Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Nuclear physics wikipedia , lookup
Old quantum theory wikipedia , lookup
History of subatomic physics wikipedia , lookup
State of matter wikipedia , lookup
Hydrogen atom wikipedia , lookup
Time in physics wikipedia , lookup
Photon polarization wikipedia , lookup
Condensed matter physics wikipedia , lookup
Relativistic quantum mechanics wikipedia , lookup
Theoretical and experimental justification for the Schrödinger equation wikipedia , lookup
Investigation of an Atom-Ion Quantum Hybrid System Lothar Ratschbacher St Catherine’s College University of Cambridge A thesis submitted for the degree of Doctor of Philosophy April 2013 This dissertation is my own work and contains nothing which is the outcome of work done in collaboration with others, except as specified in the text and/or acknowledgments. It has not been submitted in whole or in part for the award of a degree at the University of Cambridge or any other University. This thesis does not exceed the prescribed word limit of 60’000 words. To Natalie, Karin, Jutta and Leo Abstract Investigation of an Atom-Ion Quantum Hybrid System In this thesis I report on experiments that study the chemistry and spin evolution of single trapped Yb+ ions immersed in an ultracold cloud of 87 Rb atoms. The investiga- tions are performed in a hybrid setup that overlaps the two species in the same physical location by combining a linear Paul trap for ions and an apparatus to produce and magnetically/optically trap atomic quantum gases. With both systems well isolated from the environment, we study in detail the effects of binary inelastic collisions to identify the fundamental interaction processes between atoms and ions at low temperatures. Our experiments on reactive collisions in Yb+ +87 Rb make use of the beneficial properties of the ion-neutral hybrid system to investigate chemical processes at the most elementary level. We use state-selected single particles, precise control of the interaction, and detailed analysis of the reaction products to characterize the rate, branching ratio and energy release of various inelastic reaction channels. In particular, the charge exchange reaction rate 174 Yb+ +87 Rb→Yb+87 Rb+ is shown to have a very strong dependence on the internal electronic quantum state of the ion. We use this effect to tune the exchange reaction rate of a single ion over several orders of magnitude by exerting control over its internal quantum states. Interestingly, a high sensitivity of the charge exchange process on the hyperfine state of the atoms is also detected, which highlights the importance of the hyperfine interaction in atom-ion collisions. The investigation of the spin evolution in the hybrid quantum system represents another important part of this work. We perform measurements on the spin dynamics and decoherence of two different kinds of ion spin qubits during interaction with the neutral atomic cloud environment. For this purpose we implement quantum state preparation, manipulation and detection techniques for the hyperfine and Zeeman spin systems of the isotopes 171 Yb+ and 174 Yb+ , respectively. We observe spin coherence times T1 and T2 on the timescale of the Langevin collision rate and identify the spin-exchange interaction and an unexpectedly strong spin-nonconserving coupling mechanism as the sources of spin decoherence in our system. We further detect a strong ion-heating mechanism for atomic clouds prepared in the upper hyperfine manifold that we attribute to the release of energy due to 87 Rb hyperfine spin relaxation. The insights into inelastic atom-ion interaction and the experimental techniques described in this thesis should help to realize proposed quantum simulation and quantum information experiments in this apparatus and the steadily increasing number of atom-ion hybrid systems worldwide. Contents 1 Introduction 9 2 Theory of Ion-Atom interaction 15 2.1 Introduction to Binary Atomic Interactions . . . . . . . . . . . . . . . . . . 15 2.2 Atom-Ion Potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 The Polarization Potential, V (R → ∞) ∝ −1/R4 . . . . . . . . . . . 18 Ab Initio Potential Energy Curves V (R) . . . . . . . . . . . . . . . . 18 Atom-Ion Collisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.3.1 Elastic Collisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2.3.2 Inelastic Collisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 2.2.1 2.2.2 2.3 3 Theory of Ions and Atoms in a Paul Trap 3.1 33 Radio Frequency Traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 3.1.1 Pseudo Potential Approximation . . . . . . . . . . . . . . . . . . . . 34 3.1.2 The Linear Paul Trap . . . . . . . . . . . . . . . . . . . . . . . . . . 35 3.1.3 Linear Coulomb Crystals . . . . . . . . . . . . . . . . . . . . . . . . 37 3.1.4 Trap Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 3.1.5 Micromotion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 3.2 Micromotion Heating in Ion-Atom Collisions . . . . . . . . . . . . . . . . . 42 3.3 Effect of the Paul Trap on the Atoms 46 . . . . . . . . . . . . . . . . . . . . . 4 Experimental Setup and Methods 4.1 4.2 Preparation of Ultracold 87 Rb 49 Atoms . . . . . . . . . . . . . . . . . . . . . . 49 4.1.1 Magneto-Optical-Trap and Molasse Cooling . . . . . . . . . . . . . . 50 4.1.2 Magnetic Transport and Evaporative Cooling . . . . . . . . . . . . . 52 4.1.3 Tunability of the Atomic Bath . . . . . . . . . . . . . . . . . . . . . 53 87 Rb 4.1.4 Absorption Imaging of Atoms . . . . . . . . . . . . . . . . . . . 56 4.1.5 Density Measurement of the Neutral Bath . . . . . . . . . . . . . . . 59 Trapping of 4.2.1 Yb+ Ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 Loading of Ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 174 Yb+ 4.2.2 Doppler Cooling and Fluorescence Detection in . . . . . . . 65 4.2.3 Experimental Determination of the Ion Electronic State Occupation 67 171 Yb+ 4.2.4 Doppler Cooling of . . . . . . . . . . . . . . . . . . . . . . . 69 4.2.5 Optical Layout of the Experiment . . . . . . . . . . . . . . . . . . . 71 7 Yb+ Laser System . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Experimental Cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 4.2.6 4.3 5 Cold Chemistry with Single Particles 5.1 Chemistry in the 81 . . . . . . . . . . . . . . . . . . 82 Measurement Sequence . . . . . . . . . . . . . . . . . . . . . . . . . 82 Inelastic Collisions in Electronically Excited States . . . . . . . . . . . . . . 88 5.2.1 Collisional Quenching . . . . . . . . . . . . . . . . . . . . . . . . . . 90 5.2.2 Inelastic Collision during Fluorescence Detection . . . . . . . . . . . 91 5.2.3 Error Estimation for the Survival Method . . . . . . . . . . . . . . . 93 5.2.4 Error Estimation for the Fluorescence Method . . . . . . . . . . . . 94 5.1.1 5.2 Yb+ + Rb Ground State 6 Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms 99 6.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 6.1.1 6.1.2 6.2 Zeeman Qubit in 174 Y b+ Hyperfine Qubit in 171 Yb+ . . . . . . . . . . . . . . . . . . . . . . . . 101 . . . . . . . . . . . . . . . . . . . . . . . 108 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 6.2.1 Spin-Exchange and Spin-Relaxation in 171 Yb+ 174 Yb+ . . . . . . . . . . . . 115 6.2.2 Hyperfine Spin-Relaxation in 6.2.3 Spin Relaxation Heating . . . . . . . . . . . . . . . . . . . . . . . . . 121 6.2.4 Spin Coherence of the 171 Yb+ . . . . . . . . . . . . . . . . . . 119 Qubit . . . . . . . . . . . . . . . . . . 123 7 Outlook 127 Appendix A A.1 Spins of 133 174 Yb+ , 171 Yb+ , 87 Rb in a Magnetic Field . . . . . . . . . . . . . . 133 A.2 Rate Equation Model for Spin Dynamics in 171 Yb+ . . . . . . . . . . . . . . 135 A.3 Axial Trap Frequencies of Small Ion Crystals . . . . . . . . . . . . . . . . . 136 Chapter 1 Introduction “The whole is greater than the sum of its parts” [1]. The motivation for combining two very successful experimental platforms, single trapped atomic ions and ultracold neutral atoms to a single atom-ion hybrid system is rooted in the wealth of experimental possibilities and physical phenomena that has been predicted for this cold matter composite. For its realization, the hybrid system can resort to the precise manipulation techniques for the internal and external degrees of freedom that have been developed for both species over the last few decades. For ultracold neutral atoms, the realization of Bose-Einstein-Condensates (BEC)1 of dilute alkali vapors [2, 3] in 1995 and the observation of the first fermi-degenerate quantum gas [4] in 1999 have been key moments. The possibility to access and tune the microscopic parameters of macroscopic quantum objects has made dilute quantum gases valuable tools for simulating and testing quantum many-body models of condensed matter theory. Prominent examples of this process are the observations of superfluid flow [5], vortices [6] and matter wave interference [7] in BECs. The investigations of the superfluid to Mott-insulator quantum phase transition [8] and the BEC-BCS crossover [9] in strongly correlated bosonic and fermionic atomic gases, respectively, are remarkable successes and the field is now advancing towards the local probing of quantum phases [10]. Registers of ultracold neutral atoms have been envisioned for large scale quantum information processing and the demonstration of collisional entangling gates [11] and the recent focus on single-particle addressing [12, 10] are important steps in this direction. At the same time, the excellent coherence properties of ultracold neutral atoms have enabled applications in metrology, ranging from gravitational [13] and electromagnetic field [14] sensors to high precision atomic clocks [15, 16]. Single trapped ions have seen an equally impressive evolution in their state manipulation techniques and now arguable constitute the best-controlled single particle quantum systems2 . The long trapping times, excellent coherence properties, and exquisite control over the internal and external quantum states have put single trapped ions at the forefront of ultrahigh resolution spectroscopy [17, 18] and quantum information science. With the fidelity of fundamental quantum logic gates [19, 20] routinely exceeding the thresholds for 1 2 Nobel prize in Physics in 2001. Nobel prize in Physics in 2012. 9 Chapter 1. Introduction efficient quantum error correction [21], the biggest remaining challenge for universal quantum computation [22] in ion traps is the scaling of current architectures to system sizes beyond a few ions [23]. By harnessing the mutual interaction between these two quantum systems, an integrated ion-atom hybrid setup has the potential to enhance the capabilities of the two individual constituent quantum systems and at the same time give rise to new physical phenomena. Theoretical predictions have ranged from the sympathetic cooling [24] of the ion due to momentum transferring elastic collisions, to the modifications of the local density distribution of bosonic gases, caused by the presence of ion impurities [25, 26, 27]. Assuming accurate control of the atom-ion interaction, single trapped ions inside a degenerate quantum gas have further been suggested as powerful local probes for characterizing the properties of strongly correlated many-body quantum systems [28, 29]. Advanced future applications could also include the use of state dependent ion-atom interactions for entangling gates in ion-atom quantum processors [30, 31] and to control the tunneling of atoms through a Josephson junction [32]. Experimentally, ultracold quantum gases and trapped ions have only been combined very recently [33, 34]. Indeed, the apparatus, employed for the experiments in this thesis, has realized the first insertion of a single ion into a BEC in 2009 [33]. On immersion into the degenerate gas of Rb atoms, the sympathetic cooling of single trapped Yb+ ions has been observed [35] in combination with significant atom losses from the neutral cloud. The results have also provided a first reality check for theoretical proposals that require ions and the atoms to interact at ultralow energies. The steady state energies of the ion during elastic collisions have been found to be far in excess of the atomic bath temperature and instead limited by collisional heating effects due to uncompensated electric fields in the ion trap [36, 37]. Recent theoretical studies suggest that immersion cooled ions in Paul traps are fundamentally prevented from entering the s-wave collision regime [38] for most ion-atom combinations, except for the ones with the largest ion-atom mass ratios [39]. Current ion-atom hybrid systems, including the growing number of experiments that combine single trapped ions with ultracold atoms in magneto-optical-traps [40, 41, 42, 43], are therefore limited to the multiple partial wave regime, where collisional dynamics is well described by semiclassical theory [44]. In this regime the elastic scattering properties are indeed completely independent of the short range potential and can be understood from the asymptotic behaviour of the long range atom-ion interaction potential. In addition to elastic scattering colliding ions and atoms can also undergo inelastic events that change their internal electronic state or spin state. Charge exchange [45, 40, 46, 35, 42, 41, 47] and molecule formation [48, 25, 49, 41, 50], which belong to the first kind of inelastic events, represent simple prototypes of chemical reactions. The study of neutral-ion 10 reactions [51, 52] and molecules [53, 54] in ion traps offers the same richness of chemical interactions as the chemistry between neutral atoms, but at the same time has the key advantage that charged particles can be manipulated and analyzed for extended periods of time in the trap. In recent years the study of chemical reactions at low temperature has attracted considerable interest [55, 56, 42, 41, 57, 58, 50, 47] as the excellent level of control in cold and ultracold matter experiments has the potential to enhance the understanding and control over chemical reactions [59] at the most fundamental level. In this thesis I present experiments that investigate in detail the dynamics of the internal states of the pioneering hybrid quantum system of Yb+ ions and neutral Rb atoms. We develop experimental techniques to obtain control over the quantum states of ions and atoms in the complex setting of a hybrid experiment and apply our system to the study of cold chemistry and the dynamics of spin-bath decoherence. Our experiments on reactive collisions in Yb+ +87 Rb parallel similar efforts in a number of other hybrid combinations [41, 42, 58] that control the motional degrees of freedom of both reactants. We add to this the exact manipulation of the internal quantum state and perform studies with quantum state-selected single particles, precise control of the interaction, and detailed analysis of the reaction products to characterize the rate, branching ratio and energy release of various inelastic reaction channels [47]. In particular, the charge exchange reaction rate 174 Yb+ +87 Rb→Yb+87 Rb+ is shown to have a very strong dependence on the electronic quantum state of the ion and the hyperfine state of the neutral atoms. We show that cold reactive processes can be steered by exerting control over the internal quantum states of the ion and the atoms. At the same time, the immersion of a single ion with a spin degree of freedom into the spin polarized neutral atomic cloud also realizes the coveted concept of an isolated, controllable single qubit that couples to a well-defined and adjustable environment. The dynamics of a spin-1/2 particle interacting with its surroundings is a problem that has been intensely studied in theory [60, 61]. It represents a model system for the understanding of the quantum to classical transition [62] and plays an important role in impurity effects in solid state systems [63, 64]. Experimentally, a high level of control over the spin impurity, the tunability of the properties of the environment and a good understanding of the interaction are the key tools required to address these questions [65, 66, 67]. In our hybrid system the spin-polarized neutral cloud constitutes such a widely tunable bath environment which has both continuous (motional) and discrete (spin) degrees of freedom. The well-defined atomic nature of both the impurity and the bath offers the possibility of a precise microscopic description of their interaction. The localized spin impurities are implemented using one of two complementary ion spins: the ideally isolated two-level Zeeman qubit in the electronic ground 11 Chapter 1. Introduction state of 174 Yb+ and the clock-transition hyperfine qubit in 171 Yb+ [68]. Our experiments are the first study of the decoherence and spin dynamics in this all-atomic spin-bath system formed by the combination of ions and atoms in a hybrid setup. The thesis is structured as follows I begin in Chapter 2 by presenting the theoretical framework that exists for the description of atom-ion interactions at low collision energies and explain how it can be applied to the experimental conditions of our Yb+ + 87 Rb hybrid system. Starting with a short motivation of the binary interaction Hamiltonian, the atom-ion interaction potential and its approximations at long (the polarization potential) and at short distance (the ab initio Born-Oppenheimer potentials) will be introduced. Based on this model the predictions for the elastic and the various inelastic processes in the classical, semiclassical or quantum mechanical approximation will then be reviewed in detail. Chapter 3 discusses the physics of both ions and atoms in the linear Paul trap, which is at the centre of our hybrid experiment. In the following Sections the aspects most important for ion trap chemistry, the stability of the trap for different particle masses and the mass sensitive oscillation modes of ion crystal, are discussed in some detail. Moreover, we present both an intuitive picture and quantitative estimates for the crucial problem of micromotion induced radio frequency heating of ions in neutral buffer gas experiments, that is limiting the collision energies in current hybrid systems. In Chapter 4 the experimental apparatus for producing ultracold trapped Yb+ 87 Rb clouds and single ions will be reviewed. We will give an overview of the pre-existing experi- mental setup [69, 70] and describe the additions which have been made for the experiments in this thesis. We explain the techniques for preparing and controlling the ultracold bath of neutral atoms and the basic operation of the ion trap. Chapter 5 contains the experimental description and the results of the chemistry experiments in the ion-atom hybrid system. The techniques for state selective preparation and analysis, including in-trap mass spectroscopy are explained in the first part of the chapter. This is followed by a detailed presentation of the the various electronic state changing inelastic processes in the Yb+ + Rb hybrid system. We find a strong dependence of the charge exchange process on both the electronic state of the ion and the hyperfine state of the atoms and demonstrate how control over the internal state of single ions and neutral atoms can be employed to tune cold reaction processes. Our investigations of the spin interaction of a single ion qubit with its ultracold atomic bath are reported in Chapter 6. First, we introduce the methods we have developed for the preparation, manipulation and detection of the two different ion spin qubits in this experiment. We then study the spin dynamics and decoherence of the single spin qubit coupled 12 to its environment of spin polarized neutral atoms and observe spin coherence times T1 and T2 on the timescale of the Langevin collision rate. Spin-exchange interaction and an unexpectedly strong spin-nonconserving relaxation mechanism are identified as the sources of spin decoherence. The strong spin relaxation in our system is further shown to give rise to significant motional heating for certain atomic states of our hybrid system. In Chapter 7, I conclude this report with an outlook onto future investigations using the atom-ion hybrid setup, including some ideas for the experimental route towards the holy grail [71, 72] of atom-ion interactions: s-wave scattering. Publications in the context of this thesis Decoherence of a single ion qubit immersed in a spin-polarized atomic bath Ratschbacher, L.; Sias, C.; Carcagni, L.; Silver, J.; Zipkes, C.; Köhl, M. accepted for Physical Review Letters (2013) Controlling chemical reactions of a single particle Ratschbacher, L.; Zipkes, C.; Sias, C.; Köhl, M. Nature Physics, 8, 649 (2012) Kinetics of a single trapped ion in an ultracold buffer gas Zipkes, C.; Ratschbacher, L.; Sias, C.; Köhl, M. New Journal of Physics, 8, 053020 (2011) Cold heteronuclear atom-ion collisions Zipkes, C.; Palzer, S.; Ratschbacher, L.; Sias, C.; Köhl, M. Physical Review Letters , 105, 133201 (2010) 13 Chapter 1. Introduction 14 Chapter 2 Theory of Ion-Atom interaction A detailed understanding of the fundamental interactions of atoms and ions in the experimental regime of our cold matter apparatus is the essential foundation on which further investigations and applications can build. In this chapter I will present the theoretical framework that exists for the description of atom-ion interactions and apply it to the experimental conditions of our Yb+ +87 Rb hybrid system. I start by motivating the simplified quantum mechanical Hamiltonian of binary atomic interaction that generally serves as the starting point for the treatment of the collisional phenomena and molecular dimer physics. Next, the polarization potential V (r → ∞) ∝ −1/r4 and the numerical results of Yb+ +Rb ab initio potential curves will be introduced as the approximations for the atom-ion interaction potential at large and small atom-ion internuclear distances. Based on this model, we explore the theoretical predictions for the elastic and the various inelastic processes in the classical, semiclassical and quantum regime of Yb+ + Rb collisions. 2.1 Introduction to Binary Atomic Interactions In the context of atomic physics, the complete quantum mechanical description of the interaction of two atoms in field-free space is a many-body Schrödinger equation involving the two nuclei and all electrons of the system. Its Hamiltonian consists of the kinetic energy and the mutual interactions of all particles as well as the interactions arising from the electron spin and nuclear spin couplings. For all but the lightest atoms, the computational requirements for calculations with the complete many-body Hamiltonian are prohibitively high. In order to arrive from this first principle model to the simplified single particle Hamiltonian, that is used in the field of molecular chemistry and ultracold collisions, a number of approximations are performed, that we will shortly review based on [73]. The first method to reduce the complexity is to approximate the effect of core electrons by introducing effective core potentials for both nuclei and consider only the valence electrons explicitly. 15 Chapter 2. Theory of Ion-Atom interaction In the case of the Rb+Yb+ interaction such a Schrödinger equation would look like ~ ~ ~ ~ HΨ = − ∇2e1 − ∇2e2 − ∇2Rb − ∇2 2me 2me 2mRb 2mY b Y b e2 2e2 e2 2e2 − − − − 4πǫ0 |r e1 − RRb | 4πǫ0 |r e1 − RY b | 4πǫ0 |r e2 − RRb | 4πǫ0 |r e2 − RY b | e2 2e2 + Ψ = EΨ. (2.1) + 4πǫ0 |r e1 − r e2 | 4πǫ0 |RRb − RY b | Atom (r) = Here, we have assumed the very crude effective core potentials UECP Ion (r) = UECP 2e2 4πǫ0 |r| , e 4πǫ0 |r| ) and that correspond to a maximal shielding of the nuclei charges by the core electrons. We do not consider the spin degree of freedom at this point. Another important concept is the Born-Oppenheimer approximation. The basic idea behind it is that the electrons, due to their smaller masses, move faster than the nuclei and therefore are able to immediately adapt to the slower nuclear motion. By making use of the different time scales the problem can be split into two parts1 . First the solutions for the electronic motion are calculated while holding the position of the nuclei fixed ~ ~ e2 e2 2e2 HΨe = − ∇2e1 − ∇2e2 + − − 2me 2me 4πǫ0 |r e1 − r e2 | 4πǫ0 |r e1 − RRb | 4πǫ0 |r e1 − RY b | 2 e 2e2 − − Ψe = W Ψe . (2.2) 4πǫ0 |r e2 − RRb | 4πǫ0 |r e2 − RY b | For each fixed separation R = |RRb − RY b | of the two nuclei2 Equation 2.2 is solved by sets of electronic states with energy values W (|R|) and distinguished by their corresponding quantum numbers. These instantaneous electronic-state dependent energies W (|R|) act as potential energy curves for the motion of the nuclei and give rise to the Schrödinger equation effective core HΨn = z }| { electronic z }| { ~ 2e2 ~ 2 2 + W (|R|) Ψn ∇ − ∇ + − 2mRb Rb 2mY b Y b 4πǫ0 |R| | {z } = ẼΨn . (2.3) effective potential The result of the Born-Oppenheimer approximation is a description of the interaction of two multi-electron atoms as a problem of two structureless point particles interacting 1 Formally one arrives at the individual Equations 2.2 and 2.3 by making a product ansatz of electronic and nuclear wave functions for the total wave function in Equation 2.1 Ψ(re1 , re2 , RY b , RRb ) = Ψe (re1 , re2 , RY b , RRb )Ψn (RY b , RRb ), then neglect terms which are smaller by the order of mmReb and mmYeb and finally consider the separation of timescales of the electron and nuclei motion. 2 A change of coordinates, for example by r → r + RY b makes it evident that Equation 2.2 indeed only depends on the difference |RRb − RY b | and not on RRb , RY b individually. 16 2.2. Atom-Ion Potentials through an electronic-state dependent effective potential of radial dependence V (|R|) = UECP (|R|) + W (|R|). The introduction of centre-of-mass and relative coordinates Rcm = RRb − RY b allows to separate the wave functions Ψn (RRb , RY b ) mRb RRb +mY b RY b , R mRb +mY b iK·R cm · ΨR (R). = e = It follows that the energy eigenvalues of the system are the combination of the free particle energy spectrum ~2 K 2 2mcm for the centre-of-mass motion and the eigenvalues described by the effective one-particle Schrödinger equation HΨR = ~ 2 − ∇ + V (|R|) ΨR 2µ R = EΨR (2.4) for the relative motion of the nuclei. Equation 2.4 is the Schrödinger equation for a single mRb mY b mRb +mY b in a spherically symmetric potential. By decom∞ P (R) Pl (cos(θ)) ulR posing the wave functions into the radial and angular parts ΨR (R) = particle with reduced mass µ = l=0 the problem of binary atomic interaction is finally described by the one-dimensional radial partial-wave equation 2 2 ~ ∂ ~2 l(l + 1) − + + V (R) ul (R) = E · ul (R), 2µ ∂R2 2µR2 (2.5) that, in addition to the adiabatic Born-Oppenheimer potential energy curves V (R) specific to the electronic state of the atomic system, also includes the repulsive centrifugal term ~2 l(l+1) , 2µR2 which depends on the rotational state as specified by its azimuthal quantum number l. 2.2 Atom-Ion Potentials In the Born-Oppenheimer approximation the ion-atom interaction problem has been reduced to a set of one-dimensional second-order linear differential equations that can be solved for the scenarios of interest, once its effective potential energy terms are known. The central challenge therefore lies in the accurate determination of the atom-ion interaction potentials. For convenience the conversions of the various units commonly used by atomic physicist and chemists for energy and length scales shall be mentioned here. 1eV ⇔ 8065.73 cm−1 ⇔ 11604.9 K ⇔ 0.0367 Eh , a0 = 0.529 · 10−10 m, (2.6) where the Bohr radius a0 and the Hartree energy Eh are the units of length and energy in atomic units. 17 Chapter 2. Theory of Ion-Atom interaction The notation that we use in the following Section for the characterization of heteronuclear molecular states is the molecular term symbol 2S+1 (+/−) ΛΣ , (2.7) where S is the total spin quantum number, Λ is the projection of the orbital angular momentum along the internuclear axis, Σ is the projection of the total angular momentum along the internuclear axis and (+/−) is the reflection symmetry along an arbitrary plane containing the internuclear axis. 2.2.1 The Polarization Potential, V (R → ∞) ∝ −1/R4 At large internuclear separations the dominant interaction between an ion and a neutral atom is the attractive polarization interaction that can be understood as follows. The charge of the ion generates the inhomogeneous electric field E(R) = Q R̂ 4πǫ0 R2 at the location R of the atom. Due to the static polarizability α of the atom, the field induces the electric dipole moment p = α · E that itself interacts with the field according to F = p · ∇E = αE · ∇E. The asymptotic R → ∞ interaction between an ion and an atom is therefore described by the attractive polarization potential Uind (R) = Z∞ R F (R′ )dR′ = − αQ2 α|E(R)|2 =− ≡ −C4 /(2R4 ). 2 8πǫ0 R4 (2.8) The polarization potential is independent of the electronic state of the ion, but varies for different electronic states of the atom due to the state-dependent static polarizability α. For the electronic ground state 2 S1/2 of 87 Rb it has the value α0 = h · 0.0794(16)Hz/(V/cm)2 [74]. 2.2.2 Ab Initio Potential Energy Curves V (R) At shorter atom-ion separations the electronic wave functions of the atom and the ion are significantly modified and the interaction potentials become strongly state dependent, until at distances on the order of several Bohr radii the interaction turns repulsive. In this regime the molecular potential energy curves for different electronic states can be estimated by numerical ab initio calculations. One method to assess the accuracy of such computational models is to compare their predictions of the individual atomic spectroscopic properties with experimental results. This procedure shows that Equation 2.2 is in fact a bad starting point for the calculation of adiabatic potentials, as it would approximately produce the atomic spectra of He+ and H, instead of Yb+ and Rb for well separated atoms. However, even the most advanced ab initio quantum chemical calculation are currently not precise 18 2.3. Atom-Ion Collisions enough to predict the collisional properties of atoms and ions in the quantum regime and the energy of weakly bound molecular states. In Figure 2.1 the potential energy curves of the few lowest electronic excitations of the (Yb + 87 Rb)+ system are shown from two recent calculations [75, 76]. The direct comparison of the potential energy curves from the two independent quantum chemical calculations reveals differences at the level of ≈ 1000 cm−1 between the two sets of data. We interpret this as an estimate for the accuracy of the current state of the art quantum chemical calcu- lations. The discrepancies between the calculated potential energies at large internuclear separations and the well-know atomic energy levels of the separated atom and ion are in the range of ≈ 100 cm−1 . It is evident from Figure 2.1 that the entrance channel of electronic ground state Yb+ ions and 87 Rb atoms, which we prepare in most of our experiments, is in fact not the absolute electronic ground state of (Yb + exchanged channel Yb(6s2 1 S) 87 Rb)+ + 87 Rb+ system. The R → ∞ asymptote of the charge- is located 16752 cm−1 (2.08 eV, kB · 24216 K) lower in energy as a result of the difference in the ionization potentials of Yb and Rb. All other molecular potential curves in Figure 2.1 are electronic excitations of the chargeexchanged combination Yb + 87 Rb+ . These states correspond asymptotically to excitations of the neutral Yb atom, since the excitations of Rb+ amount to transitions in the extreme ultraviolet due to the ion’s Kr-like noble gas shell structure. The lowest of the chargeexchanged excitations, which dissociates to the fine structure triplet Yb(6s6p, 3 P0J ) and Rb+ , lies asymptotically just above the entrance channel but crosses it at shorter internuclear distances. The Bohr-Oppenheimer potential shown have been obtained from scalar-relativist calculations and therefore do not account for spin-orbit coupling effects. In addition, the quantum chemical calculations are limited to excited state potentials with energies below the level of Yb+ (4f 13 )(6s2 ) + Rb(6s) (see Figure 5.3). This state, that represents the lowest electronic excitation of the Yb+ ion cannot be efficiently accounted for by the theoretical techniques used in the two studies due to its open f -shell. Its presence also prevents the accurate computation of potential energy curves for any higher-lying excited states [75]. 2.3 Atom-Ion Collisions The potential energy curves of Figure 2.1 indicate, that even for low energy collisions of Yb+ ions with 87 Rb atoms, several different types of processes can occur. Our theoretical description will focus on the description of binary collisions of a single ion and a single atom (Table 2.1). Three-body collisions of one ion and two atoms, that have recently been observed [77], are not included into our treatment as they do not play an observable role 19 Chapter 2. Theory of Ion-Atom interaction 40000 30000 Energy (cm-1) Rb+ + Yb (6s6p 1 P) Rb+ + Yb (6s6p 3 P) 20000 Rb(5s) + Yb+ (6s) 1 Π 1 + Σ 10000 3 Π 3 + Σ Σ 3 + Σ 1 + Σ 1 + 0 Rb+ + Yb (6s2 1S) ) 5 10 15 20 25 30 R (a.u.) Figure 2.1: Potential energy curves for the (Yb + 87 Rb)+ system. The molecular singlet (1 Σ+ ) and triplet (3 Σ+ ) channels that asymptotically connect to the state of separate Yb+ (6s)+87 Rb(5s) are not the electronic ground of the system. The triplet (3 Σ+ ) channel, corresponding to the charge exchanged state Yb(6s2 1 S)+87 Rb+ , lies 16752 cm−1 (2.08 eV, kB · 24216 K) below. The graph shows a comparison of the datasets from two different ab initio calculations (lines [75]), (lines+points [76]) and yields an estimate for the precision of current quantum chemical methods on the order of 1000 cm−1 . To the right of the graph the (r → ∞) asymptotic energy levels as determined from atomic spectroscopy are indicated. The internuclear separation on the horizontal axis is given in atomic units. 20 2.3. Atom-Ion Collisions in the dynamics of the hybrid system at our experimental atomic densities. Yb+ + Rb → Yb+ + Rb Yb+ | ↑i Yb+ | ↓i + + Rb | ↓i Rb | ↑i → → Yb+ | ↓i Yb+ | ↑i + + Rb| ↑i Rb| ↓i ) spin exchange Yb+ | ↑i ··· + ··· Rb | ↑i ··· → ··· Yb+ | ↓i ··· + ··· Rb| ↑i ··· ) spin relaxation Yb+ Yb+ Yb+ + + + Rb Rb Rb Yb + Rb+ + γ Yb + Rb+ (Yb Rb)+ + γ → → → elastic radiative charge transfer non-radiative charge transfer molecule formation Table 2.1: Possible elastic and inelastic processes during binary collisions of Yb+ ions and Rb atoms. 2.3.1 Elastic Collisions In the case of purely elastic scattering the internal states of the atom and the ion do not change and the outgoing channel is per definition identical to the incoming channel. Elastic collisions in the singlet or triplet potential of the incoming channel lead to the transfer of momentum and kinetic energy in the laboratory frame and are therefore of importance for sympathetic cooling. The total elastic cross section at the collision energy E = ~2 k 2 /(2µ) is given by [78] σel. = ∞ X l=0 ∞ 4π X (2l + 1)sin2 (ηl ) σl = 2 k (2.9) l=0 as the sum over all partial-wave contributions. The partial-wave phase shifts ηl are determined by the continuum solutions ul (R) of the partial-wave equation 2.5. The asymptotic behavior of the solutions ul (R) at large distances is of the form lπ ul (R) ∼ sin kR − + ηl (k) 2 (2.10) and defines the elastic scattering phase shifts ηl . For the lowest partial-wave (l = 0, s-wave), the potential energy term in Equation 2.5 21 Chapter 2. Theory of Ion-Atom interaction simply corresponds to the adiabatic atom-ion interaction potential. The potential energy terms of higher partial waves (l ≥ 1) contain the centrifugal contribution combination with the attractive asymptotic polarization V ≈ −C4 height3 ~4 l2 (l + 1)2 Ecf (l) ≈ 8µ2 C4 at radial positions For collision energies E ≪ Ecf (l = 1) = E ∗ = rcf (l) ≈ ~4 , 2µ2 C4 /(2R4 ) s ~2 l(l+1) , 2µR2 that in form barriers of 2µC4 . + 1) ~2 l(l (2.11) only the l = 0 partial wave contributes to the elastic scattering, as for higher partial waves the effect of the inner part of the potential is suppressed by the centrifugal barrier (see Figure 2.2). In the s-wave limit the scattering cross-section approaches the energy independent value σel. = 4πa2 , where a is the zero temperature scattering length. The direct determination of a = limk→0 tan(η0 (k)) k by quantum mechanical calculations [79] using the previously presented ab initio potentials and Equations 2.5 and 2.10, is not possible. Ab initio potential energy curves are not accurate enough to calculate the scattering length from first principles. Without any knowledge about the details of the short range potential, a statistical argument can be made, that the median of the probabilityqdistribution4 of possible cross sections has the value ∼ 4πr∗2 , 4 where r∗ = rCF (l = 1) = µC is the characteristic length of the polarization potential [80]. ~2 In the case of atom-ion collisions, the characteristic energy for the occurrence of the s- wave threshold behavior is in the nano-Kelvin temperature range as a result of the long range of the 1/r4 potential5 . The value for collisions of Yb+ + Rb is E ∗ = kB · 44 nK. Whereas these temperatures can, in principle, be achieved for the ultracold buffer gases, the significantly higher energies of the ion motion in current experiments (see Section 3.2) result in the contribution of many partial waves in atom-ion collisions. Semiclassical Approximation In the many-partial wave regime the energy-dependent cross section for elastic collisions can be estimated using a semiclassical approach [44]. To this end the evaluation of the partial wave phase shifts in Equation 2.9 is split into two different parts and independent approximations are applied in each case. For large values of the angular momentum quantum numbers l the partial wave phases are assumed to be insensitive to inner part of the potential due to the centrifugal barrier. In 3 The result is readily obtained by finding the stationary points of the potential energy term Epot (R) ≈ −C4 /(2R4 ) + ~2 l(l + 1)/(2µR2 ) in Equation 2.5. 4 The actual value of the scattering length a can still take any value depending on the exact details of the short-range potential. In fact the probabilities for positive or negative scattering lengths are equal in the case of the polarization potential [80]. 