Creation and Detection of Single Photons
Creation and Detection of Single Photons
Hatim Azzouz Thesis for the degree of Doctor of Philosophy in Physics c Hatim Azzouz, Stockholm 2016
ISBN 978-91-7649-342-7
Printed in Sweden by Universitetsservice US-AB, Stockholm, Stockholm 2016 Distributor: Department of Physics, Stockholm University Cover illustration: Red fluorescence from Rhodamine 6G excited at 532 nm with the green confocal microscope. "The saddest aspect of life right now is that science gathers knowledge faster than society gathers wisdom." Isaac Asimov
Abstract A growing number of technologies employ quantum properties in order to produce solutions that surpass the performance of conventional devices, or to execute operations that are fundamentally impossible with classical sys- tems alone. In the field of optical quantum information science, photons are utilized to encode, communicate and manipulate information, making them vitally important. While photon production always constitutes the first step in any optical experiment, in the field of quantum information science, the recording of data through the process of photon detection is an equally crucial final step. This thesis deals with both the single photons generation (based on diamond color defects) and their detection, utilizing a novel type of superconducting detectors. In particular, part one of this thesis is devoted to the construction of custom-designed microscope setup, and the development of laboratory experi- ments, to enable the generation of single photons as well as the investigation of the optical and spin properties of diamond color centers. Confocal microscopy is used for this purpose, as it allows for the identification and addressing of in- dividual color centers that emit only single photons. This microscope also fea- ture an integrated self-built microwave and magnetic hardware setup, which allows for a wide range of spin environment spectroscopy studies. Single pho- ton emission is demonstrated through both photon anti-bunching and Rabi oscillations at room temperature. The second part of the thesis offers an exploration of superconducting single photon detectors through experiment. Since electronics are an essential part of these detectors, the possibility of using a novel alternative scheme based on capacitive readout combined with fast gating to enable simplified readout is demonstrated. This scheme overcomes the limitations of conventional readout schemes, which require large bandwidth amplification and complex counting electronics. Besides photon detection, the capabilities of these detectors are also expanded to include high-energy particles in the MeV energy range, and the detectors are demonstrated to not only detect single α- and β-particles, but to do so with near unity efficiency. Finally, a multipurpose testing station for superconducting detectors is demonstrated with a central objective of optimizing the coupling efficiency of light to the active area of the detector, as well as to allow for a fast exchange of the optical fiber, thereby facilitating an efficient characterization of the detector. The optimization of this coupling efficiency was demonstrated through proof-of-principle experiments.
Contents
Sammanfattning på svenska ...... vii List of accompanying papers ...... viii
Part I: Preface 1 Introduction ...... 3 1.1 Background and motivation...... 3 1.2 Ideal single photon sources...... 4 1.3 Ideal single photon detectors...... 6 1.4 Thesis outline...... 7 1.5 My contributions...... 10 1.6 Acknowledgments...... 11
Part II: Single Photon sources 2 Intensity fluctuation statistics ...... 15 2.1 Classical description of light...... 16 2.1.1 Poissonian photon statistics...... 16 2.1.2 Classical second-order correlation function...... 17 2.2 Quantum description of light...... 18 2.2.1 Hanbury Brown-Twiss setup for photons...... 18 2.2.2 Quantized second-order correlation function...... 19 2.2.3 Photon bunching and antibunching...... 20 2.2.4 Is antibunched light the same as sub-Poissonian light?...... 22 3 Nitrogen-vacancy centers in diamond...... 23 3.1 Synthesis and generation of NV defects...... 24 3.2 Zero phonon line and phonon sideband...... 25 3.3 Physical structure...... 26 3.4 Electronic structure...... 27 3.5 Spin manipulation and readout...... 30 3.5.1 Optical spin polarization and spin-dependent florescence...... 30 3.5.2 Magnetic single spin manipulation...... 31 3.5.3 CW microwave experiments...... 32 3.5.4 Pulsed microwave experiments...... 32 4 Confocal Microscopy ...... 39 4.1 Principle of Operation...... 40 4.2 Optical components...... 42 4.2.1 Laser source as illumination...... 43 4.2.2 Lens objectives...... 43 4.2.3 Optical fibers...... 44 4.2.4 Other optical components...... 45 4.3 Optical considerations and design requirements...... 46 4.3.1 Light diffraction and resolution...... 46 4.3.2 Optical sectioning and pinhole size...... 49 4.3.3 The optical fiber as a pinhole...... 51 4.4 Detection and control electronics...... 52 4.4.1 Detection and readout software...... 52 4.4.2 Positioning systems and scanning software...... 53 4.4.3 Auto-correlator and coincidence unit...... 53 4.5 Implementation and experimental setup...... 54 4.5.1 The green confocal setup...... 54 4.5.2 The red confocal setup...... 59 4.6 Beyond confocal microscopy - STED microscopy...... 62 4.6.1 Principle of operation...... 63 4.6.2 Implementation...... 65 5 Spin manipulation and readout setup ...... 67 5.1 Microwave and magnetic hardware setup...... 67 5.2 Double-pass acousto-optic modulator setup...... 68 5.3 Integrated setup and timing control...... 71 6 Experimental results - NV center characterization ...... 73 6.1 Diamond sample preparation...... 73 6.2 Confocal scans of NV centers...... 74 6.3 Optical fluorescence spectrum...... 77 6.4 Demonstration of antibunching...... 77 6.5 Demonstration of Rabi oscillations...... 79
Part III: Superconducting Single Photon Detectors 7 Background and basic concepts...... 85 7.1 Superconductivity and working principle of SSPDs...... 85 7.1.1 Superconductivity...... 85 7.1.2 Energy gap...... 85 7.1.3 Phase transition and resistance...... 86 7.1.4 Microscopic absorption model...... 88 7.1.5 Hotspot formation...... 89 7.1.6 Phenomenological model...... 92 7.2 Quantification of the performance of an SSPD...... 93 7.2.1 Detection efficiency...... 94 7.2.2 Dead time...... 97 7.2.3 Dark counts...... 98 7.2.4 Timing jitter...... 99 8 SSPD design and measurement setup...... 101 8.1 Design and fabrication requirements...... 101 8.1.1 Particle absorption...... 102 8.1.2 Advantages of a nano-structured superconductor...... 103 8.1.3 NbTiN on silicon - an alternative to the NbN on sapphire-based SSPD. 104 8.1.4 Fabrication...... 104 8.2 Readout electronics and experimental setup...... 107 8.2.1 Measurement setup...... 107 8.2.2 Characterization...... 108 9 Capacitive readout and gating of the SSPD ...... 111 9.1 Introduction...... 111 9.2 Capacitive readout scheme...... 111 9.3 Gating behavior...... 115 9.4 Conclusion...... 119 10 High-energy particle detection ...... 121 10.1 Introduction...... 121 10.2 Sample preparation...... 122 10.3 Sample mounting and setup...... 123 10.4 Particle detection...... 125 10.5 Conclusion...... 129 11 A multi-purpose testing station ...... 131 11.1 Introduction...... 131 11.2 Dipstick construction...... 132 11.2.1 Design considerations and fabrication...... 132 134subsection.11.2.2 11.3 Scanning system...... 139 11.4 Estimation of the coupling efficiency...... 143 11.5 Conclusion...... 146
Part IV: Conclusion and Outlook 12 Conclusion and outlook ...... 151 12.1 Single photon source...... 151 12.2 Superconducting single photon detector...... 156
Part V: Appendix Appendix A Diamonds - utilisation and treatments ...... 165 A.1 Utilised diamond samples...... 165 A.2 Preparation and production of the nano-diamond powder (NDP) for the Nano- Sizer measurements...... 166 Appendix B High-temperature vacuum oven for the annealing of the dia- monds ...... 167 vi
