Characterization of the retrofluorescence inhibition at the interface between glass and optically thick Cs vapor

Abstract

New experimental results and new phenomena related to the study of the retrofluorescence spectrum induced by a diode laser at the interface between glass and cesium vapor are presented. The Cs 62P3/2(Fe) hyperfine states are populated with a low-intensity tunable diode laser with a 10-MHz spectral bandwidth. We report an integrated atomic retrofluorescence spectrum in the 852.2-nm (62P3/2–62S1/2) resonance line and 917.2-nm (62D5/2–62P3/2) line generated by energy-pooling collisions. Emission spectra of the molecular fluorescence signal have been observed. At the resonance line a large proportion of the atoms excited by the laser are located in the vicinity of the surface. The interactions between the excited atomic 62P3/2 state and the surface appear as a spectral inhibition of the integrated retrofluorescence spectrum for both the atomic line and the molecular band. This spectral inhibition indicates that the nonradiative transformation process of changing atomically excited energy into thermal energy is preferred. We report the analysis of the dominant processes in the vicinity of the 852.2-nm resonance line, which can influence the retrofluorescence hyperfine spectrum at the boundary between the glass window and the saturated cesium vapor. Only the nonradiative transfer by evanescent waves toward the dissipative surface is retained. Using this mechanism, we formulate, for the first time to our knowledge, a simple model of a backscattered hyperfine fluorescence signal. The glass–vapor interface is considered as two distinct regions: a wavelength-thickness vapor layer joined to the surface and a more remote vapor region. The first region is analyzed as a spectral filter that annihilates the absorbed photons and the second one as a rich spectral light source. The experimental setup is described, and measured integrated retrofluorescence spectra are compared with predictions made by the model. The consistency between theory and experiment is remarkably good considering that the model depends only on two unknown parameters: the nonradiative transfer rate and the absorption shape line of the filtering region.