Observation of cooling-assisted velocity-selective coherent population trapping

In recent years, there has been a great deal of interest in velocity selective coherent population trapping (VSCPT). However, since VSCPT occurs via a random walk in momentum space, in order to achieve significant VSCPT in three dimensions, it is necessary to precool and confine atoms close to the recoil momentum. Shahriar et al. recently predicted that this type of cooling and velocity confinement assisted VSCPT can be achieved in one, two, or three dimensions when a Λ-system atom is excited by a pair of blue detuned Raman resonant standing waves, with a phase difference of θ = π/4. In this work, we report the observation of this effect in one dimension using a beam of metastable He atoms. Figure 1 illustrates schematically the experimental setup, where a diode laser-pumped LNA laser excites the 2³S₁ → 2³P₁ transition in helium at 1083 nm. In Fig. 2, the curves labelled 'A' show the beam profile when the laser is off. With the laser on resonance and θ = π/4, we studied first the efficiency of VSCPT as a function of the saturation parameter s = r/(1 + 4δ²/Γ²), where r is the laser intensity in units of the saturation intensity, δ is the detuning from resonance, and Γ is the natural linewidth of the transition. For s = 2, the result is shown by curve B, where VSCPT is manifested by the appearance of the two peaks at ±ℏk. When s is reduced, the degree of VSCPT is decreased as expected, since the pumping rate into the dark state is proportional to the mean excited state population, which in turn is proportional to s. Curve C shows the result for s ≅ 0.4, where the VSCPT peaks are barely visible. We then blue detune the laser to δ = Γ and adjust the intensity to get the same degree of saturation (s ≅ 0.4). The resulting distribution is shown in curve D. As can be seen, there is strong precooling, with a width of ±2ℏk, in agreement with the prediction of ref. 2. In addition, the precooling enhances the degree of VSCPT compared to that in curve C, even though the saturation is the same. Due to the constraints of the present setup, our interaction time is limited to only 6 recoil times. While this is long enough to observe the precooling, it is too short to see the full extent of the enhancement of VSCPT due to the precooling. Nonetheless, our observation validates the mechanism predicted in ref. 2, rendering credence to its claim that this scheme would enhance the rate of VSCPT in three dimensions by a factor of 200, and allow one to reach a temperature a factor of ten below the recoil limit in about 90 ms.