Appendix C Single photon spectrometer ...... 169 C.1 Grating parameters and design considerations...... 169 C.2 Spectrometer setup...... 174 Appendix D GL Theory ...... 179 Appendix E Single photon down conversion source ...... 181 E.1 Optical and setup components...... 181 E.2 Setup of the SPDC...... 184 Appendix F Setup and mounting for high-energy particle sources . . . . 187 Appendix G Additional information and photographs illustrating the cou- pling system ...... 191 G.1 Operation manual for the SSPD dipstick system...... 191 G.2 Calculated coupling value for a square-shaped active area...... 194 G.3 Additional photographs of the coupling system...... 195 Part VI: Bibliography Appendix Bibliography ...... 199 vii
Sammanfattning på svenska En växande mängd kvantteknologi använder kvantmekaniska egenskaper för att utforma och utveckla anordningar som antingen överträffar prestandan hos konventionella enheter eller utför operationer som är omöjliga med befintliga klassiska system. Specifikt när det gäller optisk kvantinformation utnyttjas fotoner för kodning, kommunikation och manipulation av information. Medan fotonproduktion utgör det första steget i ett experiment inom kvantinformation vetenskap, är registrering av information genom fotondetekteringsprocessen det sista, men lika avgörande, steg. Denna avhandling behandlar både generering av enstaka fotoner baserat på färgcentra i diamant, samt detektion som använder en ny typ av supraledande detektorer. I synnerhet är en del av denna avhandling ägnat åt konstruktionen av uppställningar och utveckling av laboratorieexperiment för att möjliggö- ra generering av enstaka fotoner samt möjliggör undersökning av färgcentra i diamant, både när det gäller deras optiska och spin egenskaper. Konfokal mikroskopiteknik används som ett medel för att identifiera och välja ett enda färgcentrum som endast emitterar en enda foton åt gången. Mikroskopet har även integrerad hårdvara för manipulation med hjälp av såvel mikrovågor som magnetfält, vilket gör det möjligt för ett brett spektrum av spin miljö spekt- roskopistudier. Fotonkällan realiserades genom demonstration av både foton antibunching och Rabi svängningar vid rumstemperatur. Den andra delen av avhandlingen syftar till att utforska och karakterisera supraledande fotondetektor genom ett par experiment. Eftersom elektroniken är lika viktigt som själva detektorn, undersökte vi möjligheten att använda en alternativ avläsningsprocedur baserad på ett nyt kapacitivt avläsningssystem i kombination med snabb gating som möjliggör en förenklad användning av detektorn, vilket gör det möjligt att övervinna begränsningarna från konven- tionella avläsningssystem som kräver en förstärkare med stor brandbredd och komplexa räknings elektronik. Förutom fotondetektering, utökade vi vårt koncept för upptäckt att även omfatta högenergipartiklar i MeV energiområdet, och visade inte bara detek- torns förmåga att effektivt upptäcka enstaka α- och β-partiklar utan också demonstrerat effektiviteten nära ett. Slutligen, har vi konstruerat och tillverkat en mångsidig teststation för de- tektorn med det huvudsakliga syftet att optimera kopplingseffektiviteten mel- lan den optiska fibern och den aktiva ytan hos detektorn, samt möjliggöra ett snabbt och lättillgängligt utbyte av den optiska fibern, varigenom en effektiv karakterisering av detektorn underlättas. Optimering av kopplingseffektivite- ten visades genom proof-of-principle experiment. viii
List of accompanying papers
This thesis is based on the following papers, which are referred to in the text by their Roman numerals.
I Capacitive readout and gating of superconducting single photon detectors Hatim Azzouz, Reinier W. Heers, Sander N. Dorenbos, Raymond N. Schouten, Valery Zwiller, Review of Scientific Instruments 84 053108 (2013).
II Efficient single particle detection with a superconducting nanowire Hatim Azzouz, Sander N. Dorenbos, Daniel De Vries, EstebanBermúdez Ureña, Valery Zwiller, AIP Advances 2, 032124 (2012).
III A highly versatile optical fibre vacuum feed-through Ian A. Davidson, Hatim Azzouz, Klaus Hueck, Mohamed Bourennane, Submitted.
Reprints were made with permission from the publishers. Part I: Preface
3
1. Introduction
1.1 Background and motivation The notion of the photon as a bullet of quantized energy hν is considered the central image of the quantum revolution. This definition stems from the quan- tization of the electromagnetic field, which necessitates the presence of an elementary excitation of a single mode of discrete energy, hence a photon [1]. This modern photon concept was initially proposed by Einstein in 1905 [2] and brought to fruition by Dirac in 1927 [3]. However, the term ’photon’ was only coined in 1926 by Gilbert N. Lewis [4]. The adoption of the photon concept has led to radical advances in physics and allowed to illustrate the quantum ’paradoxical’ nature of light through pioneering experiments, e.g., the demonstration of photon antibunching [5], Bell’s inequality and entangle- ment [6–8]. Recently, photons have been considered fundamental objects for the real- ization of a growing amount of novel quantum technologies [9]. These tech- nologies employ quantum properties in order to design and develop devices that either surpass the performance of conventional devices or execute oper- ations fundamentally impossible with existing classical systems. Especially in the field of optical quantum information (QI) science, photons are utilized for encoding, communication and information manipulation. Since they travel at the speed of light and show only a weak interaction with the surroundings over long distances, which results in low noise and loss levels, and because they can be manipulated using linear optics, photons are considered the most promising candidates for a realization of qubits for novel communication and quantum cryptography schemes. A qubit is a quantum bit (analogous to the classical bit), consisting of a two-state quantum system where information is encoded into the states of photons (or other particles), with the photon polar- ization typically used as the degree of freedom that allows for photon encod- ing. While encoding in quantum communication applications makes use of single photons, the quantum key distribution (QKD), also referred to as quan- tum cryptography, strictly requires single photons. QKD is the most mature and commercially advanced application and offers the most secure form of communication which has yet been proposed [10, 11]. It utilizes single pho- tons to allow two distant parties to establish a secret key, thereby forming a secure encryption channel. QKD does not prevent hacking attempts but detects 4 Introduction and alerts if the key security has been compromised because it is fundamen- tally impossible to clone the quantum state of a photon through eavesdrop- ping. Thus, if a pulse contains more than one photon, an eavesdropper will be able to store the extra photons and transmit the remaining single one, thereby compromising the security of the channel. While photon production constitutes the first step of any experiment in the field of QI science, the recording of data through the process of photon de- tection is the last but equally crucial step. Thus, research on single photon sources and single photon detectors has been stimulated and driven by these demands and prerequisites, in conjunction with the surging interest in the field of QI. Single photon source and -detectors are not only confined to an application in quantum technologies but are also considered for a variety of non-quantum applications ranging from metrology in remote sensing applications to medi- cal applications, e.g., imaging and clinical tomography, just to name a few. A detailed overview of a wide range of potential applications of single photon sources and -detectors is provided in [12, 13].
