Enhanced light-assisted-collision rate via excitation to the long-lived $5S_{1/2}$-$5D_{5/2}$ molecular potential in an ${}^{85}\mathrm{Rb}$ magneto-optical trap

We report measurements of a significant increase in the two-body loss rate in an ${}^{85}\mathrm{Rb}$ magneto-optic trap (MOT) caused by the addition of light resonant with the $5P_{3/2}$-to-$5D_{5/2}$ transition (776 nm) in Rb. Exposure to the additional light resulted in up to a factor of 5 decrease in the steady-state number of atoms in the MOT. This loss is attributed to more than an order of magnitude increase in the light-assisted collision rate brought about by the 776-nm light. By measuring the intensity dependence of the loss rate, the loss channel was identified to be the relatively long-lived $5S_{1/2}$-$5D_{5/2}$ molecular potential.

Figure 1

Our experimental setup. Note here that that L1 (780 nm) and L2 (776 nm) are aligned along the same path through the vacuum chamber. (AOM: acousto-optic modulator; AH: anti-Helmholtz coils; PD: photodiode; BS: beam splitting.)

Figure 2

Relevant energy levels for our experiment. (Not to scale.)

Figure 3

This is a typical fluorescence signal from the MOT before and during the application of the 776-nm light. The MOT is filled to equilibrium without the 776-nm light present. At $t=0$, the 776 nm light is applied, and the MOT fluorescence can be observed to decrease on a time scale much faster than the $~$20 s MOT fill time, indicating the presence of additional losses.

Figure 4

Density dependence of the one- and two-body loss calculations. The one-body loss calculation is represented by black circles. The two-body loss calculation is represented by red squares. The one-body loss calculations show a significant statistical difference at different densities whereas the two-body loss calculations are consistent within statistical uncertainty. A simultaneous one- and two-body calculation produced a value for the one-body loss coefficient consistent with zero.

Figure 5

Loss rates under various conditions for L2 (776 nm). (a) shows the loss rate with constant average power (shorter pulses at higher intensity for L2) for different L2 intensities (black circles). The red squares show the inferred loss rates (as if L2 was not pulsing, and on for the full duty cycle) for the different intensities. (b) shows the loss rate as a function of pulse length for a constant intensity (12 mW/cm${}^2$). (c) shows a one-body fit to the data from (b). The vertical axis is the one-body loss rate in units of ${\mathrm{s}}^{-1}$, and the horizontal axis is pulse length in ms. It can be seen here that a two-body analysis fits much better to the data.

Figure 6

A diagram of the loss channel we identify in the main text. Atom pairs excited to the 5$S$-5$P$ molecular potential by L1 accelerate until they are excited by L2 up to the 5$S$-5$D$ state. The long lifetime of the 5$S$-5$D$ state allows the atom pair to travel into a region where they can be lost from the trap upon their decay to the 5$S$-5$P$ state, due to increased acceleration along the slope of the 5$S$-5$P$ potential curve.

Figure 7

Two-body loss rate as function of L2 detuning from the point of maximum loss. (a) shows the loss rate with a 12-MHz red detuning for L1. (b) shows the loss rate for 18-MHz detuning for L1. The width of these curves is $~$15–20 MHz, which is much wider than that of the 421-nm fluorescence.