Painted loading: a toolkit for loading spatially large optical tweezer arrays

TL;DR

This work introduces painted loading, a method that expands the spatial extent of Sr-88 optical tweezer arrays by sweeping the narrow-line MOT reservoir across the array during loading. The approach yields controllable atom-number distributions, including uniform, gradient, and selectively loaded configurations, and demonstrates loading of arrays with vertical extents over 100 micrometers. A rate-equation model captures the observed trends by incorporating MOT heating, loading, and tweezer-loss dynamics, and qualitatively explains the gradient reversals as sweep speed changes. The technique promises substantial gains in scalable, high-site-count atomic arrays for quantum simulation, metrology, and computation, with clear paths toward 3D extension and automation via optimization algorithms.

Abstract

Arrays of neutral atoms in optical tweezers are widely used in quantum simulation and computation, and precision frequency metrology. The capabilities of these arrays are enhanced by maximising the number of available sites. Here we increase the spatial extent of a two-dimensional array of strontium-88 atoms by sweeping the frequency of the cooling light to move the atomic reservoir across the array. We load arrays with vertical heights of >100 μm, exceeding the height of an array loaded from a static reservoir by a factor of >3. We investigate the site-to-site atom number distribution, tweezer lifetime, and temperature, achieving an average temperature across the array of 1.49(3) μK. By controlling the frequency sweep we show it is possible to control the distribution of atoms across the array, including uniform and non-uniformly loaded arrays, and arrays with selectively loaded regions. We explain our results using a rate equation model which is in good qualitative agreement with the data.

Figures (6)

• Figure 1: Pictographic representation of painted loading. The nMOT is initially placed above the tweezer array (top left). This reservoir is then swept across the array by changing the frequency of the cooling light, moving the nMOT resonance condition across the array (top middle); tweezer sites crossed by the nMOT are loaded with atoms (top right). For illustrative purposes, the field of view of the cartoon is larger than that of our imaging system. Bottom-left inset: Image of a 90-site tweezer array loaded without sweeping. Only sites within the confines of the nMOT can be loaded. Bottom-right inset: The same array loaded using painted loading, showing atoms in all of the tweezer sites.
• Figure 2: (a, b, c) Averaged fluorescence images showing atom distributions loaded at sweep speeds of (a) $1.59(3)$ µm ms$^{-1}$ (b) $3.97(8)$ µm ms$^{-1}$ and (c) $7.95(15)$ µm ms$^{-1}$. Bar charts show the corresponding $\bar{n}_j$ by row, with model results (uncertainties) marked as diamonds (orange shaded regions). Each array was imaged 30 times. (d) Measured gradient of $\bar{n}_j$ as a function of sweep speed. Grey dashed line denotes zero gradient (uniform loading). Black dashed line (orange shaded area) shows the model predicted gradient (uncertainty) as a function of sweep speed. (e) Array-averaged number of atoms per site as a function of sweep speed. Black dashed line shows the model best fit at a scaling factor of $\kappa N_\mathrm{M}(0) =5.7(6)\times10^{-9}$, with orange shaded area showing the uncertainty.
• Figure 3: Measured tweezer lifetime $\tau_\mathrm{T}$ as a function of $\delta_\mathrm{T}$ and the corresponding nMOT-tweezer separation. Different colour crosses are used to differentiate data points for tweezers at different $z$ positions. The yellow and orange shaded regions, corresponding to regions 2 and 3 in equation \ref{['Eq:Regions']} respectively, are relevant for painted loading. A dotted black line indicates the measured nMOT lifetime for this data set, $97.0(4)$ ms, with shading denoting the uncertainty. The dashed black line shows the line of best fit for equation \ref{['Eq:Tweezer Detuning']} to the data.
• Figure 4: (a) Mean temperature of atoms in the array as a function of sweep speed $v_\mathrm{s}$. Points corresponding to figures \ref{['Fig:Atom Distributions']}(a), (b), (c) are highlighted. The initial nMOT temperature, $T_\mathrm{M}(0)=1.99(8)$ µK, and the temperature achieved for loading from a static nMOT, $0.83(6)$ µK, are marked as dashed and dash-dotted lines respectively, with the corresponding uncertainties represented by shading. (b, c) Final nMOT temperature as a function of (b) sweep speed for a fixed sweep distance of $159(3)$ µm and (c) distance swept for a fixed sweep speed of $7.95(15)$ µm ms$^{-1}$. $T_\mathrm{M}(0)$ is indicated as a dash-dotted line. Dotted lines show linear fits to the data. Dashed lines indicate the nMOT heating rate used in the model (see Section \ref{['sec:model']}), with orange shaded region showing the uncertainty.
• Figure 5: (i) Averaged fluorescence images for arbitrary sweeps targetting (a) linearly varying, (b) uniform, and (c) stepped atom distributions. Bar charts show the corresponding $\bar{n}_j$ by row, with model results marked as diamonds. The model prediction for the number of atoms in row 1 of (c) was $19.8(5)$ atoms, and as such is not displayed on these axes. Each array was imaged 250 times. (ii) Corresponding variation in the position of the nMOT resonance condition (black solid line) as a function of time. Red shading indicates the approximate region covered by the nMOT, and dashed blue lines indicate the position of each row in the tweezer array.