1.2 Ideal single photon sources A light source which ideally emits one and only one single photon at periodic intervals is referred to as a true single photon source (SPS) if it fulfills the following criteria: • The source is a solitary quantum emitter that allows to secure and realize an efficient one-photon number state. • The source emission is deterministic, i.e., it is triggered by a periodic ex- citation defined by the user at any arbitrary time, thereby guaranteeing a production of single photons on-demand. • The production of a single photon occurs with a certainty of 100 %. Con- sequently, the probability for multiphoton emission is 0%. • The source is highly polarized in both absorption and emission. • The ideal source exhibits a narrow photoluminiscnece line width and a well-defined spatial optical mode. • The emitted photons are indistinguishable. • The source exhibits a photostable emission behavior, with no bleaching or blinking. • The lifetime of the source is short (ps > τ > ns).
In reality, photon sources are often imperfect and deviate from the ideal case. But, depending on the intended application, certain key parameters can be identified and improved. In recent years, several approaches have been 1.2 Ideal single photon sources 5 adopted for single photon generation. An overview of all available SPSs is given in [12–14]. However, as a brief summary, we can distinguish single atoms and ions, single molecules, quantum dots (QDs), heralded photons and color centers, all of which share the deterministic property (to a certain extent) and can be said to be on-demand SPS systems. However, it should be men- tioned that, although the distinction between deterministic and probabilistic photon creation processes is clear in theory, for some SPSs this difference may be obscure in practice. For example, the photon generation in the fre- quently used heralded source, which is based on the production of a pair of photons via spontaneous parametric downconversion (SPDC) in a nonlinear medium, is a strictly probabilistic process. However, it can still be utilized in a deterministic manner because one photon can be used to herald the creation of the other. While both atoms and trapped ions can be used for single photon gener- ation and exhibit excellent optical properties, they are, however, difficult to isolate due to their complicated trapping schemes and the high vacuum re- quirements, making their integration into applicable physical systems a chal- lenging task. The main disadvantages of single molecules are the relatively poor antibunching behavior and the lower photostability, with photobleaching limiting the duration of the operational illumination [15], although recent stud- ies have demonstrated an improved controllability of the single photon emis- sion field [16]. QDs are more favorable candidates and usually show excellent optical properties, even though drawbacks such as the cryogenic operating temperatures (4-30 K), as well as the low coherence times combined with the difficulty of producing indistinguishable photons limit their application po- tential. Color center defects in diamond are another excellent and promising candidate in the field of single photon generation [14]. With diamond as a plat- form, the photon generation scheme adheres to the technological preferences of a solid-state SPS system which can be operated at room temperature and therefore allows for an easy integration into existing networks. In addition, with the current development of crystal fabrication techniques that enable the design and growth of high-purity diamond crystals with suitably inserted im- purities, color centers are becoming more and more attractive for an applica- tion in solid-state electronics. Diamond color defects can be operated at room temperature with coherence times of up to 2 ms and exhibit a unique photo- stable emission. The most widely studied color center is the nitrogen-vacancy (NV) center [14,17], which has been envisioned as a unique object viable not only as a robust SPS but also as a reliable spin qubit [18, 19]. Recently, it has also been more frequently applied in technologies beyond the field of QI processing, e.g., as flourescent biomarker and for nanoscale sensing applica- tions [20]. Other color centers, such as silicon- (SiV) [21] and nickel- (NE8) 6 Introduction vacancy defects [22–24], are also attracting an increasing amount of interest in the field of QI science. In the first part of this thesis I will report on the investigation of color centers in diamond. Although silicon- and nickel-vacancies are currently also receiv- ing much attention, I will only consider and demonstrate antibunching for NV centers. This will be discussed in more detail in one of the following sections.
1.3 Ideal single photon detectors Next, we consider ideal single-photon detectors for which the following re- quirements apply: • The probability of detecting an incident photon within a detection gate is 100 % over a broad spectral range, i.e., the detection efficiency (DE) is close to unity. • The number of dark counts, which describes the probability of detecting false events in the absence of an incident photon, should be zero. • The dead time, defined as the time interval during which the detector is unable to detect an incident photon, is zero. • The jitter time, which is the the time interval between the absorption of a photon and the generation of the corresponding electrical signal, is zero. • The possibility to determine the number of incoming photons, usually re- ferred to as photon-number resolution. Similar to the photon sources, the performance of current detection tech- nologies deviates from the ideal case, and only a subset of the properties listed above may be realized in a particular detector. Typically, single-photon avalanche photodiodes (SPAD)1 and photomulti- plier tubes (PMT) are most commonly used for single photon detection. How- ever, the materials utilized in these detectors are often materials with a rel- atively large bandgap, such as silicon2, making the detection inefficient at a wavelength of 1550 nm, which is the preferred wavelength for fiber optics- based telecommunication. A superconducting single photon detector (SSPD)3 is a promising and highly attractive detector candidate, which offers a better performance over a wide range of parameters when compared with existing detectors. With its very low dark count and small jitter timing, the SSPD is the popular choice for QKD applications, and it is expected to have an increasing influence on
1Also known as APD. 2With a bandgap of 1.1 eV, silicon exhibits a rather large bandgap compared to the bandgap of a typical superconductor, which is in the meV-range. 3Also known as superconducting nanowire single photon detector (SNSPD). In this thesis it is referred to as SSPD. 1.4 Thesis outline 7 the field of QI for optimized DE values. By integrating an SSPD into a low-Q optical cavity with reflecting mirrors, and depending on the combination of the selected substrate thickness and filling factor, the DE can be increased to 60-80 % [25], compared with the 36 % reported as the highest achieved DE at a wavelength of 1550 nm for conventional meander SSPDs [26]. Conventionally, SSPD have used NbN nanowires. But, in 2013, it has been shown that the DE can be improved to > 90% when using amorphous tungsten silicide (W-Si) nanowires instead [27]. A disadvantage of the SSPD, however, is that it needs to be operated at cryogenic temperatures. The second part of this thesis focuses on SSPDs and reports on an early conventional SSPD prototype with a comparatively low detection efficiency. A comparison of the performance of selected typical detectors is given in Table 1.1.
