Abstract

In this thesis I present the construction of a new apparatus aimed at studying two-component Bose-Einstein condensates (BECs) in the presence of a Rabi coupling, where the two components correspond to internal states of sodium atoms. The coherent mixture, in the miscible regime, also exhibits a metastable excitation consisting in a domain wall of relative phase connecting vortices of different components. Due to the peculiar energy dependence of such a configuration, an attractive force, independent of the vortex distance, is expected, making this system a candidate for mimicking features of quark confinement in QCD. The surface tension of the domain wall structure can be experimentally controlled via the strength of the coupling, allowing to study the system dynamics in different regimes. These include a predicted regime in which, for sufficient high coupling strength, the domain wall breaks and new vortex couples nucleate, providing by itself an interesting experimental realization of spin counterflow dynamical instabilities in a superfluid system, as well as a phenomenon analogous to string breaking in QCD. The choice of sodium as atomic species is motivated by its collisional properties that allow to obtain a perfect spatial superposition between the two miscible components |F = ±1⟩ if trapped by a spin-independent potential, avoiding the known phenomenon of ’buoyancy’. Studying the dynamics of such systems for sufficiently long times, with a mixture subject to Zeeman differential energy shifts, requires a specific effort to remove magnetic field fluctuations: a rough estimate suggests that in order to maintain the system coherence for a sufficiently long time to study its dynamics, magnetic field fluctuations should be reduced by at least three orders of magnitude compared to typical values observed in laboratory environment. Such attenuations can be obtained by means of multiple layers of μ-metal, that is incompatible with the use of ordinary magnetic traps, characterized by large magnetic field gradients on the atoms, due to residual magnetization and saturation of the shielding material. To avoid such effects it is required to either evaporatively cool atoms into an optical dipole trap loaded from a molasses stage, or a hybrid approach by means of which atoms are transferred to a low-gradient quadrupole trap superimposed to the optical trap. Producing BECs with such protocols greatly benefits from an efficient optical molasses cooling stage to prepare the sample in the best conditions of temperature and density before loading atoms into the trap. With this regard, the main limitation of ordinary laser cooling techniques is their reliance on absorption and spontaneous emission cycles, which limits the lowest temperature and highest density that can be reached as a consequence of residual heating effects and photon-reabsorption. An important resource to cope with these limits are dark states. In a broader sense a dark state is a state which does not interact with the exciting light field, and an atom in such a state would be neither subject to the beneficial cooling effects nor to the detrimental effects of light scattering. It is possible, however, to exploit the phenomenon of electromagnetically induced transparency (EIT) to induce a velocity-selective cooling mechanism for which slower atoms, that do not need further cooling, are trapped in a dark state corresponding to a coherent superposition of atomic levels whose excitation probabilities interfere destructively, while the cooling mechanism still applies to the fastest atoms. Among these techniques, gray molasses cooling allows to reach temperatures as low as a few recoil temperatures, while retaining atomic densities useful to reach quantum degeneracy in the subsequent stages of the experiment. In order to exploit this technique, an additional laser source had to be implemented during my thesis. To realize a gray molasses on the sample only |F⟩ → |F − 1⟩ or |F⟩ → |F⟩ transitions can be chosen, requiring blue-detuned laser, in contrast to ordinary (sub)Doppler laser cooling techniques. Both these requirements rule out the use of the D2 transition used for ordinary laser cooling techniques, due to its finely-spaced hyperfine structure. On the other hand, the D1 transition is characterized by a broader level spacing in the hyperfine structure and the absence of higher energy states on the blue side of the |F = 2⟩ → |F' = 2⟩ transition. As part of the work for this thesis, I successfully implemented and characterized gray molasses cooling on the D1 optical line of sodium. The buildup of the new apparatus includes the assembly of a new laser source for laser trapping and cooling on the D2 line, the assembly of the optical table devoted to the frequency and amplitude control of all the laser beams involved in the optical laser cooling procedures as well as the electronic control system. Design and assembly of the UHV and baking procedures for the stainless steel vacuum chamber are also described as well as the laser cooling techniques employed to load the atoms in a Dark-Spot MOT. Regarding the production of BEC, various strategies were attempted for different dipole beam configurations. Dipole traps typically suffer from the tradeoff between capture volume and trap depth at a given power, while hybrid traps usually take advantage of a magnetic trap stage that would not be compatible with the μ-metal shielding. Preliminary attempts to reach quantum degeneracy after directly loading the dipole trap from molasses were unsuccessful due to spurious effects. An alternative approach based on a magnetic-shield compatible hybrid trap protocol, in the absence of magnetic trap compression, was successfully implemented.