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

This paper presents a toolkit for loading spatially large optical tweezer arrays by sweeping the cooling light frequency to move a strontium-88 atomic reservoir, enabling the creation of arrays over 100 μm tall with controlled atom distributions and low temperatures.

The Gist

Imagine you are trying to fill a massive grid of tiny, invisible cups (optical tweezers) with individual marbles (atoms) to build a super-precise quantum computer or a super-accurate clock. The problem is that the "marble machine" (a cloud of cold atoms called a magneto-optical trap, or nMOT) is very flat and thin, like a pancake. If you just hold the machine still over the grid, it can only fill the cups in the very middle, leaving the top and bottom rows empty.

This paper introduces a clever new technique called "Painted Loading" to solve this problem. Here is how it works, using simple analogies:

1. The Problem: The Flat Pancake

The authors are working with Strontium atoms. These atoms are cooled down to near absolute zero and trapped in a magnetic field. However, because of how physics works with these specific atoms, the cloud of trapped atoms naturally forms a thin, vertical shell—like a hollow, vertical pancake that is only about 10 micrometers thick.

If you try to drop these atoms into a large grid of laser traps (tweezers) that is 100 micrometers tall, the "pancake" is too short to reach the top and bottom rows. In a traditional setup, you could only fill a small strip in the middle.

2. The Solution: The Paint Roller

Instead of keeping the atom cloud still, the researchers decided to move it.

Imagine you have a paint roller (the atom cloud) and a long wall with a grid of squares you want to paint (the laser tweezers).

• Traditional Method: You hold the roller still. You only paint the middle of the wall.
• Painted Loading: You roll the paint roller up and down the wall while it's spinning. As it moves, it paints every square on the wall.

In the lab, they do this by slightly changing the color (frequency) of the cooling laser light. This change makes the "gravity" of the magnetic trap shift up or down. By sweeping the laser frequency, they physically move the entire cloud of atoms across the grid of tweezers, "painting" atoms into every single spot, from the very top to the very bottom.

3. Controlling the "Paint"

The most exciting part of this toolkit is that they can control how the paint is applied just by changing how fast they move the roller:

• Moving Slow: If they move the cloud slowly, the first cups it passes get filled, but the atoms start to get "hot" and fly away before the cloud reaches the end. This results in the bottom rows having fewer atoms than the top rows.
• Moving Fast: If they move the cloud very quickly, the atoms don't have time to settle properly in the first cups, but they rush into the later cups. This flips the pattern, leaving the top rows emptier than the bottom ones.
• The "Sweet Spot": By finding the perfect middle speed, they can make the paint roller deposit an equal amount of atoms in every single cup, creating a perfectly uniform grid.
• Selective Painting: They can even stop the roller in mid-air or jump it over certain sections. This allows them to fill only specific rows of the grid while leaving others empty, creating custom patterns without needing complex hardware.

4. The Results

Using this "paint roller" method, the team successfully loaded a grid of 90 atoms that was over 100 micrometers tall. This is more than three times larger vertically than what was possible with the old, static method.

They also built a computer model (a set of equations) to predict exactly how the atoms would behave. The model matched their real-world experiments very well, confirming that the key to success is balancing the speed of the movement with how long the atoms stay trapped before flying away.

Summary

In short, the paper describes a new way to load large grids of atoms by "sweeping" a thin cloud of atoms across the grid, much like a paint roller. This allows scientists to fill much larger and more complex grids of atoms than before, giving them better control over the number of atoms in each spot, which is essential for building powerful quantum computers and ultra-precise atomic clocks.

Technical Summary

Technical Summary: Painted Loading for Spatially Large Optical Tweezer Arrays

Problem Statement
Arrays of neutral atoms in optical tweezers are a versatile platform for quantum simulation, computation, and precision frequency metrology. The performance of these applications scales favorably with the number of available sites. However, loading large two-dimensional arrays is constrained by the spatial extent of the cold-atom reservoir, typically a narrow-line magneto-optical trap (nMOT). In multivalent atoms like Strontium ($^{88}\text{Sr}$), the nMOT operates on a narrow intercombination transition ($\Gamma < 1$ Hz), resulting in a characteristic elliptical shape where atoms are confined to a thin shell (approx. 10 $\mu$m vertical thickness) due to gravity and recoil effects. This limited vertical extent restricts loaded tweezer arrays to roughly 100 sites, significantly smaller than the >1000 sites achievable in alkali atoms. Existing solutions, such as restricting arrays to the horizontal plane or using optical transport, impose geometric limitations or are incompatible with the narrow linewidths of species like Sr.