Detector type Operation Detection Dark Timing Max count Refs. temperature [K] efficiency [%] count [Hz] jitter [ps] rate [MHz] PMT (IR) 200 2 200 × 103 300 10 [28] SPAD (InGaAs) 240 10 16 × 103 55 100 [29] SPAD (Si) 250 75 @ 650 nm 25 400 100 [30] SSPD (conventional NbN) 3-4.2 1-36 < 10 60 100 [26] SSPD1 (NbN in cavity) 1.8 57 < 100 30 1000 [31] SSPD2 (NbN in cavity) 2.1 80 1000 - > 10 [25] SSPD (W-Si) 2 90 10 150 0.025 [27]
Table 1.1: Comparison of selected typical single photon detectors based on [12, 32]. The detection efficiency is given for photons with a wavelength of 1550 nm for all detectors except for the Si SPAD, which is used to detect photons with a wavelength of 650 nm. The SPPD1 features an integrated hydrogen silsesquioxane (HSQ) mir- ror structure cavity and an anti-reflection coating, whereas the NbN for SSPD2 was sandwiched between SiO and SiO2 as upper and lower cavity layers, respectively. The detector described in this thesis is the one listed as the conventional SSPD.
1.4 Thesis outline This thesis is partly based on publications in scientific journals and partly on unpublished work. My contributions are described in more detail in the following section. The thesis is divided into two main parts: PartII Single Photon Sources and Part III Superconducting Single Photon detectors. The first part focuses on SPSs, with all research performed at the Quantum Infor- mation and Quantum Optics (KIKO) lab at Stockholm University (SU). Most research associated with the second part was conducted as part of a collabo- rative research project with Prof. Valery Zwiller from the Quantum Transport (QT) group at the Delft University of Technology (TU Delft). 8 Introduction
More precisely, PartII is concerned with the construction of a custom- designed setup for the generation of single photons, as well as the manipula- tion of spin properties focused around NV centers. As an introduction, the first chapter discusses intensity fluctuation statistics, focusing on the second-order correlation function as a necessary tool for characterizing a single photon source. Subsequently, a description of the used sample, especially the physical and electronic properties of NV centers, is presented in Chapter3. The chapter concludes with a description of the spin manipulation and spectroscopy tech- niques for reading out and studying the spin properties of NV centers. Chapter 4 introduces the confocal microscopy technique as a means to identify and il- luminate NV centers, commonly referred to as an SPS setup. The chapter first compares conventional and confocal microscopy, followed by a description of the basic microscopy terminology and confocal design considerations. Next, a description and performance characterization of two custom confocal setups is given, each using a different excitation wavelength: the first is using a green laser to excite the NV centers, and the second a red laser source to excite the intended nickel and silicon vacancy centers. An alternative approach based on the concept of confocal microscopy, known as stimulated emission depletion (STED) microscopy, is described and implemented at the end of the chapter. In Chapter5, the spin manipulation and readout setup is described. This in- cludes a custom-designed microwave and magnetic resonance hardware setup, utilized for the manipulation of NV spin properties and the demonstration of Rabi oscillation. PartII concludes with a presentation and discussion of the results of the investigation of the NV centers (Chapter6), where both anti- bunching and Rabi oscillation are demonstrated. Chapters4,5 and6 are cen- tral parts of PartII. All results of the photon detection experiments presented in this part were achieved using SPADs. The second part of this thesis (Part III), describes my research performed as part of the QT group at TU Delft. The basic concepts of superconductic- ity are introduced in Chapter7, where the underlying principle behind the detection process and the quantifying performance characteristics of SSPDs are described. The SSPD design and fabrication requirements are described in Chapter8, followed by the utilized experimental readout setup. The electron- ics are as essential for the detection as the detector itself, and different readout schemes have been investigated to ensure an optimal readout. The work pre- sented in Chapter9( Paper I) covers our first collaboration project that inves- tigated the possibility of using an alternative readout scheme. In contrast to conventional readout schemes, which require the use of large bandwidth am- plification and complex pulse counting electronics, the capacitive readout of the alternative integrating scheme greatly simplifies the readout of these detec- tors and allows for an easy multiplexing. The only drawback of this approach, i.e., a loss of time information, can be compensated for through fast gating of 1.4 Thesis outline 9 the detector. At the end of the chapter, we also show that this is possible at the single-photon level by detecting heralded photons. Besides photon detection, it is also interesting to see if SSPDs can be used as high energy particle detectors with the aim of improving detection devices in the fields of high energy and nuclear particle detection physics in the same way they are currently affecting optics and communication research. Chapter 10( Paper II) presents the results of our second collaborative work, where we investigated an extension of the detection concept to include high energy particles in the MeV range, and examined the SSPD detection ability and ef- ficiency, particularly for α-, β-, and γ-particles. Because of the hazardous na- ture of the experiment, and for safety reasons, university authorities at Kavli Institute of Nanoscience would only allow the experiment to be carried out at the Reactor Institute Delft (RIH)4, under strict and constant supervision of an on-site expert by whom the handling and setting of the emitting samples was arranged. Therefore, the entire SSPD setup was dismantled, and then moved to and re-built at the reactor. The delicate and time-consuming sample prepara- tion and sample positioning processes were performed under the supervision of senior radiochemist Daniel De Vries. The results of my work at QT initiated the design and development of a novel and reliable multipurpose SSPD testing setup for our KIKO lab at Stock- holm, with the main objective of ensuring a good coupling between fiber and detector by using a 3-axis nanopositioning system. Furthermore, for a detec- tion at different wavelengths, the setup is designed to allow for an easy and swift replacement of the optical fiber. This is ensured by using a specially de- signed optical fiber vacuum feedthrough based on a modified standard optical fiber bulkhead connector with a low weak rate and a low optical loss, making it suitable for high-vacuum applications (Paper III). The design and construc- tion of the SSPD test station was informed by the previous work on confocal microscopy to merge both the SPS and SSPD into one setup. The design and construction of the multipurpose setup, as well as the characterization mea- surements, are described in Chapter 11. Although slightly modified, Chapters 9 and 10 are essentially taken from Paper I and II, respectively, whereas only the part describing the vacuum feedthrough in Chapter 11 is taken from Paper III. The thesis is ended with a conclusion and outlook (Chapter 12).
4Also referred to as the department of Radiation and Isotopes for Health (RIH), which operates the only nuclear reactor in the Netherlands, the Hoger Onderwijs Reactor (HOR). 10 Introduction
1.5 My contributions Contribution to the laboratory setup All the experimental work described in this thesis was either undertaken di- rectly by me, or by project students under my direct supervision. At the start of my PhD project, my lab was entirely empty, thus I was involved in all stages of the build-up of a modern nano-photonics lab, and was responsible for the design, purchase, and construction of all experimental setups. Among the systems I designed and constructed were: Two confocal micro- scopes, the microwave setup, the SSPD testing station, a vacuum oven system suitable for annealing the NV center samples, a highly sensitive single pho- ton spectrometer, and a compact single photon down conversion source (more details can be found in AppendixB,C andE). The most significant part of this work was the design and construction of the two confocal microscopes, including the microwave setup, which is described in detail later in this thesis.
