Large Faraday rotation of resonant light in a cold atomic cloud

During the past 2 yr, we have been theoretically and experimentally investigating coherent backscattering ~CBS! of near-resonant light in a sample of cold rubidium atoms @1–3#. CBS is an interference effect in the multiple scattering regime of propagation inside random media, yielding an enhancement of the backscattered light intensity @4#. This interference is very robust and can be destroyed only by a few mechanisms, including Faraday rotation @5# and dynamical effects @6#. In the particular case of atomic scatterers, we have shown that the existence of an internal Zeeman structure significantly degrades the CBS interference @1,3#. The breakdown of CBS due to the Faraday effect in classical samples has been recently observed and studied in detail @7#, in a situation where the scatterers are embedded in a Faraday-active matrix. We are currently exploring the behavior of CBS when a magnetic field is applied to the cold atomic cloud. Since the Faraday effect is expected to be large even at weak applied fields ~of the order of 1 G510 T!, it seems relevant to evaluate its magnitude in the particular regime of near-resonant excitation. The Faraday effect, i.e., the rotation of polarization experienced by light propagating inside a medium along an applied magnetic field, is a well-known phenomenon @8#. Faraday glass-based optical insulators are widely used in laser experiments to avoid unwanted optical feedback. Due to the presence of well-defined lines ~strong resonances!, the Faraday effect is potentially strong in atomic systems, and has been extensively studied in hot vapors @9#. In addition, light can modify the atomic gas as it propagates and induce alignment or orientation via optical pumping, yielding various nonlinear effects @10#. However, our experiment is quite insensitive to these effects and the scope of this paper will only be the linear, ‘‘standard’’ Faraday rotation. The suppression of Doppler broadening and collisions in laser-cooled atomic vapors allows us to fully exploit the strong resonance effects of atomic systems. Indeed all atoms of the sample experience the same detuning of the laser, as the Doppler broadening due to the residual velocities of the atoms is less than the natural linewidth of the transition. Furthermore only radiation relaxation of the populations and optical coherences need to be taken into account, which allows for straightforward quantitative comparison between theory and experiments. Despite offering such large resonant effects and easy