5 The characteristic energies for the s-wave threshold behavior in neutral-neutral collisions are typically in the ∼ 100 µK range due to the much shorter range of the ∼ 1/r6 van der Waals potential 22 2.3. Atom-Ion Collisions Figure 2.2: Centrifugal barriers for higher order partial waves. The red bar at 750 a0 indicates the energy distribution and spatial extend of the wavefunction for an ion-atom collision with an average kinetic energy on the order of kB · 50 mK. this case the phase shifts are determined by the long range potential and can be calculated from the semiclassical approximation for phase shifts [44] µ ηl ≈ − 2 ~ Z ∞ R0 6 V (R) , dR p 2 k − (l + 1/2)2 /R2 (2.12) where R0 is the classical outer turning point. The evaluation of this equation for the V (R) = −C4 /(2R4 ) potential yields 7 ηl ≈ πµ2 C4 E . 4~4 l3 (2.13) On the other hand, partial waves with small values of l, get affected by the inner part of the potential and their phase shifts ηl quickly become large as l decreases. As we are taking the sum over many partial waves with rapidly varying phase shifts ηl the terms sin2 (ηl ) in the elastic cross section (Equation 2.9) can be approximated by the average value 1/2. If we split the evaluation of the elastic cross section into two parts, such that Equation 2.13 and sin(ηl ) ≈ ηl apply for l > lsplit and that sin2 (ηl ) ≈ 1/2 is a good approximation for 6 The formula is derived by applying the WKB approximation to spherically symmetric potentials and includes several further approximations (see [81] p √ and references therein). √ R 7 The integral formula 1/(R3 R2 − A)dR = R2 − A/(2A · R2 ) − arctan( A/(R2 − A))/(2A3/2 ) and the approximation for the classical turning point R02 ≈ l(l + 1)/k2 ≈ (l + 1/2)2 /k2 are used during this evaluation. 23 Chapter 2. Theory of Ion-Atom interaction l < lsplit , then the cross section of Equation 2.9 reads σel. (E) = ≈ 4π k2 lsplit −1 X (2l + 1)sin2 (ηl ) + l=0 ∞ X l=lsplit lsplit −1 4π X 1 (2l + 1) + k2 2 l=0 Z∞ lsplit 2π 2 ≈ 2 lsplit (1 + ηl2split ). k (2l + 1)sin2 (ηl ) dl 2l ηl2 (2.14) It has been confirmed by quantum numerical calculations [44] that the approximations for both regimes are well satisfied, if lsplit is chosen such that ηlsplit = π/4 and as long as lsplit > 50. In this case we obtain from Equation 2.13 lsplit = µ2 C 4 E ~2 1/3 (2.15) and finally σel. (E) = π µ C42 ~2 1/3 π2 1+ 16 E −1/3 . (2.16) The cross section shows a E −1/3 dependency on the collision energy, which implies that the elastic collision rate of an ion in a neutral gas of density na γel. = na σel. (E) v ∼ E 1/6 (2.17) scales as E 1/6 in the many-partial wave scattering regime. In practice, measurements of collisional cross sections cannot be performed at single welldefined energies in our atom-ion system. The broad energy distribution8 that the ion acquires during immersion in a neutral cloud means that the theoretical results need to be energy averaged for comparison with the observed experimental rates. The approximations underlying the semiclassical calculation start to fail for collision energies such that lsplit < 50. For Yb+ + Rb this corresponds to a lower limit of E ≈ kB ·11µK for the validity of semiclassical result. In the region of collision energy between the s-wave (E ∗ = kB ·44nK) and the semiclassical (E ≈ kB ·11µK) thresholds the contributions of only few partial waves and the occurrence of shape resonances lead to a strong variation of the elastic cross section with energy. Quantum mechanical calculations based on the multi-channel treatment of atom-ion collisions, have shown that already in the semiclassical regime the effect of Feshbach resonances 8 This energy distribution is generally not thermal, as will be explained in Section 3.2. 24 2.3. Atom-Ion Collisions should be observable in ion-atom collisions [38, 82]. Feshbach resonances, that have become a valuable tool to control the interaction in ultracold quantum gases by tuning the magnetic field [83], would indeed also lead to very interesting experimental possibility in ion-neutral collisions. Theoretical studies, albeit for a different combination of ions and neutral atoms (Ca+ + Na), have demonstrated that a significant number of Feshbach resonances are expected for comparatively low magnetic field strengths in the ∼ 10 G range. Their effect at finite temperatures should be particularly pronounced for the inelastic charge exchange collision rates. 2.3.2 Inelastic Collisions Ion-atom encounters also lead to a number of inelastic processes, that change the internal states of the colliding particles. Depending on the original state of the colliding pair, these internal state changes can involve both the spin and the electronic degree of freedom of the atom and the ion. A first estimate for the energy dependent cross section of various inelastic events can be made by considering the fact that inelastic dynamics is dominated by interactions at short internuclear distances. In the semiclassical model of collisions, the inner part of the potential is only sampled, if the energy of the kinetic energy of the colliding pair of particles exceeds the height of the centrifugal barrier. From Equation 2.11 it follows, that for a given collision energy E, it is possible for partial waves with angular momentum l < lmax (E) ≈ (8µ2 C4 E)1/4 ~ to contribute to the inelastic cross section. Classically this means that collisions with a maximal angular momentum ~lmax (E) and therefore with a critical impact parameter rcritical = lmax /k = 2C4 E 1/4 (2.18) can give rise to inelastic events. The corresponding cross section and collision rate read 2 σclassic = πrcritical =π r 2C4 E and γclassic = na σclassic v = na · 2π s C4 = γLangevin . µ (2.19) Note, that a simple summation of partial waves (with sin2 (ηl ) = 1/2) up to lmax in the quantum mechanical cross section of Equation 2.5 would have resulted in a two times larger value of the cross section. We will later use this result in the estimation of the spin-exchange cross section. The classical scattering rate of Equation 2.19 is independent of the collision energy and only depends on the long range properties of the potential. It has been originally derived by Langevin [84] in 1905 in the context of the mobility of ions in gases. A purely classical solution of particle trajectories in the ∼ − Cr44 potential [85] shows that for impact param- 25 Chapter 2. Theory of Ion-Atom interaction eters r < rcritical the colliding particles attract each other to r = 0 in an inward spiraling motion, whereas for r > rcritical the particles merely deflect from each other. The probability for a specific inelastic collision to occur once the ion and the atom approach each other to close distances, depends, of course, on the details of the interaction potential. It is therefore useful to introduce the relative inelastic efficiencies ǫi,j to express the rate, at which particles in the state j undergo certain inelastic process i, as γi,j = ǫi,j · γLangevin . (2.20) In this model, the rate of inelastic processes is assumed to be independent of energy and to be limited by the Langevin rate. For Yb+ + Rb the numerical value of Langevin rate constant is γLangevin /na = 2.115 · 10−15 m3 /s, which for typical atom densities on the order of n ∼ 1018 m−3 results in Langevin collision rates on the order of 103 s−1 . Spin-Exchange Collisions Spin-exchange collisions are encounters, where electronic spin is transferred from one atom onto another and the spin orientations of both particles are altered after the event. In the case of a Rb atom and a Yb+ ion in their respective electronic ground states this simply corresponds to the exchange of spin orientation between the two valence electrons9 Rb| ↑ia + Y b+ | ↓ii → Rb| ↓ia + Y b+ | ↑ii Rb| ↓ia + Y b+ | ↑ii → Rb| ↑ia + Y b+ | ↓ii , (2.21) where | ↑i ≡ |S = 1/2, ms = 1/2i and | ↓i ≡ |S = 1/2, ms = −1/2i for both the atom and the ion. The physical origin of the effect lies in the fact that particles in symmetric electronic spin states interact differently with each other compared to those in antisymmetric states. For a given spin symmetry, the Pauli principle imposes the opposite symmetry for the spatial component, as the total wave function of the two electron system needs to be antisymmetric. The Coulomb interaction between the electrons lifts the degeneracy of the spatial configurations by the exchange energy and results in two potential energy curves. √ In Figure 2.3 the singlet potential Vs (R) for the spin state 1 Σ+ : 1/ 2(| ↑a ↓i i − | ↓a ↑i i) and √ the triplet potential Vt (R) for the spin states 3 Σ+ : | ↓a ↓i i, | ↑a ↑i i, 1/ 2(| ↑a ↓i i − | ↓a ↑i i) are highlighted for the incoming channel of electronic ground state Yb+ and Rb. These curves can be written as a single interaction potential in terms of the spin operators S a 9 In ultracold atoms coherent spin-exchange interactions have been intensely studied in spinor condensates [86] and used for entanglement generation [87]. For atoms and ions, coherent interaction is not yet possible, since in the current energy regime many partial waves contribute to collisions. 26 2.3. Atom-Ion Collisions and S i of the individual electron spins and the total spin S ≡ S a + S i : V (R) s V (R) = V (R) t for 1 Σ+ (S = 0) for 3 Σ+ (S = 1) 3 1 1 1 1 1 + 2 (S 2 − S 2a − S 2i ) + Vs (R) − 2 (S 2 − S 2a − S 2i ) 4 ~ 2 4 ~ 2 1 1 3 (2.22) = Vt (R) + Vs (R) + 2 Vt (R) − Vs (R) S a · S i , 4 4 ~ = Vt (R) where we have used S 2 ≡ (S a +S i )2 = S 2a +S 2i +2S a ·S i and S|S, ms i ≡ ~2 S(S +1)|S, ms i. In general, colliding atom-ion pairs start in superpositions of the singlet and triplet state. During collisions the phases of the singlet and triplet contribution of the wave function evolve according to their potentials, and as a result the spin directions of the atoms and ions will change. The total spin of the system S = S a + S i , however, commutes with the potential in Equation 2.22 and m = ma + mi remains a good quantum number throughout the interaction. The possible changes of the atom and ion electron spin are limited to ∆ma = −∆mi = ±1. These results can be applied to isotopes with nuclear spin by expanding the atom’s and ions’s hyperfine states in their respective electronic spin basis |F, mF i = X M =mI +mS F Cm |S, mS i|I, mI i, I mS (2.23) F where the terms Cm are the Clebsch-Gordan coefficients. The spin-exchange interaction I mS only couples the electronic and not the nuclear spins and can therefore only give rise to the transitions within the individual hyperfine structures with ∆F = −1, 0, 1 and ∆mF = ±1. However, atoms and ions colliding in fully stretched states |Fa = ±(Ia + 1/2), mF,a = Fa i + |Fi = ±(Ii + 1/2), mF,i = Fi i, only project into the electronic spin triplet state and cannot undergo spin exchange collisions. So far, we have neglected the role of the kinetic energy in the inelastic spin-exchange collisions. Indeed, if the asymptotic energy splitting between different internal states is much smaller than the collision energy of the particles, the partial wave solution for the spin-exchange cross section can be expressed in terms of elastic scattering partial wave phase shifts ηls and ηlt in the singlet and triplet potential [88] σSE = ∞ π X (2l + 1) sin2 (ηls − ηlt ) k2 (2.24) l=0 The singlet and triplet potential only differ for small internuclear distance inside the centrifugal barrier. Thus for l > lsplit the phase shifts are equal ηls = ηlt and for l < lsplit we can make again the approximation sin2 (ηls − ηlt ) ≈ 1/2. By summing up the Equation 2.24 27 Chapter 2. Theory of Ion-Atom interaction and considering the factor of 2 of Equation 2.19, we obtain the semi-classical spin-exchange cross section as σSE = 1 σLangevin . 4 (2.25) If the collision energy gets small enough to approach the Zeeman or Hyperfine splittings, the degenerate internal state solution for the total-spin exchange cross section is not a good approximation any more. In this case, the spin-exchange rates between the nonresonant internal states become asymmetric, favoring the exothermic over the endothermic spin transitions. As we will see in Chapter 3, the typical ion-atom collision energies in our hybrid system indeed fall into this intermediate range, where the energetic suppression of endothermic spin transitions plays an important role. Spin-Relaxation We have shown, that stretched polarization states, such as 174 Yb+ |1/2, 1/2i + 87 Rb|2, 2i, are immune to spin-exchange collisions. The spins of atoms in these states are still affected by spin-relaxation, if the spherical symmetry of the interaction potential that underlies the conservation of the electron spin is broken due to the presence of other interaction mechanisms10 . One such interaction, that will always be present, is the magnetic dipoledipole coupling of the two valence electronic spins [89] V (R) = µ0 µ2B gs2 µ0 (µ · µ − 3(µ · R̂)(µ · R̂)) = (S i · S a − 3(S i · R̂)(S a · R̂)), i a 4πR3 i a 4πR3 (2.26) where R = Ri − Ra , R̂ = R R, and µ0 , µB and gs are the vacuum permeability, the Bohr magneton and the Landé factor, respectively. At small internuclear separations this interaction lifts the degeneracy of the three 3 Σ+ states with spin S = 1 and spin projections Ω = 0, ±1 and leads to the spin relaxation [90]. As there are, to our best knowledge, no theoretical or experimental studies, that consider spin-relaxation phenomena in ion-atom collisions, we base much of this analysis on what is known about the analogous neutral system of two colliding alkali metal atoms. In ultracold electronic ground state collisions of light alkali atom the spin-spin interaction is indeed the leading contribution to the relaxation of spin stretched states [90]. In binary collisions of heavier alkali-metal atoms, such as cesium, a mechanism based on spin-orbit coupling becomes the dominant cause for spin-relaxation. Spin-orbit coupling, the interaction between the spin of a particle and its motion, plays 10 In the most general case only the total angular moment, which consist of the spin degree of freedom, the electronic orbital angular momentum, and the orbital angular momentum of the collision, will be a conserved quantity of the problem. 28 2.3. Atom-Ion Collisions an important role in many physical systems. The best known example of this interaction is the fine-structure splitting of the electronic energy levels of atoms and molecules. In the moving frame of an electron, the electric field of the nucleus is seen as a magnetic field that couples to the spin degree of freedom. The same principle is at work for mobile electrons and carriers in condensed matter systems, where it gives rise to the interesting physics of the spin-hall effect [91] and topological insulators. The recent implementation of effective spin-momentum coupling in ultracold atomic systems has opened the opportunity to study these effects using the excellent control techniques in the toolbox of quantum optics [92, 93, 94]. Crucially, the relativistic spin-orbit coupling mechanism also breaks the electronic spin conservation and gives rise to spin relaxation phenomena. For two atoms collide in their 2S 1/2 electronic ground states, the four molecular states 3 Σ+ and 1 Σ+ that asymptotically correspond to this atomic state, do not have any electronic orbital momentum. For this reason no first order spin-orbit coupling L · S = 0 is present in these collisions. At short internuclear distances, however, the three 3 Σ+ states with spin S = 1 and spin projections Ω = 0, ±1 are split by a second order spin-orbit coupling, that is mediated by couplings to distant electronic states with non vanishing electronic orbital angular momentum [90]. The effect has been shown to increase with atomic mass and for cesium atoms the contribution of the second order spin-orbit coupling exceeds that of the spin-spin interaction of Equation 2.26. In the context of ultracold atomic collisions, spin-relaxation can have a significant impact on the stability of ultracold gases. The spin-relaxation rate during collisions of 87 Rb atoms in the |F = 2, mF = 2ia is sufficiently small compared to the elastic collision rate to allow for efficient evaporative cooling of atoms in our magnetic trap. For 133 Cs atoms in the magnetically trapable |4, 4ia of |3, −3ia states, on the other hand, the binary spin- relaxation collisions dominate the inelastic loss behavior. The unfavourable ratio of inelastic spin-relaxation and elastic collision rates γelastic /γSR ≈ 100 has made it difficult to reach quantum degeneracy for the magnetically trapable states of 133 Cs [95]. Spin relaxation effects of similar magnitude have also been found for optically pumped alkali metal atoms in vapor cells at room temperature [96]. The comparison of different alkali metals confirms the increase of spin relaxation effects with the mass of the colliding particles. In conclusion, the analogy with the neutral two-valence electron system of alkali metal collisions could suggest that also for atom-ion collisions the ordering γelastic ≈ γSE ≫ γSR of elastic, spin exchange and spin relaxation rates should be conserved in a wide range of regimes. The experimental results in Chapter 6 will, however, show that this is strongly violated in Yb+ + Rb collisions. 29 a 30000 Energy (cm-1) Chapter 2. Theory of Ion-Atom interaction 20000 b 30000 Energy (cm-1) Σ+ 1 10000 3 + Σ Σ+ 1 20000 10000 1 + 0 5 Σ 0 10 15 20 R (a.u.) 25 30 5 10 25 15 20 R (a.u.) 30 Figure 2.3: Molecular states primarily involved in spin-exchange a and radiative association b. Potential energy curve data from [75]. b a Yb++Rb Yb +Rb+ 2+ Yb 2+ Yb + Yb+*+Rb + Yb+ +Rb Rb Rb 2+ Yb 2+ Yb + Rb + Rb Figure 2.4: Pictorial representation of the charge exchange reaction a and quenching from an excited state b. Shown are the filled core electronic shells and the relevant valence electrons. Reactive collisions Inelastic collisions that lead to transitions between electronic states in the ion and/or atom can be classified into two categories. If the inelastic event entails a chemical change of the collision partners, it constitutes a genuine reactive collision. On the other hand, if the collision only involves the (de-)excitation of the electronic state of the ion and/or the atom, we refer to it as an inelastic quenching event. In Figure 2.4 examples for each of the two processes are illustrated. There are in general three different pathways that can lead to a transition between chemically distinct states. Non-radiative charge transfer (NRCT) is mediated by non-adiabatic couplings between molecular states. In radiative charge transfer (RCT) the transition to a continuum state of a lower lying potential energy curve is induced by the emission of a photon. Finally, radiative association describes the transition to a bound state of a lower lying potential by emission of a photon. In the later case, YbRb+ molecular ions in vibra- 30 2.3. Atom-Ion Collisions tionally excited levels are formed (see Figure 2.3b and 2.5). For the entrance channel (Rb(5s) + Yb+ (6s)) the only energetically lower lying molecular potential energy curve corresponds to the charge-exchanged (Rb+ +Yb (6s2 )1 S) state. The dominant chemical reaction pathways are radiative charge transfer and radiative association between the 1 Σ+ excited state and the 1 Σ+ ground states. Radiative processes from the 3 Σ+ state are forbidden by symmetry and non-adiabatic coupling between (Rb(5s) + Yb+ (6s)) 1 Σ+ and (Rb+ +Yb (6s2 )1 S) 1 Σ+ is negligible due to the large energy gap. Nonradiative charge transfer is also not possible via the higher lying (Rb+ + Yb (6s6p 3 P)) channel. Its potential exhibits a crossing with the entrance channel potential for both the singlet and the triplet potential, but due to the asymptotic energy gap of > 500 cm−1 the channel is energetically closed for cold collisions. a b Figure 2.5: Radiative charge-exchange and radiative association. a The overall radiative cross section depends on the collision energy as ∼ E −1/2 and therefore leads to an energy independent reaction rate. The rich structure in the cross section is caused by centrifugal barrier tunneling (shape) resonances of the many partial waves that are contributing. b Frequency resolved radiative cross section. Free-free (ff) transition to continuum states above the dissociation limit of the (Rb+ +Yb (6s2 )1 S) potential correspond to radiative charge-exchange. Free-bound (fb) transitions lead to the formation of (RbYb)+ molecular ions with large vibrational quantum numbers. Figure adapted from [76]. In a recent theoretical study the charge exchange dynamics of Yb+ + Rb system has been investigated in a parameter regime relevant to current experiments [76]. The ab initio 31 Chapter 2. Theory of Ion-Atom interaction calculations reproduce the experimentally determined, energy-independent, reaction rate in the electronic ground state of ≈ 10−5 γLangevin [35, 97] (see Section 5.1). In addition, the theory predicts a relative strength of the radiative association and the radiative charge exchange pathway as 70 : 30. In Figure 2.5 the frequency resolved radiative cross section is plotted for an assumed collision energy of 0.01 cm−1 (≈ 14 mK). The data shows that all charge exchange products should have kinetic energies of less than 500 cm−1 and that all radiative associations will lead to the formation of molecules with large vibrational quantum numbers. We will compare these theoretical predictions with our experimental results on reactive collisions in Chapter 5. 32 Chapter 3 Theory of Ions and Atoms in a Paul Trap In order to investigate the various interactions between neutral atoms and charged ions without disturbance, it is necessary to isolate them from undesired effects due to the environment. The idea of studying the coupling between the two species at very low temperatures therefore relies on techniques to suspend, confine and cool both neutral and charged atoms and to overlap them in the same physical location. The development of these techniques has undergone tremendous progress since the inception of charged particle traps by Paul and Dehmelt in the 50s [98] and the pioneering works on laser cooling and neutral particle traps in the 80s [99]. In this Chapter I will give a short outline of the basic principles of the radio frequency trap that is at the heart of our cold-matter hybrid experiment. Since the general physics of ions in Paul traps is covered extensively in the literature [100, 101, 102], I will focus in the following on the aspects most relevant to ion-atom hybrid traps. 3.1 Radio Frequency Traps Ion traps are devices that confine charged particles in effectively conservative potentials over sustained periods of time. There are two main types of traps for charged particles, that due to their net charge, interact strongly with static electric and magnetic fields. In Penning traps charged particles are trapped by a combination of static electric and strong magnetic fields. In contrast, in Paul traps, such as the one implemented in our experiment, the confinement results from the time averaged action of an inhomogeneous radio frequency electric field and from static electric potentials. Quantum optic experiments routinely employ both Penning and Paul traps to trap charged particles at low energies for high precision metrology and quantum information processing purposes. In cold matter quantum hybrid experiments, so far, only Paul traps have been implemented. The main disadvantages of the use of Penning traps for atom-ion hybrid systems are the high magnetic fields, that strongly couple to the neutral atoms, and the fact that the localization of ions is more difficult to achieve [103, 104]. 33 Chapter 3. Theory of Ions and Atoms in a Paul Trap 3.1.1 Pseudo Potential Approximation A fundamental understanding of the confining effect, that ions experience in a Paul trap, can be obtained in the so called pseudo potential approximation [36]. The classical equation of motion of a particle with charge Q and mass m in an electric field E(r, t) is given by r̈ = Q E(r, t). m (3.1) It is evident from ∇ · E = 0 that charges cannot be confined in vacuum in all three dimensions only by electrostatic fields1 . However, the application of non-uniform timevarying electric fields can give rise to restoring forces in all three dimensions, under the right conditions. We start by considering a sinusoidally oscillating electric field of frequency Ω and spatial dependency E0 (r) E(r, t) = E0 (r) · sin(Ωt). (3.2) In a spatially uniform field, the mean force over one oscillation period averages to zero and returns the particle to its original position and motional state. For an inhomogeneous field, on the other hand, there remains an averaged force acting on the particle [101]. If the length scale Q mΩ2 · E(r) is small enough (see Section 3.1.4), the trajectory of the particle is well described by decomposing it into a fast oscillatory excursion δr and a slowly drifting mean position r D r(t) = r D (t) + δr(t). (3.3) The equation of motion 3.1 now reads Q (E 0 (r D + δr)) sin(Ωt) m ! ∂E 0 Q sin(Ωt), E 0 (r D ) + δr = m ∂r r=rD r̈ D + δr̈ = (3.4) (3.5) where we have Taylor-expanded the electric field around r D assuming δr ≪ r D . By making 0 and integrating Equation 3.5, the approximations δr̈ ≫ r̈ D and E 0 (r D ) ≫ δr ∂E ∂r r=r D we arrive at the solution for the fast oscillating motion δr = − 1 QE 0 (r D ) sin(Ωt). mΩ2 (3.6) The discovery of this fact predates the formulation of Maxwell’s equation and is known as Earnshaw’s theorem. 34 3.1. Radio Frequency Traps Inserting this term into Equation 3.5 and averaging over one field oscillation period 2π/Ω yields the expression for the drift motion of the particle hr̈ D + δr̈i = r̈ D QE 0 (r D ) Q ∂E 0 1 Q2 E 20 (r D ) =− = −∇ , m ∂r r=rD 2mΩ2 m 4mΩ2 (3.7) where we have introduced the conservative pseudo potential U (r D ) = Q2 E 20 (r D ) . 4mΩ2 (3.8) The time averaged force on a charged particle in an inhomogeneous oscillating electric field is also known as the ponderomotive force and is independent of the sign of the particles charge. The slow motion of the particle in the pseudo potential is called secular motion, whereas the fast superimposed oscillation at frequency Ω (Equation 3.6) is termed micromotion. 2 VDC,0 ~VRF,0 Figure 3.1: Schematic of the electrode configuration in a linear Paul trap. The radio frequency voltages are applied between the pairs of diagonally opposed linear electrodes and give rise to radial trapping of the charged particle. DC voltages are applied to the electrodes on either end of the quadrupole waveguide to provide confinement in the axial direction. 3.1.2 The Linear Paul Trap The Paul trap used in our experiments is a linear quadrupole trap, that provides radio frequency confinement in the radial direction and electrostatic trapping along the axial direction. It consists of four parallel rod electrodes surrounding the trap axis and one electrode on either end of the trap axis. A sinusoidally oscillating potential in the radio frequency domain is applied between the two pairs of diagonally opposed electrodes and equal static potentials are put on both 35 Chapter 3. Theory of Ions and Atoms in a Paul Trap end-cap electrodes (see Fig. 3.1). The electric potential at time t created by four hyperbolic electrode surfaces with infinite extension is the ideal quadrupole VRF (x, y, t) = VRF,0 ( x2 − y 2 ) cos(Ωt), 2R02 (3.9) where VRF,0 and Ω are the amplitude and frequency of the electrode voltages and 2R0 is the distance between diagonally opposed electrodes. In general, the need for optical access and Figure 3.2: Numerical simulations of the electrostatic potential generated by our ion trap geometry a, in comparison to the potential of an ideal quadrupole with identical electrode spacing b. manufacturing restrictions require more realistic geometries, which also generate higher order multipoles. The reduced quadrupole contributions of the potential can be taken into account by replacing VRF,0 with an effective geometry dependent voltage amplitude κRF · VRF,0 . Figure 3.2 shows the simulated radial electric potential inside our ion trap in comparison to that of an ideal quadrupole trap with identical electrode spacing during an antinode of the radio frequency voltage. According to Equation 3.8 the oscillating electric quadrupole of Equation 6.6 in the centre of the trap gives rise to the radial harmonic confinement U (r D ) = 2 Q2 κ2RF VRF,0 4mR04 Ω2 (x2 + y 2 ) (3.10) for a charged particle. In order to confine the particle along the axial direction, the two end-cap electrodes are held at identical static voltages. The electrostatic potential they generate at the centre of 36 3.1. Radio Frequency Traps the trap is given by VDC (x, y, z) = κDC VDC,0 2 1 2 1 [z − (x + y 2 )] = C · [z 2 − (x2 + y 2 )], 2 2 2 Z0 (3.11) where κDC is a geometry-dependent factor and Z0 is the distance to the end caps. This static electric quadrupole potential traps the ion in the axial direction, but it also leads to the weakening of the confinement in the radial direction. The total conservative potential for the secular motion of a single ion U (x, y)+QVDC (x, y, z) results in the harmonic trap frequencies ωz = 3.1.3 s 2QκDC VDC,0 , mZ02 ωx,y QκRF VRF,0 = √ 2mR02 Ω s 1− κDC VDC,0 mR02 Ω2 . 2 Qκ2RF VRF,0 Z02 (3.12) Linear Coulomb Crystals Ion Equilibrium Positions When multiple motionally cold ions are located in a single trapping potential, the ions arrange into stable ion crystals to minimize their potential energy [105]. If the radial frequency of the trap is sufficiently large compared to its axial frequency, the ions line up in a linear chain. The potential energy of such a string of N ions consists of the overall confining potential of the trap and the mutual Coulomb repulsion between ions and is given by [106, 107] V = X1 i 2 QCzi2 + 1 e2 X , 8πǫ0 |zi − zj | (3.13) i,j;i6=j where zi denotes the position of the ith ion and C (Equation 3.11) describes the strength of the axial confining potential, that traps a single ion of mass m with frequency ωz = p QC/m. The minimization constraint ∂V =0 ∂zi zi =z (0) (3.14) i leads to the set of equations ui − i−1 X j=1 N X 1 1 + = 0, 2 (ui − uj ) (ui − uj )2 (0) for the equilibrium positions zi l3 i = 1, ..., N, (3.15) j=i+1 = ui · l, where we have introduced the length scale = Q/(4πǫ0 a0 ) to obtain the normalized equilibrium ion positions ui . In general the solution to Equation 3.15 has to be found numerically, but for binary 37 Chapter 3. Theory of Ions and Atoms in a Paul Trap crystals the analytic expression z1,2 = l · u1,2 = p 3 Q/(4πǫ0 a0 ) · (± r 3 s 1 Q2 )=±3 4 16πǫ0 mωz2 (3.16) exists. The equation is useful, for example, to obtain a calibration of the scale of fluorescence images, once the axial trap frequency ωz has been measured of an ion of mass m. The equilibrium positions of ions in an axial crystal are independent of their mass and therefore do not convey information about the chemical composition of any non-fluorescing ions. In order to identify dark ions in an ion crystal we make use of the mass dependence of the ion crystal’s motional frequencies. Axial Eigenmodes At low energies the ions oscillate with small amplitudes around their respective equilibrium positions. Their motion, which is strongly coupled by the Coulomb interaction, can be described in terms of eigenmodes of the entire ion string2 . The Langrangian for small oscillations of a string of N ions, with N-1 ions of mass m and a single impurity ion of mass mn at position n is given by the expression [106, 107] N N m X 2 mn 2 1 X ∂ 2 V qi qj , q̇ − q̇i + L= 2 2 n 2 ∂zi ∂zj qi =0 (3.17) i,j=1 i=1;i6=n where q(t)i is the excursion of ion i around its equilibrium position. To obtain the displacement eigenvectors and axial eigenfrequencies of the linear crystal the equation is numerically diagonalized. The results for the lowest eigenmode of ion crystals with up to 4 ions are summarized in the Appendix A.3 The largest relative changes of eigenmode frequencies due to a single impurity ion are obtained for ion crystals of two ions, for which the analytic expression ω com breath q p = ω · (1 + µ) ∓ 1 − µ + µ2 can be derived for the centre-of-mass and the breathing-mode frequencies, where µ = (3.18) m mn is the ion mass ratio and ω is the axial trap frequency of a single ion of mass m. In Section 5.1.1 we will describe the experimental implementation of an in-trap mass spectrometry method, which is based on the mass dependence of the COM eigenmode frequency of binary ion crystals. 2 This mechanism is of great importance for ion trap quantum information tasks. In combination with the photon-recoil induced coupling of the internal and motional degrees of freedom of ions, the strong Coulomb interaction forms the basis for multi-ion entanglement gates [19] and precision spectroscopy schemes [17]. 38 3.1. Radio Frequency Traps 3.1.4 Trap Stability In order to study the range of trap parameters, for which the pseudopotential approximation is valid and stable ion trajectories are possible in our linear Paul trap [102, 108], we start with the equation of motion 3.1 for an ion in the total electric field E(x, y, z, t) = −κRF VRF,0 κDC VDC,0 xx̂ − y ŷ cos(Ωt) − (2z ẑ − xx̂ − y ŷ) 2 R0 Z02 (3.19) and rewrite it in the form ∂ 2 ri + [ai + 2qi cos(2τ )]ri = 0, ∂τ 2 1 τ = Ωt, 2 (3.20) where 4QκDC VDC,0 1 , ax = ay = − az = a = − 2 mZ02 Ω2 2QκRF VRF,0 , qz = 0. qx = −qy = q = mR02 Ω2 (3.21) (3.22) This equation is known as the Mathieu equation and its stable solutions in terms of the dimensionless stability parameters a and q are well understood [108, 102]. Figure 3.3 shows the region of stable solutions for the linear Paul trap in green. Practically all ion traps operate in the regime a, q 2 ≪ 1, for which the solutions of the Mathieu equation 3.20 to first order in a and q 2 are [102, 109] with qi ri (t) = r1i cos(ωi t + Φi ) 1 + cos(Ωt) . 2 (3.23) r 1 1 ωi = Ω ai + qi2 . 2 2 (3.24) The slow secular harmonic oscillations along the three trap axis at frequencies ωi are identical to the expressions of Equation 3.12, that were found in the pseudo potential approximation. The amplitudes r1i and phases Φi of the ion motion are determined by the starting conditions. In the radial directions, micromotion oscillations at the frequency Ω of the quadrupole field are superimposed onto the secular motion. The limits of the stability diagram for small a, q 2 in Figure 3.3 can be understood from Equation 3.24. For a > 0 the voltage on the end caps has the wrong polarity with respect to the charge of the ion and attracts the ion towards the end cap electrodes. The low boundary is approximately given by a ≈ − 21 q 2 ⇒ ωr = 0, where the radial RF confinement can just counterbalances the deconfinement due to the axial electrostatic quadrupole. The 39 Chapter 3. Theory of Ions and Atoms in a Paul Trap Table 3.1: Trap parameters of our ion trap under typical operating conditions. R0 is the distance between the ion and the radio frequency electrode, Z0 the distance between the ion and the DC end cap, and Ω/(2π) the radio frequency drive frequency. The radial (axial) trap frequencies ωr /2π (ωz /2π) and the stability parameters a0 , q0 are specified for a Yb+ ion with mass m0 = 174 amu. R0 0.466 µm Z0 4.3 mm Ω/(2π) 42.88 MHz ωr /(2π) 150 kHz ωz /(2π) 55 kHz a0 −3.3 · 10−6 q0 1.02405 · 10−2 dotted curve crossing the stability region at about a ≈ − 31 12 q 2 (⇒ ωr ≈ ωz for a single ion) represents an estimate of the lower boundary for a binary ion crystal to still crystallize along the axial direction. For large values of q 2 , Equation 3.28 is no longer a good approximation for the trajectories of the ion. The breakdown of the pseudo potential approximation on the right boundary of the a-q stability region is highlighted by the fact, that the radial confinement frequency would approach the drive frequency ωr ≈ 21 Ω for a ≈ 1 − 12 q 2 . According to the definitions of a and q in Equation 3.22, stable trapping conditions in a Figure 3.3: The stability diagram for the linear Paul trap in various levels of detail ac. The green shaded area marks regions in a-q space that give rise to stable solutions of the Mathieu’s equation. For our typical ion trap operating conditions, the stability parameters of singly charged positive ions of different masses fall onto the red line defined by Equation 3.25. specific ion trap depend on a range of physical quantities. The geometric constants R0 , Z0 , κDC , κRF and Ω are fixed by the design of the ion trap and its resonant radio-frequency voltage generating circuitry. The operating parameters VRF,0 , VDC,0 can be used to set certain stability parameters a0 , q0 for an ion of mass m0 . The a, q values for ions with 40 3.1. Radio Frequency Traps masses mn then fall onto the line an = a0 qn q0 with qn = q0 m0 . mn (3.25) 2 ωz For typical operating conditions of our trap, we have determined the values a0 = −2 Ω 2, q 1 2 √ 2 ωr − 2 ωz from ion trap frequency measurements with an 174 Yb+ ion (see Table 3.1 q0 = 8 Ω2 and Section 5.1.1). The mass relation is shown as the red line in Figure 3.3 with a selection of relevant ion masses. Our ion trap is expected to be stable for all singly charged ions with masses in the range 2 amu < m < 2500 amu and to support axial crystals of two ions of equal mass for m < 900 amu. The mass limit for the axial alignment of a binary crystal consisting of one Y b+ ion m0 = 174 and a heavier ion of mass m is m < 1300 amu. Modifications to the stability diagram due to higher order multipole fields have not been considered here, as they do not affect the stability limits of large ion masses (i.e. a, q 2 ≪ 1) [110]. 3.1.5 Micromotion In addition to the slow oscillations at the secular trap frequencies of Equation 3.24, the ion trajectories along the radial trap axes (Equation 3.28) also exhibit a fast oscillating micromotion at the frequency of the radio frequency driving field. This residual motion is proportional to the local radio frequency electric field and hence depends on the position of the ion in the quadrupole trapping field. If the ion trajectories are centered on the zero of the quadrupole the amplitude r1i q2i of the intrinsic micromotion is reduced, as the secular motion r1i decreases during laser cooling. In this case, the total kinetic energy of the ion averaged over the period of one secular oscillation EKin 1 2 2 1 2 2 qi2 1 2 1 2 2 = m ṙi = mr1i ωi + qi Ω = mr1i ωi 1 + 2 2 4 8 4 qi + 2ai (3.26) consists of a secular motion term and a contribution of similar size from the intrinsic micromotion. There are several mechanisms that can increase in the size of micromotion relative to this ideal case. If a static electric background field radially displaces the minimum of the time averaged potential ion trap potential from the centre of the radio frequency quadrupole by QE of f set = −∇Upseudo ri =r0i 41 ⇒ r0i = QEof f set , mωi2 (3.27) Chapter 3. Theory of Ions and Atoms in a Paul Trap then additional excess micromotion is superimposed on the ion motion according to qi ri (t) = (r0i + r1i cos(ωi t + Φi )) 1 + cos(Ωt) . 