Contribution to papers Paper I Capacitive readout and gating of superconducting single photon detectors Both Reinier W. Heers and I equally contributed to the design of the experiments. This holds true for the laboratory work as well. However, I analyzed the data and wrote the paper. Sander N. Dorenbos designed and fabricated the superconducting detector. Raymond N. Schouten helped with the design and construction of the readout electronics. Prof. Valery Zwiller supervised the experiments. All co-authors participated in the discussion of the results and commented on the manuscript at all stages. Paper II Efficient single particle detection with a superconducting nanowire In this work, I made major contributions ranging from coordinating and arranging the logistics with the nuclear reactor authorities at Delft, designing and performing all the described experiments, to analyzing the data and writing the paper. Sander N. Dorenbos and Esteban B. Ureña designed and fabricated the superconducting detector. Dr. Daniel De Vries was assigned as chemist to supervise the preparation of the radioactive samples. He also provided input for writing the paragraph Sample Preparation and Methods. Prof. Valery Zwiller supervised the experiments. All co-authors participated in the discussion of the results and commented on the manuscript at all stages. Paper III A highly versatile optical fibre vacuum feed-through The experiment was designed by Dr. Ian Davidson and me based on preliminary work undertaken by Klaus Hueck under my supervision. The experimental setups were jointly constructed by Ian Davidson and me, with the measurements principally being performed by Ian Davidson. Ian Davidson and I jointly analyzed the data and contributed equally to the writing of the 1.6 Acknowledgments 11 paper. Prof. Mohamed Bourennane supervised the experiment. All co-authors participated in the discussion of the results and commented on the manuscript at all stages.
1.6 Acknowledgments During the making of this thesis I have been fortunate to get to know and collaborate with many people to whom I owe my deepest gratitude for making this thesis possible. First of all I would like to thank my supervisor, Mohamed Bourennane, for giving me the opportunity to work in the intriguing field of Quantum Informa- tion and Quantum Optics, and offering me a brand new laboratory to admin- ister and run as I pleased. I would also like to express my deepest gratitude to Valery Zwiller for giving me the opportunity to work on superconducting de- tectors in his group, and for providing many ideas and much guidance during my detector work. Special thanks go to Hans Blom who enthusiastically dis- cussed and answered all my questions concerning confocal microscopy and single photon sources. I owe my gratitude to Ian Davidson for the vast amount of discussion and feedback on superconducting single photon detector theory and application, and for his thorough proofreading of this thesis. Sander N. Dorenbos and Reinier W. Heers, it was nice working with you and thanks for the successful collaboration. Reinier’s enthusiasm and Sander’s serenity was a source of great comfort during the long and at times arduous measurements. I would like to acknowledge the other team members of the KIKO group, who created an inspiring and joyful atmosphere. Especially Jan Bogdanski, Magnus Rådmark, Elias Amselem, Johan Ahrens, Nima Rafiei, Kate Blanchfield, Ally Hameedi, Muhammad Sadiq, Ashraf Abdelrazig Mo- hamedelhassan, Christian Kothe, Hannes Hübbel, and Isabelle Herbauts. I had the pleasure to supervise a number of students. First there was Per Nilsson, thanks for your programming contributions and for constructing the control software. Lars Johann Wacker, thanks for assisting me with building the prototype microwave setup - it proved very useful! A great thank should also be given to Klaus Hueck for his great efforts constructing the multi- purpose station for the detector. Although I didn’t have the pleasure to meet Christoph Dittel in person, his spin work based on our prototype microwave setup were impressive, and provided the Rabi oscillation results for this thesis. There are numerous people in both Sweden and the Netherlands, who have helped and supported me enormously during my work. Unfortunately, they are too many to name, but I am deeply thankful to all of them, and owe them all my greatest appreciation. 12 Introduction
Finally, I would like to express my deepest gratitude to my family for their endless moral support, patience, and understanding. Part II: Single Photon sources
15
2. Intensity fluctuation statistics
Since the objective of this work is to both produce and observe non-classical phenomena of light, the focus of this chapter is mainly to highlight the theo- retical differences between classical and quantum light sources, as well as to develop an experimental method for analyzing the nature of light sources, par- ticularly to identify a single photon [33–36]. However, before considering the basic concepts and terminologies of quantum optics, it is instructive to start with a brief historical overview of the research that led to the development of quantum optics. The radical change in the concept of light can be traced back to the work of Max Planck, who hypothesized that radiation from a black-body is emitted in discrete energy packets which he called quanta [37, 38]. Based on this, Einstein went on to propose the concept of the quantization of light as an explanation for the newly discovered photoelectric effect [2]. Shortly afterwards, in 1909, Einstein inspired yet another revolutionary conception of the nature of light, i.e., the wave-particle duality [39]. Partic- ularly, by applying Planck’s radiation formula, Einstein proposed a solution to the problem of the energy fluctuations observed when measuring the black body radiation, and noted that the radiation energy gave an extra term propor- tional to the average number of quanta. This was clearly incompatible with the classical wave theory of light but anticipated a new theory of photon statistics. More specifically, a deeper understanding of the full statistical properties of the various types of light became the main approach in the development of quantum mechanics as an attempt to elucidate the seemingly contradicting duality. However, it was the work of Hanbury Brown and Twiss, published in 1956 [40], who studied correlation effects in the intensity received by two separate detectors, and the invention of the laser in 1960 [41], that widened the horizon of the field, and inspired sophisticated photon statistics experiments and observations of non-classical optical phenomena. For a long time, only thermal sources had been available. But, with the new light source, a statistical distinction between thermal and laser light was experimentally demonstrated. In contrast to Planck’s formula, laser light with its coherent radiation was shown to exhibit a statistical distribution corresponding to that of a Poisso- nian distribution for the number of photons. From a classical perspective, the same statistical treatment of photons also predicted a theoretical non-classical type of light with a more narrow photon distribution, i.e., sub-Poissonian light. 16 Intensity fluctuation statistics
Although Einstein correctly hypothesized the particle nature of light in 1905, the existence of the photon was not proven until 1977 by Kimble et al., who observed the effect of photon antibunching using light emitted by sodium atoms [5]. The sub-Poissonian behaviour was then verified in 1983 by Short et al. [42]. The antibunching and sub-Poissonian phenomena are closely related and are both considered a clear proof of the quantum nature of light, as discussed later. In the following, the classifications of the light beam accord- ing to photon statistics is briefly discussed. Subsequently, the analysis tech- nique of photon statistics based on intensity fluctuation correlations and its restrictions on classical light is described. Finally, the Hanbury Brown-Twiss interferometer is presented as an experimental device enabling the analysis of light sources and re-examined considering the quantum nature of light.
