Hsu et al. (PRX Quantum 3, 030316, 2022) trapped and imaged single rubidium-87 atoms with a 0.55-numerical-aperture dielectric metalens at a 3-mm working distance, and posed the follow-up problem of bringing such flat optics to the performance envelope that bulk high-numerical-aperture objectives already provide: measured focusing efficiency of 58% versus the 78% theoretical prediction with 13% undiffracted zeroth-order light, a +300 μm chromatic focal shift between the 852-nm trapping and 780-nm imaging wavelengths, a ±11 μm field of view versus a few hundred μm for multi-element objectives, and an atom-lifetime lower bound of 10 s with unquantified loss channels. Later work has resolved several components in isolation — holographic metasurface arrays with 1.5-μm trap spacings and up to 360,000 sites, an achromatic multifunctional metalens that traps and images through one optic, and a centimeter-scale metasurface outside the vacuum cell trapping about 11,000 atoms — but no metasurface simultaneously achieves near-theoretical efficiency, achromatic dual-wavelength trap-and-image operation, objective-class field of view, vacuum-limited lifetimes beyond 10 s with quantified loss channels, and Rydberg-compatible operation at array scale. The problem remains open as an integrated optical-engineering objective for scalable neutral-atom quantum information.

Background

Arrays of optical tweezers — tightly focused far-off-resonant laser beams each confining one neutral atom — are a leading platform for quantum simulation and quantum computing, and their scale is set by the optics that generate and image the traps. Conventional experiments use bulk high-numerical-aperture (NA) multi-element objective lenses, which combine diffraction-limited focusing, achromatic operation over the trapping and imaging wavelengths, and a field of view (FoV) of a few hundred micrometers in the focal plane (about ±3 degrees of field angle), but are large, expensive, and occupy precious solid angle around the vacuum chamber. A metasurface (metalens) is a flat optic patterned with deep-subwavelength dielectric nanopillars, each acting as a small waveguide that imprints a chosen phase delay; arranging pillars with spacings below the wavelength yields a thin planar lens that can also shape polarization and multiplex several wavelengths or polarizations on one surface, functions bulk lenses do not offer. Hsu, Zhu, Thiele, Brown, Papp, Agrawal, and Regal (PRX Quantum 3, 030316, 2022) demonstrated that a 4 mm × 4 mm amorphous-silicon metalens with NA 0.55 and a 3-mm working distance can trap single rubidium-87 atoms at 852 nm and collect their 780-nm fluorescence, with trap waist 0.80 μm, a loading probability of about 52%, and a tweezer array obtained by scanning the input beam angle. Their measurements exposed a set of performance gaps. The measured focusing efficiency (58% at 852 nm, 56% at 780 nm) fell short of the 78% finite-difference-time-domain theoretical prediction, and the fraction of light transmitted undiffracted into the zeroth order was 13% versus about 6% in theory, discrepancies attributed to nonideal nanopillar cross sections and sidewall tapering from fabrication. Because a singlet metalens is dispersive by construction, the focus shifted by +300 μm between 852 nm and 780 nm (a commercial asphere used in tweezer work shifts by -40 μm), and the FoV over which the trap waist stayed within 10% (Strehl ratio — the ratio of the peak focal intensity to that of an ideal, aberration-free lens — above 0.8) was only ±11 μm, versus a few hundred μm for multi-element objectives. The measured trap lifetime without cooling light gave only a lower bound of 10 s, leaving unquantified the relative contributions of background-gas collisions at the finite vacuum level and inelastic scattering of residual trapping light, including stray and zeroth-order light associated with the patterned surface. As a path forward, the same paper designed and optically tested a polarization-multiplexed metalens (built from birefringent rectangular pillars, which have different refractive indices for orthogonal polarizations so that one surface can imprint independent phase profiles on two polarizations) that focuses 780-nm and 850-nm light of orthogonal polarization to the same diffraction-limited focus, but with efficiencies of only 42-45% and a short 0.5-mm focal length, and it identified double-layer metasurfaces, hybrid refractive-metasurface 'metaform' optics, expanded unit-cell libraries, and inverse design as routes to larger FoV and achromaticity. The overarching question raised is whether flat metasurface optics can meet the full tolerance set of single-atom trapping and imaging — tight aberration control, high efficiency for trap depth and fluorescence collection, achromatic and polarization-controlled multifunctional beams, and long atom lifetimes — at the array scale needed for neutral-atom quantum information, where high-fidelity Rydberg-state addressing (excitation of highly excited atomic states used for entangling gates, which are sensitive to stray fields and light near surfaces) and sustained ultrahigh-vacuum operation add further constraints.

