We descrbe a new method for measuring excited-state collisions between optically trapped atoms. With this method, trap-loss collision rates are deduced from the loading behavior of clouds of trapped atoms in the regime where radiation trapping limits the atom density. Our measurements indicate that
85Rb trap-loss collisions occur at significantly smaller rates than expected both from previous work on Cs and from recent models. In addition, the dependence of the trap-loss collisions on the frequency of the light used to excite the atom pairs is also different from that of Cs, suggesting that assumptions about the dynamics in these models need modification.
Collisions of trapped atoms occur over time intervals that are often longer than radiative lifetimes. This leads to radiative interruption of excited-state collision dynamics and constrains collisional interactions to atom pairs that are in close proximity at the time of excitation. Using classical dynamics I show that this causes Na associative ionization to be dominated by a four-step process in which the longer-range
Na∗-Na interaction initiates inward motion that is completed as
Na∗+
Na∗\ensuremath{\rightarrow}${\mathrm{Na}}_{2}^{\mathrm{\ifmmode^\circ\else\textdegree\fi{}}}$+e. Thus low-temperature associative ionization does not result from
Na∗+
Na∗ collisions. I predict strong single- and two-wavelength dependences, which have a molecular character due to the close proximity at the time of photoabsorption. These are amenable to experimental observation. I also calculate saturation that results from the depletion of interacting pairs. This produces very-low-power saturation and explains the associative ionization measurements of Gould et al. [Phys. Rev. Lett. 60, 788 (1988)].