2 (3.28) Since the micromotion is a driven motion, its amplitude is set by the strength of the driving force and cannot be cooled effectively. In order to minimize the amount of excess micromotion, the electrostatic background fields at the location of the ion need to be detected and compensated with the application of the corresponding electrostatic potentials to the trap electrodes [109]. Excess micromotion can additionally be caused by radio frequency fields pointing along the trap axis. The misalignment of radio frequency electrodes as well as the pick-up of radio frequency on the end-cap electrodes can both give rise to axial micromotion. A further source of radial excess micromotion can be phase differences between the potentials applied to diagonally opposite radio frequency electrodes due to, for example, a mismatch in the length of the connecting wires. For well compensated electrostatic background fields, axial micromotion and out-of-phase radial micromotion remain the dominant micromotion contributions and should therefore be minimized through careful design and precise construction of the ion trap. For many experiments with single ions in spectroscopy and quantum information science it is sufficient to reduce the amplitude of the excess micromotion uemm,i to below the amplitude of the secular motion (ground state) to suppress its detrimental effects on line shapes and gate fidelities. This condition corresponds to an excess micromotion kinetic 2 larger than the secular kinetic energy. Since both energy that can be a factor of Ω ω motions of the ion are well decoupled, there is in general no significant heating effect of the secular motion from the radio frequency driving field. 3.2 Micromotion Heating in Ion-Atom Collisions The situation is significantly altered for ions in the presence of neutral atoms, where elastic collisions strongly couple the motion of ions and neutral atoms. It has been realized early on in buffer gas experiments, that micromotion in Paul traps can give rise to both heating and cooling effects for an ion inside the neutral gas [36]. An intuitive insight into the effect of micromotion on the kinematics of buffer gas cooled ions can be obtained by considering the motion of an ion and its collisional processes in phase space. The explanation follows [111], to illustrate the problem for a simplified onedimensional geometry. In the absence of micromotion, the harmonic secular motion of an ion corresponds to a elliptical trajectory in phase space, where the enclosed area is proportional to the energy 42 3.2. Micromotion Heating in Ion-Atom Collisions a Harmonic Oscillator p p x x mi>mn p mi=mn Paul trap b p x x π/2 π/2 Figure 3.4: Illustration of collisional energy transfer of a trapped particle in one dimension. a In a harmonic potential without micromotion, collisions with neutral atoms at rest can only lead to a decrease of ion energy. The energy loss is indeed fastest for equal masses of the ion and the atoms. b A maximally rotated and stretched ellipse represents the phase space distribution of an ion in a Paul trap at the exemplary phase ξ = π2 of the radio frequency field. Depending on the size of the momentum transfer, collisions with neutral atoms at rest can lead to both heating or cooling of the secular motion of the ion. (Figure adapted from [111]) of the ion (Figure 3.4a). Following a head-on hard sphere collision with a neutral atom at rest3 , the instantaneous position of the ion remains unchanged, but some of its initial momentum c · m is transfered. The ion’s energy and momentum c′ · m after the collision are given by ∆E = E ′ 2 c ma m , − 1 = −4 c (ma + m)2 (3.29) where ma , m, c and c′ , are mass of the atom, the mass of the ion, and the instantaneous velocity of the ion in the laboratory frame before and after the collision, respectively. Due to the presence of micromotion, the distribution in phase space of the position and the total momentum of an ion in a Paul trap is represented at each phase ξ = mod(Ωt, 2π) of the radio frequency gradient field E0 cos(Ωt) by rotated and streched ellipses, as shown in Figure 3.5. During one radio frequency period the phase space trajectory of the ion motion samples the continuum of equal-area ellipses, that lie between the extremal cases of ξ= π 2 and ξ = 3π 2 , always returning to points on the same ellipse at times of identical radio frequency phase ξ. Figure 3.4b shows how a random collision event at a radio frequency phase ξ = 3 π 2 and at an ion position close to the turning point, instantaneously changes the The kinetic energies associated with buffer gas are assumed to be much lower than the motional energy of the ion. 43 Chapter 3. Theory of Ions and Atoms in a Paul Trap 3π/2 pharm pmic 3π/2 xharm p x xmic π/2 π/2 macromotion + micromotion = combined motion Figure 3.5: For any phase ξ of the radio frequency field the phase space distribution of the instantaneous velocity of the ion in a Paul trap is given by a rotated and stretched ellipse due to the presence of micromotion. The phase space trajectory of the ion (thick black line in the panel on the right) returns to a given ellipse for times of identical radio frequency phase and progresses clockwise around points on that ellipse within one period of the harmonic oscillation. (Figure adapted from [111]) area of the ellipse and therefore the secular energy of the ion. The phase-space picture demonstrates, that depending on the amount of transfered momentum (i.e. the mass ratio m ma ), the phase of the micromotion and the phase of the secular motion, collisions can lead both to an increase or a decrease of the secular ion energy. In order to arrive at a more quantitative analysis for the three dimensional case, we retrace the original treatment by Dehmelt [36]. The velocity of an ion in the laboratory frame after a purely elastic collision with a neutral gas atoms at rest is c′ = m ma |c|Θ + c, ma + m m + ma (3.30) where c is the velocity just before the collision and Θ is a unit vector that spans an angle of Θ with c in the centre-of-mass frame. Assuming a collision time much shorter than 2π Ω, the velocity before c = u + v and after the collision c′ = u′ + v can be decomposed into the secular velocities u, u′ before and after the collision and the field amplitude dependent micromotion velocity v = Q mΩ cos(Ωt). Substitution, squaring and regrouping leads to the relation u′2 − u2 = −2 2ma ma m (u2 + 2u · v + v 2 )(1 − cos Θ) + (v 2 + u · v − |c|Θ · v). (3.31) 2 (m + ma ) m + ma To describe the heating and cooling dynamics of the ions secular motion, the average secular ′2 2 per collision is the relevant quantity. Since the Langevin energy change m 2 · u −u 44 3.2. Micromotion Heating in Ion-Atom Collisions collision4 rate (Equation 2.19) is independent of the collision velocity, we can consider the random phase of the driving field during collisions to be uniformly distributed. Averaging over the radio frequency phase ξ yields the expression h∆Eion iξ = 2 m ′2 mma 2ma 2 u − u2 ξ = −2 u + v 2 (1 − cos Θ) + v (1 − cos Θ) 2 2 (m + ma ) m+a (3.32) for the secular energy change per collision, where we have made use of the fact hu · viξ = 0. The average secular energy of the ion reaches a steady state (h∆Eion iξ = 0) for hEion iξ ≡ m 2 ma m 2 u ξ= v ξ. 2 m 2 (3.33) For the illustrative example of a purely harmonically trapped ion in a homogeneous5 radio frequency field E(t) = E RF cos(Ωt), the quantity v 2 ξ is indeed independent of 2 u ξ and the average secular energy of the ion will be centered around the value hEion iξ = m ma Q2 E2 m 4m2 Ω2 RF (3.34) due to collisions with neutral atoms. The ideal unbiased hyperbolic 3D Paul trap with ax = ay = az = 0 and qz = −2qx = −2qy is another instructive case, for which a simple exact solution can be found. For the collisionless motion of an ion in a Paul trap, Equation 3.26 described the size of the intrinsic micromotion kinetic energy, relative to the secular kinetic energy, averaged over one secular oscillation. If the trap frequency of the ion is larger than its collision rate, then the relation 2 2 v ξ,sec = q2q+2a u2 ξ,sec ≈ u2 ξ,sec remains a good approximation also in the presence of neutral atoms. Inserting this expression into Equation 3.32 and considering the isotropic angular distribution of Langevin scattering for the averaging over the scattering angle Θ, we find that that the average relative change of secular energy ∆Eion Eion =2 ξ,sec,Θ m a m −1 ma m (m + ma )2 (3.35) strongly depends on ion atom mass ratio. In the presence of intrinsic micromotion, Dehmelt’s model predicts exponential heating due to collisions with the neutral buffer gas, if the ion is lighter than the atom (m < ma ). 4 Only Langevin collisions are relevant, since small angle scattering collisions do not transfer significant amounts of kinetic energy. 5 This situation could arise experimentally, if recent experiments with ions confined in optical traps [112] were combined with neutral atoms in the presence of radio frequency noise. 45 Chapter 3. Theory of Ions and Atoms in a Paul Trap No average energy change occurs for equal masses m = ma , and exponential cooling is expected, if the ion is heavier than the atom (m > ma ). The mass ratio ma m = mRB mY b = 1 2 in our experiment is clearly favourable for fast collisional cooling. The cooling process in this case would only be limited by the temperature of the ultracold buffer gas, which we have neglected so far. Due to the large difference in the trapping frequencies of the ion and the neutral atoms, this picture would result in a large ground state occupation probability of p(n = 0) = 1 − e − k~ωT B for the ion. Unfortunately, excess micromotion in a Paul trap introduces a limit for the ion energy that is many orders of magnitude higher in current ion traps . In analogy to Equation 3.34, the radio frequency field amplitude E RF at the centre point of the ion’s secular motion will give rise to a steady-state average ion energy hEion iξ = ζ · ma Q2 · m 2 2 E 2RF . m 4m Ω (3.36) The exact value of the proportionality constant ζ, which will be on the order of 1, depends on the trap stability parameters a and q, the neutral-ion mass ratio, the cause of the nonzero radio frequency field (i.e. uncompensated static offset fields, radio frequency phase mismatches, axial micromotion, etc.) and on the potentially inhomogeneous density distribution of the neutral buffer gas. In recent years Monte-Carlo simulations have become the method of choice for the calculation of heating and cooling dynamics due to elastic collisions in Paul traps [113, 37, 43, 114]. Interestingly, these calculations have revealed significant deviations of the ion steady-state energy distribution from the thermal Maxwell-Boltzmann distribution. The likelihood of higher energies is significantly enhanced in the form of long range power-law tails of the energy distribution and increases for larger neutral to ion masses [113, 37] Experimental studies in our ion trap have confirmed the quadratic dependence of the ion energy on the excess micromotion amplitude and detailed calculations have indicated average ion energies in the (≈ kB · 10 mK) range, that are limited due to uncompensatable axial micromotion in our trap [37, 69]. 3.3 Effect of the Paul Trap on the Atoms We have seen in Section 2.2.1, that atoms due to their static polarizability α are suscepti2 . In the electric ble to electric fields via the induced dipole potential Uind (r) = − α|E(r)| 2 quadrupole field of a Paul trap (Equation 3.19) the atoms experience the deconfining potential Udec (r) = hUind (r)i = 2 2 α κRF VRF,0 2 mRb 2 2 ω r r =− 2 4 2 dec R0 46 (3.37) 3.3. Effect of the Paul Trap on the Atoms in the radial direction, where we have neglected the contribution from the static axial quadrupole of order a2 q2 and we have averaged over one cycle of the radio frequency. The comparison with Equation 3.10 shows that the ions radial trap frequency6 and atoms radial deconfinement “frequency” are related according to ωRb,dec mY b =√ ωY b mRb s αΩ2 . Q2 (3.38) When turned on, the ion trap reduces q the confinement frequency of the atoms in their 2 2 optical or magnetic trap by ωON = ωOF F − ωRb,dec For our trap parameters the deconfinement effect on the neutral atoms ωRb ≈ 0.0001 · ωY b ≈ 2π · 15 Hz is starting to limit the maximally attainable ion trap frequency. Further increases of the ion confinement would impede experiments with atoms in our current magnetic trap, that has radial trap frequencies ωON, |2,2i ≈ 2π · 30 Hz and ωON, |1,−1i ≈ 2π · 10 Hz for 87 Rb atoms in their |F = 2, mF = 2i and their |F = 1, mF = −1i hyperfine states respectively. In the optical dipole trap the characteristic neutral atom trap frequencies are significantly higher (≈ 100 Hz) and ion trap frequencies in the MHz range would be possible. 6 Here, we neglect the radial deconfinement of the ion due to the static axial quadrupole potential of order a . q2 47 Chapter 3. Theory of Ions and Atoms in a Paul Trap 48 Chapter 4 Experimental Setup and Methods The experiments in this thesis are performed in a cold matter hybrid trap that has allowed for the first time the investigation of the interaction of a single trapped ion with a Bose Einstein Condensate [33]. The realization of this atom-ion hybrid system relies on an apparatus that can simultaneous prepare single trapped Yb+ ions and clouds of ultracold 87 Rb atoms and overlap them under well defined conditions in the same physical location. For this purpose the tools and methods to generate both species need to be carefully joined in a single experiment with a special focus on their interoperability. In this chapter I introduce the experimental apparatus and the techniques that support the basic functionality of the atom-ion hybrid system. The design and implementation of the hybrid trap is described in much more detail in the PhD theses [70, 69]. I begin by discussing the preparation of the neutral clouds and present the methods we use to manipulate and analyze the ultracold samples. Next, I will describe the trapping and laser cooling of Yb+ ions in our setup. We will then review the laser system of the hybrid trap and focus in particular on the modifications and additions to the original design for the experiments in the Chapters 5 and 6. 4.1 Preparation of Ultracold 87 Rb Atoms Since the first realizations of Bose-Einstein Condensates (BEC) in ultracold atomic gases almost 20 years ago [2, 3] the experimental technologies have undergone a steady development, however, the basic route for reaching ultracold temperatures and ultimately quantum degenerate have not changed much. In our experiments the critical steps, namely, laser cooling of atoms in a Magneto Optical Trap (MOT), followed by evaporative cooling of the atomic cloud under ultra-high-vaccum conditions, are implemented in the two-chamber vacuum system shown in Figure 4.1. The MOT-chamber on the left of the picture is connected to the interaction chamber, where the ion trap is located, by a differential pumping section. The laser cooled clouds of atoms are moved between both chambers by a magnetic transport over a distance of 140 mm. 49 Chapter 4. Experimental Setup and Methods z x y Figure 4.1: Layout of the central part of the vacuum chamber system and the magnetic coils in our ion-atom apparatus. Atoms are collected and laser cooled in the MOT-chamber, (enclosed by the pair of quadrupole coils shown in yellow) before they are magnetically trapped and transported through the differential pumping section. The continuous displacement of a step magnetic field minimum from the position of the MOT to the centre of the QUIC-trap (formed by the magnetic fields of the red quadrupole coils and the Ioffe coil in magenta) is aided by an overlapping pair of quadrupole coils in blue. The coil in pink, the pair of black coils and a similar pair of coils in the vertical direction (not shown), provide homogenous magnetic offset fields to align the postion of the magnetic trap with respect to the centre of the ion trap. 4.1.1 Magneto-Optical-Trap and Molasse Cooling Our experimental sequence starts with laser cooling of 87 Rb atoms in a 3D MOT [99] from thermal rubidium vapour. The rubidium pressure in this vapour cell section of vaccum chamber is kept constant at ≈ 1.5 · 10−9 mbar by firing rubidium dispensers once every experimental cycle. Six counterpropagating laser beams of 1.8 cm waist and a total power of ≈ 300 mW intersect at the centre of a quadrupole magnetic field with an axial gradient of 10.6 G/cm. The cooling laser light is red detuned by three natural line widths (3ΓD2 ≈ 18 MHz) from the 2 S1/2 , F = 2 → 2 P3/2 , F = 3 hyperfine transition of the rubidium D2-line. Off-resonant excitation to 2 P 3/2 , F = 2 state leads to the leakage of population into the 2 S1/2 , F = 1, from where it is returned into the cooling cycle with laser light that is resonant with the repump transition 2 S1/2 , F = 1 → 2 P 3/2 , F = 2 (see Figure 4.2). After collecting approximately 109 atoms at the Doppler-cooling limit of T ≈ ~ΓD2 2kB = 140 µK over a 20 s interval, the MOT magnetic field is turned off. The atom tempera- 50 4.1. Preparation of Ultracold 87 Rb Atoms ture is then further reduced during an optical molasse stage [115] in cooling light, that is 6 ΓD2 detuned from resonance. Subsequently, the atoms are optically pumped into the low field seeking 2 S1/2 , F = 2, mF = 2 state with σ + polarized light1 resonant on the 780nm F=2 52S1/2 87 6.8 GHz F=1 Rb Repumping 5 P3/2 F=3 267 MHz F=2 F=1 F=0 opt.pump. 2 Imaging trap. F = 2 → 2 P 3/2 , F = 2 transition, before they are loaded into a magnetic quadrupole Cooling 2S 1/2 , Figure 4.2: Hyperfine level structure of the 87 Rb D2-line. The various transitions relevant for the manipulation of the neutral atoms are indicated. 87 Rb Laser system system Frequency and intensity controlled coherent laser light to address the various transitions of the D2-line for trapping/cooling, optical pumping and resonant absorption imaging of 87 Rb atoms (see Figure 4.2) is produced in a purpose-built optical setup [70, 69]. For experimental convenience and in order to decouple the two tasks, the production of laser light and its final alignment on the atoms are separated on two different optical tables. Polarization maintaining single mode optical fibers are used to spatially filter the light and deliver it from the optical table housing the lasers to the second optical table, which holds the experiment’s vacuum chamber. The D2-line transitions at 780 nm are conveniently addressed with semiconductor diode lasers. In our setup, commercial Fabry-Perot laser diodes are operated in home-built Littrow-type external cavity diode lasers (ECDL), which are based on a blueprint by the Hänsch group [116]. We employ the same laser design [70] for the various Fabry-Perot laser diodes2 , that generate the light for the manipulation of the Yb+ ions (see Section 4.2.6). The laser system consists of three of these ECDL. The first Master laser is locked onto the 2 S1/2 , F = 2 → 2 P 3/2 , F = 2 transition using Doppler-free frequency-modulation 1 2 A small magnetic offset field is applied to define a quantization axis for the atoms. The collimation lenses and feedback diffraction gratings are adapted to the various wavelengths. 51 Chapter 4. Experimental Setup and Methods spectroscopy [117] in a saturated 87 Rb - vapour cell. A second ECDL is locked to this reference laser with a positive frequency offset [118] that is tunable around ≈ 266 MHz to address (2 S1/2 , F = 2 → 2 P 3/2 , F = 3) MOT cooling transition. Its light is passed through a tapered amplifier, in order to increase its power to the levels required for the MOT. The third laser provides the light for the 2 S1/2 , F = 1 → 2 P 3/2 , F = 2 repumping process during the MOT and absorption imaging and is directly locked to an independent spectroscopy vapour cell. Tunable and fast switchable light for the absorption imaging of the atoms (Section 4.1.4) on the 2 S1/2 , F = 2 → 2 P 3/2 , F = 3 transition is obtained by double-passing Master laser light through an Acousto-Optical-Modulator (AOM), that is adjustable in frequency around 133 MHz. In order to achieve short pulses at the original frequency of the Master and Repumper laser for optical pumping and imaging, we send the light through two successive single-pass AOM of opposite order. Finally, a portion of the Master laser light is coupled into an intermediate finesse transfer cavity. The length of this transfer cavity is stabilized to the frequency of the laser and is in turn used as a reference for the frequency locking of some of the infrared Yb+ lasers (see Section 4.2.6). 4.1.2 Magnetic Transport and Evaporative Cooling After the free expansion of the atoms during the molasses phase and the subsequent optically spin pumping, the magnetic field of the MOT coils is ramped up to form a magnetic quadrupole trap for the low field seeking states of 87 Rb. In order to avoid heating and losses from collisions with the room temperature background gas, the atoms are then magnetically transported within 1.4 s to the ultra-high-vacuum (∼ 2 · 10−11 ) of the interaction chamber. The continuous displacement of the centre of the magnetic quadruple trap is achieved by ramping the currents of three overlapping quadrupole pairs (see Figure 4.1). Once the atoms arrive in the interaction chamber, the magnetic trap is converted into a QUIC trap [119] configuration. This Ioffe-Pritchard type trap features a non-zero magnetic field minimum in order to avoid non-adiabatic spin flip losses of the atoms. In this trap we perform microwave induced evaporative cooling of the atoms by driving transitions from the 2 S1/2 , F = 2, mF = 2 to the untrapped 2 S1/2 , F = 1, mF = 1 hyperfine states (see Figure 4.4). The hottest atoms of the continuously rethermalizing cloud sample the highest magnetic fields and fall out of the trap as their Zeeman shifted transition becomes resonant with the gradually lowered frequency of the microwave radiation between 6.9 GHz and 6.8 GHz. After ≈ 30 s of evaporation we typically obtain ultracold clouds of up to ∼ 3 · 106 atoms at T =∼ 700 nK. 52 4.1. Preparation of Ultracold 87 Rb Atoms Transport into the ion trap At this point the atomic cloud is small enough to be transported through the φ = 600 µm hole in the end cap of the linear Paul trap. The atoms are moved into the ion trap along its axial direction by shifting the magnetic trap with additional currents in the axial coils (see Figure 4.1). The precise alignment of the final atomic cloud position inside the ion trap is achieved by the application of offset magnetic fields in all three directions. In this final magnetic configuration the trapping frequencies for atoms in the 2 S1/2 F = 2, mF = 2 state have been measured by exciting dipole oscillations as ωr = 2π · 28 Hz and ωaxial = 2π · 8 Hz in the radial and axial directions, respectively. 4.1.3 Tunability of the Atomic Bath All the steps described so far remain very much unchanged regardless of the details of the experiments we perform with the ultracold atoms. The only parameters that are routinely changed are the duration of the molasses phase and the endpoint frequency of the microwave evaporation. Increases in the free-expansion time after the MOT are used to reproducibly decrease the number of atoms in the final ultracold cloud without changing its temperature. The endpoint of the microwave evaporation, on the other hand, affects both atom number and temperature, with longer evaporations leading to a higher overall central density of the cloud in the magnetic trap. In order to prepare the atomic bath in hyperfine spin states other than the initially generated 2 S1/2 F = 2, mF = 2 state, the atoms need to be transferred from the magnetic trap into an optical dipole trap. The Optical Dipole Trap The optical dipole trap [120] is formed by two far-off-resonant Gaussian laser beams, that cross at an angle of 90◦ at their mutual focus point in the very centre of the ion trap (Figure 4.3). The laser light at 1064 nm is red detuned from both D-lines of 780 nm and 795 nm and provides a conservative attractive potential3 . 87 Rb at The horizontal laser beam is focused to an elliptical waist of dimension 160 µm × 64 µm. The beam is confined stronger in the vertical direction to provide sufficient support for the atoms against gravity, while keeping the trapping frequencies along the horizontal axis lower. In this way, the density of neutral atoms in the centre of the trap is minimized and the cloud size maximized for a given number of atoms in the optical trap. This condition is favourable for our experiments, where the atoms act as a controlled bath for the ion, since it minimizes the effects of cloud depletion and misalignment. The vertical beam has a circular focus 3 Heating due to spontaneous scattering of photons is negligible on the experimental timescale. 53 Chapter 4. Experimental Setup and Methods with a waist of 110 µm. The atoms are loaded into the optical trap by simultaneously increasing the power from 0 W to ∼ 3 W in both beams in a quadratic ramp of 1 s duration. By lowering the power in the horizontal beam, the atoms are evaporatively cooled, as the hottest atoms leave the trap in the vertical direction due to gravity. After approximately 5 s of optical evaporation, we can obtain almost pure condensates of up to 150’000 atoms in the 2 S1/2 F = 2, mF = 2 state. Bose Einstein condensates, however, are not very convenient as an atomic bath for the controlled immersion of ions due to their high atomic density and atom-ion collision rate, and their relatively small size. For this reason, all experiments reported in this thesis are performed with non-degenerate atomic clouds. The number of 87 Rb atoms in the thermal clouds, their temperature and trap frequencies, as well as their exact vertical position4 , depend on the evaporation values used in a specific experimental sequence. Typical trap frequencies for the atoms in the optical trap are ωx ≈ 2π · 50 Hz, ωy ≈ 2π · 50 Hz and ωz ≈ 2π · 75 Hz. It is worth noting that the optical dipole trap will of course also have an effect on the trapped ions. The extra potential due to the far-red detuned optical dipole trap is, however, completely negligible with respect to the ions’ strong confinement in the ion trap. In order to fine adjust the alignment of the thermal cloud in all three spatial dimensions, the horizontal and vertical dipole laser beams are each reflected off computer-controlled twoaxis piezo mirrors. For this purpose two standard optical mirror mounts have been fitted with low-voltage piezo-stacks, which are powered by computer-interfaced piezo drivers5 . The tip and tilt angles of the mirrors of α ≈ 3 µrad/V displace the laser foci by ∆d = α · ∆V · f , where f denotes the focal length of the focusing lens. The actual displacements of the trap have been determined experimentally from calibrated in-situ absorption images of the atoms. The maximum computer controlled excursions of the trap are ∆dx = 48 µm, ∆dy = 34 µm and ∆dz = 154 µm. Figure 4.3 shows a scan of the position of the dipole trap with respect to the location of the ion. The implementation of the piezo mirrors and temperature stabilization (±0.1 K) of the cooling water for the ion trap and the magnetic field coils have reduced the long term drifts of the relative position to a value smaller than our measurement precision (∼ 1 µm). The laser light for the dipole trap at 1064 nm is generated by a 500 mW YAG laser6 and then amplified by a fiber amplifier7 . The combination yields a maximum optical power of ∼12 W out of a single polarization maintaining fiber. The light is subsequently distributed 4 due to the gravitational sag Thorlabs MDT693A 6 Innolight Mephisto 7 Nufern 5 54 4.1. Preparation of Ultracold 87 Rb Atoms z y x Atom Number 150000 120000 90000 60000 30000 -15 -10 -5 0 5 10 15 Displacement y [μm] Figure 4.3: Alignment of the crossed optical dipole trap with respect to the ion. a Illustration of the relative positions of the magnetic trap coils, the ion trap and the dipole laser beams. b Loss of atoms from the neutral cloud due to elastic collisions with the ion (interaction time 150 ms). The relative position of the neutral cloud and a single ion is scanned by displacing the optical trap. Figure a adapted from [70]. into three different laser beams, two of which are used for the optical dipole trap. Each beam is sent through a single-pass Acousto-Optical-Modulator and a mechanical shutter, before it is coupled into a polarization-maintaining single mode fiber and delivered to the experiment. The operating frequencies of the AOMs are detuned by several tens of MHz in order to avoid interference effects, when the laser beams are spatially overlapped for the formation of the crossed dipole trap8 . The AOMs are used for the fast switching of the light and serve as the control elements in the closed-cycle feedback loops that stabilize the dipole beam powers at the position of the atoms to computer-controlled set values. The sensor inputs to the loops are provided by photo-diodes that detect proportional fractions of the light fields, once its polarization has been filtered after the optical fibers. The feedback algorithm is currently implemented by an analog PID controller circuit. Preparation of spin-polarized atomic clouds The use of the spin-independent optical dipole trap allows the confinement of atoms in any state of the 87 Rb electronic ground state spin manifold, shown in Figure 4.4. Of particular interest as atomic bath states are the states |F = 2, mF = −2i, |F = 2, mF = 2i, |F = 1, mF = −1i and |F = 1, mF = 1i, that are protected against spin-exchanging atom- atom collisions due to spin conservation or a hyperfine energy barrier9 . The population 8 The beams are also set to have different polarization, in order to avoid interference. The frequency detuning helps to average out any residual effects. 9 Spin-exchange collisions that require changes from the F=1 to F=2 manifold are energetically completely suppressed at the ultracold temperatures of the gas. 55 Chapter 4. Experimental Setup and Methods transfers from the initial |F = 2, mF = 2i state are performed by driving adiabatic Landau- Zener radio frequency and microwave transitions in the presence of a homogeneous magnetic field of several Gauss. The duration and range of the linear frequency chirps of constant amplitude are experimentally optimized to yield adiabatic passage efficiency of ∼ 100 % and ∼ 60 % for the radiofrequency and the microwave transition, respectively. We measure the decoherence limited transfer efficiencies by detecting the spin populations with a SternGerlach experiment. For this purpose a gradient magnetic field along the direction of the ion trap axis is applied during the time-of-flight expansion to separate the individual spin components spatially, before they are detected by resonant absorption imaging. In order to increase the polarization purity of the states in the F = 1 manifold after a microwave transfer, we selectively push out the remaining atoms in the F = 2 manifold with a resonant light pulse on the 2 S1/2 |F = 2i - 2 P 3/2 |F = 3i transition. The combination of radiofrequency and microwave population transfers with hyperfine selective push-out light pulses enables the preparation of all four polarization bath states in the optical trap with high fidelity. 87 Rb g F = 1/2 F=2 6.8 GHz g F = -1/2 F=1 mF -2 -1 0 1 2 Figure 4.4: Preparation of spin polarized atomic clouds in the optical trap. Starting from the |F = 2, mF = 2i state, the |F = 2, mF = −2i, |F = 1, mF = 1i and |F = 1, mF = −1i spin states of the 87 Rb electronic ground state are populated with high fidelity by radio frequency and microwave adiabatic passage. 4.1.4 Absorption Imaging of 87 Rb Atoms The standard tool for determining the properties of ultracold quantum gases is optical imaging. In its various forms optical imaging makes use of the absorptive or dispersive effects of atoms on light to generate a projection of the instantaneous spatial distribution of the atoms. In the case of absorption imaging, atoms are illuminated from one direction with a collimated beam of resonant laser light. The atoms absorb and randomly reemit part of that light and thereby cast a shadow that can be imaged onto a CCD camera. 56 4.1. Preparation of Ultracold 87 Rb Atoms For reasons that will become evident in this section, the standard approach of low saturation imaging is not practical for our experimental conditions. The following theoretical description of the high saturation absorption imaging is based [121] and [122]. The attenuation of the intensity of the probe beam, as it propagates through an atomic cloud is given by dI = n3d (x)σ(x)I(x), dx (4.1) where n3d (x) denotes the local density of the atoms and σ(x) = σ0 /(1 + s(x) + δ 2 ) is the light scattering cross section. Here, s(x) = I(x)/Isat is the saturation parameter and σ0 describes the resonant (detuning δ = 0) scattering cross section in the limit of very low saturation parameters. In the two-level approximation their values Isat = σ0 = 3λ20 2π hcπΓ 3λ30 and only depend on the transition line width Γ and the optical transition wave length λ0 . To calculate these quantities for the 2 S1/2 |F = 2i - 2 P 3/2 |F = 3i transition used in our setup, the polarization of the imaging light and the spin structures of the electronic states need to be considered as well. With the boundary conditions defined as the transmitted intensity IT and the initial10 intensity IT , the solution of Equation 4.1 reads [121] s0 s0 − nσ0 1 + δ2 W exp , T = s0 1 + δ2 1 + δ2 where W is the Lambert function of the first kind, T = IT I0 (4.2) is the relative transmission of light through the atomic cloud, s0 = I0 /Isat is the saturation parameter of light before R passing the atoms11 and n = n3d (x)dx is the column density. Using the defining equation X = W [Y ] ⇒ Y = XeX of the Lambert function, we can rewrite this equation to express the column density along direction of the imaging s0 1 + δ2 −ln(T ) + (1 − T ) n= σ0 1 + δ2 (4.3) as a function of the measurable intensities of probe light and in the presence (IT ) and absence of atoms (I0 ). The well-known Beer-Lambert laws n = −ln(T )/σ is in fact an approximation to this solution for the case of small saturation parameters (s0 ≪ 1) and/or small absorbance ((1 − T ) ≪ 1). The large dynamic range of in-situ atomic densities and the limited choice of time-of- flight expansion times tT OF in our quantum hybrid system make it challenging to obtain 10 11 The initial intensity equals the transmitted intensity in the case that no atoms are present. Here, we have neglected the collection of re-emitted fluorescence photons by the imaging optics and therefore underestimate the column density by a factor (1 − Ω) [121]. The fraction of the total solid angle covered by the imaging optics is Ω = 0.005 for the numerical aperture NA=0.17 of our imaging system. 57 Chapter 4. Experimental Setup and Methods satisfactory imaging conditions12 in the low saturation regime for all atomic cloud sizes. Due to the shadow cast by the ion trap electrodes, the possible time-of-flight expansion times for the horizontal imaging are restricted to the interval 0 − 7 ms, for imaging in between the electrodes, and to times > 27 ms, for imaging below the electrodes. The absolute measurement of atomic densities is critical for several of our atom-ion experiments and it is therefore important that saturation effects in the absorption imaging are correctly accounted for in Equation 4.3. Experimentally, the intensities IT and I0 are obtained from a series of three picture. The first picture records the counts in each pixel (NAtoms (yp , zp )) of the CCD image, while the atoms are present. The second picture (NBright ) is taken with an identical imaging light pulse, but without atoms. In both shots, the imaging light, will inevitably contain intensity fringes as a result of the interference of light, scattering from imperfect surfaces. In order to have similar illumination conditions in the two pictures it is essential to capture them in quick succession to avoid changes in the fringe pattern. We reduce the time between subsequent images to 5 ms by implementing the fast-kinetics function of the camera. In this mode the previous image is transferred to a shaded part of the CCD, before the next image is taken and all stored images are jointly read out at the end of the sequence. The third picture (NDark ) is recorded with neither laser light nor atoms. From these three pictures we calculate the relative transmission of imaging light through the cloud as T (yp , zp ) = IT I0 = NAtoms −NDark NBright −NDark . In contrast to the Beer-Lamberts law, Equation 4.3 also requires s0 = I0 /Isat and therefore the absolute amount of imaging light. In principle, the intensity I0 = ~ω0 M ηCCD ηO tl A · (NBright − NDark ) = C1 · (NBright − NDark ) at the location of the atoms can be calculated from the recorded pixel counts with the knowledge of the CCD’s quantum efficiency ηCCD and pixel area A = d2p , the imaging optics’ transmission ηO and magnification M and the imaging time tl . In practice, it is more convenient to determine the proportionality constant C2 between the saturation parameter and the detected pixel counts s0 = I0 /Isat = C1 (NBright − NDark )/Isat = C2 (NBright − NDark ) experimentally. To this end, a large number of identically prepared atomic clouds are imaged with varying amounts of light. The correct C2 is found as the value for which the column densities n, calculated from Equation 4.3, are as similar as possible for all measured light intensities [122]. Typical imaging parameters we use for magnetically trapped thermal clouds are tl = 40 µs, s0 ≈ 1 and tT OF = 27 ms. 12 Too little transmitted light leads to bad signal-to-noise ratios and to biases in the detection. Long imaging times, accelerate the atoms and distort their absorption cross sections due to the Doppler-effect. Ideal values for the relative transmission of light are typically T ≈ 0.3. 