Site-resolved imaging of a bosonic Mott insulator using ytterbium atoms

We demonstrate site-resolved imaging of a strongly correlated quantum system without relying on laser cooling techniques during fluorescence imaging. We observe the formation of Mott shells in the insulating regime and realize thermometry in an atomic cloud. This work proves the feasibility of the noncooled approach and opens the door to extending the detection technology to new atomic species.

Figure 1

Experimental setup. After creating a Bose-Einstein condensate and compressing it by using a combination of an optical accordion and a vertical ODT, a pair of retroreflected lattice beams is introduced to load the atoms into the two-dimensional optical lattice. The system is then driven into a Mott insulating state by gradually ramping up the intensity of the lattice beams. A photoassociation light is used to force inelastic light-assisted collisions on multiple occupied sites. Site-resolved imaging of the insulator is realized by irradiating a single excitation beam onto the atoms. No cooling mechanism is employed while the imaging system collects photons.

Figure 2

Site-resolved imaging of a bosonic ${}^{174}\mathrm{Yb}$ Mott insulator. Top row corresponds to raw images obtained with the emCCD camera during an exposure time of $40\mu \mathrm{s}$. The middle row shows the reconstructed atomic-density distribution obtained from the deconvolution algorithm. The estimated number of atoms in each image is (a) $45\pm 7$, (b) $115\pm 5$, (c) $179\pm 13$, (d) $273\pm 12$, (e) $486\pm 21$, and (f) $592\pm 27$. Bottom row shows the result of averaging 10 fluorescence raw images. Due to the presence of an external harmonic confinement in the optical lattices, the number of visible concentric Mott shells increases with the number of atoms.

Figure 3

Estimation of loss effects. (a) Histogram for the number of sites as a function of the total fluorescence at each site using 10 images taken at an exposure of 300 $\mu \mathrm{s}$. The left peak corresponds to empty sites that are affected by stray background light. (b) Typical raw image. (c) Computed complementary cumulative function of the histogram, shown as a semilog plot. Circular points are fitted with a simulation (solid line) to obtain the percentage of occupied sites. Triangular points are not considered in the fitting because they include the contribution of empty sites.

Figure 4

Estimation of hopping effects. (a) Typical raw image with a sparse population taken at an exposure time of $40\mu \mathrm{s}$. (b) Histogram of the total fluorescence per site using 100 images containing an average of 2.6 atoms per image. Left and right peaks correspond to empty and occupied sites, respectively. (c) Upper limit of the hopping probability as a function of the occupancy threshold. Error bars denote $68\%$ Clopper-Pearson confidence intervals.

Figure 5

Temperature measurement of a Mott insulator. The reconstructed atomic-density distribution is averaged azimuthally to obtain the radial profiles (points) and fitted using the grand-canonical ensemble described by the Bose-Hubbard model under the zero-tunneling approximation (solid lines). Error bars denotes $68\%$ Clopper-Pearson confidence intervals. The experimental data for (a)–(c) correspond to the reconstructed density distributions in Figs. 2, 2 and 2, respectively. From each fitting, the global chemical potential ${\mu }_0$, temperature $T$, and Mott radius $r_0$ are extracted.