2.1 Classical description of light 2.1.1 Poissonian photon statistics When considering the quantum nature of light, a common approach is to ex- amine photon statistics where the light beam is regarded as a stream of pho- tons rather than a classical wave. From the classical point of view, perfectly coherent light beams with constant intensity are the most stable form of light, thereby setting the limit for the nature of classical light. Since the photon distribution in a coherent state is Poissonian, this allows a threefold classifi- cation of light according to the variance of their photon number distributions known as: Poissonian, super-Poissonian and sub-Poissonian. For a Poissonian photon distribution, the variance (∆n)2 of the photon number distribution is related to the mean value n¯ through (∆n)2 = n¯, which is the narrowest distri- bution classical light can exhibit, whereas it is broader for a super-Poissonian distribution: (∆n)2 > n¯. While a constant and coherent light intensity is Pois- sonian, it is easy to imagine that a time-varying fluctuation in intensity would lead to a larger fluctuation in photon number, hence a super-Poissonian photon number distribution. Examples of super-Poissonian statistics are thermal and chaotic light. Experimental observations of Poissonian and super-Poissonian statistics are consistent with the classical theory of light. However, this is not true for sub-Poissonian light. Compared to Poissonian photon statistics, sub-Poissonian light features an even narrower photon number distribution (∆n)2 < n¯), suggesting a state ’more stable or quieter’ than coherent light. Sub-Poissonian light has no classical counterpart and is therefore considered a clear signature of the quantum nature of light. However, observation of sub- Poissonian light is a rather non-trivial quantum task because its sensitivity to optical losses requires fast single-photon detectors with rather high quantum efficiencies. 2.1 Classical description of light 17
2.1.2 Classical second-order correlation function An alternative approach is to analyze the statistics of photons by quantifying the intensity fluctuations using a technique known as second-order correlation function. This was conceptually introduced by the British radio astronomers Robert Hanbury Brown and Richard Twiss in their pioneering work in 1956 [40], where they adapted an intensity interferometer and applied its intensity correlation effects to the measurement of the angular diameter of stars. The restrictions for the second-order correlation function predicted by the classical and quantum descriptions are radically dissimilar. It is this discrepancy that is used to differentiate between classical and non-classical fields. Before examining in more detail the quantum description of the HBT ex- periment and the non-classical implications on the second-order correlation function, let us briefly introduce the intensity correlation measurements and the conditions they imply for classical light. Particularly, we consider the de- tection of the intensity of two light beams that have not been optically mixed. However, we impose a time delay on one of the intensities prior to detec- tion. The output of both detectors is connected to a correlator that counts and stores the measured intensity as a function of time and multiplies the intensi- ties of both photodetectors for different detection time differences. Assuming the same (point) source for the two beam intensities, i.e., one spatial mode, the output of the correlator will be proportional to the normalized temporal second-order correlation function for a classical single-mode, defined by hI(t)I(t + τ)i g(2)(τ) = , (2.1) hI(t)ihI(t + τ)i where I(t) = E(t)∗E(t) is the detected intensity at time t. The brackets h...i indicate the ensemble average calculated by integrating over a long time in- terval. In other words, g(2)(τ) describes how the joint intensity probability of measuring an intensity I at time t + τ depends on the value of the intensity at time t. Whereas the term first-order coherence function indicates that g(1)(τ) describes amplitude correlations based on the properties of the first power of the electric field, the second-order symbol in g(2)(τ) refers to correlations characterized by the intensity properties instead, which is to the second power of the electric field. For classical fields, the properties of g(2)(τ) are given by
g(2)(0) ≥ 1, (2.2) g(2)(0) ≥ g(2)(τ). (2.3)
The proofs for these classical inequalities can be found in literature [43– 45]. However it is worth addressing their heuristic derivation and how they can be intuitively understood. Since the variance of the intensity is given by ∆I2 = hI2i − hIi2, the second-order correlation can be expressed as g(2)(0) = 18 Intensity fluctuation statistics
(∆I2/hIi2) + 1 at τ = 0. Remembering that the variance obeys ∆I2 ≥ 0, we find that the value of the normalized second order coherence function is larger than or equal to 1 at zero time delays, as stated by Eq. (2.2). By applying the Cauchy-Schwartz inequality, one can relate the degree of coherence of the second-order correlation function g(2) at later times to the zero time delays. Since hI(t)i2 = hI(t + τ)i2 for stationary processes, we then get hI(t)I(t + τ)i2 ≥ hI(t)2i2, leading to Eq. (2.3). For a perfectly coherent stable source (i.e., ∆I = 0), it is easy to show that (2) g (τ) = 1 for all time delay values τ because the intensity I0 is constant. Since this is valid for all time delays including the special case of zero-time delay, we also get g(2)(0) = 1. Next, we consider any source with a fluctuating intensity1 with a time-scale determined by the coherence time of the source, i.e., light which is coherent only in time intervals shorter than τc. In this case, it can be shown that the correlation function factorizes and g(2)(τ) → 1 for τ τc [46], since the intensity fluctuations will be uncorrelated at times t and t + τ. However, if τ τc, correlations between fluctuations will occur at the two times. Particularly for τ = 0, the inequality in Eq. (2.2) applies and g(2)(0) > 1. To summarize, for any time-varying intensity, the correlation function de- creases for increasing values of τ, reaching the value of unity for τ → ∞, whereas, for a non-varying intensity, it is a constant value, g(2)(τ) = 1 for all values of τ. While Eq. (2.2) gives the upper limit for a zero time-delay of the second-order coherence function for classical fields, Eq. (2.3) conse- quently yields the restriction that the degree of second-order coherence can never exceed its value at zero time delay. In the following, we will see that the inequality equations (2.2) and (2.3) are violated by quantum light and are used to experimentally demonstrate the quantum nature of light.
2.2 Quantum description of light 2.2.1 Hanbury Brown-Twiss setup for photons The setup of Hanbury Brown and Twiss (HBT) interferometer is similar to classical interferometry systems such as the Michelson stellar interferometer. However, while the later correlates phase fluctuations of two waves, the HBT setup measures the spatial correlation between fluctuations in intensity. Since the spatial coherence is determined by the spread of the angles within which the light arrives at the interferometer, measuring a small angle requires a large distance between detectors. However, as the distance between the detectors
1Examples of classical fluctuating fields are thermal or chaotic light. It is possible to show that for all chaotic light g(2)(τ) = 1+kg(1)(τ)k, where g(1)(τ) is the first order temporal correlation function. 2.2 Quantum description of light 19 increases, it becomes more difficult to sustain the collection mirrors in sta- ble positions adequately to achieve interference patterns on the two correlated detectors. Thus, HBT proposed an intensity interferometer based on separate detectors, i.e., it does not even require a physical connection between the de- tectors − the correlations may be extracted later. More specifically, the origi- nal HBT setup allowed a lateral adjustment of one of the detectors in order to be able to investigate intensity correlations between various spatial modes at zero-time delay. However, in our case, it is more interesting to slightly modify the original HBT setup so that it allows to measure the temporal correlation instead of the spatial coherence. A schematic illustration of the HBT setup is shown in Figure 2.1. The HBT setup consists of a 50:50 beam splitter (BS) that equally distributes the incident beam intensity (stream of photons) between the trans- mission and reflection outputs of the BS, onto the detectors D1 and D2. The detectors are connected to an electronic unit that stores the measured counts from both detectors while simultaneously recording the time that elapses be- tween the counts from the two detectors. By varying the time difference, i.e., the time delay between the detection events, g(2)(τ) can be measured.
