We are currently working on producing a BEC of 87Rb. Our apparatus has several unique features which should allow us to produce Bose-condensed atoms at an unprecedented rate: First, we will use 3D Raman Sideband cooling in optical lattices as a precooling stage to evaporation in a conventional magnetic trap. We hope to demonstrate that a large BEC can be produced fast enough to permit the use of a single vapor cell with modest background pressure, removing the need for complicated double MOT or Zeeman slower-based systems. Our apparatus also incorporates a double magnetic trap design, which will give us the possibility of extreme adiabatic compression of our sample before evaporation, potentially shortening the evaporation time even further.
After producing BEC we plan to study many-body physics in optical lattices. Specifically, we plan to study quantum phase transitions, Such as the famous Mott-insulator to Superfluid transition (recently observed by the groups of T. Hansch and M. Kasevich) with a particular interest in their critical behavior at the transition itself. We are also thinking about experiments involving the effects of reduced dimensionality and disorder on the phase transition, as well as spinor condensates in optical lattices. [Top]
We are currently finishing work on sideband cooling of 87Rb atoms loaded from a dark SPOT. So far, we have succeeded in cooling (and spin-polarizing) up to 5 x 108 87Rb atoms to 3D free-space temperatures (after adiabatic release from the optical lattice) as low as 800 nK and phase-space densities as high as 10-3, at densities as high as 3 x 1011 cm-3. We have also cooled similar clouds in 2D to temperatures as low as 200 nK. Our lattices use typically 120 mW of power at a detuning of 10 GHz.
Below are shown time-of-flight traces for both cooled and uncooled atoms. Note that the 3.2 ms width of the sideband-cooled atoms is determined largely by the initial size of the cloud before it is released from the lattice. This width corresponds to a free-space kinetic temperature (after correcting for the initial size) of 780 nK. The initial temperature in the molasses, before loading into the lattice, is 10 mK. Loading efficiency into the lattice is still under investigation, but for small MOTs (~ 108 atoms) we have already observed efficiencies as high as 90%. This is especially remarkable since the lattice itself is only about 20 mK deep, and is a consequence of the fact that the cooling still works fairly efficiently even in the weakly or slightly unbound energy bands near the top of and above the lattice potential.
Because of the robustness of the cooling into the lattice, the final loading efficiency is not very sensitive to the temperature in the molasses before the sideband cooling is initiated. Thus, although the very cold molasses temperature shown above is necessary to get near unit loading efficiency, reasonable efficiency is still possible with more conventional molasses temperatures of 40 mK or so, as shown in the figure below. The blue circles are for cooling into the |1,1> state with optical pumping on the F=1 -> F'=0 transition, and the black squares are for cooling into the |2,-2> state with optical pumping on the F=2 -> F'=1 transition.
Although degenerate Raman sideband cooling has the restriction that atoms are cooled into the high-field-seeking (and therefore non-magentically-trappable state), we have demonstrated that using a non-degenerate variant of the cooling it is also possible to cool into the low-field seeking states, which may be of interest for the most efficient loading into a magnetic trap.
After sideband cooling, the atoms will be loaded into a special modematching magnetic trap, designed to simultaneously cancel the force of gravity and to provide in addition a confining potential sufficiently weak that the potential energy added by the trap is as close as possible to the extremely low kinetic energy. Without such a trap, we would be unable to take advantage of the low temperatures attainable with sideband cooling.
This trap will then be transformed into a TOP by slowly turning on a rotating bias field while increasing the curvature of the trap. This transformation is accomplished in such a way that the “circle of death” of the TOP starts far above the atoms, which are then raised up into the center as they are compressed. We elected to use the TOP in order to avoid the large anisotropy of a compressed Ioffe-Pritchard trap. In our previous work using highly anisotropic dipole force traps, we saw evidence that the thermalization rate in such traps can be strongly limited by the oscillation frequency of the weakly confining axis of the trap, possibly because of hydrodynamic behavior when the collision rate becomes much higher than that oscillation frequency.
By starting evaporation at a phase-space density three orders of magnitude higher than conventional BEC experiments, we should be able to have a much higher initial collision rate in our trap, thus dramatically reducing the evaporation time and increasing the number of atoms in the condensate. This method may speed up evaporation sufficiently to make possible a simple vapor-cell loaded BEC system which can still produce condensates of a large size in a short time.
To speed this up even further, we have constructed a double magnetic trap system which consists of a very small quadrupole pair inside a larger pair. The small coils sit on either side of a small glass finger with a 4mm ID that is attached to the top face of the glass cell. Since the TOP field rotates about an axis connecting the centers of the two traps, atoms should be able to be transported easily between them.
Below is a picture of the double magnetic trap and TOP coils. The mounting is made of polycarbonate to prevent eddy currents.
The upper, small magnetic trap can produce gradients in excess of 6000 G/cm. Atoms loaded into this trap should be able to be compressed even before evaporation to the density where 3-body recombination becomes a limiting factor. The low initial temperatures obtained with sideband cooling should allow this compression to be performed without the atoms boiling out of the trap. As evaporation proceeds, the trap can be ramped down to avoid the density limit. In this way we should be able to achieve collision rates at the start of evaporation more typical for the end of an evaporation ramp in conventional experiments.
Also shown in the picture above are our high-frequency TOP coils. Because of the large vibration frequencies that are possible in the small TOP trap, we must increase the rotation rate of the bias field, so that the oscillation period is still long compared to the rotation period. The HF TOP coils shown above are designed to operate at 60 kHz, with fields of up to 100 G.
Below is a picture of most of the apparatus. In the foreground are the two tapered amplifiers used, respectively, for our two MOTs and the 3D lattice for sideband cooling. [Top]
List of Recent Publications:
The nerve-wracking process of the first installation of the magnetic trap coils around the upper glass cell:
A picture of the ceiling above our apparatus. This will
no doubt make it harder to produce a BEC:
Back to Atomic Physics
Back to Chu Group
Last modified 04/16/2002 by Jamie Kerman