Silicon-on-sapphire metasurfaces generate arrays of dark and bright traps for neutral atoms

TL;DR

The work presents CMOS-compatible crystalline silicon-on-sapphire metasurfaces that transform a single Gaussian beam into large arrays of optical traps, including dark bottle-beam traps and interleaved bright/dark configurations, addressing scalability and noise concerns of active tweezer systems. A modified Gerchberg-Saxton algorithm is used to control complex amplitude in the focal plane, enabling 3D bottle-beam generation and uniform trap arrays. Experimentally, 7×7 dark-trap, 21×21 bright-trap, and interleaved trap arrays are demonstrated with strong agreement to simulations, and the platform supports both outside- and inside-vacuum implementations while maintaining CMOS compatibility. The passive, compact metasurfaces offer scalable, low-noise atom-trapping capabilities suitable for large neutral-atom registers and integrated quantum devices. These metasurfaces thus provide a practical path toward high-fidelity, large-scale neutral-atom quantum technologies.

Abstract

We demonstrated crystalline silicon-on-sapphire (c-SOS) metasurfaces that convert a Gaussian beam into arrays of complex optical traps, including arrays of optical bottle beams that trap atoms in dark regions interleaved with bright tweezer arrays. The high refractive index and indirect band gap of crystalline silicon makes it possible to design high-resolution near-infrared ($λ>700$ nm) metasurfaces that can be manufactured at scale using CMOS-compatible processes. Compared with active components like spatial light modulators (SLMs) that have become widely used to generate trap arrays, metasurfaces provide an indefinitely scalable number of pixels, enabling large arrays of complex traps in a very small form factor, as well as reduced dynamic noise. To design metasurfaces that can generate three-dimensional bottle beams to serve as dark traps, we modified the Gerchberg-Saxton algorithm to enforce complex-amplitude profiles at the focal plane of the metasurface and to optimize the uniformity of the traps across the array. We fabricated and measured c-SOS metasurfaces that convert a Gaussian laser beam into arrays of bright traps, dark traps, and interleaved bright/dark traps.

Figures (14)

• Figure 1: Two schemes for positioning a trap-forming metasurface with respect to the vacuum cell. (A) Atom-trapping setup with the metasurface placed outside of the vacuum cell. The dark traps (shown in blue) trap the atoms at the intensity minima when the laser frequency is blue-detuned from the atomic resonance, whereas bright traps (shown in red) trap the atoms at the intensity maxima when the laser is red-detuned. Relay optics re-image the intensity profiles into the vacuum cell. (B) Schematic of the atom trapping setup with the metasurface integrated inside the vacuum cell. The color convention is the same as (A).
• Figure 2: Modified Gerchberg-Saxton (G-S) algorithm for generating the bottle-beam array (A) Flowchart of the modified G-S algorithm. Each iteration goes through four steps: (1) forward propagation from the metasurface to the image plane; (2) enforcement of both the target amplitude and phase profiles in the fully constrained domain, while only the phase profile is enforced in the partially-constrained domain; (3) backward propagation from the image plane to the metasurface; (4) enforcement of the amplitude profile based on the input (usually Gaussian) beam. When the target amplitude $A'(x,y)$ converges, the resulting metasurface phase $\phi(x,y)$ can be implemented in hardware. (B) The simulated intensity profiles in the image plane forming an array of bottle-beam traps, generated using the algorithm in (A). The inset shows a magnified view of several bottle-beam traps. (C) The same profile as (B), but saturated to better show energy leaking into the partially constrained region.
• Figure 3: Realization of the designed phase profile using silicon-on-sapphire (c-SOS) metasurfaces. (A) Simulated transmittance and phase of a single unit cell as a function of the silicon cylinder radius. The height of the cylinder is 500 nm, with a period of 360 nm, on a sapphire substrate. The free-space wavelength of the incident beam in this simulation is $770~\mathrm{nm}$. The transmittance here assumes no reflection at the air-sapphire interface. (B) The fabrication processes of the silicon-on-sapphire metasurface. An SiO$_2$ hard mask is first deposited on the silicon-on-sapphire substrate. The patterns are defined by electron-beam lithography and transferred to a hard mask via SiO$_2$ etching. Then silicon etching creates the silicon pillars, which remain covered with the SiO$_2$ mask. Since the refractive index of the amorphous SiO$_2$ is low, the residual SiO$_2$ disk has negligible influence on the performance of the metasurface. (C) SEM image of the metasurface on a tilted stage. (D) Photo of the fabricated metasurfaces mounted on a sample holder.
• Figure 4: Generation and measurement of a single dark trap. (A) Experimental setup for imaging the intensity profile after the metasurface. A Gaussian beam at $780~\mathrm{nm}$ is launched from a single-mode fiber with a collimator and expanded to the desired beam waist. The metasurface is mounted on a motorized XYZ stage combined with a tip-tilt stage. The intensity profiles of the optical traps are acquired with a microscope, while scanning the metasurface along the Z-direction. (B) Experimental (top row) and simulated (bottom row) intensity profiles. Left plots show the XY intensity profiles in the focal plane at $Z = 350~\mu\mathrm{m}$, and right plots show the YZ intensity profiles at $X = 0~\mu\mathrm{m}$. All profiles are normalized to the maximum intensity of the corresponding XY intensity profile.
• Figure 5: Experimental and simulated intensity distributions for dark, bright, and dual-species trap arrays. Experiments (top row) and simulations (bottom row). The left plots of each panel show the XY intensity profiles in the focal plane, at $Z = 350~\mu\mathrm{m}$. The right plots of each panel show the YZ intensity profiles across trap sites. In panels (A) and (B), the profiles are along $X = 0~\mu\mathrm{m}$; for panel (C), the profiles are along $X = 0~\mu\mathrm{m}$ (bright traps) and $X = 3.5~\mu\mathrm{m}$ (dark traps). All profiles are normalized to the maximum intensity of the corresponding XY intensity profile, with line profiles shown in Fig. \ref{['fig:si6']}. (A) 7$\times$7 dark-trap array (B) 21$\times$21 bright-trap array (C) 7$\times$7 bright-trap array interleaved with 6$\times$6 dark-trap array.