We analyze the implosion and subsequent explosion of a trapped condensate after the scattering length is switched to a negative value. Our results compare very well qualitatively and fairly well quantitatively with the results of recent experiments at JILA.
Evolution and collapse of a trapped Bose condensate with negative scattering length are predetermined by 3-body recombination of Bose-condensed atoms and by feeding of the condensate from the nonequilibrium thermal cloud. The collapse, starting once the number of condensate atoms reaches the critical value, ceases and turns to expansion when the density becomes so high that the recombination losses dominate over attractive interparticle interaction. As a result, we obtain a sequence of collapses, each of them followed by dynamic oscillations of the condensate. In every collapse the 3-body recombination burns only a part of the Bose-condensed atoms.
We solve the Gross-Pitaevskii equation numerically for the collapse induced by a switch from positive to negative scattering lengths. We compare our results with experiments performed with Bose-Einstein condensates of 85Rb, in which the scattering length was controlled using a Feshbach resonance. Building on previous theoretical work we identify quantitative differences between the predictions of mean-field theory and the results of the experiments. In addition to the previously reported difference between the predicted and observed critical atom number for collapse, we also find that the predicted collapse times systematically exceed those observed experimentally.
Phenomena associated with the controlled collapse of a Bose-Einstein condensate described in the experiment of Donley et al [1] are explained here as a consequence of the squeezing and amplification of quantum fluctuations above the condensate by the condensate dynamics. In analyzing the changing amplitude and particle contents of these excitations, our simple physical picture provides excellent quantitative fits with experimental data on the scaling behavior of the collapse time and the amount of particles emitted in the jets. In the experiments described by Donley et al. [1], a Bose-Einstein condensate (BEC) [2] in a cold (3nK) gas of Rubidium atoms is rendered unstable by a sudden inversion of the sign of the interaction between atoms. After a waiting time tcollapse, the condensate implodes (called Bose-Nova), and a fraction of the condensate atoms are seen to oscillate within the magnetic trap which contains the gas. These atoms are said to belong to a ‘burst’. After a time τevolve the interaction is suddenly turned off. For a certain range of values of τevolve, new emissions of atoms from the condensate are observed. They are called ‘jets’. Jets are distinct from bursts: they are colder, weaker, and have a characteristic disk-like shape. In the specific conditions of the experiment described in [1], the interaction is attractive in the relevant regime, corresponding to a negative scattering length, whose magnitude is typically tens of the critical value when collapse is first observed. This is different from a later JILA experiment [3] which explores the regime of positive scattering lengths thousands of times the critical one. There is a variety of theoretical explanations to these experiments, ranging from mean field dynamics of the condensate [4] to bosonic amplification [5]. The second experiment may already have a satis-