58 4.1. Preparation of Ultracold 4.1.5 87 Rb Atoms Density Measurement of the Neutral Bath In a harmonic trapping potential 1 V (r) = ma (ωx2 x2 + ωy2 y 2 + ωz2 z 2 ) 2 (4.4) the density distribution of an ideal Bose gas above the BEC transition temperature follows [123] n(r) = 1 g3/2 (z(r)), λ3dB (4.5) where λdB = (2π~2 /ma kB T )1/2 is the thermal deBroglie wavelength, z(r) = exp((µ − V (r))/kB T ), µ is the chemical potential and T is the temperature of the gas. Compared to a thermal gas of distinguishable particle, the density of the Bose gas is enhanced by the P factor g3/2 (z)/z, with the function gj (z) = i z i /ij being the polylogarithm. For the phase space densities of thermal clouds used in our ion-atom experiments, the Bose enhancement is negligible even in the centre of the cloud and the neutral density distribution is well described by n3d (r) = n3d (0) · e V (r) BT −k 3/2 2 2 +ω 2 y 2 +ω 2 z 2 y z 2kB T Na ma ωx ωy ωz −ma ωx x ·e = (2πkB T )3/2 , (4.6) where Na is the number of atoms in a neutral cloud of temperature T . By integrating the R density along the imaging direction n(r) = n3d (r)dx the column density 2 2 2 2 Na ma ωy ωz −m ωy y2k+ωTz z B ·e n(y, z) = (2πkB T ) (4.7) of the cloud is obtained. In-situ resonant absorption imaging of atoms in their trapping potential is hampered by the high optical densities n ≫ 1 of ultracold clouds and the limited spatial resolution of the imaging optics (≈2.5 µm). The solution two both limitations, is time-of-flight imaging, where the atoms are abruptly release from the trap and expand, before they are imaged. Once the trap is switched off, the atoms fly ballistically13 from their in-trap position r 0 with their original velocity v to reach the points r(t) = r 0 + p/ma · t after the time t. As a consequence, the 3D density of the cloud evolves during the time-of-flight expansion as [123] x,y,z 3/2 − 2kmaT Na ma ωx ωy ωz Y 1 B q )e n3d (r, t) = ( (2πkB T )3/2 2 2 1+ω t i x,y,z P i ωi2 x2i . (4.8) i 13 This neglects the effect of interactions during the expansion, which do not play a role due to the low atomic densities in our neutral clouds. 59 Chapter 4. Experimental Setup and Methods The self-similar expansion does q not change the shape of the density distribution, but scales its width in the direction ri by 1 + ωi2 t2 . In conveniently reparameterized units, the column density of the time-of-flight cloud n(yp , zp , t) = n0 · e − (yp −y0 )2 2 2wy ·e − (zp −z0 )2 2 2wz (4.9) serves as the fitting function for the experimental column density pixel maps (Equation 4.3), obtained from the absorption imaging. The pixel fit parameters n0 , wy , wz relate to the quantities of interest T and Na of the imaged atomic cloud through k B Ti = ma ωi2 wi2 A n0 (1 + ωi2 t2 )M 2 and Na = πn0 wy wz A , M2 (4.10) where M is the magnification of the imaging system, A = d2p is the area of a pixel, and the lengths yp , zp , wy and wz are measured in pixels. The central density of an atomic cloud, while it is held in the trap, 3/2 Na ma ωx ωy ωz . n3d (0) = (2πkB T )3/2 (4.11) follows from the Equation 4.6 and the fitted temperature and atom number. When imaging identically prepared, large thermal clouds of neutral atoms, we typically observe fluctuations in the detected atom number and temperature of 15% and 5% standard deviation, respectively. Theses fluctuations correspond mainly to real variations in the properties of the prepared clouds due to imperfections in the repeatability of the experiment. For very small atomic samples, fitting errors due to image imperfects also start to contribute significantly to the observed variations. Considering the uncertainties of all quantities entering into Equation. 4.11, we estimate that the absolute values of the central densities derived from absorption imaging have a systematic uncertainty on the order of 40%. The main contributions to the error budget result from the determination of the polarization-dependent optical cross section, the estimation of the saturation intensity, and the corrections to the expansion of neutral atoms due to the deconfining effects of the ion trap. 4.2 Trapping of Yb+ Ions The second ingredient in our quantum hybrid system are trapped Ytterbium ions. In recent years singly ionized Yb+ ions have become an increasingly popular choice for ion trap experiments in quantum metrology and quantum information science due to their versatile 60 4.2. Trapping of Yb+ Ions atomic level structure. Being the member of the lanthanide (rare earth) series of chemical elements with atomic number 70, singly ionized Ytterbium atoms have the electron ground state [Xe](4f 14 )(6s)S1/2 . In its lowest energy state the ion has a single electron in an open s-shell, which gives it an electronic level structure, similar to that of the heavier earth-alkali ions. In Figure 4.5 the electronic level scheme, the most prominent optical transitions and the laser wavelengths used in our experiment are summarized. The electronic ground state is strongly coupled to the fine structure split (4f 14 )(6p) 1 τ[3/2]=37.7ns 2 2 P3/2 P1/2 3 [3/2]1/2 1.8% 98.2% τp=8.12ns 935 nm 2 0.5% 99.5% [5/2]1/2 τD5/2=7.2ms D3/2 D5/2 2 τD3/2=51ms 369 nm 638nm 2 83% τF≈years F7/2 17% 411 nm 2 S1/2 Yb+ Figure 4.5: Energy level scheme of the electronic ground and excited states of Yb+ (drawn to scale). The transitions driven by diode lasers in our Experiment are marked as straight lines. Life times and decay branching ratios are taken from [124]. 2P 1/2 and The 2 S1/2 2P 369nm and 329nm, respectively. - of Γ369 = 3/2 states via dipole allowed transitions at 2P 1/2 transition with a natural linewidth 2π τp = 1 8.12ns = 2π · 19.60(5)MHz [125] serves as the cycling transition for Doppler cooling and fluorescence detection (Section 4.2.2) in most Yb+ ion trapping experiments14 . Dipole allowed decay from the 2 P 1/2 state leads to the leakage of population into the (4f 14 )(6d) 2D 3/2 state with a branching ratio of 0.501(15)%. In order to close the fluorescence cycle, repump laser light at 935 nm depopulates the metastable state with a radiative lifetime of τD3/2 = 51 ms via the short lived 2 D[3/2]1/2 state and returns the ion into the ground state. A detailed description of the relative state populations during the 4-level fluorescence cycle will be given in Section 4.2.3. 14 In [126] the narrower 2 S1/2 - [3/2]1/2 transition was recently explored for Doppler cooling. 61 Chapter 4. Experimental Setup and Methods The second state of the metastable fine structure doublet, the 2 D5/2 state has a lifetime of 7 ms [127] and couples to the lower lying (4f 13 )(6d)2 2 F7/2 state (83 %) and via a quadrupole transition (17 %) to the 2 S1/2 ground state. The later state exhibts an extremely long natural lifetime of several years due to the fact that its radiative relaxation mechanism is an electric octupole transition to the 2 S1/2 ground state. The (4f 13 )(6s)2 2 F7/2 state is usually not populated during Doppler cooling, as neither the 2 P 3/2 , nor 2 D5/2 state form part of the 4-level fluorescence cycles. However, collisions of electronically excited ions with hot background gas atoms can lead to the occupation of the F-state. At background gas pressures of ≈ 10−11 mbar these collisional transfers occur several times per hour. The F-state is also efficiently populated when we drive the narrow 2 S1/2 - 2 D5/2 quadrupole transition with laser light at 411 nm. In order to return ions from the F-state back into the fluorescence cycle we apply a further repump laser at 638 nm in our experiment. 4.2.1 Loading of Ions Whereas the fast and efficient loading of a deterministic number of ions is a beneficial feature for many trapped ion experiments, it is a critical requirement for the investigations of our atom-ion system. In the hybrid trap, the reactive collisions of ions with ultracold neutral atoms make ion reloading rates necessary, that can exceed those of pure ion trap experiments15 by several orders of magnitude. In addition, the frequent reloading process must not lead to any deterioration of the vacuum, in order to not affect the lifetime of the ultracold quantum gas. I will briefly summarize the ion loading procedure we use in our ion trap to satisfy these stringent criteria. The oven design and the experimental sequence underlying this functionality are described in detail in [69]. Isotope Selective Loading Ytterbium features seven stable isotopes, of which five (171, 172, 173, 174, 176) have a natural abundance of larger than 10% and their respective nuclear spins are (1/2, 0, 5/2, 0, 0). In our experiment we primarily use spin-free isotopes, and 171 Yb+ , 174 Yb+ , since it is the most abuntant of the nuclear which is of interest for our quantum information experi- ments due to its hyperfine structure (see Section 6.1.2). The preparation of isotope-pure ion crystals of a selected isotope is achieved by means of resonance enhanced two-photon ionization [128]. For this purpose a beam of thermal atoms is generated by resistively heating a thin steel tube containing Yb metal of natural abundance with a short current pulse of 90 A and 100 ms duration [69]. After being collimated by a small aperture, the beam 15 In pure ion trap experiments the need to replace ions usually arises from losses due to background gas collisions. 62 4.2. Trapping of Yb+ Ions crosses the centre of the ion trap, where it is intersected by two laser beams. The first laser beam at 399 nm addresses the 1 S0 - 1 P1 transition of the neutral ytterbium atoms. The laser crosses the beam at the Doppler-free angle of 90◦ , in order to resolve the isotope shifts of the line, that are on the order of hundreds of MHz. The laser resonantly excites atoms of a single isotope species to 1 P1 , from where they are ionized by a second photon that exceeds the energy gap to the continuum. The ion cooling laser at a wavelength of 369 nm naturally provides the light required for this second, non-resonant step of the process, so that no further lasers are required. Ion Number Preparation The average number of ions that are simultaneously loaded into the trap in a single loading attempt, depends on several experimental parameters. The most important parameters are the size and length of the current pulse, that determine the temperature of the Yb in the oven and therefore the amount and velocity distribution of neutral Yb in the atomic beam. Since the atoms get ionized on passing the very centre of the trap, the energy of newly formed ions is given by the thermal Maxwell-Boltzmann distribution of the initial atomic beam 16 . The kinetic energy of the ions (∼ kB · 1000 K) is significantly smaller than the maximum radial trap depth of ∼ 0.250 eV (≈ kB · 3000 K), which explains why for strong axial confinements, the loading efficiency saturates in our experiment. Since every ionized atom will stay in the trap and will eventually be cool, the number of loaded ions is directly proportional to the number of atoms in the neutral beam. The resonant light at 399 nm is not a critical parameter of the loading process either, because the 1 S0 - 1 P1 transition is saturated even for moderate laser powers (> 10 µW) due to the small laser focus size (< 100 µm). We observe, however, a gain in the ion loading efficiency by increasing the intensity of 369 nm cooling light beyond the ion’s Doppler cooling saturation value, which results from the enhancement of the second step in the two-colour ionization process. In order to increase the efficiency for preparing a desired number of ions N beyond the probabilistic value of the Poissonian loading distribution, an automated feedback scheme has been implemented for our experiment previously [69]. For this purpose the trap loading parameters of the trap are adjusted to produce a number of ions per loading pulse that on average exceeds N . Following a loading pulse the ions are Doppler cooled to form a Coulomb crystal and the number of ions is automatically detected from a CCD camera fluorescence image. If the ion count is found to be larger than N , the axial confinement of the trap is reduced to a experimentally adjusted value for a short period of time to expel 16 The fact, that the energy of the second photon at 369 nm significantly exceeds the ionization threshold corresponding to 394 nm does not lead to a relevant kinetic energy increase in the ion. Its surplus energy is almost exclusively taken up by the electron due to the large ion to electron mass ration. 63 Chapter 4. Experimental Setup and Methods ions from the trap. The success of the ion dump is checked by another ion image and, if necessary, a limited number of further attempts is made to reach N . Electrostatic Offset Field Drifts In practice, the capability of preparing fresh quasi-deterministic ion crystals in every experimental cycle (∼80 s) comes at the price of the creation of significant electrostatic offset field at the location of the ion. These fields are the result of the formation of electric charges on the ion trap electrodes due the deposition of neutral atoms from the oven beam. The presence of cooling laser light at 369 nm (∼3.36 eV) is sufficient to extract photo electrons from metallic Yb, that has a work function of only 2.6 eV [129]. The extraction of electrons can lead to the charge-up of isolated patches on the blade electrodes. The problem is evident in Figure 5.2 that shows the drift of the axial equilibrium position of ions due to the creation of uncompensated static offset fields. The data in theses plots, that contain 830 oven firing events over the course of 20 hours, can be used to quantify the change of the electrostatic offset field per single oven firing. In order to convert the pixel position into ion displacements the scale of the camera images first needs to be established. According to Equation 3.16, the separation between the two ions of a binary crystal in a harmonic trap of trap frequency ωz = 2π · 51.40(2) kHz is d=24.81(1) µm. The comparison with the measured ion separation of 10.1(2) pixels on the CCD camera yields a scale of 4.9(1) µm/pixel. We approximate the change of ion positions by a linear drift with a fitted velocity of 0.14(1) pixel/h or ∆z = 0.016(1) µm per oven firing event. The electrostatic field change is therefore estimated to be ∆E = ∆z · mωz2 /Q = 0.0033(3) V/m per oven firing event for the specific loading conditions of that measurement. Changes in the micromotion compensation voltages (Section 3.1.5) during times of intensive ion loading have shown that electrostatic fields of similar size are also generated in the radial direction. As expected from the accumulation of localized charges, the loadinginduced offset fields always increase, pointing in the same spatial direction. We also observe that during long intervals without continuous loading the electrostatic offset fields slowly return to their pre-loading values. This indicates the existence of a mechanism that lets accumulated charges diffuse over long timescales. The detrimental effects due to loading induced contamination could be reduced in a future setup by improving the collimation of the hot atomic beam as it emerges from the oven. A more elegant solution to the challenge of efficient and clean ion loading, however, is the idea of sourcing the ions from a cold cloud [130]. The method is particularly convenient for atom-ion hybrid systems, that already have a vacuum system with two connected chambers. In this way, the contaminating part of ion trap loading process can take place in the 64 4.2. Trapping of Yb+ Ions chamber that houses the magneto-optical-trap for the neutral atoms. After the loading of a small MOT with Yb atoms, the cold cloud will have to be transported optically17 through the differential pumping section. Once the cloud has been positioned in the centre of the ion trap, the loading of ions by two-photon ionization should be both efficient and clean. 4.2.2 Doppler Cooling and Fluorescence Detection in 174 Yb+ The laser cooling of trapped atoms and ions in the regime Γ ≫ ω, where the natural linewidth of the addressed atomic transition Γ is much larger than the mechanical oscillation frequency of the trapped particle ω, is known as Doppler cooling. In this weak binding limit the coherence of the atomic state over one oscillation period in the harmonic potential can be neglected and the interaction of light with the atom can be treated as if occurring at an instant of time. The steady-state rate of absorption and spontaneous emission of photons by a two-level atom R(v) = Γ · ρee (v) = Γ 1 I 2 Isat 1+ I Isat 2 + ( 2∆(v) Γ ) (4.12) depends on the light’s frequency detuning from the resonant atomic transition ∆(v) = ∆Laser + ∆Doppler (v) and on its intensity I. The saturation intensity Isat , specifies the intensity light that is needed on resonance to maintain an excited state population ρee = 1/4. Importantly, the detuning from resonance includes a frequency shift that is caused by the instanteous motion of the particle due to the doppler effect. As in the case of free atoms18 , the preferential absorption of momentum-transfering photons by an atom with velocities v · k ≈ ∆Laser can be used to modify the motional state of trapped ions. For light that is red-detuned from the atomic resonance, it can be shown that the energy distribution of an ion during Doppler cooling reaches a lower limit of TD = ~Γ 2kB for the optimal detuning ∆Laser = Γ/2. (4.13) The application of Doppler cooling to the 2 S1/2 - 2 P 1/2 transition in Yb+ results in Dopplerlimited ion temperatures of ≈ 470 µK. The energy states of the ion motion in its trap potential are the discrete energy eigenstates E = ~ω(n + 1/2) of the quantum harmonic oscillator. With the knowledge of the radial and axial trap frequencies (ωAxial ≈ 2π·55 kHz, ωRadial ≈ 2π · 150 kHz), we obtain the average energy occupation numbers nAxial ≈ 180 17 Magnetic transport is not possible due to the lack of a magnetic moment in the ground state of neutral Yb atoms. 18 In Section 4.1 we have described the use of Doppler cooling as the method to decrease the temperature of neutral atoms in the first step of the experimental preparation of ultracold rubidium clouds. 65 Chapter 4. Experimental Setup and Methods and nRadial ≈ 65. The corresponding spatial extend of the ion wave function is given by √ zi = z0i 2ni + 1, where z0i = r ~ 2mωi (4.14) is the size of the harmonic ground state wave function. The numbers for our current trap are z0 Radial = 23 nm, z0 Axial = 14 nm, z Radial ≈ 160 nm and z Axial ≈ 430 nm. Fluorescence Detection The detection of ions in our trap and the readout of information about their internal states relies on the collection of fluorescence photons that are scattered by the ion, while it is driven on the 2 S1/2 - 2 P 1/2 Doppler cooling transition. A simple estimate for the rate of photons from a cold ion can be made from the two state approximation in Equation 4.12. The photon scattering rate is directly proportional to the occupation probability of the excited state and takes the form of a power broadened Lorentzian curve as a function of cooling laser intensity and detuning from resonance. For typical Doppler cooling parameters I Isat = 2 and ∆ = Γ369 /2 the ion scatters photons into free-space at a rate of R = Γ · ρee ≈ 40 · 106 s−1 . In order to obtain a more accurate model of the fluorescence behaviour of Yb+ , it is necessary to take into account the ion’s multilevel nature, that can lead to the formation of dark. Optical pumping into the electronic 2 D3/2 state due to the weak dipole allowed decay from the excited 2 P 1/2 state is countered by the application of repump laser light at 935nm. The fact that the additional higher lying electronic 3 [3/2]1/2 state is used to close the fluorescence cycle has the benefit that coherent population trapping effects between the 2S 1/2 and the 2 D3/2 cannot arise19 . However, the formation of coherent dark states due to quantum interference effects is also possible, if a single laser field is driving a Zeemandegenerate transition with a high multiplicity in the ground state than in the excited state. This is the case for the 2 D3/2 to 3 [3/2]1/2 repump transition in both as well as for the 2 S1/2 (F = 1) to 2 P 1/2 (F = 0) cooling transition in 174 Yb+ 171 Yb+ and 171 Yb+ [131]. The two common experimental methods for the destabilization of these dark states [132] are the fast modulation of the polarization of the driving laser field or the application of a static magnetic field to rotate the ground-state spin-polarization out of the dark state. In our experiment we make use of the second method and apply static offset fields in the range of several Gauss to achieve efficient ion fluorescence detection. As a result of the complex dark state dynamics, the ion fluorescence rate is a function of the detunings, and the polarization and propagation angles of both cooling and repump beams as well as the 19 The physics of Λ three level system has of course also many interesting properties that can for example be used for electromagnetically induced transparency cooling of ions, etc.. 66 4.2. Trapping of Yb+ Ions magnetic field strength and direction. 4.2.3 Experimental Determination of the Ion Electronic State Occupation For our experiments on the role of electronically excited ion states in the inelastic collisions of Yb+ +Rb in Chapter 5 it is essential to accurately determine the average electronic state population during fluorescence scattering of the ion. Traditionally, this is done by trying to adapt the various laser and magnetic field parameters in a many-level/two-laser optical-Bloch-equation model such that the model’s predicted fluorescence rates best fit the experimental results over a large range of laser detunings. Here, we develop an experimental method that allows us to directly determine the relative state occupation for a single experimental condition. Most importantly the technique is significantly more sensitive to repumping efficiency and therefore the relative time the ion spends in the 2 D3/2 state. We start by defining the average state populations during fluorescence cooling as px = Tx /Tc , (4.15) where we have introduced the average fluorescence cycle time Tc = X Tx (4.16) x as the average time between two spontaneous decays on the 2 P 1/2 - 2 D3/2 transition and Tx as the average time spent in state x ǫ {S, P, D, [3/2]} during one fluorescence cycle. The time TP = τP /BP D is set by the natural lifetime τP = 8.12 ns of the 2 P 1/2 and the branching ration BP D = 0.005 of the decay from the P to the D state. The same statement can be made for T[3/2] = τ[3/2] /B[3/2]S with τ[3/2] = 37.7 ns and B[3/2]S = 0.98. The times TS and TD are determined by the dynamics of the cooling and repump transition and depend on the laser parameters and the magnetic field. With these definitions the fluorescence photon rate of a single ion is given by R= 1 B P D · Tc . (4.17) Since in Yb+ the relation TS & TP ≫ TD & T[3/2] is obeyed during typical fluorescence detection conditions, the scattering rate is very insensitive to the repump parameters. For this reason we determine TS and TD by using the pulsed sequence in Figure 4.6, which permits the direct measurement of the times for a single set of laser parameters. The sequence consist of alternating pulses of cooling (369 nm) and repump (935 nm) light with 67 Chapter 4. Experimental Setup and Methods 369nm 935nm N369 Int xegrat5000 Repetitions ed Count s Count s per [ µs] sum m ed over 5000 repet it ions 0 2000 1000 800 τ1 τ2 τ3 C1 C2 C3 τ4 TS+TP =8 μs 0 100 200 Tim e[ μs] s] C4 300 400 C4 0.5% C1 99.5% 369 nm C2 τT935D = 0.4[µs] μs b 600 400 1.8% C3 200 0 τ3 0 d τ2 τ1 1.5 Pulse Tim e 935nm t935 [μs]light [ µs] 0.5 98.2% τ4 1 2 935 nm c Figure 4.6: Determination of the electronic state populations during fluorescence detection. Our measurement sequence alternates between the 369 nm and 935 nm light periods, in order to directly observe, how long the ion will stay on average in the cooling and repump part of the continous fluorescence cycle. The time TS + TP is observed as the exponential decay constant of fluorescence after the 369 nm laser is pulsed on. The pulse length is set to get complete depletion of the S-P system into the D-state. d For a given setting of intensity and detuning of the 935-nm laser, the average number of photons N369 per 369-nm pulse depends on the duration t935 of the preceding repump pulse. We vary the length of the repump pulse t935 to determine the average time TD that is needed to repump the ion for a given set of experimental conditions. During the 8 s measurement 16 different repump times are repeated 5000 times each in order to gather sufficient statistics. The solid line is a fit to the data with Equation 4.18. the same detuning and intensity as the fluorescence steady-state for which the population of electronic states shall be determined. After the start of the 80 µs cooling pulse the ion scatters fluorescence on the 2 S1/2 to 2 P 1/2 until it is completely pumped into the 2 D3/2 dark state. The decay constant of the recorded ion fluorescence curves directly measures the average time TS + TP the ion spends in the S - P system. The efficiency of a repump light pulse, on the other hand, decides the total number of fluorescence counts N369 during the following cooling pulse. Varying the length t935 allows us to retrieve both TD and the photon detection efficiency ηDet from the fit N369 = NDark + ηDet (1 − e−t935 /TD ) BP D 68 (4.18) 4.2. Trapping of Yb+ Ions to the recorded photon counts. Here, NDark denotes the experimentally determined average number of stray light dark counts during the 80 µs detection interval. Experimentally, Acousto-Optical-Modulators are used to time accurately switch the cooling and repump lights according to pulse sequences that are generated by an FPGA. The population measurement for a single fluorescence condition is performed in a single 8 s interval, by repeating a sequence of 16 different 935 nm repump durations, 5000 times in order to gather statistics. The experimental results for the measurement shown in Figure 4.6 are TD = 0.40(2)µs, TS = 8.0(2)µs − TP = 6.4(2)µs and Tc = 8.4(2)µs. The value of TD = 0.40(2)µs represents in fact the shortest repump time observed in our experiments and was obtained for several mW of resonant 935 nm light. It is still significantly longer than the minimal theoretical two-level value of TD ≈ T[3/2] ≈ 38.5 ns and is limited due to the formation of coherent dark states. According to Equation 4.15 the state populations in this case are pS = 0.76(1), pP = 0.19(1), pD = 0.047(2) and p[3/2] = 0.0044(1). From the fit to the data in Figure 4.18 we also find ηDet = 2.1(1) × 10−3 for the probability that an emitted fluorescence photon is detected by the photon counter. The value is in good agreement with the simple estimate of the overall detection efficiency ηDet = 12 (1 − √ 1 − N A2 ) · ηT · ηP C ≈ 0.0019, where we have considered the design numerical aperature N A ≈ 0.27 of the light collecting optics, the measured overall transmission ηT ≈ 0.75 of all optical components and the specified quantum efficiency ηP C ≈ 0.14 of the photon multiplier tube at 369 nm. 4.2.4 Doppler Cooling of 171 Yb+ The Doppler cooling scheme for the nuclear spin 1/2 isotope is very similar to one of 174 Yb+ 171 Yb+ , shown in Figure 4.7, and other ions with nuclear spin 0, that we have previously presented in Section 4.2.2. In order to efficiently Doppler cool 171 Yb+ ions two additional transitions need to be addressed to cover all hyperfine sublevels of the 2 S1/2 and 2 D3/2 states. The main cycling transition is driven by 369 nm cooling light that is Doppler detuned (∆369 = −Γ369 /2) from the 2 S1/2 , F = 1 → 2 P 1/2 , F = 0 resonance. From the excited 2P 1/2 , 2D 3/2 , F = 0 state, the ion can only return to the 2 S1/2 , F = 1 state or decay into F = 1 due to selection rules. As mentioned in Section 4.2.2, in 171 Yb+ ions the photon scattering rate on the cooling transition is limited by the formation of coherent dark states [131]. For this reason, the typical fluorescence rate we obtain for 171 Yb+ ions in a magnetic field of several Gauss and with optimized polarization angles is only about 1/3 of the fluorescence rate of ions without nuclear spin like 69 174 Yb+ . Chapter 4. Experimental Setup and Methods During fluorescence scattering the cooling light also off-resonantly excites the 2.105 GHz detuned 2 P 1/2 , F = 1 state, from where the ion can leak into the dark 2 S1/2 , F = 0 state. There are two possible ways to clear out the 2 S1/2 , F = 0 state. The use of the 2S 1/2 , F = 0 → 2 P 1/2 , F = 1 optical transition to repump the population requires laser light that is 14.7 GHz blue detuned from the 369 nm cooling light. Alternatively, the dark 2S 1/2 , F = 0 state can be coupled to the bright 2 S1/2 , F = 1 state by driving the hyperfine transition at 12.6 GHz with microwave radiation. We have tested both options and found that the fluorescence counts during Doppler cooling are reduced by ∼30% for the microwave in comparison to the laser repumping method. Due to its experimental convenience we, nevertheless, currently use the microwave rather than the additional optical transition in our setup. To return the ion from the metastable D3/2 -states back to the 2 S1/2 , F = 1 ground state, the 935 nm repump laser is tuned to the 2 D3/2 , F = 1 → 2 [3/2]1/2 , F = 0 transition. In order to clear out population from the 2 D3/2 , F = 2 hyperfine state, we use an Electro- Optic-Modulator (EOM) operating at 1.535 GHz to imprint sidebands onto the repump laser. The light of the red detuned second-order sideband at 3.07 GHz is resonant with the 2D 3/2 , F = 2 → 2 [3/2]1/2 , F = 1 repump transition. 3 F=0 F=1 2 P1/2 2.105 GHz D[3/2]1/2 2.2095(11) GHz F=1 F=0 F=2 F=1 2 D3/2 0.86(2) GHz 2 S1/2 F=1 12.643 GHz F=0 171 Yb+ Figure 4.7: Doppler cooling of ions. Compared to the nuclear spin-0 isotopes, two additional transitions need to be addressed to avoid optical pumping into dark hyperfine states. Adapted from [68]. 70 4.2. Trapping of Yb+ Ions Ion Microwave Setup Microwave radiation resonant with the 12.6-GHz hyperfine transitions is used for coherent qubit manipulation on the clock transition, for Rabi π-flips on the |0, 0i → |1, −1i and |0, 0i → |1, 1i transitions (see Section 6.1.2), as well as for repumping during Doppler cooling. The coherent manipulation of hyperfine states requires a system that can generate microwave pulses with high power, accurate timing, and well-defined phase relations. The microwave source in our setup is an Agilent 5183A synthesizer that is locked to a 10-MHz rubidium atomic clock20 for long term stability. The system is not referenced to any primary frequency standard, since our experiments are not concerned with absolute frequency accuracy21 . The synthesizer’s continuous wave output at 12.24 GHz is frequency mixed22 with the tunable radio frequency signal of a computer-interfaced DDS solution23 to provide timing, amplitude and frequency-controlled microwave pulses around 12.64 GHz. The signal is amplified24 up to 10 W and delivered to a SMA-waveguide adapter25 that serves as a horn antenna. The microwave antenna is pointed at the ion trap through a vacuum chamber window from a distance of about 120 mm (see Figure. 4.8). 4.2.5 Optical Layout of the Experiment The simultaneous manipulation and detection of both atoms and ions in the same physical location presents a formidable challenge in terms of optical access in our experiment. Figure 4.8 and Figure 4.9 show the layout of laser beams and optical detectors with respect to the location of the ion trap and the surrounding vacuum chamber. In Figure 4.8 the experiment has been cut open horizontally at the height of the ion trap and is now viewed from the top. In Figure 4.9 the experiment is seen from the side of the camera CCD1. For simplicity the six laser beams of the magneto optical trap have not been included into the drawing. All other laser beams, optical detectors, and microwave antennas are aligned to the interaction region in the centre of the ion trap. Basic Ion Trapping The laser light at 399 nm, that addresses the resonant transition in the neutral Yb atoms, and the 369 nm Horiz. Cooling beam, that ionizes the excited atoms in the second step of the two-photon ionization, are coupled out of the same optical fiber. The beams are located 20 Efratom LPRO Indeed the comparison of two identical atomic clock-synthesizer combinations in our lab has shown a stable frequency offset between both systems of several Hz at the 12 GHz level. 22 Minicircuits ZX05-153MH+ 23 designed by Christoph Zipkes http://decads.scondaq.com/ 24 Microwave Amplifiers Ltd AM51-12.4-12.8-40-40 25 Flann microwave 18094-SF40 21 71 Chapter 4. Experimental Setup and Methods 369nm Detection 935 nm 369 nm Horiz. Cooling 399 nm Figure 4.8: Overview of the optical layout. In this rendering the experimental chamber has been cut open horizontally at the height of the ion trap and is now viewed from the top. The black components form part of the vacuum chamber, the µ-metal box that is used for magnetic shielding is shown in grey and the arm holding the ion trap is displayed brown. The 1064 nm Horiz. Dipole Trap and the 369 nm Pol. Pump. laser beams are both aligned with the ion trap axis and enter the trap through the holes in its end caps. Details about the individual beams and detection devices are given in the text. in the horizontal plane where they form an angle of 60◦ with the ion trap axis and at the same time intersect the beam of hot atoms from Yb oven at the Doppler-free angle of 90◦ . Both beams are shaped into an elliptical focus that extends along the axial direction of the ion trap in order to better illuminate long axial ion crystals, while minimizing the amount of stray light from the blade electrodes of the ion trap. A second UV beam, 369 nm Vertical Cooling, crosses the ion trap in the vertical direction from the bottom and is needed to perform micromotion compensation measurements along the second radial trap axis. The ion repump light at 935 nm and the second ion repumper at 638 nm again originate from the same optical fiber and cross the ion trap axis at an angle of 60◦ . We employ two devices to detect fluorescence photons at 369 nm wavelength from the ion (CCD3 and PMT ). In order to record pictures of single ions and spatially resolved small ion crystals we image the central region of our ion trap from the top with a home-built multi-lens objective and the camera26 CCD3. The objective lens with a numerical aperature 26 Andor iXon DV-885 72 4.2. Trapping of Yb+ Ions of 0.15 is designed from of-the-shell spherical lenses to enable diffraction limited imaging at both 369 nm and 780 nm wavelength with magnifications M369 = 5.7 and M780 = 8.0, respectively. A dichroic mirror is used to separate the UV light from light at 780 nm, with the later casting a resonant absorption image of the neutral atom cloud onto a second camera27 (CCD2 ). In the horizontal plane a second home-built multi-lens objective [133] of numerical aperture 0.27 is used to collect ion fluorescence photons onto a photo multiplier tube28 (PMT ). In order to reduce stray light counts, we spatially filter an intermediate image with a small aperture and block nonresonant light with a narrow band pass filter29 at 370 nm. The absolute detection efficiency of the system has been measured in Section 4.2.3 to be ηDet = 0.0021(1). Individual photon detection events are time tagged30 with a time resolution of 25 ps and made available to the control software for real time processing. CCD3 CCD2 1064 nm Vertical. Dipole Trap λ/4 Wave Plate 1064 nm Horiz. Dipole Trap 369nm Pol. Pump. z x 780nm Vertical Imaging 411nm 369nm Vertical Cooling Figure 4.9: Overview of the optical layout. The experiment is seen from the side from the position of the camera CCD1 (see Figure 4.8). Different laser beams and detection lights are separated by the use of dichroic mirrors, polarizing beam splitters or by spatial displacement. The vertical absorption imaging with camera CCD2 is only used for alignment purposes. Details about the individual beams and detection devices are given in the text. 27 Apogee Alta U1 Hamamatsu H7360-01 29 Semrock FF01-370/10-25 30 RoentDek TDC8HP 28 73 Chapter 4. Experimental Setup and Methods Ion Zeeman Qubit Manipulation The experiments on the dynamics of the 174 Yb+ Zeeman qubit (Section 6.1.1) additionally require the 369 nm Polarization Pumping beam. The beam propagates along the ion trap axis, which is during these measurements also the direction of the homogeneous magnetic offset field. With the help of a computer-controlled motorized λ/4 waveplate the polarization of the beam can be set to σ + or σ − polarization in order to initialize the ion in its upper or lower Zeeman state. The frequency of the UV light in this beam is the same as in the two 369 nm Cooling beams. The propagation direction and the linear polarization of the 411 nm light were chosen to be mutually orthogonal with each other and the direction of the magnetic field in order to only address electric quadrupole transitions with ∆m = ±2. Ion Hyperfine Qubit Manipulation For the preparation, manipulation and detection of the hyperfine qubit in 171 Yb+ we use a further pair of laser beams with 369 nm light in the horizontal plane of the experiment. The 369 nm Detection light is applied during the hyperfine detection. Compared to the 369 nm Cooling beams it has a significantly smaller waist to minimize the amount of stray light background counts and operates closer to the resonance frequency of the ion in order to maximize the number of resonantly scattered photons before off-resonant pumping occurs. The function of the 369 nm Frequency Pumping light is to initialize the hyperfine qubit through frequency selective optical pumping. The microwave radiation resonant with the ground state hyperfine transitions at 12.6 GHz is emitted by a waveguide-SMA adapter that is located at a distance of ≈120 mm from the ion. Rb Manipulation The optical dipole trap consists of a horizontal beam, that is aligned to the trap axis and propagates through the holes in the end-caps of the ion trap, and a vertical beam that crosses the ion trap from the top. The linear polarizations of both beams are orthogonal and both lie in the horizontal plane of the experiment shown in Figure 4.8. The spiral antenna for microwaves at 6.8 GHz is used during the microwave evaporation in the QUIC trap as well as for Landau-Zener sweeps to transfer atoms in the dipole trap from the F = 2 to the F = 1 manifold of the Rb ground state. Resonant absorption imaging of the Rb atoms can be performed, both in the vertical and horizontal direction. For the imaging in the horizontal direction resonant light at 780 nm is launched from the tip of a bare single mode fiber, that is mounted in the centre of the PMT objective. The image of the cloud is captured by the camera CCD1 that serves as 74 4.2. Trapping of Yb+ Ions our main quantitative imaging, both in-situ and after time-of-flight expansion. 