Figure 2.1: Conceptual schematic illustration of the interferometer setup proposed by Hanbury Brown and Twiss. The electronic unit consisted of an auto-correlator whereas the delay line was set by adding coaxial cables connected between detector D2 and the auto-correlator.
2.2.2 Quantized second-order correlation function In the classical picture, where the incoming beam is considered as light in- tensity, the 50:50 BS will divide the incident intensity such that half of the intensity is going to each arm. However, this is no longer valid when con- sidering a quantized field. When assuming a stream of photons instead, with 20 Intensity fluctuation statistics long time intervals between successive photons, an impinging photon on the BS will have a 50% probability to be directed onto only one of the detectors, but not both. Hence the correlator will record no event at τ = 0, i.e., there will be no correlation between the two detectors. Only when a subsequent photon is directed towards the counter-detector, an event is recorded. We realize that, while we very well might have events for larger time delay values, we will never have any at τ = 0. Now, because the correlation function, i.e., Eq. (2.1), was expressed in terms of intensity, it needs to be rewritten so that the value of g(2)(τ) is expressed in terms of coincidences between photon count rates. g(2)(τ) is then given by
hn (t)n (t + τ)i g(2)(τ) = 1 2 , (2.4) hn1(t)ihn2(t + τ)i where ni(t) is the number of photon counts on detector i at time t. When focusing on single-mode fields, the quantum interpretation of the expression in Eq. (2.4) at τ = 0 can be simplified to
haˆ†(t)aˆ†(t + τ)aˆ (t + τ)aˆ (t)i hnˆ(nˆ − 1)i 2 g(2)(0) = 1 2 2 1 = , (2.5) † † hnˆi haˆ1(t)aˆ1(t)ihaˆ2(t + τ)aˆ2(t + τ)i where nˆ = aˆ†aˆ is the number operator and hnˆi denotes the expected photon number detection values. Remembering the definition ∆nˆ ≡ nˆ − hnˆi, Eq. (2.5) can be rewritten as a function of the variance of the photon number h(∆nˆ)2i:
h(∆nˆ)2i − hnˆi g(2)(0) = 1 + . (2.6) hnˆi2
Thus, if the single mode field is in a coherent state, i.e., h(∆nˆ)2i = n¯ for a Pois- sonian photon number distribution, we get g(2)(0) = 1. This was also predicted by the classical theory of light. A non-classical beam containing n photons is known as Fock- or number states. Since the number state |ni is an eigenstate of the number operator, Eq. (2.5) can be rewritten as n(n − 1) g(2)(0) = , n ≥ 1 (2.7) n2
2.2.3 Photon bunching and antibunching Let us more closely examine the results of the quantum description and the implications of Eq. (2.7), and compare them with the classical theory. The classical limit for a perfectly coherent light is g(2)(τ) = 1. Now, if the timing between the photons is more narrow, i.e., the photons are bunched together, the probability of detecting photons arriving together in pairs is higher around 2.2 Quantum description of light 21
τ = 0 than at later times, leading to g(2)(0) > 1. Subsequently, we expect g(2)(τ) to be larger for small values of τ than for higher values, such that g(2)(0) > g(2)(τ). This phenomenon was already predicted by the classical theory, for instance, by Eqs. (2.2) and (2.3) for thermal light, and is referred to as photon bunching. The converse case, where g(2)(0) < g(2)(τ), is referred to as antibunching. Here, the light field consists of photons more equally spaced, and the probability of detecting two photons simultaneously is lower than the probability of detecting them at different times. Furthermore, we have seen that the no-event observation at τ = 0 discussed for the case of a beam con- sisting of individual photons leads to the non-classical result of g(2)(τ) = 0. Hence, we can conclude that g(2)(0) < 1. Clearly, this is a violation of the classical restrictions of both Eq. (2.2) and Eq. (2.3). In fact, antibunching is considered a conclusive signature of the quantum nature of light. While we expect that g(2)(0) = 0 for an ideal single-photon source emitting one photon, Eq. (2.7) implies that the expected g(2)(0) value is 0.5 for photon number states with n = 2. Because of the experimental challenges and diffi- (2) 1 culties in arranging a single-photon source setup, the condition g (0) < 2 is used as the upper limit to identify photon antibunching. Table 2.1 summarizes the conclusions of this chapter, and the resulting typical curves for g(2)(τ) are shown in Figure 2.2.
Figure 2.2: The normalized second-order function for various single-mode fields. The fluctuating field corresponds to a bunched photon stream (blue curve) whereas the per- fectly coherent field corresponds to a random photon stream. The dashed line indicates the level for Poissonian light). Only the quantum one-photon state shows antibunch- ing. 22 Intensity fluctuation statistics
Model Light source g(2) property Photon stream (∆n)2 property I(t) (for all τ) (for all T) Classical All g(2)(0) ≥ 1 g(2)(0) ≥ g(2)(τ) Coherent g(2)(τ) = 1 Random Poissonian = n¯ Constant Chaotic/thermal g(2)(0) > 1 Bunched super-Poissonian > n¯ Time-dependent g(2)(0) > g(2)(τ) Quantum 1-photon state g(2)(0) < 1 Antibunched sub-Poissonian < n¯ Constant g(2)(0) < g(2)(τ)
Table 2.1: Classification of light according to the second-order correlation function and photon statistics.
2.2.4 Is antibunched light the same as sub-Poissonian light? As mentioned earlier, although antibunching and a sub-Poissonian photon distribution are closely related, it is important to note that they are differ- ent manifestations of the same optical quantum phenomenon. It is true that, from Eq. (2.6), sub-Poissonian statistics indicates that g(2)(0) < 1 which, as we already established, also means antibunching for some range of τ, unless g(2)(τ) is time-independent. However, this cannot be a general conclusion, since g(2)(0) < g(2)(τ) does not necessarily entail that g(2)(0) < 1. In fact, the first experimental demonstration of antibunching [5] is an example of a source for which g(2)(0) < g(2)(τ), but which also exhibited a super-Poissonian dis- tribution over some time interval. More specifically, since the measured cor- relation function, although increasing, had its initial value at unity, it did not fulfill the inequality g(2)(0) < g(2)(τ) for all values of (τ). The inequality was true only for a finite delay (τ). Hence, although the light was strictly super- Poissonian, it was considered antibunched for a certain time period. Also, it has been experimentally demonstrated that non-classical light, while bunched, still can be sub-Poissonian for a certain time interval T [47]. Thus, although the two effects (antibunching and sub-Poissonian) are not causally related, it is natural that they frequently coincide for certain fields, since they are both expressions of the same quantum nature of light2.
