established DUV laser light sources and their emission wavelength λ, typical average optical output power Popt, and power consumption Pcon...... 3 1.3 Overview of diode laser based DUV light sources. SHG: second harmonic generation, SFG: sum frequency generation, TA: tapered amplifier, MOPA: master oscillator power amplifier, SP: single-pass, CE: cavity-enhanced...5 1.4 Targeted specifications for the diode laser based DUV light source...... 6

2.1 Transmission cut-off wavelength λcut-off, minimum SHG wavelength at which phase matching with type I SHG can be achieved at room temperature (T ≈ 293 K), and effective nonlinear coefficient deff for selected crystals... 12 4.1 ECDL parameters from a selection of publications on low-power GaN based ECDLs in Littrow configuration. PLD: nominal maximum laser diode output power, PECDL: maximum ECDL output power, ∆ν: ECDL emission bandwidth, ∆λtun,contin: mode-hop-free or continuous tuning range, ∆λtun,coarse: manual or coarse tuning range...... 42 4.2 Spectral bandwidth ∆λ and diffraction efficiency DE for the three gratings with different groove densities G with the laser diode polarization E perpendicular to the grating grooves (∆λ⊥, DE⊥), or parallel (∆λk, DEk). The measurement uncertainty for the diffraction efficiency values is specified by Richardson Gratings to be ±3 % [137]...... 46 4.3 Experimental parameters used for the calculation of the grating dispersion (figure 4.8) for a grating with 3600 grooves/mm...... 47 4.4 Summary of the measured spectral parameters of the macroscopic ECDL system emitting at 445 nm...... 53 4.5 Parameters used to simulate the diffraction efficiency of the reflecting VBG applied in the µECDL module and the obtained values for DEmax at 445 nm and the spectral selectivity ∆λ. Lower part: manufacturer (Optigrate Corp.) specifications for DEmax and ∆λ...... 58 4.6 Optical path length and free spectral range for the three resonators CLD, Cinner, and Couter formed in the µECDL module...... 60 4.7 Summary of the electro-optical, spectral and spatial parameters of the free-running laser diode (LD), the macroscopic ECDL and the µECDL module. 70 5.1 Simulated beam waist radii in the phase matching plane (PM) and perpendicular to the phase matching plane for different focal lengths of lens L3...... 78 5.2 Normalized SHG conversion efficiency η for the different focal lengths of lens L3...... 79

117 118 List of Tables

5.3 Overview of the different applied lateral beam expansions. fL1, fL2: focal length of lens L1 and L2. Mlat: lateral magnification factor. d0,PM, d0,PM: beam diameter in phase matching and lateral plane. d0,PM : d0,lat: aspect ratio...... 87 5.4 Summary of the focus parameter for the different applied focusing conditions.

w0,PM: beam waist radius in the PM plane. w0,lat: beam waist radius in the lateral plane. zR,PM: Rayleigh length in the PM plane. zR,lat: Rayleigh length in the lateral plane. η: SHG conversion eﬃciency (LBBO = 7.5 mm, ◦ TBBO = 40 C)...... 89 A.1 Electro-optical parameters of the laser diode applied in this work for a ◦ case temperature of Tcase = 25 C. Taken from the datasheet published by OSRAM Opto Semiconductors GmbH...... 97