Problem Statement

Develop and experimentally validate a metasurface-lens optical tweezer system that simultaneously meets the performance envelope that bulk high-numerical-aperture objectives already provide for neutral-atom experiments, at array scale. The concrete, source-faithful targets are: (1) close the fabrication-limited gap between measured and theoretically predicted focusing efficiency (58% versus 78% in the source demonstration) while suppressing undiffracted zeroth-order light (13% measured versus about 6% predicted); (2) remove the chromatic focal shift (+300 μm between the 852-nm trapping and 780-nm imaging wavelengths) and extend the field of view (±11 μm in the source, versus a few hundred μm at about ±3 degrees for multi-element objectives); (3) quantify the atom-loss channels in metasurface traps — vacuum-limited background-gas collisions versus inelastic scattering of residual trapping and stray light, including zeroth-order light from the patterned surface — and demonstrate atom lifetimes beyond the reported 10-s lower bound; and (4) validate metasurface optics in dense tweezer arrays and for high-fidelity Rydberg addressing near the optic. These deliverables share one scientific objective — establishing flat metasurface optics as a drop-in replacement for bulk high-NA objectives in scalable neutral-atom quantum-information hardware — and are coupled by the same efficiency, dispersion, and nanofabrication trade-offs, so the problem is solved only when a single metasurface-based system meets the full envelope, not when the components are met separately by different devices. Here 'a single metasurface-based system' means one integrated metasurface-based optical system that replaces the bulk high-NA objective and performs both trapping and imaging; consistent with the routes the source paper itself identifies, it may incorporate multilayer or doublet metasurfaces or hybrid refractive-metasurface (metaform) elements, and is not required to be a single patterned surface.

References

  1. T.-W. Hsu, W. Zhu, T. Thiele, M. O. Brown, S. B. Papp, A. Agrawal, and C. A. Regal, 'Single-Atom Trapping in a Metasurface-Lens Optical Tweezer', PRX Quantum 3, 030316 (2022). DOI: 10.1103/PRXQuantum.3.030316; arXiv:2110.11559. https://doi.org/10.1103/PRXQuantum.3.030316
  2. A. Holman, Y. Xu, X. Sun, J. Wu, M. Wang, Z. Zhu, B. Seo, N. Yu, and S. Will, 'Trapping of single atoms in metasurface optical tweezer arrays', Nature 649, 859-865 (2026). DOI: 10.1038/s41586-025-09961-5; arXiv:2411.05321. https://doi.org/10.1038/s41586-025-09961-5
  3. Y. Wang, Z. Zhang, T. Zhang, Y. Liao, H. Wang, Y. Tian, B. Ji, Y. Wu, L. Ma, C. Qing, C. Li, W. Zhang, Y. Huang, W. Zhang, X. Feng, W. Chen, and H. Zhai, 'Trapping 11,000 Atoms in a Tweezer Array Generated by a Single Metasurface' (2026). arXiv:2606.02715. https://arxiv.org/abs/2606.02715
  4. C. Fang, M. Kim, M. Saffman, J. T. Choy, and M. A. Kats, 'Metasurfaces for neutral-atom trapping' (review, 2026). arXiv:2605.30498. https://arxiv.org/abs/2605.30498
  5. G.-J. Chen, D. Zhao, Z.-B. Wang, Z. Li, J.-Z. Zhang, L. Chen, Y.-L. Zhang, X.-B. Xu, A.-P. Liu, C.-H. Dong, G.-C. Guo, K. Huang, and C.-L. Zou, 'Multifunctional Metalens for Trapping and Characterizing Single Atoms', Laser & Photonics Reviews 19(6) (2024). DOI: 10.1002/lpor.202401595; arXiv:2411.05501. https://doi.org/10.1002/lpor.202401595
  6. D. Li, Q. Liao, B. Xu, T. Zentgraf, E. Narvaez Castaneda, Y. Zhou, K. Qin, Z. Xu, H. Shen, and L. Huang, 'In vacuum metasurface for optical microtrap array' (2024). arXiv:2407.05918. https://arxiv.org/abs/2407.05918
  7. C. Fang, M. Kim, H. Mei, X. Yang, Z. Yu, Y. Xiao, S. Deshpande, P. Huft, A. M. Dibos, D. A. Czaplewski, M. Saffman, J. T. Choy, and M. A. Kats, 'Silicon-on-sapphire metasurfaces generate arrays of dark and bright traps for neutral atoms' (2026). arXiv:2601.01038. https://arxiv.org/abs/2601.01038
  8. Y. Wang, Y. Liao, T. Zhang, Y. Tian, Y. Wu, W. Zhang, W. Zhang, Y. Huang, H. Zhai, W. Chen, X. Feng, and Z. Zhang, 'Direct Generation of an Array with 78400 Optical Tweezers Using a Single Metasurface', Chinese Physics Letters 43, 010606 (2025). DOI: 10.1088/0256-307X/43/1/010606; arXiv:2512.08222. https://doi.org/10.1088/0256-307X/43/1/010606
  9. C. Fang, J. Miles, J. Goldwin, M. Lichtman, M. Gillette, M. Bergdolt, S. Deshpande, S. A. Norrell, P. Huft, M. A. Kats, and M. Saffman, 'Interleaved dual-species arrays of single atoms using a passive optical element and one trapping laser', Science Advances 11(29) (2025). DOI: 10.1126/sciadv.adw4166; arXiv:2412.20261. https://doi.org/10.1126/sciadv.adw4166