Atomic mixtures in radiofrequency dressed potentials

We have developed experimental techniques to confine two species of ultracold atoms in trapping potentials that are independently controlled by applied radiofrequency (rf) fields. Elementary species-selective manipulations are demonstrated for the isotopes 85Rb and 87Rb, such as changing the relative positions of the species. More detailed manipulations are possible, such as creating a double-well potential for one species that is overlapped with a single harmonic well of the other. We observed that both isotopes have long lifetimes in the rf-dressed potentials when separate; this is well known for 87Rb which has very favourable collision properties, but the stability of 85Rb in rf-dressed potentials had not previously been demonstrated. A large rate of inelastic inter-species collisions was observed when the clouds were brought together, and our results identify this as being due to two-body collisions between 85Rb and 87Rb. We explain the origin of these losses using a qualitative model. Our experimental data is compared to computations of the inelastic loss rate coefficients that have been carried out using coupled-channel calculations by theoretical chemists at Durham University. We have created double-well potentials for ultracold atoms by dressing with three rfs, as reported in our published work. Since this first demonstration we have eliminated sources of noise from the experimental apparatus using a systematic method, enabling a quantum degenerate gas of 87Rb to be loaded into a double-well potential with 6.7 micron separation between the wells. This is an order of magnitude improvement over our previously published work, and this separation is confirmed by matter-wave interference observed between atoms released from the two wells. In the near future, this apparatus will be able to split a quantum degenerate, two-dimensional gas, as has been performed for one-dimensional gases elsewhere. Moreover, many technical improvements have been implemented, allowing large amounts of data to be measured under highly reproducible conditions.