A cavity array microscope for parallel single-atom interfacing

Abstract

Neutral atom arrays and optical cavity QED systems have developed in parallel as central pillars of modern experimental quantum science. While each platform has demonstrated exceptional capabilities-such as high-fidelity quantum logic in atom arrays, and strong light-matter coupling in cavities-their combination holds promise for realizing fast and non-destructive atom measurement, building large-scale quantum networks, and engineering hybrid atom-photon Hamiltonians. However, to date, experiments integrating the two platforms have been limited to spatially interfacing the entire atom array with one global cavity mode, a configuration that constrains addressability, parallelism, and scalability. Here we introduce the cavity array microscope, an experimental platform where each individual atom is strongly coupled to its own individual cavity across a two-dimensional array of over 40 modes. Our approach requires no nanophotonic elements, and instead uses a new free-space cavity geometry with intra-cavity lenses to realize above-unity peak cooperativity with micron-scale mode waists and spacings, compatible with typical atom array length scales while keeping atoms far from dielectric surfaces. We achieve homogeneous atom-cavity coupling, and show fast, non-destructive, parallel readout on millisecond timescales, including through a fiber array as a proof-of-principle for networking applications. As an outlook, we realize a next-generation iteration of the platform with over 500 cavities and a nearly 10 times improvement in finesse. Our work unlocks the regime of many-cavity QED, and opens an unexplored frontier of large-scale quantum networking with atom arrays.

Figures (14)

• Figure 1: The cavity array microscope.a. We introduce a new type of cavity architecture, the cavity array microscope, which transduces between single atoms and single photons in parallel across a two-dimensional array of atoms. Our design uses intra-cavity lenses to engineer micron-scale cavity mode waists at the atom location, and an intra-cavity microlens array to engineer an array of such modes with micron-scale spacing, compatible with typical atom array geometries. b. Average atomic fluorescence image read out via each individual cavity mode. c. Cross-cavity correlations between measured photons are on average ${\lesssim}1\%$, indicating each cavity-atom pair is independent.
• Figure 1: Detailed schematic of the cavity array microscope and surrounding experiment.a. Full ray-tracing simulation of a one-dimensional slice of the cavity array microscope. b. Intra-cavity optics with manufacturer and part numbers. For the flat end-mirror we either use a reflectivity of $98\%$ for measurements of the finesse (Fig. \ref{['fig:cooperativity']}), or $90\%$ for measurements of atoms to increase the collection efficiency (Figs. \ref{['fig:imaging']} and \ref{['fig:fiberarray']}). c. Schematic of the experimental system surrounding the cavity. d. Simplified schematic of cavity array generation showing relevant telescopic distances.
• Figure 2: Achieving a degenerate array of cavities.a. Beams input to the cavity with some displacement from the central axis are not, by default, stably trapped due to intrinsic aberrations coming from the intra-cavity optics. This is evidenced by ray optics simulations (with lossless optics) of the number of round-trips an incident ray with zero slope makes within the cavity before escaping or clipping. However, adding an intra-cavity microlens array (MLA) provides local regions of stability even for large positional displacements. b. Inverted reflectance spectra over a free spectral range (FSR) of measured resonances for driving the fundamental Gaussian mode for three adjacent cavities (yellow, red, and blue; inset shows a partial round-trip for each mode) as the position of the intra-cavity spherical lens is tuned. For an arbitrary lens position the modes are separated, but by tuning to the position that creates a perfect telescope all modes are made simultaneously degenerate. The physical variable being swept on the $x$-axis is the voltage of the piezo-driven mirror, see Methods. c. Detunings (defined in b) between the Gaussian modes of adjacent cavities grow linearly as a function of lens displacement, with the sensitivity increasing as a function of the cavity's distance from the central axis.
• Figure 2: Doubled imaging and two-atom loading.a. In the cavity geometry we introduce, inversion on the curved mirror leads to each cavity mode having two wavelength-scale waists, and two outcoupling ports. Also annotated are the directions and transitions of the beams driving fluorescence and atom repumping. b. The conjugate imaging ports are inverted around the center of the array (red cross), indicated here for a few select pairs of ports (colored boxes) on the average fluorescence image. The black dashed box indicates the central 21 cavities which we consider for most in-depth array characterizations in the main text. c. Single-shot fluorescence images show an inversion-symmetric pattern of fluorescence due to the doubling effect (conjugate ports for which no atom has been flagged are boxed with the same color to highlight the inversion). d. The doubling is clear from the near-perfect photon correlations between conjugate ports of fluorescence spots (ordering is column-major); in Fig. \ref{['fig:schematic']}c of the main text, the correlation is taken after imaging counts have been summed over both output ports. e. By default, this doubling effect will also induce two-atom loading, one in each wavelength-scale waist, as evidenced by a tri-modal fluorescence histogram. 0-, 1-, and 2- atom loading fractions are annotated in each section. f. We intentionally lower the loading efficiency by lowering the trap depth during loading and pulsing traps on and off to quadratically suppress the double loading effect by a factor of $10\times$ compared to the single loading probability.
• Figure 3: Cavity array microscope performance.a. Spectra showing three consecutive resonances for the central 21 cavity modes (measured sequentially), where the physical variable being swept is the voltage of the piezo-driven mirror. b. Degeneracy of the cavity modes. c. Finesses fit from cavity spectra. d. Cavity mode waists at the atom location, measured via trap depth and trap frequency spectroscopy. e. Calculated peak cooperativity. In b-e the array-averaged value (in b, absolute value) and standard deviation are under each subtitle.