4.2.6 Yb+ Laser System The cooling and manipulation techniques for Yb+ ions, described in the previous Sections, require the generation of a significant number of intensity and frequency controlled coherent light fields. The purpose-built laser system that supports the basic functionality of the ion experiment, has been designed and built by the former PhD student Stefan Palzer [70]. Here, I will present an overview of the laser system and describe the most important additions that we have made. The laser system for Yb+ currently consists of the six external cavity diode lasers (ECDL), two tapered amplifiers, one frequency-doubling cavity and two frequency transfer cavities. A schematic drawing of the complete setup, also including the three ECDL and the tapered amplifier of the Rb laser system described in Section 4.1.1, is shown in Figure 4.10. All Yb+ lasers are intermediate line width lasers, that require frequency stabilization to (near-) resonantly address the various optical transitions in the ion over long timescales. Several different ways of frequency referencing diode lasers are currently implemented in our laser system. Frequency locking of Lasers The light at 399 nm for the resonant first step in the two-colour photo-ionization process of Yb, is provided by an ECDL that is stabilized by frequency modulation spectroscopy to an Ytterbium hollow cathode lamp. The Doppler-free saturation spectroscopy resolves the hyperfine structure of the 1 S0 →1 P1 transition in neutral Yb, so that the laser can be locked to a specific hyperfine transition for isotope selective loading. Part of the spectroscopystabilized 399 nm laser light now serves as a frequency reference onto which a blue transfer cavity is actively stabilized. With the exception of the mirror coatings, the blue cavity is identical in design to the infrared transfer cavity in our setup, which is described in detail in [70]. The blue transfer cavity can be used to referenced lasers in the wavelength range 365 − 440 nm to the Yb-spectroscopy. The cooling laser light at 369 nm is created in our setup by frequency-doubling laser light at 739 nm in a nonlinear LBO crystal inside a bow tie enhancement cavity. The infrared light is provided by amplifying the light from an ECDL in two successive tapered amplification stages. The seeding ECDL is locked with a tunable frequency offset to another diode laser, which itself is Pound-Drever-Hall locked to the infrared transfer cavity. The Master and Seed laser solution allows to continuously frequency tune the cooling laser light over a range 75 1064nm Horiz. Dipole 1064nm Fiber Ampifer Mirror Beam splitter 1064nm Vert. Dipole 1064nm Seed Yb+ Repump 935nm and 638nm Yb+ 411nm Yb+ 369nm Freq. Pump. 76 Chapter 4. Experimental Setup and Methods Rb Repump MOT Blue Transfer Cavity Doubling Cavity Yb+ 369nm Pol. Pump. or 369nm Detection Rb Img Hori Locking Feedback Mechanical Shutter Yb+ 369nm Horiz. Cooling and Yb 399nm Yb+ 369nm Vert. Cooling Rb SigmaPump Laser Tapered Amplifier Fiber Coupler Polarizing Beam Spiltter 1064nm Lattice Rb Img Vert Optical Isolator AOM Rb Cool MOT Offset Lock 780nm TA MOT Rb Repump Img Rb Pi-Pump WM Lock 739nm TA Red Transfer Cavity Offset Lock Vapor Cell Lock 780nm Rb Repumper 780nm Rb Repump 780nm Rb Seed WM Lock Yb Hollow Cathode Lock Vapor Cell Lock 780nm Rb Master 780nm RB Master 935nm Yb+ Repump 740nm Yb+ Master 740nm Yb+ Seed 399nm Yb 411nm Yb+ Seed 369nm Yb+ Diode 638nm Yb+ Repumper Figure 4.10: Simplified schematic of the laser system of the atom-ion hybrid setup. Details are given in the text. 4.2. Trapping of Yb+ Ions of several hundred MHz. The main output beam of the cavity passes through an AOM for intensity control and fast switching and is then divided into the 369 nm Horiz. Cooling and the 369 nm Vertical Cooling beam. A second output beam of the cavity, that results from the reflection of UV light on the output facet of the nonlinear crystal, is sent through an independent AOM and is used for either the 369 nm Detection or the 369 nm Polarization Pumping beam, depending on the experimental setting. The light for the repump transition at 935 nm is provide by an ECDL, that is frequency locked to the infrared transfer cavity. A double-pass AOM is used to switch and frequency tune the repump light over a range of 50 MHz. In order to conveniently lock laser that do not require the ∼ MHz stability provided by the transfer cavities, we have implemented a digital frequency lock based on our commercial High Finesse WS7 wave meter. The wavemeter’s programming interface is used to make the wave readings available to the ion control software for real-time feedback and control purposes. The measurement shown in Figure 4.11 characterizes the short- and intermediateterm stability of the wavemeter for various wavelengths by logging the apparent frequency drifts of the spectroscopy referenced ion lasers. In general, wavemeter drifts are on the order of several MHz per hour with worse performance in the ultraviolet and for low light intensities. Over timescales of days and weeks and the wavemeter drifts within the specified absolute accuracy of 60 MHz. Frequency [MHz] We currently implement a simple digital integrator loop by feeding the analog-channel 3 0 -3 740nm Seed 3 0 -3 9 6 3 0 -3 -6 -9 740nm Master 935nm 399nm 15 12 9 6 3 0 -3 -6 -9 -12 0 1 2 3 4 5 6 7 8 9 Time [h] Figure 4.11: Short- and intermediate-term stability of the commercial wavemeter for the experimental conditions in our laboratory. 77 Chapter 4. Experimental Setup and Methods voltage Vn+1 = Vn +α·(νSet −νW avemeter ) back to the piezo of the ECDL grating. The linear frequency-voltage tuning slope is determined experimentally for each grating stabilized diode laser. The small loop frequencies on the sub-Hz scale31 are sufficient for efficient frequency feedback due to the intrinsic short term stability of the diode lasers and the excellent linearity of the grating feedback (α(ν) = const.). The wavemeter locking technique is used for the repumper laser at 638 nm. In order to efficiently repump from all Zeeman states of the 2 F7/2 -state, that span, for example, 175 MHz at a magnetic field of 13.8 G, a triangular modulation voltage at 200 Hz is added to the grating feedback voltage to continuously sweep the laser. The stability of the wavemeter feedback is also sufficient for the locking of the 369 nm diode laser that generates the light for the 369 nm Frequency Pumping beam. In the long term, direct 369 nm diode lasers could completely replace the current frequency-doubling system in our experiment, that requires a relatively high maintenance effort due to the rapid aging of the inefficient 739 nm Fabry-Periot laser diodes. The Yb spectroscopy stabilized blue transfer cavity could be used as a stable frequency reference in this case. The direct generation of 411 nm light for the 2 S1/2 - 2 D5/2 transition is achieved with an ECDL that employs a frequency selected blue-ray Fabry-Periot diode, which is heated to 45 ◦ C. Since the laser is not intended for coherent manipulation of the 139 Hz narrow electric quadrupole transition (see Section 6.1.1), its frequency is only referenced to the blue transfer cavity. We have determined a instantaneous linewidth of 380 kHz for this diode laser using a self heterodyne measurement [134]. Intensity and Pulse Control The generation of light pulses with computer-controlled duration and intensity is implemented in the same way for the 369 nm Horiz. Cooling, 369 nm Vert. Cooling, 935 nm, 411 nm, 369 nm Detection, 369 nm Pol. Pump.32 and the 369 nm Freq. Pump.32 laser beams. Each control loop includes an Acousto-Optical-Modulator as the control element, a logarithmic photo diode as the sensor, and a FPGA-controlled unit33 which combines the functionality of a digital PID and a DDS-sourced AOM driver. The typical extinction rations of ∼ 104 − 105 , that we obtain, when switching UV laser beams with AOMs are not sufficient for some of our experimental pulse sequences. Whereas the ion should scatter on the order of 107 photons during detection and Doppler cooling, it must not scatter a single photon during coherence time measurements that can extend over several hundred milliseconds. In order to increase the extinction ration we synchronize the 31 The loop frequencies are in practice limited to the typical light integration times of the wavemeter. Since the intensity is not critical we only use open loop control for these beams. 33 developed by Christoph Zipkes http://decads.scondaq.com/. 32 78 4.3. Experimental Cycle Figure 4.12: Experimental cycle of the hybrid trap. During time needed for the production of ultracold neutral cloud, a range of operations are performed on the ion to compensate for long term drifts of laser frequencies and electrostatic offset fields. The controlled interaction window between the ion and atoms is only ∼ 8 s out of a total experimental cycle time of ∼ 80 s. Figure adapted from [69]. open and closing of mechanical shutters in the beams with the AOM light pulse sequences. 4.3 Experimental Cycle The production of an ultracold cloud of neutral atoms as described in Section 4.1 takes about 65 s. After this time the atomic sample is available for several seconds for interactions with the ion until it is finally destructively analyzed by resonant absorption imaging. The total experimental cycle time of the experiment is 80 s (Figure 4.12). On the ion side, the time before and after the interaction is used for the preparation and the detection of the ion state as well as for a number of compensation steps to keep the conditions of the experiment constant from one experimental cycle to the next. The low rates of data acquisition in the experiments on ion-atom interaction, that only obtain a single binary outcome once every experimental cycle, make it necessary to accumulate data over long time scales. For this reason it is important to guarantee the repeatability of the experiment with continuous measurements that compensate for drifts in the micromotion [109], the frequency of the lasers and the relative positioning of the ion and the atom cloud [69]. With only ∼ 1100 repetitions of the experiment available in 24 hours, the autonomous operation of the experiment, including the automatic reloading of ion (Section 4.2.1)) is essential for many of our experiments. The flexible experiment control software has been 79 Chapter 4. Experimental Setup and Methods designed34 with this objective in mind from the very start35 . 34 35 by Christoph Zipkes http://decads.scondaq.com/,[69] Our recent implementation of a Skype interface in the software has made it now possible for the experiment to call, if a laser falls out of lock or a problem arises with one of the interlocks of the experiment. 80 Chapter 5 Cold Chemistry with Single Particles In Chemistry the control over the structure of matter and its properties, and the study of the underlying principles are traditionally performed under conditions dominated by the laws of thermodynamics. Chemical reactions are analyzed and steered by the choice and amount of agents and by varying macroscopic variables such as temperature and pressure. In recent years the booming field of cold and ultracold, dilute matter experiments has handed researchers new tools to investigate and control chemical reactions. At low temperatures the influence of externally applied radiation or static electromagnetic fields on chemical reactions is not attenuated by the averaging effect of large motional and internal state distributions of the reactants [59]. One can imagine, that the ultimate degree of control and understanding of a particular chemical reaction could be achieved, if all motional and internal quantum states of the reacting particles were manipulated precisely and the states of the final reaction products were analyzed in detail. In the context of neutral atoms, photo association, for instance, has been employed to efficiently transfer colliding atoms into electronically excited bound molecular states [135]. Using Feshbach resonances ultracold atoms have also been associated into weakly bound molecules [83]. The interaction between ions and neutral atoms as outlined in Chapter 2 is predicted to be subject to the same range of experimental control methods. In this Chapter we describe our first use of the Yb+ +Rb hybrid trap as a tool to investigate cold chemical processes and inelastic collisions1 . To obtain a clean and wellcontrolled experimental system, we use state-selected single particles, precise control of the interaction parameters and finally analyze the reaction outcomes in detail using in-trap mass spectrometry. This allows us to study the large spectrum of electronic state changing collisions that we have introduced in Section 2.3.2, including the various types of charge exchange and collisional quenching. We demonstrate how control over the internal electronic state of a single ion and the hyperfine state of neutral atoms can be employed to tune cold exchange reaction processes. Our measurements show a large sensitivity of the charge-exchange reaction rates to the atomic hyperfine state, highlighting the influence of the nuclear spin on atom-ion collisions. 1 The first results on ground state charge exchange collisions were published in [35] and also form part of C. Zipkes thesis [69]. The main Sections of this Chapter are based on the publication [47]. 81 Chapter 5. Cold Chemistry with Single Particles 5.1 Chemistry in the Yb+ + Rb Ground State The theoretically considerations of Chapter 2 have given an idea of the complexity of inelastic interactions in the heteronuclear Yb+ + Rb system. We therefore start our investigation into cold chemistry with the simplest scenario and considering collisions of ions with atoms in the electronic ground state. In the absence of resonant light, collisions in the Yb+ (6s)+87 Rb(5s) state natural arise from inserting a Yb+ ion into a neutral bath of Rb atoms. In addition to sympathetic cooling [33] by elastic scattering the immersed ion can also undergo collision induced transitions to the single lower lying binary molecular state that corresponds asymptotically to Rb+ +Yb(6s2 )1 S (see Section 2.3.2). The existence of a lower lying charge exchange channel is a feature Yb+ +87 Rb(5s) shares with several other atom-ion combinations of current quantum hybrid experiments [34, 42, 41]. It is the result of the generally higher ionization potential of earth alkali metals used as the ion species, compared to alkali metals that make up the ultracold neutral gas. 5.1.1 Measurement Sequence Experimentally, we measure the inelastic loss rate γℓ at the single particle level by immersing single ions for a time t into the neutral atom cloud and determining the survival probability [33, 35, 136, 65], Ps = exp (−γℓ t). The measurement sequence starts with the confirmation of the presence of a single bright Yb+ ion in our trap (see Section 4.2.1). Prior to the arrival of the neutral cloud, the ion is displaced from the interaction region along the ion trap axis, in order to avoid atom-ion interactions during the transport of the atoms. By ramping the end cap voltages, the ion is then moved into the centre of the cloud within a few ms to start interactions. The interaction interval is terminated by releasing the atoms from the magnetic or optical trap. We infer the central in-trap density na ≡ n3d (0) of the neutral cloud at the end of the interaction from the time-of-flight ab- sorption images according to Equation 4.11. Depletion and heating effects due to elastic collisions play a minor role for the large, low-density clouds of magnetic trapped atoms. For optically trapped atoms the changes in the neutral bath during the interaction interval are taken into account by a simple model that linearly interpolates between the measured cloud conditions before and after ion interaction. Following the time of flight imaging of the neutral cloud the ion cooling laser light is turned on and the presence or absence of a fluorescing Yb+ ion is detected. In order to describe the measured loss rates we make use of the Langevin capture model 82 5.1. Chemistry in the Yb+ + Rb Ground State |2,2ia of Equation 2.20, that introduce the inelastic collision loss efficiency ǫS of collisions of electronic ground (S) state 174 Yb+ ions with 87 Rb for the case atoms prepared in the |2, 2ia hyperfine spin state of their electronic ground state manifold. We have shown that |2,2ia the measured loss rate with a value of ǫS = 10−5±0.3 is indeed energy independent and scales linear with the neutral atom density in [35]. The measurements in Chapter 6 will show that the spin of the 174 Yb+ ion is in an almost completely mixed state and that col- lisions happen three times more often in the triplet potential than in the singlet potential. Similarly low inelastic rates have been found for other ion-atom combinations in hybrid traps [34, 42, 58] except for resonant charge exchange [40, 43]. The permanent loss of fluorescence can come about in two different ways. On the one hand, it can mean that the ion has undergone a chemical reaction, such that a chemically distinct, dark ion, such as (RbYb)+ , Rb+ or Rb+ 2 is left in the trap after the interaction. Whereas ions (RbYb)+ and Rb+ are the result of radiative association and radiative charge exchange, respectively, the Rb+ 2 ion can be formed in the secondary process (RbYb)+ + Rb → Yb + Rb+ 2 (5.1) from a previously associated (RbYb)+ ion2 . On the other hand the absence of fluorescence can indicate the loss of the charged particle from the trap. If in an exothermic inelastic collision the kinetic energy imparted on the charged particle exceeds its trap depth, then the particle will be lost from the trap. From numerical simulations of our ion trap we estimate the depth of our ion trap to be on the order of 250 meV. For Yb+ and Rb colliding in their electronic ground state the only inelastic processes that lead to ion loss are the various pathways of the charge exchange reaction, since the release of energy in spin changing collisions is only on the order of µeV (see Chapter 6). To distinguish the two events, ion loss and the presence of a dark ion, we perform inelastic fluorescence loss experiments with binary ion crystals. The four possible outcomes after the interaction are shown in Figure 5.1a. In the case of a single bright ion left in the trap, the position of the ion in the trap clearly reveals the presence of a further dark ion. Mass Spectrometry In order to further determine the chemical composition of the dark ion we identify the ion by its mass. For this purpose we perform in-trap mass spectrometry[137, 35] on the binary ion crystal consisting of one bright ion and one dark ion of unknown mass. The principle of this in-trap mass spectroscopy technique is based on probing the mass 2 For molecules in the absolute electronic ground state of the system, this is the only two-body decay channel, as the dissociation of molecules by the process (RbYb)+ + Rb → Rb2 + Yb+ is energetically not allowed (see Table 5.1). 83 Chapter 5. Cold Chemistry with Single Particles a b Fluorescence counts [a.U.] dependent eigenmode frequencies (see Section 3.1.3) by resonant excitation. In our experi- 40 30 20 10 0 (RbRbYb) 30 + 35 (RbYb) + 40 + Yb 45 + Rb 50 55 Common mode excitation frequency [kHz] Figure 5.1: Identification of the inelastic reaction outcome. a Fluorescence images of the binary bright ion crystal before interaction (top) and images of the four possible reaction outcomes (bottom). b Example light curves of mass spectroscopy scan for two different dark ion mass, a dark Yb+ ion (black) and a charge exchange Rb+ ion (red). We do not observe RbYb+ nor RbRbYb+ in our mass spectroscopy scans. ment we perform the mass spectroscopy by monitoring the fluorescence light of the binary crystal while exciting its centre-of-mass mode frequency through light pressure modulation. To this aim a frequency tunable square wave is imprinted onto the intensity of the Doppler detuned 369 nm Horiz. Cooling light, that illuminates the ion pair, by chopping the laser beam with an Acousto Optical Modulator3 . In a typical measurement as seen in Figure 5.1 the frequency of the modulation is linearly swept over the region of possible centre-of-mass frequencies during a scan time of several seconds. As the modulation of the light pressure becomes resonant with the centre-of-mass resonance of the trap, the oscillation amplitude of the collective ion mode becomes large and we observe a reduction in the number of detected fluorescence counts. Numerical simulations show that for our experimental parameters the decrease in the photon rate is the result of two different effects. In addition to the direct reduction of photon scattering rate of the ion due to the Doppler shift of the cooling laser transition, the detected photon rate is also suppressed because the bright ion spends less time in the ∼60 µm wide field of view of the photon counter. The large ion oscillation amplitudes needed to achieve a detectable reduction in the fluorescence intensity also explain why the eigenfrequencies of linear ion crystals beyond the COM mode cannot be detected with this method and why the dip in the fluorescence rapidly becomes smaller for the centre-of-mass mode of crystals with increasingly unequal ion masses. In both cases 3 The laser beam is aligned at 60◦ with respect to the trap axis and can therefore be used to excite both the axial and the radial trap frequencies. 84 5.1. Chemistry in the Yb+ + Rb Ground State the large amplitudes of the relative motion between ions, which are required for a fluorescence drop, exceed the validity region of the linear approximation in Equation 3.17 and the motion becomes unharmonic. Compared to the alternative method of electronic excitation, our optical excitation scheme is self regulating so that ions can not be heated out of the trap by the mass spectrometry measurement. By repeatedly sweeping a small frequency range we can detect the COM mode frequency with a relative error of less than 10−3 , which is sufficient to resolve mass differences of 1 amu at ion masses of ≈ 174 amu (see Figure 5.2). Molecular Ions The fluorescence image and mass spectroscopy analysis shows that the fluorescence loss in the collisions of S-state 174 Yb+ ions with 87 Rb atoms in |2, 2ia is the result of ion loss from the trap in 65% of all events, whereas the presence of Rb+ is detected in 35%. We do not observe any ions with masses that would correspond to radiatively associated RbYb+ ions. We have previously considered the 65:35 splitting as the branching ratio of non-radiative to radiative charge exchange [35]. In light of a recent theoretical study [76] of charge exchange in our system this interpretation is becoming increasingly unlikely. The study confirms that radiative charge exchange indeed will lead to trapped Rb+ as the kinetic energies of the Rb−1 ion (< 2/3 · 500 cm−1 ≈ 20 meV, see Figure 2.5) in this case are expected to be considerably smaller than our trap depth of 250 meV. The simulated occurence of radiative charge exchange exchange (30%) also agrees well with the observed 35% of trapped Rb−1 ions within experimental and theoretical uncertainties. The study, however, predicts the formation of bound RbYb+ molecule for 70% of reactive events and negligible non-radiative charge exchange, whereas we detect ion loss in 65% of all cases. The formation of molecular ions has recently been observed for excited state collisions of Ca+ in Rb [41] and Ba+ in Rb [50]. The analysis in Section 3.1.4 has shown that binary ion crystals are formed for a very large range of ion mass and we can therefore exclude ion trap stability as a possible reason for the absence of detected molecules. It has been suggested [42] that RbYb+ molecules could be collisionally dissociated in a secondary collision with Rb to form Rb+ 2 ions and a neutral Yb atom (Equation 5.1). of nucleon as 174 Yb+ 87 Rb+ 2 molecules incidentally have the same number so that in order to distinguish the molecule from the Yb+ ions4 by mass spectrometry, we have performed experiments with the isotope measurement results in Figure 5.2 could indicate the creation of 4 176 Yb+ . YbH+ Whereas the molecules in two This measurement predates the implementation of the 638 nm repump laser, which is now used to efficiently avoid the creation of dark Yb+ in the F-state. 85 Chapter 5. Cold Chemistry with Single Particles cases, we have not detected the presents of Rb+ 2 in any of our measurements. Position along trap axis [Pixel] In the case of very high neutral atom densities, three-body processes have to be considered, 8 a 6 4 2 0 -2 -4 Axial trap frequency [kHz] -6 51.6 b 51.5 51.4 51.3 51.2 51.1 0 4 8 Time [h] 12 16 20 Figure 5.2: Search for Rb+ 2 molecules. Over the course of 20 hours we successfully prepare 487 binary ion crystals of 176 Yb+ . The ions in their 2 S1/2 electronic ground state interact with a magnetically trapped cloud of 87 Rb atoms in the |2, 2ia state. a The pixel positions of the ions along the trap axis after interaction. Two bright ions (light blue), single ion loss (dark blue) and the presence of a bright and a dark ion (orange) are clearly distinguished by their position. b Axial trap frequency measurements for the data shown in a. Single ions and binary crystals of 176 Yb+ have the same centre-of-mass frequency. The dark ions are identified as 176 Yb+ in the 2 F7/2 state and two possible cases of 176 YbH+ . The orange line indicates the expected trap frequency for particles with mass m = 174 amu. We do not observe the formation of Rb+ 2 molecular ions. Both ion positions and axial trap frequencies drift over time, since loading induced offset fields are not compensated during this measurement. that give rise to further inelastic collision processes. Recent experiments in Ulm [138] have + demonstrated that Rb+ 2 molecules and Rb ions can also be created from ultracold clouds of atoms alone, if neutral Rb2 , created from three-body recombinations in the ultracold gas is ionized in a multi-photon process by dipole trapping laser light at 1064 nm. For our dipole trap parameters we do observe the occasional loading of an additional Rb+ ion for our largest and most dense Rb clouds. Further studies by the group [77] have also shown that ions can act as a three body reaction centre that functions as a catalyst for the formation of Rb2 . The release of the binding energy has been observed to impart kinetic energies on the order of several hundred meV 86 5.1. Chemistry in the Yb+ + Rb Ground State Table 5.1: Ionization energies Ei of selected neutral atoms and dissociation energies De of relevant molecules. The dissociation energies are given for the deepest molecular potential of the electronic ground state dimers. The full information about the Born-Oppenheimer potentials of the dimers can be found in the listed references. Ionization Energy Ei /cm−1 Rb Yb Ba Li 33691 50443 42035 43487 [140] [140] [140] [140] Dissociation Energy De /(cm−1 ) Rb2 Rb+ 2 YbRb+ 4269 [141] 5816 [142] 3496 [76] Figure 2.1 on the ion5 , which in our trap could lead to the loss of the Yb+ ion from the trap. These three-body effects are avoided in our experiments by working with thermal clouds with low neutral atom densities. We demonstrate the binary nature of the observed inelastic collisions by measuring their linear dependence on the neutral atom density in Figure 5.6 and in [35]. One possible reason why we do not detect the predicted RbYb+ ions in our trap is the dissociation of the molecule by cooling light at 369 nm (⇔ 3.36 eV). The electronic ground state molecules with binding energies of up to ≈ 2000 cm−1 (⇔ 0.25 eV) can be excited into the region of densely populated excited state of RbYb+ shown in Figure 5.3. By driving transitions to anti-binding states [139] or to the repulsive wall of binding states the presence of 369 nm light could potentially lead to a rapid dissociation of the RbYb+ molecular ion and loss of the charged particle. The significant fraction of ion loss events observed in our experiment would in this case correspond to dissociated molecules. In order to test this hypothesis the presence of molecular ions without 369 nm light could be shown indirectly in the following way. After immersion of a binary ion crystal in the neutral gas, the interaction outcomes are subjected to a strong electronic excitation at the axial COM frequency corresponding to a binary ion cyrstal with an RbYb impurity ion. If indeed a molecular ion is in the trap both ions should be resonantly heated out of the trap. After the excitation the 369 nm cooling light is turned on and the ion fluorescence images are taken. In comparison to a reference sequence without the electronic excitation, this measurement should lead to a sharp increase in double ion losses from the trap, thereby confirming the presence of RbYb+ ions in the trap6 . 5 6 The study was conducted with Rb+ ions. The idea about the possible dissociation of RbYb+ molecules by UV light was developed after the completion of our experiments on cold chemistry. It has therefore not been investigated in this thesis and will have to be addressed in future experiments. 87 Chapter 5. Cold Chemistry with Single Particles Hyperfine Spin Dependence We can compare these results now to inelastic electronic ground state collisions in the absolute lowest hyperfine state |F = 1, mF = 1ia of the neutral atom. For this purpose the atoms are prepared in the optical trap by an adiabatic passage microwave transition (Section 4.1.3). The rest of the preparation sequence is identical to one for |F = 2, mF = 2ia atoms described above. The hyperfine energy difference between the |F = 1, mF = 1ia and the |F = 2, mF = 2ia states of Rb is 30 µeV, which is larger than the collision energy but negligible on the scale of the molecular potentials or the trap depth (250 meV). We find a significantly enhanced probability for inelastic collisions with the ion in the electronic |1,1ia ground state ǫS |2,2ia = (35±11)×ǫS , which highlights the important role of the hyperfine interaction. We also observe a significantly increase in the fraction of reactive events that result in a trapped Rb+ ion compared to collisions with |F = 2, mF = 2ia atoms. The large differences in the inelastic collision behaviours for the two hyperfine states are not understood [143] and will require further theoretical efforts. In the next Chapter we will shed further light on the role of hyperfine and Zeeman structure in cold atom-ion collisions. 5.2 Inelastic Collisions in Electronically Excited States After the analysis of electronic ground state reactions we turn our attention to the electronically excited states of the Yb+ + Rb system. The study of electronic states is interesting both for fundamental and practical reasons. The control over the internal states can be used to steer chemical reaction and provides an additional experimental degree of freedom to further the understanding of chemical reactions at low temperatures. The stability of excited state is also of importance for buffer-gas cooled ion clocks [144] and for the prospect of quantum information protocols in hybrid traps. We start by determining the inelastic collision loss rate coefficients for the long-lived 2 D3/2 (radiative lifetime 52 ms) and 2 F7/2 (radiative lifetime 10 years) states of the ion. By optical pumping, both states are prepared as “dark” states, in order to study pure two-body collisions without the presence of light. This approach differs fundamentally from other experiments in atom-ion[41] excited state collisions, which always have been in the presence of near-resonant laser light, where effects of photo association can complicate the picture. For the measurement the ion and the neutral cloud are first overlapped in the 2 S1/2 ground state. The ion is then prepared in the 2 D3/2 state in less than 10 µs by optical pumping from the 2 S1/2 state using the 369 nm cooling laser. To end the interaction interval the 369 nm is switched off and the 935 nm repump laser light is turned on to repump the ion back into the ground state within microseconds. The fast pumping times, which are 88 3 } 5.2. Inelastic Collisions in Electronically Excited States [3/2]1/2 E (eV) 2 935 nm P1/2 2 2 D5/2 D3/2 369 nm 3 2 F7/2 2 2 411 nm P3/2 2 P1/2 1 2 Yb + S1/2 2 S1/2 0 Rb Yb++Rb -2 Yb+Rb+ Figure 5.3: Level scheme of the ion-neutral system. a) Left: Yb+ level scheme with the transitions used for optically pumping the ion (to scale). Middle: Level scheme of a Rubidium atom. Right: Asymptotic level schemes of the two channels Yb+ +Rb and Yb+Rb+ . The collision is initiated in the Yb+ +Rb manifold and the Yb+Rb+ manifold can be populated by charge-exchange processes. 89 Chapter 5. Cold Chemistry with Single Particles significantly faster than the inverse of the Langevin collision rate of a few 10−3 s are needed |2,2ia because of the high inelastic collision loss efficiency of ǫD collisions of 174 Yb+ = 1.0 ± 0.2 that we measure for with Rb atoms in the |F = 2, mF = 2ia hyperfine ground state. The constant ǫ = 1 corresponds to the largest allowed inelastic collision rate in the semiclassical model, and even in near-resonant charge-exchange between equal elements of atoms and ions [40] it was expected and approximately found to be ǫ = 1/2. For the inelastic rate measurement of the extremely long-lived 2 F7/2 state (radiative lifetime 10 years) we first populate the state outside the neutral cloud by optical pumping with the 411 nm laser on the 2 S1/2 2D 5/2 line [127] with spontaneous decay on the 2 D5/2 2F 7/2 transition and then ramp the ion into the centre of the cloud. The end of the interaction is defined by the release of the atomic cloud from its trap. The inelastic collision loss efficiency of the F-state |2,2ia is ǫF = 0.018 ± 0.004. We also perform the mass spectrometry for the two excited states and find ion loss as well as significant Rb+ production in both cases (see Table 5.2). It is important to emphasize that in the case of electronically excited states both reactive collisions and quenching collisions can contribute to ion loss (see Figure 2.4). The large number of densely packed molecular levels make a multitude of inelastic pathways possible. However, we do not observe the formation of molecular ion in the case of excited state collisions either. |2,2ia In addition to ǫD |1,1ia we have also determined ǫD for collision of with atoms in the |F = 1, mF = 1ia state, which is identical to the former within experimental errors. Once calibrated in this way, D-state inelastic loss rate measurements can be used to directly measure the local density of neutral atoms at the position of the ion without resorting to time-of-flight imaging of the atomic cloud. We will make use of this technique for the experiments in Chapter 6. 5.2.1 Collisional Quenching With regards to collisional quenching between different electronic levels of the ion, we are principally able to detect two different quenching scenarios: 2 F7/2 → 2 S1/2 and 2 D3/2 → 2F 7/2 . We directly observe quenching of 2 F7/2 → 2 S1/2 as shown in Figure 5.4, where we exemplarily show two different experimental runs, with and without quenching. In this measurement we prepare two ions in a small Coulomb crystal at time t = 0 in the 2 S1/2 ground state. The optical pumping into the 2 F7/2 state using light at 411 nm is switched on at t = 500 ms for 500 ms and typically within 100 ms the ion is pumped into the desired state. As a result, the fluorescence counts on the 369-nm transition drop to zero and the 369-nm cooling laser light is then turned off. After the interaction with the neutral atoms (effective duration typically 16 ms) and the subsequent removal of the neutral atom cloud, the laser is switched on again and the fluorescence on the 2 S1/2 − 2 P1/2 transition (together 90 5.2. Inelastic Collisions in Electronically Excited States Table 5.2: Measured proportionality constant ǫ and branching ratios. Loss events of the charged particle are always chemical reactions for the S−states, whereas for D− and F − states quenching and chemical reactions can contribute. 87 Rb 2 S 1/2 174 Yb+ ǫ Charged particle lost Rb+ identified Dark Yb+ identified Hot ion (unidentified) Number of events |F = 2, mF = 2ia 2S 1/2 −5±0.3 10 65% 35% 283 2D 2F 7/2 3/2 1.0 ±0.2 0.018 ±0.004 754 1% 225 87% 12% < 1% 84% 15% |F = 1, mF = 1ia 2P 1/2 0.1 ±0.2 2S 1/2 (35 ± |2,2i 11)ǫS a 50% 48% < 1% 2% 236 2D 3/2 1.0 ±0.2 with repumping at 935 nm) is probed from t = 5000 ms onwards. If the ion is still in the 2F 7/2 state it will not scatter photons, however, if it has been quenched to 2 S1/2 , we observe fluorescence. At t = 6000 ms, the ions are optically pumped from the F -state back into the S-state to ensure that no ions have been lost. The fluorescence count rate during the preparation part of the sequence is lower owing to the presence of an offset field from the magnetic atom trap, which is turned off when the atoms are released. The dips in the ion fluoresecence after the quenching event in Figure 2.4 indicate a large temperature of the ion due to the release of kinetic energy in the quench. The only way the ion can still remain in trap after the quench from the F- into the S-state is if the Rb atom gets excited and takes up most of the internal energy of the ion. Given a trap depth of 250 meV the only possible Rb states are 4d 2 D5/2 and 6s 2 S1/2 just below the F-state level (see Figure 5.3). The quenching rate from 2 D3/2 → 2F 7/2 was observed not to be detectable above our background rate. The background rate for transfers into the 2 F7/2 results from collisions with background gas during the fluorescence detection cycle. 5.2.2 Inelastic Collision during Fluorescence Detection Having established the inelastic collision parameters of the metastable D- and F -states without resonant laser light, we now turn our attention to inelastic collisions in the presence of laser light on both the 2 S1/2 − 2 P1/2 (369 nm) and the 2 D3/2 − 3 [3/2]1/2 (935 nm) transitions. The purpose of the light is to experimentally tune the rates and to observe the occurrence of inelastic collisions in real time. The radiative lifetime of the 2 P1/2 state (8 ns) is too short compared to the collision rate to provide pure P-state measurements. Here and in the following, we assume that the F -state is unoccupied since the 411-nm light 91 Fluorescence counts S1/2 → P1/2 (kHz) Chapter 5. Cold Chemistry with Single Particles Optical pumping 2 S1/2 → 2F7/2 Optical pumping 2 F7/2 → 2S1/2 Time (ms) Figure 5.4: Collisional quenching from 2 F7/2 to 2 S1/2 . A two-ion Coulomb crystal is prepared in the 2 F7/2 dark state by optical pumping, then interacts with the neutral atoms, and is subsequently probed by laser fluorescence. An early appearance of laser fluorescence indicates collisional quenching (red curve) as compared to no quenching (blue curve). The large fluorescence dips indicate a high temperature of the ion crystal. is turned off and collisional quenching into the F -state has been measured to be negligible. Therefore, the overall inelastic collision rate is determined by a mixture of S-, P -, D-, and D[3/2]-states γℓ = 2π p C4 /µ na pS ǫS + pP ǫP + pD ǫD + pD[3/2] ǫD[3/2] . (5.2) Here, px is the occupation probability of state x, which we determine experimentally for different settings of laser intensities and detunings. The relative state population measurement is performed with the method described in Section 4.2.3. In Figure 5.5a we show the state populations of the S, P , D and 3 D[3/2]1/2 states as we vary the frequency of the laser at 935 nm on the 2 D3/2 − 3 D[3/2]1/2 transition. Figure 5.5b shows the associated change in the reaction rates in the presence of the neutral atoms. We demonstrate tun- ing of the reaction rate by one order of magnitude and we fit our data with the model of Equation 5.2. The only fit parameter is the inelastic collision efficiency ǫP = 0.1 ± 0.2 of the P-state, which is small and consistent with zero. Figure 5.6 shows the linear scaling of the inelastic collision rate with neutral atom density over almost three order of magnitude in density, confirming the picture of binary collisions. 