Experimental realization of two-isotope collision-assisted Zeeman cooling

Abstract

The work presented in this thesis focuses on the demonstration and initial evaluation of a novel non-evaporative cooling method called collision-assisted Zeeman cooling. For this realization, an ultracold gas consisting of a mixture of 87Rb and 8Rb was used. Cooling was accomplished through interisotope inelastic spin-exchange collisions that converted kinetic energy into magnetic energy. Continual optical pumping spin polarized the 85Rb which ensured that only kinetic energy reducing collisions occurred and the scattered pump photons carried entropy out of the system. Thus, cooling of the ultracold gas can be achieved without requiring the loss of any atoms in order to do so. This represents a theoretical advantage over forced evaporative cooling, which is the current state-of-the-art cooling technique in most experiments. This thesis discusses the details of collision-assisted Zeeman cooling, as well as how the theory of the technique has been extended from cooling a single species to cooling with two species. There are many predicted advantages from using two rather than one species of atom in this type of cooling: greater flexibility in finding favorable spin-exchange collision rates, easier requirements on the magnetic fields that must be used, and an additional means to mitigate reabsorption (the primary limitation in many if not most non-evaporative cooling techniques). The experimental considerations needed to prepare a system that simultaneously trapped two isotopes to be able to perform collision-assisted Zeeman cooling are discussed. Because this cooling scheme is highly reliant on the initial conditions of the system, a focused experiment examining the loading of the optical trap with both isotopes of Rb was conducted and the results of that experiment are described here. The first experimental observations of spin-exchange collisions in an ultracold gas mixture of Rb are described as a part of this work. The experiments where collision-assisted Zeeman cooling were demonstrated are then described and evaluated. In this first implementation of the cooling technique the initial densities were too low and optical-pump-induced heating and loss too high for achieving the full predicted performance of the cooling technique. Through additional modeling, these limitations were understood and the necessary improvements for the next iteration of CAZ cooling experiments are laid out at the end of this work.