2000-Times Repeated Imaging of Strontium Atoms in Clock-Magic Tweezer Arrays

Editors' Suggestion

2000-Times Repeated Imaging of Strontium Atoms in Clock-Magic Tweezer Arrays

Jacob P. Covey, Ivaylo S. Madjarov, Alexandre Cooper, and Manuel Endres

Phys. Rev. Lett. 122, 173201 – Published 1 May 2019

Abstract

We demonstrate single-atom resolved imaging with a survival probability of 0.99932(8) and a fidelity of 0.99991(1), enabling us to perform repeated high-fidelity imaging of single atoms in tweezers thousands of times. We further observe lifetimes under laser cooling of more than seven minutes, an order of magnitude longer than in previous tweezer studies. Experiments are performed with strontium atoms in 813.4 nm tweezer arrays, which is at a magic wavelength for the clock transition. Tuning to this wavelength is enabled by off-magic Sisyphus cooling on the intercombination line, which lets us choose the tweezer wavelength almost arbitrarily. We find that a single not retroreflected cooling beam in the radial direction is sufficient for mitigating recoil heating during imaging. Moreover, this cooling technique yields temperatures below $5 μ K$ , as measured by release and recapture. Finally, we demonstrate clock-state resolved detection with average survival probability of 0.996(1) and average state detection fidelity of 0.981(1). Our work paves the way for atom-by-atom assembly of large defect-free arrays of alkaline-earth atoms, in which repeated interrogation of the clock transition is an imminent possibility.

Received 11 December 2018

DOI:https://doi.org/10.1103/PhysRevLett.122.173201

Physics Subject Headings (PhySH)

Atomic, optical & lattice clocks Cooling & trapping Metrology Quantum information with atoms & light Quantum sensing Time & frequency standards

Atoms Exotic atoms & molecules

Atomic, Molecular & OpticalQuantum Information, Science & TechnologyGeneral Physics

Authors & Affiliations

Jacob P. Covey, Ivaylo S. Madjarov, Alexandre Cooper, and Manuel Endres^*

Division of Physics, Mathematics and Astronomy, California Institute of Technology, Pasadena, California 91125, USA

^*mendres@caltech.edu

Article Text (Subscription Required)

Click to Expand

Supplemental Material (Subscription Required)

Click to Expand

References (Subscription Required)

Click to Expand

Issue

Vol. 122, Iss. 17 — 3 May 2019

Reuse & Permissions

Access Options

Article Available via CHORUS

Download Accepted Manuscript

Author publication services for translation and copyediting assistance advertisement

