Trapping and Imaging Single Dysprosium Atoms in Optical Tweezer Arrays

We report the preparation and observation of single atoms of dysprosium in arrays of optical tweezers with a wavelength of 532 nm, imaged on the intercombination line at 626 nm. We use the anisotropic light shift specific to lanthanides and in particular a large difference in tensor and vector polarizabilities between the ground and excited states to tune the differential light shift and produce tweezers in near-magic or magic polarization. This allows us to find a regime where single atoms can be trapped and imaged. Using the tweezer array toolbox to manipulate lanthanides will open new research directions for quantum physics studies by taking advantage of their rich spectrum, large spin, and magnetic dipole moment.

Figure 1

(a) Simplified diagram of the beams used to trap and image single atoms in the tweezers. MOT beams are not shown here. (b) Relevant energy levels of Dy and their associated optical transitions used in this work. (c) Anisotropic light shift experienced by the Zeeman states in both the ground $\mathrm{\Delta }{\nu }_{m_J}^G$ and (d) excited $\mathrm{\Delta }{\nu }_{m_{J^{\prime }}}^E$ manifolds in magic conditions. The values plotted here are obtained by diagonalizing the full Hamiltonian including the Zeeman effect and the trap light shift. We then subtract the Zeeman shift to the eigenvalues of the Hamiltonian, to keep only the light shift. (e) Measured frequency difference between $|g⟩$ and $|e⟩$ with respect to their unperturbed frequency as a function of tweezer power and for different ellipticities $\theta$ of the tweezer polarization; shaded areas are linear fits with confidence interval.

Figure 2

(a) Single shot and (b) average picture of $5\times 5$ trap arrays for exposure time of 30 ms. The halo that can be seen between traps on the average image is due to the fluorescence of the microscope objective. (c) Histogram of the fluorescence of the central trap for 30 ms (d) and 100 ms exposure time. The line is a fit to the sum of two peak distributions joined by a “bridge” (see text). Dashed lines indicate the chosen threshold to maximize the imaging fidelity $F$.

Figure 3

(a) Imaging fidelity and loss probability as a function of the exposure time. (b) Loss probability ($P_{\mathrm{loss}}$) and average number of 626 nm photons detected before a loss ($N_{\mathrm{ph},\mathrm{loss}}$) as a function of imaging power for 30 ms exposure. In the shadowed area, losses are due to less efficient cooling. (c) The inverse of $N_{\mathrm{ph},\mathrm{loss}}$ as a function of tweezer power. The dashed gray line is a linear fit.

Figure 4

(a) Number of collected photons over 30 ms for a given trap under continuous illumination but with no background gas to reload the trap. This shows events where the atom is pumped to metastable states and events where it decays back to the ground state. (b) Probability to reimage an atom that previously became dark after having applied an imaging pulse of 1.5 s. The dashed line is a fit to an exponential saturation, with decay time 0.48(8) s.