Methodology: Painted Loading
The authors introduce "painted loading," a technique that dynamically sweeps the atomic reservoir across the tweezer array during the loading phase.

• Mechanism: The vertical position of the nMOT resonance condition is controlled by sweeping the detuning ($\delta$) of the cooling light. By increasing the detuning, the resonance condition moves downward (in the $-z$ direction), effectively moving the nMOT "paint roller" across the static tweezer array.
• Control Parameters: The method allows for control over the sweep speed ($v_s$), sweep distance, and sweep shape (linear or non-linear).
• Experimental Setup: The experiment utilizes a 2D array of optical tweezers at the magic wavelength (813.427 nm) for the $^{88}\text{Sr}$ clock transition. The nMOT is generated using retroreflected beams red-detuned from the $5s^2\ ^1S_0 \to 5s5p\ ^3P_1$ transition (689 nm) in a quadrupole magnetic field.
• Modeling: A rate equation model was developed to describe the loading process, accounting for three time regions: pre-loading, loading, and post-loading. The model incorporates the loading rate (dependent on nMOT atom number and temperature), loss rates (dependent on tweezer lifetime and detuning), and the heating of the nMOT during the sweep.

Key Results

1. Extended Spatial Extent: The technique successfully loaded a 90-site rectangular array with a vertical height of >100 $\mu$m. This exceeds the height of an array loaded from a static reservoir by a factor of >3, filling a 200 $\mu$m $\times$ 100 $\mu$m field of view.
2. Controlled Atom Number Distribution: By varying the sweep speed, the authors demonstrated control over the site-to-site atom number distribution:
   • Slow sweeps: Resulted in more atoms in sites loaded later (lower in the array).
   • Fast sweeps: Resulted in more atoms in sites loaded earlier.
   • Intermediate sweeps: Achieved a roughly uniform distribution.
   • Non-linear sweeps: Enabled the creation of linear gradients, uniform arrays in reduced time (20 ms), and selectively loaded regions (e.g., skipping specific rows via frequency steps).
3. Temperature and Lifetime Characterization:
   • Temperature: The average temperature across the array was 1.49(3) $\mu$K. A minimum temperature was observed at intermediate sweep speeds, while fast and slow sweeps approached the initial nMOT temperature. The authors attribute this to the interplay between Sisyphus cooling (effective only in a narrow detuning range) and heating from the nMOT beams outside this range.
   • Lifetime: The tweezer lifetime was found to be strongly dependent on the detuning relative to the nMOT resonance. A narrow region of enhanced lifetime (attributed to Sisyphus cooling) was observed at $\delta \approx -10\Gamma$, while a broad reduction occurred near resonance due to heating.
4. Model Agreement: The rate equation model showed good qualitative agreement with experimental data, successfully predicting the trends in atom number gradients and the dependence on sweep speed. The model identified that the sign change in the atom number gradient arises from the interplay between nMOT heating (reducing loading rate) and the nMOT lifetime being longer than the tweezer lifetime under nMOT illumination.

Significance and Claims
The paper claims that painted loading provides a simple and effective toolkit for loading spatially large optical tweezer arrays without requiring complex rearrangement or optical transport.

• Scalability: The method is not limited by the static size of the nMOT, allowing for the loading of arrays significantly larger than previously possible for Sr. The authors note that related techniques have loaded lattices >1 mm in length, suggesting similar vertical extents are achievable here.
• Versatility: The technique enables the creation of non-uniform arrays (gradients) and selectively loaded regions, which is useful for studying density-dependent effects (e.g., clock shifts, spin squeezing) or preparing specific sub-arrays for quantum computing without complex sorting.
• Generalizability: While demonstrated with $^{88}\text{Sr}$, the authors assert the technique is readily extendable to other multivalent species using nMOTs (e.g., Yb, Dy, Er) and potentially to any MOT-based loading experiment with magnetic field control.
• Future Outlook: The authors suggest that combining this method with collisional loss steps could enable large single-atom arrays. They also propose that integrating optimization algorithms (e.g., machine learning) could further improve loading speed and control over atom number distributions.

The work establishes painted loading as a practical method to overcome the spatial limitations of narrow-line MOTs, offering a pathway to larger, more complex neutral atom arrays for quantum science applications.