92 Inelastic collision rate (s-1) State population px 5.2. Inelastic Collisions in Electronically Excited States a D3/2 S1/2 D[3/2]1/2 P1/2 b Figure 5.5: Inelastic collision control by laser light. a State population measurement of the ion in the absence of neutral atoms as a function of the detuning of the repump laser at 935 nm. b Inelastic collision rate for the same detuning data in presence of neutral atoms at na = 1 × 1018 m−3 . The solid line shows the theoretical result of Eq. 5.2. 5.2.3 Error Estimation for the Survival Method Since every single repetition of the experiment takes ∼ 80 s, the optimization of the inelastic loss rate estimation is of considerable importance for our measurements. The single ion inelastic rate estimation method used so far calculates the unknown rate γℓ from the survival statistics of many binary outcomes after a fixed interaction time tf . A sequence of n independent repetitions of the experiments gives rise to a distribution of k survivals and n − k losses that is described by binomial statistics with survival probability Ps = e−γℓ tf . For large numbers of trials and probabilities Ps not too close to 0 and 1, the binomial proportion and its 100 (1 − α)% confidence interval are well described by their normal approximation values k Pˆs = , n Pˆs ± z1−α/2 s Pˆs (1 − Pˆs ) n (5.3) where z1−α/2 is the 1 − α/2 percentile of a standard normal distribution. For simplicity of notation we will from now on only consider the 16 and 84 percentiles, that correspond to a 1σ confidence interval. The estimated inelastic loss rate and its confidence interval then 93 Inelastic collision rate (s-1) Chapter 5. Cold Chemistry with Single Particles 1017 1018 -3 Neutral atom density (m ) Figure 5.6: Density dependence of the inelastic collision rate with the repump laser on resonance. The exponent of the power-law fit (solid line) is 0.98 ± 0.02. follow from linear error propagation 1 γ̂ℓ = − ln Pˆs τ 1 ∂γℓ ˆ ∆Ps = − ∆γ̂ℓ = ∆Pˆs ∂Ps τ Pˆs s (1 − Pˆs ) 1 1 ∆Pˆs ∆γ̂ℓ = = γ̂ℓ Pˆs · n ln Pˆs Pˆs ln Pˆs (5.4) In this approximation the relative error on the estimated rate exhibits a minimum at P̂s,opt ≈ 0.20 independent of n. In order to obtain the smallest error for the least number of runs, the interaction time tf should be set such that the survival probability is Ps ≈ Ps,opt . 5.2.4 Error Estimation for the Fluorescence Method In the case of fluorescence scattering during inelastic collisions, we use a more efficient way to determine the unknown inelastic rate at the single particle level. The method makes use of the detected photon counts of the ion’s 369 nm fluorescence to extract the time when the ion undergoes an inelastic reaction. If the exact times tion,0 , ..., tion,j ..., tion,n of the inelastic events in a series of n experimental runs can be determined without error, then the maximum likelihood estimate (MLE) of the 94 5.2. Inelastic Collisions in Electronically Excited States 2.00 Bootstrap Number of Summed Samples 50 25 5 2 1 100 80 60 40 Theory Median Mean 16 % 84 % 1.75 Normalized Rate Summed Counts (50μs bins) 120 1.50 MLE Exp. Fit 16% MLE Exp. Fit 84% Ps=Ps,Opt Binomial 16% Ps=Ps,Opt Binomial 84% Ps=0.5 Binomial 16% Ps=0.5 Binomial 84% 1.25 1.00 20 0.75 0 0 2 4 6 8 0 10 20 Time [ms] 40 60 80 100 120 Number of Summed Samples (n) Figure 5.7: Inelastic rate estimation from the decay of ion fluorscence. a We fit the 369 nm fluorescence curves that result from the summation over n identical runs of the experiment with the exponential decay in Equation 5.10 to retrieve the inelastic loss rate. b Theoretical and experimental parameter estimation confidence intervals for the fluorescence decay method and binomial survival method as a function of the number of experimental repetitions n. inelastic loss rate γℓ in the ion survival probability function Ps (t) = e−γℓ t is given by [145] γˆℓ = n n P (5.5) tion,j j=1 The 100 (1 − α)% confidence interval of the rate estimate [145] γˆℓ 2n χ21−α/2,2n < γℓ < γˆℓ 2n , χ2α/2,2n (5.6) is limited by the sample size n. Here, χ2a,v is the 100 (1 − a) percentile of the chi squared distribution χ2 (v) with v degrees of freedom. By approximating the chi square distribution by a normal distribution and considering the 16 and 84 percentiles (1σ error interval), we obtain γˆℓ 2n 2n √ < γℓ < γˆℓ √ 2n + 2 n 2n − 2 n (5.7) for the uncertainty of the inelastic loss rate after n experimental runs. In practice, however, the accuracy by which the exact ion loss times tion,j can be determined is limited by the Poissonian nature of the fluorescence photons. At a detected photons count rate R from the ion the uncertainty of a single loss time is on the order of 1/R with back ground counts further complicating the estimation error. Instead of explicitly extracting 95 Chapter 5. Cold Chemistry with Single Particles the times tion,j , we construct a fluorescence decay curve by binning the photon arrival times and summing the photon histograms of many consecutive runs of the experiment. For a single run the photon detection probability is uniform f (t) = R, 0 ≤ t ≤ tion 0, (5.8) tion < t until the ion is lost. Over a large number of experimental runs, the summed photon histogram will approach the normalized photon detection probability (see Figure 5.7a) f (t) = 1 −γℓ t e . γℓ (5.9) We therefore fit the total binned light curve with the function f (t) = A · e−γℓ t + C, (5.10) which contains the desired inelastic loss rate γℓ and considers the amplitude A(R, n) and the back ground count level C(n) as additional parameters (see Figure 5.7). To take into account the Poissonian rather than Gaussian statistic of the binned photon counts we fit the histogram data (tph,i , f (tph,i )) minimizing the function [146] χ2mle = 2 nBins X j=1 f (tph,i ) − tph,i − 2 nBins X tph,i ln(f (tph,i )/tph,i ) (5.11) j=1,tph,i 6=0 instead of the more familiar least square residuals. In order to assess the efficiency of the method, we investigate how the parameter estimation error of the fluorescence decay method scales with the number of experimental runs n for typical experimental conditions. We compare the experimental result to the theoretical values for the ideal scenario of expontial maximum likelihood estimation in Equation 5.7. For this purpose we determine the parameter estimation errors for the inelastic loss rate by bootstrapping [147] a sample of 131 experimental runs. The uncertainties in the inelastic loss rate as a function of the number of experimental runs n are shown in Figure 5.7a for the ideal exponential maximum likelihood estimation (Equation 5.7), the experimental fluorescence decay fit method, the optimal binomial survival method, and a binomial survival method with survival probability Ps = 0.5. √ For a large number of experimental runs n ≫ 1, it follows from Equations 5.7 and 5.11 that the confidence intervals for the inelastic loss rate determined by the binomial methods are (Ps = Ps,opt ) 1.24 and (Ps = 0.5) 1.44 times larger than for the fluorescence decay 96 5.2. Inelastic Collisions in Electronically Excited States method. In other words, the binomial method with Ps = 0.5 requires three times more experimental runs than the fluorescence decay method to obtain the same precision. For inelastic collision rates γℓ on the order of the photon count rate R, the fluorescence method will clear not be efficient any more, since barely any photons are detected before the ion is lost. The method can, in principle, be extended to inelastic collisions without fluorescence. Instead of probing the survival of the ion after a single long interaction interval, the ion can be immersed for shorter intervals, several times (nc ) during the 8 s interaction window of a single experimental cycle. If the probing times and repetitions are chosen such that the individual interaction intervals have a large survival probability Ps,c , but the total survival probability Ps = (Ps,c )nc is low, then the times of the inelastic loss can be reconstructed and used for the computation of the inelastic loss rate. a b c d e Figure 5.8: Monitoring of inelastic atom-ion collisions. a-c Recorded fluorescence for selected events. After a quick initial loss, the fluorescence reoccurs after random times when the kinetic energy released in the inelastic collision has been removed by cooling. Then, a second collision occurs and the ion disappears again. The curves are vertically offset for clarity. d Sum of 343 repetitions of the experiment, which is fitted with an exponential decay for short times. e Zoom-in on the initial decay of plot d. The solid line is an exponential fit to the data. Reoccurence of fluorescence Finally, the presence of fluorescence during collisions can also be used to observe the kinematics of collision products. The Subplots a-c of Figure 5.8 show individual experimental 97 Chapter 5. Cold Chemistry with Single Particles runs, in which the initial sharp loss of detected 369 nm fluorescence results from an inelastic collision. In 4% of all experimental runs7 we observe that the fluorescence of the ion reoccurs after a certain time, before it undergoes a second inelastic collision and is dark again. The relatively slow increase of the reoccurring fluorescence indicates that the ion has significant amounts of kinetic energy and is on a large trajectory in the trap (much larger than the size of the atomic cloud). From the number of photons scattered during the recooling we determine a lower bound for the released energy of ≈ 8 meV. Since we keep the Yb+ ion in the trap, these processes cannot be charge-exchange reactions. One possibility are quenching processes with a kinetic energy release for the ion of less than the trap depth ≈ 250 meV. Since the ion is colliding in highly excited electronic states, such as P1/2 and D3/2 with internal energies of ≈ 3 eV, this suggests a mostly radiative decay into the ground state S1/2 . A further possibility could be the dissociation of a potential molecular ion by the 369 nm cooling light suggested in Section 5.1.1. We have also investigated if the reoccurrence events are linked to the dissociation of a potential molecular ion in a secondary collision with neutral atoms [42]. To this end, we have performed this measurement also with extremely short interaction times between neutral atoms and the ion (on the order of a few collision times) and performed mass spectrometry on the dark ion after the collision confirming the absence of a molecular ion. In light of a possible dissociation of molecules by laser cooling light in our current mass measurement these results should be reconfirmed using the electronic excitation mass spectrometry suggested in Section 5.1.1. 7 We consider this number a lower estimate since very short periods of reoccurence might go undetected. 98 Chapter 6 Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms In addition to the changes in the internal electronic quantum states considered in the previous Chapter, the interaction between ions and atoms also affects the spin degree of freedom (see Section 2.3.2). The spin state of the ion impurity during immersion in the ultracold spin-polarized atomic cloud has been in fact one of our longest standing questions since the inception of our hybrid system. The spin dynamics and decoherence of the ion are indeed of wider interest as the hybrid system realizes the coveted concept of an isolated, controllable single qubit that couples to an adjustable environment. The evolution of single spin-1/2 particle interacting with its surroundings is a problem that has been intensely studied in theory [60, 61]. It represents a model system for the comprehension of the quantum to classical transition [62] and is important for the suppression of qubit decoherence in practical applications of quantum mechanics, like atomic clocks and quantum information processors. In this Chapter we investigate the properties of the all-atomic spin-bath system that is formed by the immersion of the single Yb+ ions into the ultracold 87 Rb cloud in the hybrid trap1 . In the first part we introduce the qubit preparation, manipulation and detection methods, we have developed for controlling two different ion spin qubits. The description of the ideally isolated and tunable two-level Zeeman qubit in the electronic ground state of 174 Yb+ is followed by the discussion of the 171 Yb+ clock-transition hyperfine qubit. We use these two complementary single spin qubits to study spin dynamics and decoherence in the tunable environment of spin-polarized ultracold atoms and find spin coherence times T1 and T2 on the time scale of the Langevin collision rate. In addition to the spin-exchange interaction, we identified an unexpectedly strong spin-nonconserving coupling mechanism as the main source of spin decoherence. We further detect a strong ion-heating mechanism for atomic clouds prepared in the upper hyperfine manifold that we attribute to the release of energy due to 1 87 Rb hyperfine spin relaxation. This Chapter is mainly based on the publication [47]. 99 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms a b 174 Yb+ 171 Yb+ F=1 12.6 GHz F=0 mF -1 0 mJ -1/2 1/2 c 87 1 Rb F=2 6.8 GHz F=1 mF -2 -1 0 1 2 Figure 6.1: a Illustration of the trapped single ion spin coupled to a spin-polarized neutral atom cloud. b Level structure of the Yb+ electronic ground state in a weak magnetic field. To implement the spin-1/2 system, we use either the Zeeman qubit |mJ = ±1/2i in the isotope 174 Yb+ or the magnetic field insensitive hyperfine qubit |F = 0, mF = 0i and |1, 0i in the isotope 171 Yb+ . c The cloud of neutral 87 Rb atoms is prepared in one of the four atomic spin states |F = 2, mF = 2ia , |2, −2ia , |1, 1ia or |1, −1ia . A quantitative description of all three spin systems is given in Appendix A.1. 6.1 Methods The long trapping times, excellent coherence properties, and precise control over the internal and external quantum states make single trapped ions an excellent physical system for the realization of qubits [22]. The internal level structure of Yb+ ions offers the possibility to encode spin qubits in two levels of the Zeeman and hyperfine structure of the electronic ground state2 for the isotopes 174 Yb+ and 171 Yb+ , respectively (see Figure 6.1). The experimental challenge part is to realize state preparation, state manipulation and state detection for the two different spin qubits in a way that is compatible with the requirements of the hybrid system. There are several factors that need to be considered for the implementation of single ion qubit operations that are unique to the hybrid environment. They include the relatively high temperatures of the ions (see Section 6.2.3) and the low repetition rate of pulse sequences due to the long timescales of atom-ion interaction. During overlap with the atoms, the high inelastic losses efficiencies of electronically excited states (Chapter 5) furthermore prevents the use of protocols that require the occupation of metastable states. In some of our experiments also the strength and direction of the magnetic offset field is 2 We do not consider optical qubits for the investigation of spin dynamics because of the high reactivity of electronically excited states in the hybrid system. Moreover, due to its relatively short (D-state) or extremely long (F-state) metastable state lifetimes, Yb+ ions is currently a rather inconvenient candidate for the implementation of an optical qubit. 100 6.1. Methods imposed by the magnetic trap of atoms. 6.1.1 Zeeman Qubit in 174 Y b+ The first qubit we investigate is the Zeeman qubit, formed by the |mJ = 1/2i ≡ | ↑i and the |mJ = −1/2i ≡ | ↓i states of the 2 S1/2 electronic ground state of absence of hyperfine structure in the nuclear spin-0 isotope 174 Yb+ 174 Yb+ . The offers the advantage of studying an ideally isolated two-level spin system. By changing the external magnetic field the energy separation between the two qubit states can furthermore be tuned according to ∆E = 2.8 MHz/G. The qubit’s susceptibility to the first order Zeeman effect, however, also limits its coherence time due to dephasing by fluctuations in the magnetic field. For our typical experimental magnetic field stabilities this dephasing time is only on the order of hundreds of microseconds (see Section 6.1.2) and too short compared to the time scale of atom-ion collisions. We therefore conduct our investigations of coherent spin evolutions with the first order magnetic field insensitive hyperfine qubit in 171 Yb+ (see Section 6.1.2). Preparation of Zeeman Qubit In order to initialize the S1/2 ground state zeeman qubit we employ polarization selective optical pumping on the S1/2 -P1/2 cycling transition. Near resonant light of left (right) circular polarization, aligned along the direction of the magnetic field, drives the σ + (σ − ) transition. In this way, the ion is pumped within a few scattered photons into the upper or lower Zeeman state, which is then a dark state for ideally polarized pump photons (see Figure 6.2). In practice the fidelity of the initialization is limited by imperfections in the 174 Yb + P 1/2 g J = 2/3 gJ = 2 S1/2 mJ -1/2 1/2 Figure 6.2: State initialization of the Zeeman qubit by optical pumping. In the example shown the lower state is prepared by σ − polarized light. polarization preparation of the 369 nm Pol. Pump. beam. In order to characterize the 101 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms efficiency of the preparation in our setup we determine the fluorescence rate as a function of the motorized quarter wave plate angle in the polarization pumping beam. We fit the observed fluorescence rate in Figure 6.3a with the steady state solutions of the 4-level (S1/2 mJ = ±1/2, P1/2 mJ = ±1/2) rate equation model that considers the offset magnetic field strength of 13.8 G, the frequency detuning of ≈ −Γ/2 = 2π · 10 MHz and the imperfect retardance (0.2374) of the quarter waveplate at 369.5 nm. The free parameters are the misalignment angle of the beam propagation with respect to the magnetic field, and a small amount of beam displacement as the wave plate is rotated. The model allows us to calculate the preparation fidelity (see Figure 6.3b) and lets us test its robustness for different ion temperatures, magnetic fields, and laser detunings. The significantly higher state preparation efficiency of 99.8% for the lower (mJ = −1/2) as compared to 97% for the higher (mJ = +1/2) Zeeman state is caused by the additional frequency pumping effect of the Doppler detuned polarization pump light. Whereas this is not an issue for our current experiments, significant improvements to the state preparation fidelities can be made by optimizing the frequency detuning, increasing the magnetic field during preparation and using a better matched quarter-wave plate in the future. Counts in 2[s] 15000 12000 9000 6000 3000 a State Occupation Probability 0 1.0 0.97 0.998 0.97 0.5 b 0.0 0.00 0.25 0.50 0.75 1.00 1.25 1.50 Quarter Wave Plate Angle (rad/ ) Figure 6.3: Fidelity of the state initialization by optical pumping. a The photon scattering rate exhibits minima, when the alignment of the quarter wave plate results in σ + ( 14 π and 5 − 3 4 π) or σ ( 4 π) polarized light. The horizontal line (100 ± 5) shows the background counts without ion fluorescence. b The state populations of the qubit states after preparation, determined from the fit of the rate equation model to a. 102 6.1. Methods Detection of Zeeman Qubit Internal state detection in ions is based on state selective fluorescence detection [148, 149]. For this purpose the ion is addressed with a laser light on a cycling transition such that it will scatter many photons if it is in one internal state (| ↑i) and practically no photons if it is in the other internal state (| ↓i). A small fraction of the scattered photons will reach a photo-detector during the detection interval and will be recorded. Based on the distinctly different photon detection behaviours for the two quantum states, the original state of an ion can be inferred from a specific set of photon detection events by statistical arguments. For qubits encoded in the Zeeman split electronic ground state 2 S1/2 , the energy difference between the two levels is only on the order of the linewidth of the dipole-allowed cycling transition and therefore both states will resonantly scatter photons when illuminated by the fluorescence laser. In order to distinguish the two states the electron shelving technique [150] is employed3 , whereby one of the two states gets selectively transferred into a metastable state not occupied during the ion’s fluorescence cycle before the fluorescence detection is applied. In the case of the 2 S1/2 Zeeman qubit, the “dark” qubit state is mapped into the 2 D5/2 manifold by a laser that coherently excites the narrow electronic quadrupole transition [149, 152]. In order to reconstruct the original state of the ion for a given number of recorded photon counts n during the detection interval tdet we need to known the probability distributions p̃↑ (n) and p̃↓ (n) for detecting n photons in our detection setup if the ion starts in the qubit state | ↑i and | ↓i, respectively. Formally, we can decompose the probability distributions p̃↑ (n) = ηDark,↑ p̃Dark (n) + (1 − ηDark,↑ ) p̃Bright (n) p̃↓ (n) = ηDark,↓ p̃Dark (n) + (1 − ηDark,↓ ) p̃Bright (n) (6.1) into the probabilities ηDark,↑ , ηDark,↓ that the ion gets mapped into the metastable (dark) state from a particular initial state and the probability distributions p̃Dark (n) and p̃Bright (n) in the (dark) mapped and (bright) original state. In order to infer the state of an ion as | ↓i, if the number of detected photons n is smaller than a discriminator value nd or as | ↑i, if n ≥ nd , with high fidelity, the probability distributions in Equation 6.1 are required to have minimal overlap. This requires both a good distinguishability between the probability distributions p̃Dark (n) and a high state selectivity of the shelving sequence p̃Bright (n) and a ηDark,↑ ≈ 0 and ηDark,↑ ≈ 1. In ion trap experiments dedicated to quantum optic experiments the shelving errors have been shown to robustly reach the 10−3 or 10−4 level [152, 149] level. As will be explained in the following, the experimental conditions in the hybrid system do not permit electron 3 A notable exception is the method based on electromagnetically induced transparency in [151]. 103 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms shelving at this level. Whereas high fidelity state detection is essential for fault tolerant quantum computation and quantum limited metrology, we are interested in investigating intrinsically probabilistic processes in the hybrid system. This purpose requires the detection-error corrected spin state occupation probabilities, but does not demand perfect single shot readout. Experimental photon number distributions, determined from many identical repetitions with identical spin preparation, will be described by p̃Det (n) = p↑ · p̃↑ (n) + p↓ · p̃↓ (n) = (p↑ · ηDark,↑ + p↓ · ηDark,↓ ) p̃Dark (n) + (p↑ · (1 − ηDark,↑ ) + p↓ · (1 − ηDark,↓ )) p̃Bright (n), (6.2) where p↑ and p↓ are the occupation probabilities of the ion in the states | ↑i and | ↓i before the state measurement. Figure 6.4 shows the example of a histogram of the number of 0.07 0.06 Probability 0.05 0.04 0.03 0.02 0.01 0.00 0 20 40 60 80 100 120 Photons counted in detection interval Figure 6.4: Normalized histogram of the detected photon numbers during 174 Yb+ fluorescence detection. The number of photons is counted during a 2 ms fluorescence interval after electron shelving. The dashed line is a fit assuming the sum of two Poissonian distributions. The fit shown as a solid line, considers Gaussian intensity noise of 8% standard deviation in the detection laser beam4 . pDark = 508/1968. photons that are detected during a 2 ms fluorescence interval following a mapping into the 2 D5/2 state. The histogram is well-described by the sum of two clearly separable Poissonian distributions for p̃Dark (n) and p̃Bright (n) with average photon numbers nDark = 2.5 and nBright = 88.2 that represent photon scattering in the dark and bright state (in Equation 6.2). The detection events within p̃Dark (n) do not originate from the ion but are caused by stray light photons from the detection laser and detector dark counts (≈ 20 s−1 ). 4 This measurement predates the implementation of an efficient PID for the UV laser light. 104 6.1. Methods By introducing the discriminator level nd , we define bright and dark events and their <n ∞ P Pd p̃Det (n) . probabilities are pBright = p̃Det (n) and pDark = n=nd n=0 From Equation 6.2 it follows that the probability of dark detection events is determined by pDark = p↓ · ηDark,↓ + p↑ · ηDark,↑ , where the error made by the approximations <n Pd n=0 negligible for our discriminator nd = 20. p̃Dark (n) ≡ 1 and (6.3) <n Pd n=0 p̃Bright (n) ≡ 0 is If we rewrite Equation 6.3 as p↓ = 1 − p↑ = pDark − ηDark,↑ ηDark,↓ − ηDark,↑ (6.4) it becomes evident that the spin occupation probability can be reconstructed for any measurement from the experimentally determined fraction of dark fluorescence detection outcomes pDark and the independently calibrated shelving efficiencies ηDark,↑ , ηDark,↓ . Electron Shelving by Optical Pumping The state-selective electron shelving is performed in our experiment by driving the narrow 2S 1/2 - 2 D5/2 electric quadrupole transition [127] at 411 nm. The Zeeman level structures of the S1/2 and D5/2 states are depicted in Figure 6.5. The strengths of the ten possible transitions between these states are described by the Rabi frequencies [106] ΩE2 0 eE ′ = hS, m|(ǫ · r)(k · r)|D, m i 2~ 2 X (2) eE 1/2 2 5/2 (q) c ǫ i n j , = hS1/2 |C |D5/2 i ij 2~ m′ q=−2 −m q (6.5) (6.6) where hS1/2 ||C (2) ||D5/2 i denotes the reduced matrix element, the 2x3 matrices are the Wigner 3-j symbols and we have for now neglected the motional state of the ion in the trap5 . (q) The terms cij ǫi nj contain the dependence on the relative orientation of the polarization vector ǫ of the light, the wave vector of the light n and the magnetic field vector defining (q) the quantization axis. The 2nd order tensors cij are explicitly stated in [106]. In our experiments the geometry is chosen such that all three vectors are mutually orthogonal, resulting in the suppression of transitions other than ∆m = ±2. In contrast to previous experiments [152, 149], however, we do not use these transitions for coherent population transfer but instead perform a scheme that is based on incoherent 5 This corresponds to a vanishing Lamb-Dicke parameter ηLD , see Equation 6.7. 105 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms 174 Yb + D5/2 g J = 6/5 S1/2 mJ gJ = 2 -5/2 -3/2 -1/2 1/2 3/2 5/2 729 476 869 [MHz] -1/2 to 3/2 1/2 to -3/2 -1/2 to - 5/2 1/2 to 5/2 Figure 6.5: Zeeman level structure of 2 S1/2 −2 D5/2 quadrupole transition for 174 Yb+ . The linear polarization and the wave vector of the 411 nm light exciting the transition are aligned mutually orthogonal with the direction of the magnetic field, so that only transitions with ∆m ± 2 are excited. The relative transition strength and the Zeeman splitting of the individual components of the line are shown below. optical pumping. I will first briefly explain, why the standard approach to electron shelving on the electric quadrupole transition does not work well in our hybrid setup. A minor complication in this context is the fact that the natural lifetime of the 2 D5/2 in Yb+ ions is only τD5/2 = 7.2 ms. The branching ratio from the 2 D5/2 state is 0.17 for decay back into the 2 S1/2 ground state and 0.83 for spontaneous decay to the long lived 2 F7/2 . For typical experimental conditions the decay into the ground state during the fluorescence detection interval will therefore lead to a detection error at the few percent level. The main factor preventing coherent population transfer are the high temperatures of the ions during the qubit measurement. As we will see in Section 6.2.3, neutral atom flip heating can lead to ion equilibrium temperatures that are on the order of 200 mK. The light-ion interaction on the narrow quadrupole transition takes place in the strong binding regime (Γ ≪ ωtrap ), where the spontaneous decay rate is much smaller than the trap frequency and the motional side-band spectrum can be resolved. The coupling of the light to the ion, this time considering both its internal (Equation 6.6) and motional states, leads to the Rabi frequencies [100] † iηLD (a+a ) Ωn+∆n,n = ΩE2 |ni 0 hn + ∆n|e 106 (6.7) 6.1. Methods for transitions between the initial motional state n and the q final state motional state n+∆n. The Lamb-Dicke parameter ηLD = k · z0 = 2π λ · z0 = 2π λ ~ 2mωr takes the value 0.21 for the transition wavelength λ = 411 nm, the radial trap frequency ωr ≈ 2π · 150 kHz and the mass of the Yb+ ion m. For ion temperatures T ≈ 200 mK ⇔ n =≈ 27700 the light-ion 2 ·(n+1) ≈ interaction is far outside the Lamb-Dicke regime hΨmotion |k 2 z 2 |Ψmotion i1/2 = ηLD 2500 ≫ 1. The great variations in the Rabi frequencies Ωn,n (Equation 6.7)6 of the many thermally occupied initial states n, even for the carrier transition (∆n = 0) averages out any coherent behaviour on timescales much shorter than the π-Rabi flip time. In order to exceed the transfer efficiency limit of 0.5 for saturated incoherent excitation, 174 Yb + D5/2 F7/2 S1/2 mJ -7/2 -5/2 -3/2 -1/2 1/2 3/2 5/2 7/2 Figure 6.6: Electron shelving in the F-state by optical pumping. We frequency selectively drive the 2 S1/2 (m = −1/2) - 2 D5/2 (m = −5/2) transition to optically pump the population of the lower Zeeman level into the long-lived F-state. we employ the optical pumping scheme depicted in Figure 6.5. For this purpose we apply a magnetic offset field of 13.8 G and frequency selectively drive the 2 S1/2 (mJ = −1/2) 2D 5/2 (mJ = −5/2) transition with a intermediate linewidth laser (see Figure 6.6). Spon- taneous decay from the D5/2 state back into the S1/2 can only return the ion into the mJ = −1/2 state, thereby avoiding the leaking of population into the other qubit state. The timescale for optical pumping into the long lived 2 F5/2 is set by the radiative lifetime τD5/2 = 7.2 ms state and the branching ratio BF D5/2 = 0.83. The fidelity of the shelving is generally limited by our finite optical pumping time and for high ion temperatures also off-resonant pumping on the Doppler broadened transitions S1/2 (mJ = 1/2) D5/2 (mJ = −3/2) (17 MHz detuned) and the S1/2 (mJ = 1/2) - D5/2 (mJ = −1/2) (∆m 6= ±2 polarization suppressed and 10 MHz detuned) starts to contribute as an error source. Since the shelving efficiencies are temperature dependent, we need to determine the detection errors experimentally for different spin-bath configurations. Another important aspect 6 The explicit relations for evaluating the Rabi frequencies outside the Lamb-Dicke regime are given in [153]. 107 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms of 174 Yb+ qubit detection is that it needs to be performed in the absence of neutral atoms. The high inelastic loss efficiencies of electronically excited shelving states will otherwise lead to loss of the ion during the qubit detection. 3 3 D[3/2]1/2 2.2095(11) GHz F=0 F=1 b D[3/2]1/2 2.2095(11) GHz S =1 1/2 F=1 a F=0 F=1 2 2 P1/2 2.105 GHz 2 P1/2 2.105 GHz F=1 F=0 2 S1 /2 F=0 =1 =0 2 D3/2 0.86(2) GHz F=2 F=1 ' F=2 F=1 2 D3/2 0.86(2) GHz & ! % $ # " ! 2 S1/2 1 2 S1/2 F=1 F=1 12.643 GHz 12.643 GHz F=0 F=0 Figure 6.7: Initialization and Fluorescence detection of the hyperfine qubit in 171 Yb+ . a Initialization of the qubit by frequency selective optical pumping into the dark |F = 0, mF = 0i hyperfine ground state. b Qubit state detection. The large hyperfine splittings enables the efficient distinction of the states in the upper hyperfine manifold (F=1, “bright”), from the single lower hyperfine state (F=0, “dark”) by direct fluorescence detection on the UV transition without the need of electron shelving. Adapted from [68]. 6.1.2 Hyperfine Qubit in 171 Yb+ The second ion qubit we employ for our studies of spin dynamics and decoherence is the hyperfine qubit in electronic ground state of 171 Yb+ ions. The nuclear spin-1/2 ion has the simple hyperfine structure shown in Figure 6.7. In recent years 171 Yb+ ions have found widespread use in ion trap quantum metrology and quantum information experiments [154, 155]. Several 1st order magnetically insensitive (mF = 0 → m′F = 0) tran- sitions in 171 Yb+ are currently investigated for use in future ion-based optical frequency standards [156, 157, 158]. In our experiment we encode the qubit in the hyperfine states |F = 0, mF = 0i − |F = 1, mF = 0i of the 2 S1/2 electronic ground state that are connected by the microwave “clock” [159] transition at 12.6 GHz (see Figure 6.1 and Appendix A.1). Due to its low magnetic field sensitivity, the spin dephasing times of this qubit are significantly longer than the typical collisional timescales in our experiment. 108 6.1. Methods Initialization The initialization of the ion in the |0, 0i state is performed by the frequency selective optical pumping scheme shown in Figure 6.1. With the microwave repump field turned off, a short (250 µs) 369 nm Pol. Pump. laser pulse resonantly excites the ion on the S1/2 |F = 1i → P1/2 |F = 1i transition and pumps it into the dark |0, 0i state [68]. Microwave radiation resonant with the 12.6 GHz hyperfine transitions is used for the coherent qubit manipulation on the clock transition and for Rabi π-flips on the |0, 0i → |1, −1i and |0, 0i → |1, 1i transitions. Before we describe the coherent manipulation of the qubit in more detail, the qubit detection scheme needs to be introduced. Detection In contrast to the 174 Yb+ Zeeman qubit discussed in Section 6.1.1, the large hyperfine splittings and transition selection rules (F = 0, mF = 0 6→ F = 0, mF = 0) in 171 Yb+ enable state detection without the need for electron shelving into metastable electronic states [68]. For direct fluorescence detection the ion is probed on the 2 S1/2 |F = 1i →2 P1/2 |F = 0i UV transition, which together with the 2 D3/2 |F = 1i →2 [3/2]1/2 |F = 0i 935 nm transition forms a closed fluorecence cycle for ions initially in one of the three states |1, −1i,|1, 0i, |1, 1i. The probability for detecting n photons when starting in one of the F = 1 states is [148]: p̃|F =1,mi (n) = Here, P (a, x) ≡ α2 /η e−(1+α2 /ηDet )λ0 λn0 + P (n + 1, (1 + α2 /ηDet )λ0 ) n! (1 + α2 /ηDet )n+1 1 (a−1)! Rx (6.8) e−y y a−1 is the incomplete Gamma function and λ0 denotes the 0 average number of detected photons that would be recorded during the detection interval, if the ion did not leak into the dark state. α2 is the probability per scattered photon to leak from the |1, −1i, |1, 0i and |1, 1i into the |0, 0i state. Finally, ηDet is the detection efficiency of the photon counting system. The derivation [148] of Equation 6.8 does not consider the effect of background counts due to stray light and detector dark counts, which play a negligible role for the bright state. In contrast, when the ion starts in the lower hyperfine state |0, 0i, the probability for 109 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms recording n photons7 p̃|0,0i (n) = e λ0 −α1 η Det α1 α1 e−λD λD α1 ηDet + )λ0 )−P (n+1, (1− )λD ) P (n+1, (1− 1 n! (1 − ηαDet ηDet ηDet )n+1 (6.9) is dominated by background counts λD with an additional contribution that results from the leakage probability α1 from the |0, 0i into the |1, −1i, |1, 0i, |1, 1i states during the detection window. Figure 6.8a and b show the experimentally determined photon number distributions for ions prepared in the |0, 0i and the |1, 0i, respectively. Under the assumption that the state preparation errors should be negligible compared to other detection errors due to the very high efficiency of the optical pumping and the microwave Rabi π-pulse, we can fit both histograms directly by Equation 6.9 and 6.8, respectively. In direct analogy to the 1.00 0.10 0.08 Probability Probability 0.99 0.98 0.02 0.01 0.06 0.04 0.02 0.00 0.00 0 5 10 15 0 20 5 10 15 20 25 30 Photons counted in detection interval Photons counted in detection interval Figure 6.8: Normalized histograms of the detected photon numbers during 171 Yb+ fluorescence detection. a Distribution of recorded photon counts for ions prepared in the |0, 0i state and b for ions initialized the |1, 0i state during a fluorescence detection interval of 1 ms. The red curves represent fits of the data with the functions in Equation 6.9 and 6.8 for a and b, respectively. The fits use ηDet = 0.002 and yield α1 = 2.3· 10−6 and α2 = 3.8· 10−5 (see text for details). discussion of fluorescence detection in 174 Yb+ , and such that the detection error Err = <n Pd n=0 <n Pd 0 we can introduce a discriminator value nd <n ∞ P Pd p̃|F =1i (n)+ p̃|F =1i (n)+(1− p̃|0,0i (n) = n=nd n=0 p̃|0,0i (n)) = ηDark,|F =1i +(1−ηDark,|0,0i ) is minimal. The optimal discriminator is nd = 2 and for the measurement in Figure 6.8 the detection errors are ηDark,|F =1i = 0.038(6) and 7 Here we have extended the treatment in [148] to include the contribution from background counts. 