2A more thorough discussion of this topic can be found in [34, 48]. 23
3. Nitrogen-vacancy centers in diamond
Color centers in diamond are fluorescent point defects in the crystal lattice. They consist of one or more foreign atoms or vacant lattice sites replacing the lattice atom, and are uniquely determined by their optical properties (emission and absorption spectra), hence the term color center [49]. Numerous color centers have been identified in diamond and studied extensively in order to understand their structure and to control their physical properties [14]. One of these color centers that has received considerable attention and has undergone a renewed exploration is the nitrogen-vacancy (NV) center, due to its favor- able photo-luminescence properties. Trapped (as an artificial atom) in the di- amond lattice, the NV center proved as a reliable photo-stable single photon source, which can be operated at room temperature with no signs of bleach- ing. Spurred by the first demonstration of anti-bunching in 2000 [50, 51], the NV center was soon envisioned as a suitable candidate for quantum optical network applications. However, it is the exquisite magnetic and spin prop- erties that are the most interesting aspect of the NV center and what sets it apart from other color centers. Since establishing the coupling between lu- minescence and the electronic spin state of the center [18, 19, 52], and the demonstration that the luminescence intensity can be modulated and manipu- lated at room temperature by applying a magnetic field [53,54], the NV center was identified as an attractive qubit for many quantum information processing applications [17]. Only a few color centers possess this unique magneto-optic property1. However, if the spin properties are not relevant for the desired ap- plication, many other color centers, such as nickel- and silicon-vacancies can serve as fluorescent emitters [14]. As mentioned elsewhere in this work, nickel- and silicon-vacancy centers were also among the first defects proposed for investigation. However, since their photo-luminescence properties are quite similar to that of the NV center in most respects, we will only focus on the NV center. Thus, in this chapter, we will only consider the generation of NV defects along with their various host crystals, as well as the basic electronic energy level structure of the NV
1Recently, the silicon-vacancy center has also received increasing attention because it displays similar electron spin and photo-luminescence properties, although it shows a sharper optical spectrum than the NV center, suggesting it as yet another promising potential qubit [21]. 24 Nitrogen-vacancy centers in diamond center. The chapter ends with a description of spin manipulation and readout techniques required for the realization of a single qubit.
3.1 Synthesis and generation of NV defects Most of the diamond samples used for research applications are synthesized [14, 17, 20] and come in different forms and shapes such as bulk crystals, thin films and nanocrystals. According to the employed cultivation method, i.e., chemical vapor deposition (CVD) or a high-pressure high-temperature (HPHT) crystal formation technique, synthetic diamonds are generally re- ferred to as CVD or HPHT diamonds [49], respectively. CVD is a relatively new diamond fabrication technique that allows for the diamond to be grown from a hydrocarbon gas mixture, and produces thin films of polycrystalline di- amond with dimensions in the range from several microns to a few millimeter. The HPHT method has been used since 1950 and is the most widely employed method due to its low cost. The method is an attempt to mimic the natural ge- ological process conditions. The majority of the commercially used synthetic diamonds are HPHT crystals, typical with dimension of a few millimeter, and are considered useful as hard-wearing edges on cutting and drilling tools. Syn- thetic nanocrystalline diamonds can be obtained by grinding large crystals to sub-100-nm-sized particles. The two predominant defects in both natural and synthetic diamonds are nitrogen impurities and vacancy2 aggregations, and based upon concentrations of the nitrogen impurities, diamonds are categorized in two major groups, namely type I and type II [55–58]. While type II refers to diamonds without any detectable nitrogen impurities, diamonds are classified as type I when they contain detectable amounts of nitrogen3. For type I diamonds, two major sub- divisions are further identified. Diamonds that contain nitrogen in aggregated forms (comprised of a pair or more nitrogen atoms) are known as type Ia, whereas those with isolated substitutional nitrogen atoms are referred to as type Ib. While very rare in natural diamonds, virtually all synthetic diamonds fall into the type Ib category. Although many types and forms of diamonds4 have been tested during the course of this work, we are only focusing on type Ib synthetic diamonds in this thesis. Due to the stress caused by the single nitrogen atom inside the lattice struc- ture, the nitrogen defect is efficient in trapping a moving vacancy injected
2When denoting a crystal defect, lattice vacancy refers to a missing atom leaving a hole on its regular atomic site. In this case, by a vacancy we mean a missing carbon atom. 3The sensitivity limit of conventional IR absorption spectroscopes used to detect the impurities is around 1 ppm (parts per million). However, typical nitrogen concentrations are >10 ppm and may be as high as 3000 ppm [55]. 4All tested samples are listed and compared in Table A.1. 3.2 Zero phonon line and phonon sideband 25 during growth, thereby reducing the lattice tension and producing the NV cen- ters [49,59]. The concentration of NV centers can be increased by high-energy radiation5 followed by annealing at high temperatures6 [61]. The high radia- tion damage produces cracks in the diamond and thereby vacancies in the diamond lattice, whereas annealing at > 700◦C makes the vacancies mobile7 leading to the migration of vacancies and their ultimate trapping by nitrogen atoms, resulting in the formation of NV centers. A suitable NV concentration can be reached by controlling the irradiation dose. However, for the samples used in this study, no alternation of the NV center concentration was needed as it was readily fabricated. This will be described in more detail in section 6.1.
3.2 Zero phonon line and phonon sideband Before considering the physical and electronic structures of the NV center, it is instructive to briefly introduce two key concepts: the zero phonon line and the phonon sideband. In a solid where atoms are locked in a lattice, the internal energy determining its temperature primarily manifests itself as vi- brational energy. In quantum mechanics, a collective vibrational energy mode in a lattice is expressed as a phonon. Thus, while an excited isolated atom releases the energy difference between the ground and the excited state as a photon, in a solid, the same energy is converted into internal heat energy rep- resented by the creation of a phonon. However, in a solid, it is also possible for the energy to be released as a combination of both phonons and photons, with the phonon release being the dominating process. The de-excitation between two energy levels generates a broad light spectrum in a solid because different fractions of released energy are transferred to the phonon for each electron de-excitation event. The broad spectrum consists of both a zero phonon line (ZPL) and a phonon sideband (PSB). While the ZPL is a narrow peak with Lorentzian line shape representing the photons that received all the available energy without the involvement of phonons, the PSB represents photons that had to share their energy with a phonon. The PSB is characterized by the ob- servation of Stokes-shifted lines, with absorption and emission at higher and lower energies, respectively. 26 Nitrogen-vacancy centers in diamond