Reservoir spectroscopy of $5s5p$ ${}^{3}P{}_2$–$5snd$ ${}^3{D}_{1,2,3}$ transitions in strontium

We perform spectroscopy on the optical dipole transitions $5s5p{}^3{P}_2–5snd{}^3{D}_{1,2,3}$, $n\in (5,6)$, for all stable isotopes of atomic strontium. We develop a spectroscopy scheme in which atoms in the metastable ${}^3{P}_2$ state are stored in a reservoir before being probed. The method presented here increases the attained precision and accuracy by two orders of magnitude compared to similar experiments performed in a magneto-optical trap or discharge. We show how the state distribution and velocity spread of atoms in the reservoir can be tailored to increase the spectroscopy performance. The absolute transition frequencies are measured with an accuracy of 5 MHz. The isotope shifts are given to within 200 kHz. We calculate the $A$ and $Q$ parameters for the hyperfine structure of the fermionic isotope at the MHz level. Furthermore, we investigate the branching ratios of the ${}^3{D}_J$ states into the ${}^3{P}_J$ states and discuss immediate implications on schemes of optical pumping and fluorescence detection.

Figure 1

Schematic illustration of the energy levels and transitions in strontium relevant for this work. Atoms in the metastable 5s5p3P2 state (indicated by the small solid arrow) are repumped using the 5s5p3P2–5snd3D1,2,3 transitions around 497 nm (n=5, dotted line) and around 403 nm (n=6, dash-dotted line).

Figure 2

Burst in MOT fluorescence after repumping of atoms in the 3P2 reservoir state back into the MOT cycle. The data is approximated by the sum of two exponential functions (solid red line), consisting of a fast growing (τrepump, dotted blue line) and a slowly decreasing contribution (τrMOT, dashed blue line). The horizontal dashed line indicates the small and stable background signal, caused by residual stray light.

Figure 3

Reservoir spectroscopy scheme. (a) The reservoir spectroscopy signal is background free and rises many orders of magnitude above the background noise. By contrast, the fluorescence signal of a MOT (inset) is increased by only a small factor on resonance. A Lorentzian profile is fit to each data set, and the fluorescence curves are taken at a repump laser intensity of I=1, 3, and 10×10−3Isat, from below. (b) Dependence of the width of the transition on the depth of the magnetic trap. (c) Width (black squares) and atom number (red circles) in dependence of repump light intensity. The repump pulse duration is 10 ms. (d) Accumulation of a certain hyperfine state by optical pumping increases the signal. (e) Depletion of a certain hyperfine state allows for a separation of otherwise overlapping hyperfine transitions. Measurements (a) through (c) are taken on the 5s5p3P2–5s5d3D2 transition of 88Sr, measurements (d) and (e) employ 87Sr; see the text for details.

Figure 4

Reservoir spectroscopy data obtained with the absorption imaging detection method, shown here for the 5s5p3P2–5s5d3D2 transition. (a) Transitions of the bosonic isotopes 84Sr (black, left peak), 86Sr (red, center), and 88Sr (blue, right). (b) A scan across the transitions of the fermionic 87Sr isotope. The separate measurement of the two overlapping lines around 5.75 GHz is discussed in Sec. III D.

Figure 5

Transition probabilities Aik, given in units of 106s−1, between triplet states of Sr relevant for this work. Values are taken from Ref. [86]. The fine structure splitting, indicated here only for the lowest 3PJ states, is otherwise smaller than the thickness of the horizontal lines. Decay paths with negligible contribution are omitted.

Figure 6

Repump rate in dependence of light intensity. The number of atoms repumped into the 1S0 ground state is proportional to the repump time before reaching saturation (inset, taken on the green 5s5p3P2–5s5d3D3 transition at I=1×10−2Isat). This rate is measured for a large range of intensities on both of the green 3P2–3D2 and 3P2–3D3 transitions.

Figure 7

Characterization of the wavemeter error. The wavemeter is calibrated at 729 nm (open diamond) and used to measure light at various well-known frequencies (filled circles). We plot the difference between the obtained values and the literature values, where the error bars are dominated by the calibration and measurement uncertainty of the wavemeter. The stars denote wavelengths at which additional consistency checks were performed, however at a smaller precision. The spectroscopy measurements presented in this work were performed at 403 and 497 nm, denoted by arrows.