110 6.1. Methods (1 − ηDark,|0,0i ) = 0.018(4). In most measurements the detection errors are even somewhat larger as we do not individually optimize the orientation of the detection light polarization vector [131] for the different atom trap configurations. The magnetic bias fields for our various measurements with the hyperfine qubit are 5.64(1) G in the optical trap, 6.91(1) G in the magnetic trap of the 87 Rb |2, 2ia and 7.56(1) G in the magnetic trap of the 87 Rb |1, −1ia atoms and they are approximately aligned with the ion trap axis. As in the case of 174 Yb+ , the imperfections in the detection scheme limit the direct correspondence of the detection of a bright or dark state, as determined from the statistics of photon events, and the occupation of a specific quantum spin quantum state. However, the statistically correct value of the spin probability p|0,0i = 1 − p|F =1i = pDark − ηDark,|F =1i ηDark,|0,0i − ηDark,|F =1i (6.10) can be calculated from the measured probability pDark and the detection errors. The direct fluorescence detection can only distinguish the |0, 0i state from the three hy- perfine states in the F = 1 manifold. During collisions with neutral atoms, also the |1, −1i and |1, 1i state can get populated. In order to detect a specific hyperfine state |1, mi we perform a π Rabi flip on the |0, 0i - |1, mi microwave transition just before the fluorescence detection. For quantum hybrid experiments, the 171 Yb+ ion has the great advantage that at no point during the state initialization, evolution or detection, significant amounts of time are spent in the electronically excited states. This opens the possibility to study ion spin dynamics in the atomic bath without the need to repeatedly merge and separate both species for successive interaction and detection cycles. We have not taken any quantitative data on the inelastic loss efficiencies of the various electronic states of the values similar as for detections as Ps = 174 Yb+ , Nc Ps,c ≈ e 171 Yb+ , but assuming we estimated the ion survival probability after Nc successive − n·BP D γLangevin (TP ǫP +TD ǫD )·Nc ηDet ≈ 0.99Nc , where we have used γLangevin ≈ 200s−1 , an average number of detected photons n ≈ 15 and the results of Chapter 5. This allows us to perform on the order of 100 qubit detections before the ion is lost. Coherent Manipulation The coherent manipulation of the qubit is performed with 12.6 GHz microwave radiation resonant with the hyperfine |0, 0i → |1, −1i clock transition. Due to the extremely small Lamb-Dicke factor of the microwave transition, the homogeneous magnetic fields at the position of the ion and the low magnetic field sensitivity of the clock transition, we do not observe any ion temperature dependent effects for the coherent qubit manipulation. 111 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms 1.0 0.8 pBright 0.6 0.4 0.2 0.0 0 20 40 60 80 100 Rabi pulse time ( s) Figure 6.9: Resonant microwave Rabi oscillations on the qubit clock transitions |F = 0, mF = 0i - |1, 0i in the optical for the maximum microwave power available in our system. Rabi frequency is Ω0 = 2π· 17.9 kHz, corresponding to a Rabi π flip time of 28.0 µs. Each point is the result of more that 2000 measurements. The coherent evolution of the two-level system during interaction with the microwave √ 2 +∆2 ·t Ω Ω20 0 radiation gives rise to Rabi oscillations p|0,0i (t) = Ω2 +∆2 sin2 in the qubit state 2 0 population, where ∆ is the detuning of the microwave frequency from the atomic transition frequency, Ω0 is the resonant Rabi frequency and the system is initially prepared in the state √ |0, 0i at t = 0. Figure 6.9 shows resonant (∆ = 0) Rabi oscillations (Ω0 ∼ P ) between the two qubit states for the maximally available microwave power P in our experiment. In order to probe the coherence time of our qubit in the absence of atomic collisions, we perform Ramsey spectroscopy. The spin population after a Ramsey sequence is described by [160] !2 q " ∆T 2 2 2 sin p|1,0i (∆, T ) = p 2 Ω0 + ∆ · τ /2 cos cos 2 Ω0 + ∆2 !# p Ω20 + ∆2 T ∆T ∆ sin −p 2 sin , 2 2 Ω0 + ∆2 Ω0 ! p Ω20 + ∆2 T 2 (6.11) where τ is the duration of the first and second microwave pulse, T the time in between the pulses and the ion is initialized in the state |0, 0i before the first microwave pulse. For short Rabi π/2-pulses (τ · Ω0 = π/2) and long waiting times T ≫ τ in between, the central section of the Ramsey fringe pattern in Equation 6.11 is well-approximated by 1 p|1,0i (∆, T ) = (1 + V · cos(∆ · T )), 2 112 (6.12) 6.1. Methods where we have introduced the visibility contrast V in order to account for the effects of p(|0,0>) decoherence. In order to determine the coherence properties of the ion under the exact 0.9 0.7 0.5 0.3 0.1 0.9 0.7 0.5 0.3 0.1 0.9 0.7 0.5 0.3 0.1 a b c 0.9 0.7 0.5 0.3 0.1 -10 -5 0 5 10 Frequency Detuning [Hz] Figure 6.10: Ramsey fringes for an ion in the magnetic field of the optical trap without neutral atoms. The decay of fringe visibility with increasing Ramsey time T a 83 ms, b 133 ms, c 258 ms. In a single experimental cycle one point can be measured Nc ≈ 5s/T times and every point shown in a-c is the average over several experimental cycles. In d we have plotted the Ramsey fringe for 258 ms only including data from a single experimental cycle for each point. The comparison with c shows that the long term drift of the magnetic field between one experimental cycle and the next is the largest contribution to the dephasing. conditions relevant for the later interaction measurement with the atoms, we perform the complete experimental cycle but without atoms8 . To this end, we prepare the ion in the |0, 0i hyperfine ground state and apply a π/2-pulse on the 12.6 GHz clock transition, √ creating a superposition state (|0, 0i + i|1, 0i)/ 2. After a varying waiting time we apply a second π/2-pulse, followed by a readout of the |F = 1i spin state. Figure 6.10 shows the Ramsey fringes of the ion measured for the optical trap configuration. We determine the transverse coherence time from the decay of the visibility with waiting time T and find T2 ≈ 130 ms. To identify the dominant source of qubit decoherence in our system we have additionally performed spin echo measurements and obtained coherence times of TRephase ≈ 700 ms. We can therefore exclude the phase stability of our microwave oscillator and 369 nm leakage light as important decoherence contributions. The limiting factor for the coherence time T2 in our experiment is dephasing due to the poor stability of the 8 The magneto-optical trap light is blocked for this purpose. 113 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms magnetic field. We can estimate the magnetic field stability by considering 1 dE 1 (gJ − gI )2 µ2B B 1 ∼ ∆ν ≡ (B) · ∆B = ∆B T2 ~ dB ~ E HF S ∆B ~E HF S ⇒ ≈ 10−4 , ∼ 2 2 2 B T2 (gJ − gI ) µB B where we have used Equation A.10 to evaluate the linear Zeeman shift (6.13) (6.14) dE dB (B) at the mag- netic bias field value of 5.64(1) G in the optical trap. The magnetic field stability is consistent with the specifications of our high current power supplies. For the magnetic trap configurations the coherence times of the qubit are about a factor of 3 shorter due to the higher magnetic bias field values and the fact that several power supplies contribute to generate the magnetic field. The dephasing effect is particularly strong in the 1st order magnetically sensitive transitions |F = 0, mF = 0i − |F = 1, mF = −1i and |F = 0, mF = 0i − |F = 1, mF = 1i as shown in Figure 6.11, where it is limiting the fidelity of the Rabi π-flips that we use for the detection of the individual states in the F = 1 manifold. 1.0 0.8 pBright 0.6 0.4 0.2 0.0 0 20 40 60 80 100 120 0 20 40 60 80 100 Rabi pulse time [ 120 0 20 40 60 80 100 s] Figure 6.11: Resonant microwave rabi flips on the transitions between the hyperfine ground state |F = 0, mF = 0i and the |1, −1i a, |1, 0i b and |1, −1i c states. The measurements were performed in the magnetic trap configuration for the |F = 1, mF ia atoms, hence the different Rabi frequency for the clock transition compared to Figure 6.9. Decoherence is starting to limit the fidelity of our Rabi π-flips on the 1st order magnetically sensitive transitions. 114 120 6.2. Experiments 6.2 6.2.1 Experiments Spin-Exchange and Spin-Relaxation in 174 Yb+ We start our investigation of ion impurity spin dynamics in the neutral cloud with measurements on the 174 Yb+ Zeeman qubit, which represents an ideal two-level system. As a first step, we determine the steady state distribution of the ion spin for all four atomic bath configurations |2, −2ia , |1, −1ia , |1, 1ia and |2, 2ia . For this purpose the neutral 87 Rb atoms are loaded into the optical trap and prepared in one of the four polarization states using the techniques described in Section 4.1.3. During this time a homogeneous magnetic bias field pointing along the trap axis is ramped up to 13.8 G. The single 174 Yb+ is then moved into the centre of the neutral cloud and one of two different sequence shown in Figure 6.12a is applied. The first sequence (case b) serves as a direct measurement of the detection errors. After an ion-atom interaction time of 40 ms, that is chosen such that the ion acquires both its spin and motional energy steady state, the neutral atom cloud is displaced from the interaction position by 45 µm in the vertical direction. Once the atoms have been removed in this way and the ion-atom interaction has ceased, the ion is prepared in the upper or lower spin state by a 500 µs long polarization pumping pulse. Since only a few UV photons are scattered during the optical pumping the temperature, that the ion has acquired during interactions with the atoms, is not changed during state preparation. This is followed by the quantum state detection sequence described in Section 6.1.1 that includes electron shelving with the quadrupole laser at 411 nm and subsequent fluorescence scattering. After the detection repump light at 638 nm is used to return the population in the shelved F-state back into the ground state and only then is the neutral cloud ramped on top of the ion again. The entire pulse sequence takes 430 ms and is repeated 8 times during a single experimental interaction cycle in order to increase the data acquisition rate. Significantly more repetitions could be achieved in the future by ramping the ion from the interaction region9 instead of the current method that moves the atoms in the dipole trap using the piezo mirrors. As shown in Figure 6.12b, the measured spin distributions pBright differ between the neutral atom bath states |1, −1ia , |1, 1ia and |2, −2ia , |2, −2ia . Given the known preparation efficiencies from Section 6.1.1 we can calculate the detection errors from Equation 6.4. For measurements with 87 Rb |F = 2ia , the detection errors are (1 − ηDark,↓ ) = 0.19 ± 0.01 and ηDark,↑ = 0.03 ± 0.01, and for measurements involving the |F = 1ia manifold we find 8 We repeat 500 µs long 411 nm pulses every 5 ms for a total optical pumping time of 60 ms. Numerical simulations have since shown that a continuous excitation would lead to higher transfer fidelities. 9 Since this requires in practice photon collection and laser addressing of the ion at two different positions along the ion trap we have chosen to move the atoms 115 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms (1 − ηDark,↓ ) = 0.10 ± 0.01 and ηDark,↑ = 0.00 + 0.01. The larger detection errors of the Zeeman qubit, when the atoms are in the |F = 2ia state result from the increased ion kinetic energy due to release of atomic hyperfine energy (see Section 6.2.3). c 369nm Pol. Pum p. b c b c 369nm Det ect ion 411nm 638nm 0[ µm] At om Posit ion -45[ µm] a 0.0 0.2 0.4 0.6 1.0 0.8 Tim e[ s] 1.0 0.6 0.6 p ↑, 8 pBright 0.8 0.4 0.2 0.0 b c |2, -2>a |1, -1>a |1, 1>a |2, 2>a Rb state |2, -2>a |1, -1>a |1, 1>a |2, 2>a Rb state 0.5 0.4 d |2, -2>a |1, -1>a |1, 1>a |2, 2>a Rb state Figure 6.12: Measurement of the equilibrium spin states of the 174 Yb+ Zeeman qubit for all four atomic bath configurations |2, −2ia , |1, −1ia , |1, 1ia and |2, 2ia . a Timing diagram of the two different sequences that are used to determine the detection errors b and the equilibrium spin states c for each bath state. The sequence starts with the ion immersed in the centre of the neutral cloud. After 40 ms of interaction, the atoms in the dipole trap are ramped away from the position of the ion in a 80 ms ramp and the lower Zeeman state is shelved during a 60 ms long optical pumping sequence10 . After 2 ms of fluorescence detection, the ion is repumped from the F-state for 160 ms and the atoms are ramped back on top of the ion for the next interaction cycle. For the measurement of the detection efficiency b and the steady state spin state c the preparation of the qubits in the | ↑i (full symbols) or | ↓i (open symbols) with the 250 µs UV pulse is performed after and before the interaction with neutral atoms, respectively. b The detection errors of the ion spin state are different for the |2, 2ia , |2, −2ia states compared to the |1, 1ia , |1, −1ia bath states due to the unequal ion temperature in the two cases (Section 6.2.3). c and d show the ion spin steady-states for all four atomic bath configurations |2, −2ia , |1, −1ia , |1, 1ia and |2, 2ia before and after the correction for the detection efficiencies according to Equation 6.4. The second sequence (case c) is used for the actual measurement for the ion spin steady state. It only differs from the first sequence by the fact that the ion is prepared in the upper or lower state before the atom-ion interaction. The fact that the measured spin distributions are identical for the two different initial state preparations demonstrates that indeed the collisional steady state distribution is measured. The corrected steady state spin dis- 116 6.2. Experiments tributions p↑,∞ , shown in Figure 6.12d are calculated from the measured spin distributions pBright (Figure 6.12c) and the previously determined detection errors, with Equation 6.4. In order to determine the timescale of the spin dynamics of the ion due to interactions with the neutral atoms, we measure the coherence time T1 for two different bath configurations |F = 1, mF = 1ia and |2, 2ia . To this aim the immersed ion is initialized in the upper or lower qubit state by a short 50 µs polarization pump pulse. After a varying time t the ionatom interaction is abruptly ended by pushing the neutral atom cloud away with resonant imaging light in less than 50 µs. During the interaction period, the ion undergoes binary collisions with the atoms, and we normalize interaction times by the Langevin time constant tL = 1/γL . The Langevin collision rates of the ion in the centre of the atomic clouds are ex|F,mF ia perimentally determined (γLangevin = γℓ /ǫD ) from additional experimental runs that measure the previously calibrated inelastic loss rate in the electronically excited D3/2 state (see Section 5.2). The results for the two different bath configurations |F = 1, mF = 1ia 1.0 a b |1,1>a |2,2>a 0.8 pBright 0.6 0.4 0.2 0.0 0 2 4 6 8 10 0 2 4 6 8 10 t/tL t/tL Figure 6.13: Spin-relaxation in the 174 Yb+ Zeeman qubit. The probability pBright of the ion to occupy the bright | ↑i state after preparation in | ↑i (full symbols) or | ↓i (open symbols) in a bath of a |1, 1ia or b |2, 2ia atoms. Error bars denote one standard deviation uncertainty intervals resulting from approximately 3000 measurements per spin state. and |2, 2ia are shown in Figure 6.13a and b, respectively. The spin relaxation behaviour towards the steady state is well described by solutions of the general two-level rate equation model dp↑ = −p↑ γ↑ + p↓ γ↓ = −p↑ γ↑ + (1 − p↑ )γ↓ ⇒ p↑ (t) = p↑,0 e−t/T1 + p↑,∞ (1 − e−t/T1 ) (6.15) with the longitudinal coherence time T1 = 1/(γ↑ + γ↓ ) and the equilibrium spin state p↑,∞ = γ↑ /(γ↑ + γ↓ ). The decay rates γ↑ = γ↑,SE + γ↑,SR and γ↓ = γ↓,SE + γ↓,SR can 117 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms be decomposed into a spin-exchange part depending on the atomic bath spin and a spinrelaxation part affected by spin-orbit coupling. For collision with the maximally polarized bath state |2, 2ia at energies much larger than the Zeeman splittings the relation γ↓,SR = γ↑,SR should hold true, and since the atoms are in a spin stretched state, the spin-exchange rate form the upper state is γ↑,SE = 0. From our fitted steady-state population data p↑,∞ = 0.609±0.015 and T1 = (2.50±0.39)tL for atoms in the |2, 2ia state, we can determine the spin-exchange and spin-relaxation rates to be γ↓,SE = (2p↑,∞ −1)/T1 = (0.22±0.03)/T1 and γ↑,SR = γ↓,SR = (1 − p↑,∞ )/T1 = (0.39 ± 0.02)/T1 . An identical result within errors follows from the measurement on the |2, −2ia state. The steady-state spin values for the |1, −1ia and |1, 1ia bath states in Figure 6.12 can be qualitatively understood by looking at the allowed spin-exchange processes in the two different spin environments. For this purpose, we consider the collision of an ion in both the | ↑i and the | ↓i with atoms in the |1, 1ia bath state and expand the atom’s hyperfine state in the electronic spin basis (Equation 2.23): √ 3 3 3 1 | ↑i ⊗ | ↓ia | , ia + | ↑i ⊗ | ↑ia |3/2, 1/2ia 2 2 2 2 √ 3 3 1 3 3 1 | ↓i ⊗ | ↑ia | , ia + | ↑i ⊗ | ↑ia | , ia ⇒− 2 |{z} | 2 2 } 2 2 2 {z | ↑i ⊗ |1, 1ia = − SE √ ✟ |2,2i ✟ 3 3 3 1 | ↓i ⊗ | ↓ia | , ia + | ↓i ⊗ | ↑ia |3/2, 1/2ia 2 2 2 2 √ 3 3 3 1 3 1 | ↓i ⊗ | ↓ia | , ia + | ↑i ⊗ | ↓ia | , ia ⇒− 2 2 2 2 |{z} | 2 2 } {z | ↓i ⊗ |1, 1ia = − SE (6.16) √ 3 ✟2 ✟ ✟ |2,1i + 12 |1,1i Spin-exchange collisions leading to a transition of the atomic hyperfine state from the |F = 1ia to the |F = 2ia manifold are energetically strongly suppressed. We therefore expect γ↑,SE < γ↓,SE for |1, 1ia and analogous γ↑,SE > γ↓,SE for |1, −1ia , which is in accordance with the observed ordering of the steady state spin values in Figure 6.4. All measurements so far have been performed at an energy splitting of the ion Zeeman qubit of 37.5 MHz (at a magnetic field of 13.8 G). In order to investigate if the spin behaviour varies with different Zeeman splittings, we measured the steady-state spin distribution of the ion in the |1, 1ia environment for Zeeman splittings in the range from 0.39 MHz to 139 MHz, as shown in Figure 6.14. Following the interaction with the atoms the Zeeman qubit splitting is ramped from the interaction value to a state detection value just above or below the ion trap drive frequency for the values 139 MHz and 0.39 MHz, respectively. This is necessary, because the qubit transition is driven by the ion trap radio frequency 118 6.2. Experiments |1, 1>a 0.60 p↑, 8 0.55 0.50 0 50 100 150 Zeeman Splitting [MHz] Figure 6.14: Steady-state spin population for different qubit Zeeman splittings. The measurement is taken in the neutral |1, 1ia environment. We do not observe any dependence on Zeeman energy splittings within the statistical uncertainties (1σ) of our measurements. field if it becomes resonant with the Zeeman splitting. We do not observe any dependence of the steady-state spin population on the qubit energy splitting within the statistical uncertainties (1σ) of our measurements. The observed spin dynamics of the Yb+ ion in a spin polarized cloud of 87 Rb atoms is inconsistent with the picture of dominant spin-exchange and negligible spin-relaxation, that has previously been reported in He+ + Cs collisions [161], and with alkali-metal atomatom collisions, where for various regimes the hierarchy γc ≈ γSE ≫ γSR is fulfilled (see Section 2.3.2). In both cases the dominant spin-exchange interaction leads to a spin steady state of near perfect polarization, if the environment is polarized. The measured spin-exchange rate γSE = (0.09 ± 0.02) γLangevin of the ion is factor of 2.5 smaller than the theoretically estimate γSE = 1 4 γLangevin in the degenerate state approximation of Equation 2.25. However, it is the unexpectedly large spin-relaxation rate that dominates the spin dynamics in the hybrid system and that is responsible for almost complete mixture of ion spin states. The main reason for this strong spin-relaxation could be a strong spin-orbit interaction, that is caused by the level crossings of both the incoming singlet 1 Σ+ and triplet 3 Σ+ channels with the strongly spin-orbit coupled 3 Π channel at short internuclear separation (see Figure 2.1) [75, 76]. Level crossings of the excited chargeexchange 3 Σ+ state with the incoming collision channel are not unique to the combination Yb+ +Rb, but also occur, for example, in Ba+ +Rb [50] and Sr+ +Rb [49]. This could indicate similarly strong spin-relaxation behaviour in several other hybrid-systems, that are currently under investigation. 6.2.2 Hyperfine Spin-Relaxation in 171 Yb+ Next, we study the spin dynamics in the hyperfine spin qubit of 171 Yb+ . The measurements are performed with |1, −1ia or |2, 2ia atoms confined in the magnetic trap, which offers the 119 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms possibility of a large atomic reservoir with a low central density. After the immersion of the ion into the centre of the cloud, the measurements sequence is started by a 250 µs long 369 nm Freq. Pump. pulse (and a microwave π-pulse) to prepare the ion in the |0, 0i (|1, 0i) qubit state. Following 25 ms of ion-atom interaction time the state of the qubit is detected during a 1 ms fluorescence interval after which the preparation for the next interaction interval can start again. In this way we achieve up to 100 measurements during a single experimental cycle of the experiment which greatly improves the measurement statistics compared to the 174 Yb+ Zeeman qubit. We vary the effective interaction time t/tL by increasing and decreasing the number of atoms in the atomic cloud rather than changing the actual time of the interaction interval. The atomic density at the location of the ion in the centre of the neutral bath is determined from the atom number and temperature measured in absorption imaging and the harmonic trap frequencies. Figure 6.15a shows 1.0 a pBright 0.8 |1,-1>a 0.6 0.4 0.2 0.0 1.0 b pBright 0.8 |2,2>a 0.6 0.4 0.2 0.0 0 2 4 t/tL 6 8 1.0 pDark 0.35 c |1,-1> 0.30 0.35 d 0.8 |1,0> 0.30 0.25 0.6 0.25 0.20 0.4 0.20 0.2 0.15 0.15 0.10 0 2 4 6 8 10 12 0.0 10 0 2 4 6 t/tL t/tL 8 10 12 0.10 e 0 |1,1> 2 4 6 8 10 12 t/tL Figure 6.15: Hyperfine spin relaxation in 171 Yb+ . a The probability pBright after preparation in the |1, 0i (full symbols) or the |0, 0i (open symbols) state vs. the interaction time for atoms in the |1, −1ia state. Error bars denote one standard deviation uncertainty intervals resulting from a total of 5000 measurements. The fit values at t = 0 are limited by detection errors. b Similar data for collisions with 87 Rb atoms in the |2, 2ia state and a total of 19000 measurements. c-e Zeeman-resolved detection within the F = 1 manifold after preparation in |1, 0i (atoms in |2, 2ia ) with 1200 measurements per state. The measurements are performed by applying resonant π pulses, exchanging the population of the dark state |0, 0i with |1, −1i, |1, 0i or |1, 1i immediately before the detection of the probability of the ion being in the dark state pDark . 120 6.2. Experiments the results for ions in a neutral |1, −1ia bath. We observe an exponential decay of the population in the upper hyperfine manifold towards the |0, 0i ground state with a time constant T1 of a few Langevin collision times. This is in strong contrast to the case of the neutral |2, 2ia bath (Figure 6.15b), where interactions result in a steady-state occupation P in the excited |F = 1i manifold of p1 = m p|1,mi = 0.16(1). Here, we have already corrected the probability pBright for the detection errors of (1 − ηDark,|0,0i ) = 0.02 ± 0.01 for the hyperfine ground state |F = 0i and ηDark,|F =1i = 0.07 ± 0.03 for the |F = 1i hyperfine states (see Equation 6.10). 6.2.3 Spin Relaxation Heating We can assign to this non-zero steady-state occupation of the energetically excited states a spin-temperature Ts by considering p1 p0 HF S = 3 exp − EkB Ts , where E HF S = h × 12.6 GHz is the hyperfine splitting of the qubit and the factor 3 considers the multiplicity of the F = 1 manifold. The steady-state of the ion for the interaction with atoms in the |2, 2ia state corresponds to a spin temperature of Ts ≈ 200 mK. Since spin selection rules do not play an important role in the spin dynamics of the ion due to the dominance of spin relaxation, this spin temperature will correspond to a similar kinetic energy of the ion. The micromotion heating effect discussed in Section 3.2 can be neglected at the high temperatures under consideration. Instead the kinetic energy Ekin of the ion results from the balance of elastic and inelastic collision processes between the ion and the atoms. The average kinetic energy intake of the ion per Langevin collision due to a hyperfine flip in the neutral atom is δEheat = ǫEaHF S ma ma + mi (6.17) Here, we have considered the probability 0 ≤ ǫ ≤ 1, that an atomic hyperfine flip from the |2, 2ia to the F=1 manifold occurs during a Langevin collision, which releases internal hyperfine energy of the atom EaHF S = h × 6.8 GHz in the centre-of-mass frame of the two particles. The ion receives the fraction ma ma +mi as kinetic energy, where ma is the mass of the atom and mi the mass of the ion. The average ion energy loss by elastic ion-atom momentum transfer on the other hand is δEcool = − 2ma mi Ekin (ma + mi )2 (6.18) per Langevin collision, which follows directly from Equation 3.31 by neglecting the micromotion v = 0 and integrating over the isotropic scattering angle. As a result, the 121 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms steady-state average kinetic energy of the ion is given hEkin i = ǫEaHF S ma + mi = ǫkB × 240 mK. 2mi (6.19) For ǫ near unity, which corresponds to a neutral atom spin relaxation event in every Langevin collision, we hence obtain hEkin i ≈ kB Ts , which signals an equidistribution of energy between kinetic and spin degrees of freedom. The spin dynamics of the neutral cloud is not directly observable in the hybrid system, because regardless of the spin change of the atom during a Langevin collision, the atom is always lost as the amount of elastically transfered kinetic energy exceeds the trap depth. If the atomic bath is prepared in its hyperfine ground state |F = 1ia , no energy-releasing hyperfine changes can occur. The steady-state energies expected from the release of Zeeman energy in the |1, −1ia bath is smaller than the ion energies due to micromotion heat- ing [161, 37] for our experimental parameters. However, in order to reach the s-wave scattering regime for ion-atom combinations with a similarly large spin relaxation rate, the atoms will need to be prepared in the absolute spin ground state. To independently confirm the high kinetic energies of the ion, we perform a temperature measurement using the 174 Yb+ ion. We have previously observed (Figure 6.12) that the detection efficiency of the Zeeman qubit is different for the |1, −1ia and |1, 1ia compared to the |2, −2ia and |2, 2ia and have attributed this to the temperature sensitivity of the 411 nm electron shelving pulse. In order to quantify the effect we perform spectroscopy on the shelving transition for three different scenarios. The measurement sequence shown in Figure 6.16 is identically repeated for ions immersed in a neutral bath of |1, 1ia atoms, a bath of |2, 2ia atoms and no atoms at all. Whereas the transition line shape is dominated by the laser linewidth for Doppler-cooled ions and ions immersed in the |1, 1ia bath, we observe significant Doppler broadening for atoms after interactions with |2, 2ia atoms. Assuming a thermal energy distribution of the ion we estimate an ion temperature of T ≈ 120 mK from q the fit of σ(T ) = kB T ν411 m c to the line shape. This measurement independently confirms the high temperature the ion acquires during interactions with a |2, 2ia as a result of atomic hyperfine energy release during inelastic collisions. Spin dynamics within the Hyperfine manifold Returning to the spin dynamics of the 171 Yb+ 171 Yb+ ion, we can additionally study spin transfer within the excited |F = 1i manifold. We prepare the ion in the |1, 0i qubit state and detecting the individual excited states by exchanging the population of the dark state |0, 0i with |1, −1i, |1, 0i or |1, −1i with a resonant microwave π-pulse as discussed in Section 6.1.2. The data is shown in Figure 6.15c-e and compared to the four-level rate equation model 122 6.2. Experiments 369nm Pol. Pum p. 369nm Det ect ion 411nm 638nm 0[ µm] At om Posit ion -45[ µm] a 0.0 0.2 0.4 0.6 1.0 0.8 Tim e[ s] 0.4 b pDark 0.3 0.2 0.1 0.0 -30 -25 -20 15 -10 Δ -5 0 5 10 15 20 25 30 ν411[MHz] Figure 6.16: Ion energy estimate from the spectroscopy of the 411 nm electric quadrupole transition. a The experimental sequence is similar to the one we use for state detection in Figure 6.12a, but with a single 400 µs short pulse of 411 nm light instead of the extended optical pumping phase. After an atom-ion interaction time of 40 ms during which the ion reaches its steady state energy distribution for the respective environment, the quadrupole transition is probe by the 411 nm light. b The transfer efficiency as a function of the detuning of the 411 nm light from the resonance of the S1/2 (m = −1/2) − D5/2 (m = −5/2) transition. The transition is significantly Doppler-broadened for ions after the |2, 2ia bath (red markers and line) as compared to the |1, 1ia bath (black) or ions at Dopplertemperature (blue). described in Appendix A.2. The spin dynamics within the hyperfine system of 171 Yb+ is consistent with the strong spin non-conserving mechanism observed for the Zeeman qubit in 174 Yb+ . Interaction with the neutral atoms rapidly transfers population from the |1, 0i state into both |1, ±1i states, where the occupation correspondingly rises. Eventually, the |F = 1i states decay to their steady-state populations with roughly the same occupation in all three excited states. 6.2.4 Spin Coherence of the 171 Yb+ Qubit With the investigation of clock qubit we answer the question if the transverse coherence time is limited by the longitudinal coherence time or if it is additionally affected by forward scattering events. We have shown earlier that spin-exchange and spin-relaxing inelastic collisions happen at the timescale of the Langevin rate. The full quantum mechanical cross 123 Frequency shift [Hz] p(|0,0>) Visibility Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms t/tL Frequency offset [Hz] Figure 6.17: Spin coherence of the hyperfine clock transition |0, 0i ↔ |1, 0i in 171 Yb+ . a-d Ramsey fringes recorded for increasing atomic density. e Ramsey contrast as a function of t/tL displays an exponential decay with a time constant of T2 = (1.4 ± 0.2)tL . f No frequency shift of the clock transition is observed within the experimental errors. Error bars denote one standard deviation uncertainty intervals. section of Equation 2.17, which also includes forward scattering events, however, is a factor of 10 larger than the Langevin rate for a collision energy of E = kB · 100 mK. A good understanding of the influence of atomic collisions on hyperfine clock transitions is also of practical interest for buffer gas cooled ion microwave clocks [159], where clock shifts have previously been studied for inert buffer gases at room temperature [162]. We characterize the effect of atom-ion interaction on the transition by measuring its coherence time and line shift using the Ramsey technique described. The time between the two π/2- pulses is set to 27 ms and the interaction time in units of the Langevin timescale is again varied by changing the density of the magnetically trapped cloud of neutral atoms in the |2, 2ia state. The subplots a-d in Figure 6.17 show sample traces of Ramsey fringes with increasing atom-ion interaction strength. Even without neutral atoms, the visibility of the Ramsey fringes is limited to (55 ± 2)% owing to magnetic field fluctuations of the bias field. We find that the spin coherence decays within T2 = (1.4 ± 0.2)tL (see Figure 6.17e). This is on the time scale of the population relaxation of the |1, 0i state, which is driven by the spin-exchange and spin-orbit interaction. The measurement hence identifies spin-relaxation as the leading mechanism of spin decoherence in our system with a non-detectable contribution from elastic (forward) scattering. Finally, the phase of the Ramsey fringe pattern as a function of increasing atomic density is used to detect interaction induced frequency shifts in the clock transition. The 124 6.2. Experiments frequency shift ∆ν of the clock transition is below our measurement resolution of ∆ν ≤ 4 × 10−11 EiHF S /h (see Figure 6.17e). It is worth noting that the feasibility to detect fre- quency shifts is largely determined by the intrinsic properties of ion-atom collisions itself. On the one hand, the expected frequency shift ∆ν ∼ na from binary collisions between the ion and atoms is directly proportional to the atomic density. On the other hand the frequency resolution δν of the Ramsey measurement is proportional to the inverse of spinrelaxation limited interrogation time, and therefore also proportional to the atomic density. For spin-1/2 alcali metal buffer gases in hybrid setups, the decoherence effect due to spinrelaxation is strong compared to the collision induced frequency shift, in contrast to the case of inert spin-0 buffer gases. Table 6.1: Summary of the decoherence times (T1 and T2 ) and steady-state spin distributions of all spin-bath combinations. In addition to the statistical errors quoted, all times have systematic uncertainties of 40% due to the determination of the absolute atomic densities at the location of the ion. All values are compensated for detection efficiencies. Ion isotope Atom spin state T1 /tL 171 Yb+ |1, −1ia |2, 2ia |2, 2ia |2, −2ia |1, 1ia |1, −1ia 1.73 ± 0.17 3.39 ± 0.16 2.50 ± 0.39 171 Yb+ 174 Yb+ 174 Yb+ 174 Yb+ 174 Yb+ p↑,∞ 1.60 ± 0.24 125 0.609 ± 0.015 0.423 ± 0.026 0.563 ± 0.017 0.457 ± 0.021 P p|1,mi,∞ 0.000 + 0.005 0.163 ± 0.013 T2 /tL 1.4 ± 0.2 Chapter 6. Spin Dynamics of a Single Ion inside a Polarized Bath of Neutral Atoms 126 Chapter 7 Outlook In this thesis I have presented our investigations into a recently realized [33] quantum hybrid system of single trapped Yb+ ions inside ultracold clouds of neutral Rb atoms. In order to explore its numerous innovative features we have developed and implemented various manipulation methods to exert detailed control on the hybrid system’s internal quantum states. By applying these techniques to the study of cold reactive collisions, we have demonstrated the capabilities of our apparatus for the investigation of chemical processes at the most fundamental level [47]. Our results on Yb+ +Rb highlight the strong internal state dependence of reactive collisions at low energies. We have used the control over the internal quantum states to tune chemical reactions at the single particle level. From a practical perspective, the measurements have helped us to characterize the role of electronic state changing collisions for our particular hybrid system combination of Yb+ + Rb. The understanding of chemical processes in our system has been essential to suppress these events during our measurements of the more subtle spin degree of freedom. Our investigations of how a single ion spin impurity interacts with its environment of neutral atom spins represent a significant step forward for the study of spin-bath physics in atomic systems and for the potential use of hybrid systems in quantum information processing [31]. We study spin dynamics and decoherence in well-understood setting of an all-atomic spin-bath system at low temperatures [97]. For this purpose, we implement techniques for the state preparation, controlled manipulation and spin detection of two different ion qubit systems, the tunable ground state Zeemann qubit in clock-transition qubit in 171 Yb+ . 174 Yb+ and the We measure longitudinal and transverse decoherences on the Langevin timescale and discover in addition to the expected spin-exchange interaction an intriguingly strong spin relaxation mechanism behind spin decoherence. Neutral atomic hyperfine spin relaxation is further identified as a significant source of motional heating for certain states of our hybrid system. The understanding and control of the internal degrees of freedom complements our earlier insights into the role of elastic collisions [35, 37] and provides a comprehensive picture of a atom-ion hybrid system in the semiclassical regime. An interesting open question that remains in this context is the formation and stability of molecules in collisions. The mea- 127 Chapter 7. Outlook surements suggested in Section 5.1.1 could confirm or exclude photo-induced dissociation as the cause for the absence of molecules in our setup. Future studies in the hybrid system have a full portfolio of experiments to choose from. They will include the characterization of the atom-ion interaction potential by photoassociation spectroscopy [135] and the search for Feshbach resonances [83]. We have also started to perform the first measurements in our setup that look at the interactions of ions with Rb atoms that are confined in three dimensional optical lattices. All of these studies will benefit greatly from technical improvements that are currently under way. The implementation of a new ion trap with increased ion trapping frequencies and reduced excess micromotion heating, will extend the accessible collision energy range to lower limits in the future. The Route Towards S-Wave Ion-Atom Scattering The logical next long-term step for quantum hybrid systems is the quest for the s-wave regime of atom-ion scattering, which is needed to tap into the full potential of coherent atom-ion interaction. Of paramount importance for such a future hybrid system will be the choice of the atom-ion combination. The ever increasing list of laser cooled ion species together with the considerable number of degenerate quantum gases species results in principle in an impressive plethora of possible combinations to choose from. The selection is limited by the fact that approaching the the s-wave regime by the sympathetic cooling of an ion in the ultracold neutral cloud demands a heavy ion and a light atom since the micromotion limited final ion energies need to be smaller than the s-wave energy threshold E ∗ ≫ Eem , Ef m . A large s-wave energy threshold (Equation 2.11) requires a small reduced mass, whereas the steady state energies due to excess micromotion Eem (Equation 3.36) and the fundamental limit Ef m discussed in [39] become small for large ion to atom mass ratios. Although a large ion-atom mass ratio also reduces the relative amount of energy transfered per collision and therefore the cooling efficiency for the ion, this is not expected to limit the attainable temperature during immersion cooling. Assuming collision rates1 on the order of 1000 s−1 the cooling rate should sufficiently dominate the typical heating rates of several quanta per second in large macroscopic ion traps [163]. In the current generation of atom-ion hybrid setups with degenerate quantum gases, Rb has been chosen as the neutral component for experimental convenience. The inexpensive laser system and the relatively small efforts needed to bring Rb to quantum degeneracy have been the main arguments in favor. Its comparatively large mass, however, prevents 1 The collision rate scales as E 1/2 in the threshold regime and its magnitude depends on the yet unknown s-wave scattering length. 