Images

Figure 1
Low-loss imaging scheme for ${}^{88}{Sr}$ . (a) Our single-atom imaging scheme requires only a single not retroreflected cooling beam at 689 nm for narrow-line Sisyphus cooling (red single arrow) to compensate recoil heating from fluorescence generated by exciting atoms with a single retroreflected 461 nm beam (blue double arrow). Microscope objectives with $NA = 0.5$ are used to generate 813.4 nm tweezers and to collect the fluorescence light. (b) Previous studies found an imaging loss channel via the decay of ${{}^{1}P}_{1}$ to ${{}^{1}D}_{2}$ with a branching ratio of $\sim 1 : 20000$ , where atoms left the tweezers since ${{}^{1}D}_{2}$ was strongly antitrapped [21, 22]. Crucially, the ${{}^{1}D}_{2}$ state is now trapped in 813.4 nm tweezers and our results show that atoms are recovered into the ${}^{3}{P_{J}}$ manifold with very high probability. Two lasers (679 and 707 nm) repump atoms to ${}^{3}{P_{1}}$ , which decays back into the ground state, thus closing the ${}^{3}{P_{J}}$ loss channel. (c) We use narrow-line attractive Sisyphus cooling on the $m_{J} = \pm 1$ states [26], originally proposed in Refs. [27, 28]. This mechanism is based on the excited state being more strongly trapped than the ground state (in contrast to repulsive Sisyphus cooling demonstrated recently by us [21]). Atoms at the bottom of the trap are excited and have to climb up a steeper potential than they would in the ground state, leading to an average reduction in energy after spontaneous emission. Cooling results from the trapping potential mismatch, and not from photon recoil, thus requiring only a single cooling beam. (d) Average image (top) and single-shot image (bottom) of atomic fluorescence from twenty-five uniformized tweezers with an imaging time of one second.
Reuse & Permissions
Figure 2
Low-loss high-fidelity imaging. (a) Histogram of fluorescence counts from a single representative tweezer. We find a detection fidelity of 0.99991(1) and an average survival probability of 0.99932(8), demonstrating simultaneous high-fidelity and low-loss imaging. Results are for an imaging time of $t = 50 ms$ under simultaneous repumping and Sisyphus cooling. (b) The survival fraction as a function of hold time in minutes under these imaging conditions (blue squares and fitted line). Importantly, we find a lifetime of $τ = 126 (3) s$ , while only needing $t ≲ 50 ms$ imaging time for reaching high detection fidelity, leading to small loss fractions consistent with $\exp (- t / τ)$ . Moreover, we find a lifetime of 434(13) s under Sisyphus cooling alone (without 461 nm), demonstrating a vacuum-limited lifetime greater than seven minutes (red circles and fitted line). (c) Survival fraction versus image number for 2000 repeated images. The dark red line represents the mean over 40 realizations, with the lighter red lines showing the standard error of the mean. Atoms are imaged with high fidelity for 50 ms followed by a 29 ms cooling block. (d) A representative realization of atom detection over the course of the 2000 images. Detected atoms are plotted in red versus the image number, where the rows represent the 25 tweezers. Note that we find no occurrences of atoms returning after they are lost. A video of the 2000 images is available in the Supplemental Material (SM) [37].
Reuse & Permissions
Figure 3
Sisyphus cooling to low temperatures (a) The survival fraction in an array in a release-and-recapture measurement performed by diabatically turning off the traps for a variable time followed by a sudden switch-on [38]. We show data after imaging (blue squares) and after adding a dedicated cooling block with Sisyphus cooling alone (red circles). Results are compared with classical Monte Carlo simulations for a three-dimensional thermal distribution at $5 μ K$ (dashed line). Note that the release-and-recapture method is mostly sensitive to the energy distribution in the radial direction. (b) Survival fraction in an array after release-and-recapture for $60 μ s$ off-time versus the red frequency during Sisyphus cooling for 25 ms with an intensity of $I / I_{s} \approx 200$ . The dashed line represents the case without a dedicated cooling block. We find that atoms are cooled for appropriately chosen red detunings, and heated for detunings further to the blue. This is consistent with an understanding of Sisyphus cooling as an attractor in energy space [27, 28]. Data in (a) are at $- 2.6 MHz$ detuning.
Reuse & Permissions
Figure 4
Low-loss state-resolved detection. (a) A statistical mixture of optical clock qubit states is represented as a circle, where the green section at ${{}^{1}S}_{0} \equiv | g ⟩$ represents the population in $| g ⟩$ and the purple section at ${{}^{3}P}_{0} \equiv | e ⟩$ represents the population in $| e ⟩$ . To measure the population in $| g ⟩$ , we image without the 679 nm repumper during which $| e ⟩$ remains dark. The accuracy of measuring the population in $| g ⟩$ is limited both by off-resonant scattering of the tweezer light, which pumps $| e ⟩$ back to $| g ⟩$ , and by pathways that pump $| g ⟩$ to $| e ⟩$ such as the ${{}^{1}D}_{2}$ decay channel and the off-resonant scattering from ${{}^{3}P}_{1}$ during cooling. As a result, the average state detection fidelity is 0.981(1). (b) We perform a second image that includes the 679 nm repumper, which pumps $| e ⟩$ to $| g ⟩$ via the ${{}^{3}S}_{1}$ state and the 707 nm repumper, such that both states are detected. The pumping process is illustrated by the purple arrow. This image measures the population in $| g ⟩$ and $| e ⟩$ , and informs whether the atom has been lost as a result of the first image. We find that the average survival probability of this double-imaging sequence is 0.996(1).
Reuse & Permissions

Physical Review Letters