128 Rb from being used as an ultracold buffer gas in experiments that seek to obtain the s-wave limit. The most attractive alternative are Li atoms due to their low mass, the choice of isotopes to create both quantum degenerate Fermi gases and small2 BECs and their large ionization energy. On the ion side, the heavy ions Yb+ and Ba+ are the prime contenders. Immersed into Li both ions should be cooled below their respective s-wave thresholds (Equation 3.36) of ≈ 6400 nK and ≈ 6500 nK for a well compensated ion trap. The combination of Ba+ +Li has the rare property that the entrance channel is the absolute electronic ground state and the charge exchange channel Li+ +Ba is energetically closed. As a consequence, electronic ground state Ba+ can never be lost from the ion trap due to two-body collisions with the buffer gas. At high atomic densities, three-body collisions could nevertheless amount to quite significant Ba+ loss and heating rates [77], respectively3 In the case of Yb+ + Rb the entrance channel is well isolated from both the charge exchange ground state and its first electronic excitation by asymptotic energy gaps of 7000 cm−1 and 10000 cm−1 and should therefore exhibit both small charge exchange and spin relaxation rates. The large number of available isotopes for Yb+ allows not only flexibility in the implementation of the spin qubits, but also provides a range of different scattering length options. Additional control of the scattering properties should be possible through the predicted presence of Feshbach resonances at low magnetic fields. In summary, a hybrid system of Li atoms and Yb+ or Ba+ ions with adequate micromotion compensation should make the investigation of s-wave collisions and the immersion cooling of ions into the ground state possible. 2 At small magnetic fields the atom number is limited to ∼ 1400 due to the attractive interaction of 7 Li atoms. 3 The use of a fermi degenerate quantum gas could potentially suppress these three-body effects. 129 Chapter 7. Outlook 130 Acknowledgements One of the great privileges of pursuing a PhD in experimental physics are the inspiring people one happens to work alongside every day. This could not be more true for my supervisor Michael Köhl. In an ever growing research group, he manages to give his 24/7 support to every single one of his students and postdocs. His enthusiasm and knowledge about physics and all things technical have been invaluable during the many discussions and hours of hands-on help in the lab. With his friendly and relaxed way (and his considerable table football skills) he has been with us during many after-work fun activities. Thank you! The experiments in this thesis build on the excellent work of former PhD students Stefan Palzer and Christoph Zipkes, who designed and constructed the apparatus that supports the basic operation of the hybrid trap. Stefan was responsible for the laser system and the BEC side of the setup, while Christoph took charge of the ion trap. Together with Michael and the postdoc Carlo Sias, who joined in 2008, they pursued the project with the great foresight, care and efficiency that was needed to make the idea of placing a single trapped ion inside a ultracold atomic bath a success. Christoph and Carlo continued to stay on the experiment with me until autumn 2012 and I fondly remember the passionate discussions the three of us have had about new measurements, exciting results and the physics behind it all. To Christoph I am particularly grateful for passing on some of his extensive electronics and programming knowledge. His skills remain evident in many of the devices that are now essential parts of our experimental setup. To Carlo, my long time lab colleague, I not only owe my vocabulary of Italian/Sardinian curse words, but also a lot of the fun and many of the insights that developed during the long lab hours. His undefeatable optimism and his practical approach helped our experiment through the tough times and made the good times even better. With Jonathan Silver and Leonardo Carcagni’ the Ion-Bec experiment got a new generation of talented PhD students that will carry the experiments with ions and atoms to Bonn and beyond. I am very grateful to the current and former members of the neighboring experiments, Bernd Fröhlich, Michael Feld, Enrico Vogt, Hendrik Meyer, Matthias Steiner, Marco Koschorreck, Luke Miller, Daniel Pertot, Daniel Sigle, Eugenio Cocchi, Johanna Bohn, Tim Ballance and Alexandra Behrle for the atmosphere of companionship in our group, where advice and helpful discussions, the loan of an essential piece of equipment and some much needed after-work entertainment are never far away. 131 Thanks are due also to the staff of the Cavendish laboratory, in particular to our secretary Pam Hadder for her kind assistance with administrative tasks, to Peter Norman for his help with the realization of mechanical designs as head of the workshop and to Nigel Palfrey for his supervisions during the work in the students mechanical workshop. My deepest gratitude goes to my family and friends for their encouragement, trust, love and friendship throughout the years. I acknowledge financial support by the Cambridge European Trust and the Postgraduate Scholarship of the Austrian Bundesministerium für Wissenschaft und Forschung. Appendix A A.1 Spins of 174 Yb+ , 171 Yb+ , 87 Rb in a Magnetic Field The following discussion applies to all atomic systems with a single valence electron. The total Hamiltonian of an atom in a magnetic field B H = H0 + µ · B (A.1) contains the atomic Hamiltonian H0 and a magnetic term that describes the interaction of the atoms magnetic moment µ with the magnetic field B. The magnetic moment of an atom µ has contributions from the orbital angular momentum L and the spin angular momentum S of the electron, as well as, the spin angular momentum of the nucleus I µ = µB (gL L + gS S + gI I)/~, (A.2) with gi being the Landè factor of the respective angular momentum. For experimentally relevant magnetic fields the spin orbit coupling dominates the effect due to external fields. Therefore the total electric angular momentum J = L + S is well conserved and the Zeeman interaction can be treated as a perturbation. The joint Landè factor for the electron angular momentum is thus gJ = gL J(J + 1) − S(S + 1) + L(L + 1) J(J + 1) + S(S + 1) − L(L + 1) + gS 2J(J + 1) 2J(J + 1) (A.3) with gL = 1 and gS ≈ 2.0023192. The Zeeman energy splitting is given by E = µB gJ mJ B, mJ = mL + mS . (A.4) and applies to the spin manifolds of all electronic states of the nuclear spin-0 isotopes of Yb+ (see Table A.1) In atoms with nuclear spin the hyperfine interaction, which results from the coupling of electron angular momentum J to the total nuclear angular momentum I, leads to a further energy splitting of states. The total spin Hamiltonian is given by H = hA I · J − µB (gJ J + gI I) · B, 133 (A.5) Appendix A. where gI = −9.95 · 10−4 and A is the magnetic dipole constant. Here, we have ignored hyperfine orders beyond the magnetic dipole term. In first approximation a similar argument as for the fine structure coupling can be made for the much smaller hyperfine structure. If we consider the Zeeman interaction a perturbation to the total atomic angular momentum F = J + I eigenstates, then the energy splitting is E = µB gF mF B (A.6) with gF = gJ F (F + 1) + J(J + 1) − I(I + 1) F (F + 1) − J(J + 1) + I(I + 1) + gI . 2F (F + 1) 2F (F + 1) (A.7) For the case of J = 1/2 an analytic solution of the Hamiltonian of Equation A.5 exists for all magnetic field strength, that is known as the Breit-Rabi formula EF =I±1/2,mF hE HF S hE HF S + µB gI mF B ± =− 2(2I + 1) 2 x≡ For mF = −(I + 1/2) the expression interpreted as (1 − x). µB B(gJ − gI ) hE HF S q 1+ 2mF x I+1/2 s 2mF x + x2 I + 1/2 1 =A I+ . 2 1+ E HF S (A.8) (A.9) + x2 is an exact square and needs to be The Expansion to lowest non-vanishing order of the Breit-Rabi formula around zero magnetic field yields a quadratic magnetic field dependence for the |F = 0, mF = 0i − |F = 1, mF = 0i hyperfine “clock” transition of ~ωclock (B) − ~ωclock (0) = EF =I+1/2,mF =0 − EF =I−1/2,mF =0 − E HF S = 134 (gJ − gI )2 µ2B 2 B 2E HF S (A.10) A.2. Rate Equation Model for Spin Dynamics in Table A.1: Summary of spin properties of relevant electronic states in 171 Yb+ 171 Yb+ , 174 Yb+ , 87 Rb. Atomic Species Nuclear Spin EHF S /h 171 Yb+ 12.6428121185 GHz [164] 174 Yb+ 1/2 0 87 Rb 3/2 6.8346826109 GHz [165] A.2 Electronic (& Hyperfine) State 2S 1/2 (F = 2S 1/2 2P 1/2 2D 3/2 2D 5/2 2F 5/2 2S 1/2 (F = 2S 1/2 (F = Low Field Landè factor 1) 1 2 2/3 4/5 6/5 8/7 -1/2 1/2 1) 2) Rate Equation Model for Spin Dynamics in 171 Yb+ In order to model the data in Figure 6.15 we start with the general case of a four level rate equation model ṗ00 = −p00 · (γ00,10 + γ00,1−1 + γ00,11 ) + p10 · (γ10,00 ) + p11 · (γ11,00 ) + p1−1 · (γ1−1,00 ), ṗ1−1 = −p1−1 · (γ1−1,10 + γ1−1,00 + γ1−1,11 ) + p10 · (γ10,1−1 ) + p11 · (γ11,1−1 ) + p00 · (γ00,1−1 ) ṗ10 = −p10 · (γ10,1−1 + γ10,00 + γ10,11 ) + p00 · (γ00,10 ) + p11 · (γ11,10 ) + p1−1 · (γ1−1,10 ) ṗ11 = −p11 (γ11,10 + γ11,00 + γ11,1−1 ) + p10 · (γ10,11 ) + p1−1 · (γ1−1,11 ) + p00 · (γ00,11 ), (A.11) where γi,j is the rate for a transition from state i to state j and the initial conditions in our experiment are p10 (0) = 1 , p00 (0) = p1−1 (0) = p11 (0) = 0. Given data quality and the size of the parameter space, an unconstraint fit will not provide much insight, we therefore choose to make some assumptions based on our earlier results on spin-dynamics to minimize the size of the parameter space. We take into account that the hyperfine rates are dominated by energy considerations and that spin selection rules play a minor role γ1−1,00 = γ1−1,00 = γ11,00 ≡ γH γ00,1−1 = γ00,10 = γ00,11 ≡ γH e 135 (A.12) HF S − Ek T B , (A.13) Appendix A. where we use T = 200 mK. For the transitions with the F = 1 manifold we consider the result γ↑,SE ≈ 0.5 γSR that we have obtained for 174 Yb+ interacting with |2, 2ia γ10,1−1 = γ11,10 ≡ γSR (A.14) γ1−1,10 = γ10,11 ≡ (1 + 0.5) γSR e Zeeman −E k T B ≈ 1.5 γSR γ1−1,11 = γ11,1−1 ≈ 0, (A.15) (A.16) where we assume that spin relaxation events that change the ion spin by ∆m = ±2 are strongly suppressed. The hyperfine relaxation rate for the neutral bath |2, 2i has been obtained in Section 6.2.2; γH = 1/T1 = (0.29 ± 0.02) γLangevin and we make the assumption that the spin relaxation rate for 174 Yb+ also holds true for the 171 Yb+ isotope γSR = γSR = (0.1 ± 0.02) γLangevin . As a last step, we take into account the state detection efficiencies and π-flip efficiencies on the magnetically sensitive hyperfine transitions ηDark,|0,0i ≈ 0.98, ηDark,|1i ≈ 0.07 and ηπ,1−1 ≈ ηπ,11 ≈ 0.95. The final model without free parameters describes the evolution of the spin populations in Figure 6.15 well. A.3 Axial Trap Frequencies of Small Ion Crystals Number of Ions 2 3 4 Impurity Position Identical Ions 171 Yb+ Edge Centre Edge Centre 1.000 1.000 1.000 1.000 1.000 1.00432 1.00287 1.00288 1.00215 1.00216 87 Rb+ Impurity (174 Yb87 Rb)+ (174 Yb87 Rb87 Rb)+ 1.12603 1.07914 1.08694 1.05821 1.06373 0.885861 0.917615 0.921832 0.935862 0.939734 0.796225 0.843392 0.854744 0.87329 0.884995 Table A.2: Centre-of-mass (COM) mode frequencies of small linear 174 Yb+ ion crystals with relevant impurity ions. The frequencies are given as multiples of the COM frequency of the pure ion crystal. 136 Bibliography [1] Aristotle, Metaphysics. Book H 1045a 8-10. [2] M. H. Anderson, J. R. Ensher, M. R. Matthews, C. E. Wieman, and E. A. Cornell, “Observation of bose-einstein condensation in a dilute atomic vapor,” Science, 269, 198–201, 1995. [3] K. B. Davis, M. O. Mewes, M. R. Andrews, N. J. van Druten, D. S. Durfee, D. M. Kurn, and W. Ketterle, “Bose-einstein condensation in a gas of sodium atoms,” Phys. Rev. Lett., 75, 3969–3973, 1995. [4] B. DeMarco and D. Jin, “Onset of fermi degeneracy in a trapped atomic gas,” Science, 285, 1703, 1999. [5] R. Onofrio, C. Raman, J. M. Vogels, J. R. Abo-Shaeer, A. P. Chikkatur, and W. Ketterle, “Observation of superfluid flow in a bose-einstein condensed gas,” Phys. Rev. Lett., 85, 2228–2231, 2000. [6] M. R. Matthews, B. P. Anderson, P. C. Haljan, D. S. Hall, C. E. Wieman, and E. A. Cornell, “Vortices in a bose-einstein condensate,” Phys. Rev. Lett., 83, 2498, 1999. [7] M. R. Andrews, C. G. Townsend, H. J. Miesner, D. S. Durfee, D. M. Kurn, and W. Ketterle, “Observation of interference between two bose condensates,” Science, 275, 637–641, 1997. [8] M. Greiner, O. Mandel, T. Esslinger, T. Hänsch, and I. Bloch, “Quantum phase transition from a superfluid to a mott insulator in a gas of ultracold atoms,” Nature, 415, 39, 2002. [9] C. A. Regal, M. Greiner, and D. S. Jin, “Observation of resonance condensation of fermionic atom pairs,” Phys. Rev. Lett., 92, 040403, 2004. [10] J. F. Sherson, C. Weitenberg, M. Endres, M. Cheneau, I. Bloch, and S. Kuhr, “Singleatom-resolved fluorescence imaging of an atomic mott insulator,” Nature, 467, 68–72, 2010. [11] O. Mandel, M. Greiner, A. Widera, T. Rom, T. W. Hansch, and I. Bloch, “Controlled collisions for multi-particle entanglement of optically trapped atoms,” Nature, 425, 937–940, 2003. 137 Bibliography [12] W. S. Bakr, J. I. Gillen, A. Peng, S. Folling, and M. Greiner, “A quantum gas microscope for detecting single atoms in a hubbard-regime optical lattice,” Nature, 462, 74–77, 2009. [13] M. de Angelis, A. Bertoldi, L. Cacciapuoti, A. Giorgini, G. Lamporesi, M. Prevedelli, G. Saccorotti, F. Sorrentino, and G. M. Tino, “Precision gravimetry with atomic sensors,” Measurement Science and Technology, 20, 022001+, 2009. [14] P. Böhi and P. Treutlein, “Simple microwave field imaging technique using hot atomic vapor cells,” Applied Physics Letters, 101, 181107+, 2012. [15] G. Wilpers, T. Binnewies, C. Degenhardt, U. Sterr, J. Helmcke, and F. Riehle, “Optical clock with ultracold neutral atoms,” Phys. Rev. Lett., 89, 230801, 2002. [16] M. Takamoto, F.-L. Hong, R. Higashi, and H. Katori, “An optical lattice clock,” Nature, 435, 321–324, 2005. [17] P. O. Schmidt, T. Rosenband, C. Langer, W. M. Itano, J. C. Bergquist, and D. J. Wineland, “Spectroscopy using quantum logic,” Science, 309, 749–752, 2005. [18] C. W. Chou, D. B. Hume, J. C. J. Koelemeij, D. J. Wineland, and T. Rosenband, “Frequency comparison of two high-accuracy Al+ optical clocks,” Phys. Rev. Lett., 104, 070802, 2010. [19] J. I. Cirac and P. Zoller, “Quantum computations with cold trapped ions,” Phys. Rev. Lett., 74, 4091–4094, 1995. [20] F. Schmidt-Kaler, H. Haffner, M. Riebe, S. Gulde, G. P. Lancaster, T. Deuschle, C. Becher, C. F. Roos, J. Eschner, and R. Blatt, “Realization of the cirac-zoller controlled-not quantum gate,” Nature, 422, 408–411, 2003. [21] J. Chiaverini, D. Leibfried, T. Schaetz, M. D. Barrett, R. B. Blakestad, J. Britton, W. M. Itano, J. D. Jost, E. Knill, C. Langer, R. Ozeri, and D. J. Wineland, “Realization of quantum error correction,” Nature, 432, 602–605, 2004. [22] D. P. DiVincenzo, “The physical implementation of quantum computation,” Fortschr. Phys., 48, 771–783, 2000. [23] D. Kielpinski, C. Monroe, and D. J. Wineland, “Architecture for a large-scale ion-trap quantum computer,” Nature, 417, 709–711, 2002. [24] M. Krych, W. Skomorowski, F. Pawlowski, R. Moszynski, and Z. Idziaszek, “Sympathetic cooling of the Ba+ ion by collisions with ultracold Rb atoms: Theoretical prospects,” Phys. Rev. A, 83, 032723, 2011. 138 Bibliography [25] R. Côté, V. Kharchenko, and M. D. Lukin, “Mesoscopic molecular ions in boseeinstein condensates,” Phys. Rev. Lett., 89, 093001, 2002. [26] P. Massignan, C. J. Pethick, and H. Smith, “Static properties of positive ions in atomic bose-einstein condensates,” Phys. Rev. A, 71, 023606, 2005. [27] J. Goold, H. Doerk, Z. Idziaszek, T. Calarco, and T. Busch, “Ion-induced density bubble in a strongly correlated one-dimensional gas,” Phys. Rev. A, 81, 041601, 2010. [28] C. Kollath, M. Köhl, and T. Giamarchi, “Scanning tunneling microscopy for ultracold atoms,” Phys. Rev. A, 76, 063602, 2007. [29] Y. Sherkunov, B. Muzykantskii, N. d’Ambrumenil, and B. D. Simons, “Probing ultracold Fermi atoms with a single ion,” Phys. Rev. A, 79, 023604, 2009. [30] Z. Idziaszek, T. Calarco, and P. Zoller, “Controlled collisions of a single atom and an ion guided by movable trapping potentials,” Phys. Rev. A, 76, 033409, 2007. [31] H. Doerk, Z. Idziaszek, and T. Calarco, “Atom-ion quantum gate,” Phys. Rev. A, 81, 012708, 2010. [32] R. Gerritsma, A. Negretti, H. Doerk, Z. Idziaszek, T. Calarco, and F. Schmidt-Kaler, “Bosonic josephson junction controlled by a single trapped ion,” Phys. Rev. Lett., 109, 080402, 2012. [33] C. Zipkes, S. Palzer, C. Sias, and M. Kohl, “A trapped single ion inside a bose-einstein condensate,” Nature, 464, 388–391, 2010. [34] S. Schmid, A. Härter, and J. H. Denschlag, “Dynamics of a cold trapped ion in a bose-einstein condensate,” Phys. Rev. Lett., 105, 133202, 2010. [35] C. Zipkes, S. Palzer, L. Ratschbacher, C. Sias, and M. Köhl, “Cold heteronuclear atom-ion collisions,” Phys. Rev. Lett., 105, 133201, 2010. [36] H. Dehmelt, “Radiofrequency spectroscopy of stored ions I: Storage,” Adv. At. Mol. Phys., 3, 53, 1967. [37] C. Zipkes, L. Ratschbacher, C. Sias, and M. Köhl, “Kinetics of a single trapped ion in an ultracold buffer gas,” New Journal of Physics, 13, 053020+, 2011. [38] Z. Idziaszek, T. Calarco, P. S. Julienne, and A. Simoni, “Quantum theory of ultracold atom-ion collisions,” Phys. Rev. A, 79, 010702, 2009. 139 Bibliography [39] M. Cetina, A. T. Grier, and V. Vuletić, “Micromotion-induced limit to atom-ion sympathetic cooling in paul traps,” Phys. Rev. Lett., 109, 253201, 2012. [40] A. T. Grier, M. Cetina, F. Oručević, and V. Vuletić, “Observation of cold collisions between trapped ions and trapped atoms,” Phys. Rev. Lett., 102, 223201, 2009. [41] F. H. J. Hall, M. Aymar, N. Bouloufa-Maafa, O. Dulieu, and S. Willitsch, “Lightassisted ion-neutral reactive processes in the cold regime: Radiative molecule formation versus charge exchange,” Phys. Rev. Lett., 107, 243202, 2011. [42] W. G. Rellergert, S. T. Sullivan, S. Kotochigova, A. Petrov, K. Chen, S. J. Schowalter, and E. R. Hudson, “Measurement of a large chemical reaction rate between ultracold closed-shell 40 Ca atoms and open-shell 174 Yb+ ions held in a hybrid atom-ion trap,” Phys. Rev. Lett., 107, 243201, 2011. [43] K. Ravi, S. Lee, A. Sharma, G. Werth, and S. A. Rangwala, “Cooling and stabilization by collisions in a mixed ion-atom system,” Nature Communications, 3, 1126+, 2012. [44] R. Côté and A. Dalgarno, “Ultracold atom-ion collisions,” Phys. Rev. A, 62, 012709, 2000. [45] O. P. Makarov, R. Côté, H. Michels, and W. W. Smith, “Radiative charge-transfer lifetime of the excited state of (NaCa)+ ,” Phys. Rev. A, 67, 042705, 2003. [46] P. Zhang, A. Dalgarno, and R. Côté, “Scattering of Yb and Yb+ ,” Phys. Rev. A, 80, 030703, 2009. [47] L. Ratschbacher, C. Zipkes, C. Sias, and M. Kohl, “Controlling chemical reactions of a single particle,” Nature Physics, 8, 649–652, 2012. [48] B. Gao, “Universal properties in ultracold ion-atom interactions,” Phys. Rev. Lett., 104, 213201, 2010. [49] M. Aymar, R. Gurout, and O. Dulieu, “Structure of the alkali-metal-atom + strontium molecular ions: Towards photoassociation and formation of cold molecular ions,” J. Chem. Phys., 135, 064305, 2011. [50] F. H. J. Hall, M. Aymar, M. Raoult, O. Dulieu, and S. Willitsch, “Light-assisted cold chemical reactions of barium ions with rubidium atoms,” arXiv:1301.0724, 2013. [51] B. Roth, P. Blythe, H. Wenz, H. Daerr, and S. Schiller, “Ion-neutral chemical reactions between ultracold localized ions and neutral molecules with single-particle resolution,” Phys. Rev. A, 73, 042712, 2006. 140 Bibliography [52] P. F. Staanum, K. Højbjerre, R. Wester, and M. Drewsen, “Probing isotope effects in chemical reactions using single ions,” Phys. Rev. Lett., 100, 243003, 2008. [53] P. F. Staanum, K. Hojbjerre, P. S. Skyt, A. K. Hansen, and M. Drewsen, “Rotational laser cooling of vibrationally and translationally cold molecular ions,” Nature Physics, 6, 271–274, 2010. [54] T. Schneider, B. Roth, H. Duncker, I. Ernsting, and S. Schiller, “All-optical preparation of molecular ions in the rovibrational ground state,” Nature Physics, 6, 275–278, 2010. [55] S. Willitsch, M. T. Bell, A. D. Gingell, S. R. Procter, and T. P. Softley, “Cold reactive collisions between laser-cooled ions and velocity-selected neutral molecules,” Phys. Rev. Lett., 100, 043203, 2008. [56] E. R. Hudson, “Method for producing ultracold molecular ions,” Phys. Rev. A, 79, 032716, 2009. [57] S. T. Sullivan, W. G. Rellergert, S. Kotochigova, and E. R. Hudson, “Role of electronic excitations in ground-state-forbidden inelastic collisions between ultracold atoms and ions,” Phys. Rev. Lett., 109, 223002, 2012. [58] F. H. J. Hall and S. Willitsch, “Millikelvin reactive collisions between sympathetically cooled molecular ions and laser-cooled atoms in an ion-atom hybrid trap,” Phys. Rev. Lett., 109, 233202, 2012. [59] R. V. Krems, “Cold controlled chemistry,” Phys. Chem. Chem. Phys., 10, 4079, 2008. [60] A. J. Leggett, S. Chakravarty, A. T. Dorsey, M. P. A. Fisher, A. Garg, and W. Zwerger, “Dynamics of the dissipative two-state system,” Reviews of Modern Physics, 59, 1–85, 1987. [61] N. V. Prokof’ev and P. C. E. Stamp, “Theory of the spin bath,” Reports on Progress in Physics, 63, 669+, 2000. [62] W. H. Zurek, “Decoherence, einselection, and the quantum origins of the classical,” Rev. Mod. Phys., 75, 715–775, 2003. [63] A. V. Gorshkov, M. Hermele, V. Gurarie, C. Xu, P. S. Julienne, J. Ye, P. Zoller, E. Demler, M. D. Lukin, and A. M. Rey, “Two-orbital SU(n) magnetism with ultracold alkaline-earth atoms,” Nature Physics, 6, 289–295, 2010. 141 Bibliography [64] F. M. Cucchietti and E. Timmermans, “Strong-coupling polarons in dilute gas boseeinstein condensates,” Phys. Rev. Lett., 96, 210401, 2006. [65] N. Spethmann, F. Kindermann, S. John, C. Weber, D. Meschede, and A. Widera, “Dynamics of single neutral impurity atoms immersed in an ultracold gas,” Phys. Rev. Lett., 109, 235301, 2012. [66] R. Hanson, V. V. Dobrovitski, A. E. Feiguin, O. Gywat, and D. D. Awschalom, “Coherent dynamics of a single spin interacting with an adjustable spin bath,” Science, 320, 352–355, 2008. [67] T. Fukuhara, A. Kantian, M. Endres, M. Cheneau, P. Schausz, S. Hild, D. Bellem, U. Schollwock, T. Giamarchi, C. Gross, I. Bloch, and S. Kuhr, “Quantum dynamics of a mobile spin impurity,” Nature Physics, advance online publication, 2013. [68] S. Olmschenk, K. C. Younge, D. L. Moehring, D. N. Matsukevich, P. Maunz, and C. Monroe, “Manipulation and detection of a trapped Yb+ hyperfine qubit,” Phys. Rev. A, 76, 052314, 2007. [69] C. Zipkes, A Trapped Single Ion Inside a Bose-Einstein Condensate. PhD thesis, University of Cambridge, 2011. [70] S. Palzer, Single Impurities in a Bose-Einstein Condensate. PhD thesis, University of Cambridge, 2010. [71] “Indiana jones and the last crusade,” 1989. [72] “Monty python and the holy grail,” 1975. [73] H. C. Wolf and H. Haken, Molecular Physics and Elements of Quantum Chemistry. Springer-Verlag Telos. [74] D. A. Steck, “Rubidium 87 d-line data,” http://steck.us/alkalidata, 2010. [75] O. Dulieu, “private communication.”. [76] E. R. Sayfutyarova, A. A. Buchachenko, S. A. Yakovleva, and A. K. Belyaev, “Charge transfer in the cold Yb+ + Rb collisions,” arXiv:1303.3093, 2013. [77] A. Härter, A. Krükow, A. Brunner, W. Schnitzler, S. Schmid, and J. H. Denschlag, “Single ion as a three-body reaction center in an ultracold atomic gas,” Phys. Rev. Lett., 109, 123201, 2012. 142 Bibliography [78] J. J. Sakurai, Modern Quantum Mechanics (Revised Edition). Addison Wesley, 1 ed., 1993. [79] J. Weiner, V. S. Bagnato, S. Zilio, and P. S. Julienne, “Experiments and theory in cold and ultracold collisions,” Rev. Mod. Phys., 71, 1–85, 1999. [80] G. F. Gribakin and V. V. Flambaum, “Calculation of the scattering length in atomic collisions using the semiclassical approximation,” Phys. Rev. A, 48, 546–553, 1993. [81] A. Dalgarno, M. R. C. McDowell, and A. Williams, “The mobilities of ions in unlike gases,” Philosophical Transactions of the Royal Society of London. Series A, Mathematical and Physical Sciences, 250, 411–425, 1958. [82] Z. Idziaszek, A. Simoni, T. Calarco, and P. S. Julienne, “Multichannel quantumdefect theory for ultracold atomion collisions,” New Journal of Physics, 13, 083005+, 2011. [83] C. Chin, R. Grimm, P. Julienne, and E. Tiesinga, “Feshbach resonances in ultracold gases,” Rev. Mod. Phys., 82, 1225–1286, 2010. [84] P. Langevin, “A fundamental formula of kinetic theory.,” Ann. Chim. Phys., 5, 245288, 1905. [85] E. Vogt and G. H. Wannier, “Scattering of ions by polarization forces,” Phys. Rev., 95, 1190–1198, 1954. [86] H. Schmaljohann, M. Erhard, J. Kronjäger, K. Sengstock, and K. Bongs, “Dynamics and thermodynamics in spinor quantum gases,” Applied Physics B, 79, 1001–1007, 2004. [87] M. Anderlini, P. J. Lee, B. L. Brown, J. Sebby-Strabley, W. D. Phillips, and J. V. Porto, “Controlled exchange interaction between pairs of neutral atoms in an optical lattice,” Nature, 448, 452–456, 2007. [88] A. Dalgarno, “Spin-Change Cross-Sections,” Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences, 262, 132–135, 1961. [89] J. Dalibard, “Collisional dynamics of ultra-cold atomic gases,” Proceedings of the International School of Physics Enrico Fermi, Course CXL: Bose – Einstein condensation in gases, Varenna, 1998. [90] P. S. J. Frederick H. Mies, Carl J. Williams and M. Krauss, “Estimating bounds on collisional relaxation rates of spin-polarized 87 Rb atoms at ultracold temperatures,” J. Res. Natl. Inst. Stand. Technol., 101, 521, 1996. 143 Bibliography [91] Y. K. Kato, R. C. Myers, A. C. Gossard, and D. D. Awschalom, “Observation of the spin hall effect in semiconductors,” Science, 306, 1910–1913, 2004. [92] Y. J. Lin, K. Jimenez-Garcia, and I. B. Spielman, “Spin-orbit-coupled bose-einstein condensates,” Nature, 471, 83–86, 2011. [93] P. Wang, Z.-Q. Yu, Z. Fu, J. Miao, L. Huang, S. Chai, H. Zhai, and J. Zhang, “Spin-orbit coupled degenerate fermi gases,” Phys. Rev. Lett., 109, 095301, 2012. [94] L. W. Cheuk, A. T. Sommer, Z. Hadzibabic, T. Yefsah, W. S. Bakr, and M. W. Zwierlein, “Spin-injection spectroscopy of a spin-orbit coupled fermi gas,” Phys. Rev. Lett., 109, 095302, 2012. [95] J. Söding, D. G. Odelin, P. Desbiolles, G. Ferrari, and J. Dalibard, “Giant spin relaxation of an ultracold cesium gas,” Physical Review Letters, 80, 1869–1872, 1998. [96] S. Kadlecek, T. Walker, D. K. Walter, C. Erickson, and W. Happer, “Spin-axis relaxation in spin-exchange collisions of alkali-metal atoms,” Phys. Rev. A, 63, 052717, 2001. [97] L. Ratschbacher, C. Sias, L. Carcagni, J. Silver, C. Zipkes, and M. Köhl, “Decoherence of a single ion qubit immersed in a spin-polarized atomic bath,” Accepted in Phys. Rev. Lett., arXiv:1301.5452, 2013. [98] W. Paul, “Electromagnetic traps for charged and neutral particles,” Rev. Mod. Phys., 62, 531–540, 1990. [99] E. L. Raab, M. Prentiss, A. Cable, S. Chu, and D. E. Pritchard, “Trapping of neutral sodium atoms with radiation pressure,” Phys. Rev. Lett., 59, 2631–2634, 1987. [100] D. J. Wineland, C. Monroe, W. M. Itano, D. Leibfried, B. E. King, and D. M. Meekhof, “Experimental issues in coherent quantum-state manipulation of trapped atomic ions,” Journal of Research of the National Institute of Standards and Technology, 103, 259, 1998. [101] P. K. Ghosh, Ion Traps (International Series of Monographs on Physics, 90). Oxford University Press, USA, 1996. [102] D. Leibfried, R. Blatt, C. Monroe, and D. Wineland, “Quantum dynamics of single trapped ions,” Rev. Mod. Phys., 75, 281–324, 2003. [103] M. König, G. Bollen, H. J. Kluge, T. Otto, and J. Szerypo, “Quadrupole excitation of stored ion motion at the true cyclotron frequency,” International Journal of Mass Spectrometry and Ion Processes, 142, 95–116, 1995. 144 Bibliography [104] H. F. Powell, S. R. de Echaniz, E. S. Phillips, D. M. Segal, and R. C. Thompson, “Improvement of laser cooling of ions in a penning trap by use of the axialization technique,” Journal of Physics B: Atomic, Molecular and Optical Physics, 36, 961+, 2003. [105] F. Diedrich, E. Peik, J. M. Chen, W. Quint, and H. Walther, “Observation of a phase transition of stored laser-cooled ions,” Phys. Rev. Lett., 59, 2931–2934, 1987. [106] D. F. V. James, “Quantum dynamics of cold trapped ions with application to quantum computation,” Applied Physics B: Lasers and Optics, 66, 181–190, 1998. [107] D. Kielpinski, B. E. King, C. J. Myatt, C. A. Sackett, Q. A. Turchette, W. M. Itano, C. Monroe, D. J. Wineland, and W. H. Zurek, “Sympathetic cooling of trapped ions for quantum logic,” Phys. Rev. A, 61, 032310, 2000. [108] M. Drewsen and A. Brøner, “Harmonic linear paul trap: Stability diagram and effective potentials,” Phys. Rev. A, 62, 045401, 2000. [109] D. J. Berkeland, J. D. Miller, J. C. Bergquist, W. M. Itano, and D. J. Wineland, “Minimization of ion micromotion in a paul trap,” Journal of Applied Physics, 83, 5025–5033, 1998. [110] R. Alheit, S. Kleineidam, F. Vedel, M. Vedel, and G. Werth, “Higher order nonlinear resonances in a paul trap,” International Journal of Mass Spectrometry and Ion Processes, 154, 155–169, 1996. [111] S. Schwarz, Simulations for Ion Traps Buffer Gas Cooling. Lecture Notes in Physics 749, Springer Berlin Heidelberg, 2008. [112] C. Schneider, M. Enderlein, T. Huber, and T. Schaetz, “Optical trapping of an ion,” Nature Photonics, 4, 772–775, 2010. [113] R. G. DeVoe, “Power-law distributions for a trapped ion interacting with a classical buffer gas,” Phys. Rev. Lett., 102, 063001, 2009. [114] D. S. Goodman, I. Sivarajah, J. E. Wells, F. A. Narducci, and W. W. Smith, “Ionneutral-atom sympathetic cooling in a hybrid linear rf paul and magneto-optical trap,” Phys. Rev. A, 86, 033408, 2012. [115] P. D. Lett, R. N. Watts, C. I. Westbrook, W. D. Phillips, P. L. Gould, and H. J. Metcalf, “Observation of atoms laser cooled below the doppler limit,” Phys. Rev. Lett., 61, 169–172, 1988. 145 Bibliography [116] L. Ricci, M. Weidemüller, T. Esslinger, A. Hemmerich, C. Zimmermann, V. Vuletic, W. König, and T. W. Hänsch, “A compact grating-stabilized diode laser system for atomic physics,” Optics Communications, 117, 541–549, 1995. [117] G. C. Bjorklund, M. D. Levenson, W. Lenth, and C. Ortiz, “Frequency modulation (FM) spectroscopy,” Applied Physics B, 32, 145–152, 1983. [118] U. Schünemann, H. Engler, R. Grimm, M. Weidemüller, and M. Zielonkowski, “Simple scheme for tunable frequency offset locking of two lasers,” Review of Scientific Instruments, 70, 242–243, 1999. [119] T. Esslinger, I. Bloch, and T. W. Hänsch, “Bose-einstein condensation in a quadrupole-ioffe-configuration trap,” Phys. Rev. A, 58, R2664–R2667, 1998. [120] R. Grimm, M. Weidemüller, and Y. B. Ovchinnikov, “Optical dipole traps for neutral atoms,” Advances in Atomic, Molecular, and Optical Physics, 42, 95–170, 2000. [121] M. Pappa, P. C. Condylis, G. O. Konstantinidis, V. Bolpasi, A. Lazoudis, O. Morizot, D. Sahagun, M. Baker, and W. von Klitzing, “Ultra-sensitive atom imaging for matter-wave optics,” New Journal of Physics, 13, 115012+, 2011. [122] G. Reinaudi, T. Lahaye, Z. Wang, and D. Guéry-Odelin, “Strong saturation absorption imaging of dense clouds of ultracold atoms,” Opt. Lett., 32, 3143–3145, 2007. [123] W. Ketterle, D. S. Durfee, and D. M. Stamper-Kurn, “Making, probing and understanding bose-einstein condensates,” arXiv:cond-mat/9904034, 1999. [124] S. Olmschenk, Quantum Teleportation between Distant Matter Qubits. PhD thesis, University of Michigan, 2009. [125] S. Olmschenk, D. Hayes, D. N. Matsukevich, P. Maunz, D. L. Moehring, K. C. o Younge, and C. Monroe, “Measurement of the lifetime of the 6p 2 P 1/2 level of Yb+ ,” Phys. Rev. A, 80, 022502, 2009. [126] H. M. Meyer, M. Steiner, L. Ratschbacher, C. Zipkes, and M. Köhl, “Laser spectroscopy and cooling of Yb+ ions on a deep-uv transition,” Phys. Rev. A, 85, 012502, 2012. [127] P. Taylor, M. Roberts, S. V. Gateva-Kostova, R. B. M. Clarke, G. P. Barwood, W. R. C. Rowley, and P. Gill, “Investigation of the 2 S1/2 −2 D5/2 clock transition in a single ytterbium ion,” Phys. Rev. A, 56, 2699–2704, 1997. 146 Bibliography [128] M. Johanning, A. Braun, D. Eiteneuer, C. Paape, C. Balzer, W. Neuhauser, and C. Wunderlich, “Resonance-enhanced isotope-selective photoionization of Yb I for ion trap loading,” Applied Physics B: Lasers and Optics, 103, 327–338, 2011. [129] M. V. Nikolić, S. M. Radić, V. Minić, and M. M. Ristić, “The dependence of the work function of rare earth metals on their electron structure,” Microelectronics Journal, 27, 93–96, 1996. [130] M. Cetina, A. Grier, J. Campbell, I. Chuang, and V. Vuletić, “Bright source of cold ions for surface-electrode traps,” Phys. Rev. A, 76, 041401, 2007. [131] S. Ejtemaee, R. Thomas, and P. C. Haljan, “Optimization of Yb+ fluorescence and hyperfine-qubit detection,” Phys. Rev. A, 82, 063419, 2010. [132] D. J. Berkeland and M. G. Boshier, “Destabilization of dark states and optical spectroscopy in zeeman-degenerate atomic systems,” Phys. Rev. A, 65, 033413, 2002. [133] W. Alt, “An objective lens for efficient fluorescence detection of single atoms,” Optik - International Journal for Light and Electron Optics, 113, 142–144, 2002. [134] L. Richter, H. Mandelberg, M. Kruger, and P. McGrath, “Linewidth determination from self-heterodyne measurements with subcoherence delay times,” Quantum Electronics, IEEE Journal of, 22, 2070–2074, 1986. [135] K. M. Jones, E. Tiesinga, P. D. Lett, and P. S. Julienne, “Ultracold photoassociation spectroscopy: Long-range molecules and atomic scattering,” Rev. Mod. Phys., 78, 483–535, 2006. [136] B. Ueberholz, S. Kuhr, D. Frese, D. Meschede, and V. Gomer, “Counting Cold Collisions,” J. Phys. B: At. Mol. Opt. Phys., 33, 1996. [137] M. Drewsen, A. Mortensen, R. Martinussen, P. Staanum, and J. L. Sørensen, “Nondestructive identification of cold and extremely localized single molecular ions,” Phys. Rev. Lett., 93, 243201, 2004. [138] A. Härter, A. Krükow, M. Deiß, B. Drews, E. Tiemann, and J. H. Denschlag, “Shedding light on three-body recombination in an ultracold atomic gas,” arXiv:1301.5518, 2013. [139] A. K. Hansen, M. A. Sørensen, P. F. Staanum, and M. Drewsen, “Single-ion recycling reactions,” Angew. Chem. Int. Ed., 51, 7960–7962, 2012. 147 Bibliography [140] NIST, “Handbook of basic atomic spectral data,” http://www.nist.gov/pml/data/handbook/index.cfm. [141] D. Edvardsson, S. Lunell, and C. M. Marian, “Calculation of potential energy curves for Rb2 including relativistic effects,” Molecular Physics, 101, 2381–2389, 2003. [142] M. Aymar, S. Azizi, and O. Dulieu, “Model-potential calculations for ground and ex+ + cited Σ states of Rb+ 2 , Cs2 and RbCs ions,” Journal of Physics B: Atomic, Molecular and Optical Physics, 36, 4799+, 2003. [143] P. S. Julienne, “Ultracold atoms: Cool ion chemistry,” Nature Physics, 8, 642–643, 2012. [144] M. M. Schauer, J. R. Danielson, A.-T. Nguyen, L.-B. Wang, X. Zhao, and J. R. Torgerson, “Collisional population transfer in trapped Yb+ ions,” Phys. Rev. A, 79, 062705, 2009. [145] K. S. Trivedi, Probability and Statistics with Reliability, Queueing, and Computer Science Applications, 2nd Edition. Wiley-Interscience, 1 ed., 2001. [146] T. A. Laurence and B. A. Chromy, “Efficient maximum likelihood estimator fitting of histograms,” Nature Methods, 7, 338–339, 2010. [147] B. Efron, “Bootstrap methods: Another look at the jackknife,” Annals of Statistics, 7, 1–26, 1979. [148] M. Acton, K.-A. Brickman, P. C. Haljan, P. J. Lee, L. Deslauriers, and C. Monroe, “Near-perfect simultaneous measurement of a qubit register,” Quantum Info. Comput., 6, 465–482, 2006. [149] A. Keselman, Y. Glickman, N. Akerman, S. Kotler, and R. Ozeri, “High-fidelity state detection and tomography of a single-ion zeeman qubit,” New Journal of Physics, 13, 073027+, 2011. [150] W. Nagourney, J. Sandberg, and H. Dehmelt, “Shelved optical electron amplifier: Observation of quantum jumps,” Phys. Rev. Lett., 56, 2797–2799, 1986. [151] M. J. McDonnell, J.-P. Stacey, S. C. Webster, J. P. Home, A. Ramos, D. M. Lucas, D. N. Stacey, and A. M. Steane, “High-efficiency detection of a single quantum of angular momentum by suppression of optical pumping,” Phys. Rev. Lett., 93, 153601, 2004. 148 Bibliography [152] U. G. Poschinger, G. Huber, F. Ziesel, M. Deiß, M. Hettrich, S. A. Schulz, K. Singer, G. Poulsen, M. Drewsen, R. J. Hendricks, and F. Schmidt-Kaler, “Coherent manipulation of a 40 Ca+ spin qubit in a micro ion trap,” Journal of Physics B: Atomic, Molecular and Optical Physics, 42, 154013+, 2009. [153] C. F. Roos, Controlling the quantum state of trapped ions. PhD thesis, University of Innsbruck, 2000. [154] N. Timoney, I. Baumgart, M. Johanning, A. F. Varon, M. B. Plenio, A. Retzker, and C. Wunderlich, “Quantum gates and memory using microwave-dressed states,” Nature, 476, 185–188, 2011. [155] R. Islam, E. E. Edwards, K. Kim, S. Korenblit, C. Noh, H. Carmichael, G. D. Lin, L. M. Duan, C. C. Joseph Wang, J. K. Freericks, and C. Monroe, “Onset of a quantum phase transition with a trapped ion quantum simulator,” Nature Communications, 2, 377+, 2011. [156] M. Roberts, P. Taylor, S. V. Gateva-Kostova, R. B. M. Clarke, W. R. C. Rowley, and P. Gill, “Measurement of the 2 S1/2 −2 D5/2 clock transition in a single 171 Yb+ ion,” Phys. Rev. A, 60, 2867–2872, 1999. [157] K. Hosaka, S. A. Webster, A. Stannard, B. R. Walton, H. S. Margolis, and P. Gill, “Frequency measurement of the 2 S 1/2 -2 F 7/2 electric octupole transition in a single 171 Yb+ ion,” Phys. Rev. A, 79, 033403, 2009. [158] N. Huntemann, M. Okhapkin, B. Lipphardt, S. Weyers, C. Tamm, and E. Peik, “High-accuracy optical clock based on the octupole transition in 171 Yb+ ,” Phys. Rev. Lett., 108, 090801, 2012. [159] Y. Y. Jau, H. Partner, P. D. D. Schwindt, J. D. Prestage, J. R. Kellogg, and N. Yu, “Low-power, miniature 171 Yb ion clock using an ultra-small vacuum package,” Ap- plied Physics Letters, 101, 253518+, 2012. [160] J. Vanier and C. Audoin, The Quantum Physics of Atomic Frequency Standards: Volume 1. Institute of Physics Publishing, 1989. [161] F. G. Major and H. G. Dehmelt, “Exchange-collision technique for the rf spectroscopy of stored ions,” Phys. Rev., 170, 91–107, 1968. [162] S. K. Chung, J. D. Prestage, R. L. Tjoelker, and L. Maleki, “Buffer gas experiments in mercury (Hg+ ) ion clock,” in Frequency Control Symposium and Exposition, 2004. Proceedings of the 2004 IEEE International, 130–133, IEEE, 2004. 149 Bibliography [163] G. Poulsen, Y. Miroshnychenko, and M. Drewsen, “Efficient ground-state cooling of an ion in a large room-temperature linear paul trap with a sub-hertz heating rate,” Phys. Rev. A, 86, 051402, 2012. [164] P. T. H. Fisk, M. J. Sellars, M. A. Lawn, and G. Coles, “Accurate measurement of the 12.6 GHz clock transition in trapped 171 Yb+ ions,” IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, 44, 344–354, 1997. [165] S. Bize, Y. Sortais, M. S. Santos, C. Mandache, A. Clairon, and C. Salomon, “Highaccuracy measurement of the 87 Rb ground-state hyperfine splitting in an atomic fountain,” Europhysics Letters, 558+, 1999. 150