A single optical waveguide was reported to act as a nanoscale lens for detecting light from individual atoms. In a laboratory experiment, the probability of detecting a single-atom fluorescence signal at zero transport displacement was 20% in the nanolensing condition, which the authors reported as four times greater than without nanolensing.
The work combined electromagnetic-field simulations with measurements of cold 87 Rb atoms transported near a GaN waveguide in a vacuum cell. The collection path crossing the waveguide was compared with a path positioned 7.5 µm away, where the study described the light as unmodulated by nanolensing.
The findings are a laboratory proof of principle for shaping the collection field close to an integrated photonic structure. They concern fluorescence collection and signal detection in this particular setup, rather than a demonstrated system for trapping atoms or enhancing atom–photon coupling.
A focus just above the chip
The proposed lens was tested first in a numerical model. The calculation used the angular spectrum method, which follows how an optical field propagates from its spatial pattern. Its parameters included a 700 nm waveguide width, a 200 nm thickness, a refractive index of 2.36 and a wavelength of 780 nm.
The model produced a Gaussian-like focus with a 280 nm beam waist about 250 nm above the pixel. It also predicted an input-intensity magnification of approximately 2.5-fold.
Those two figures describe the modeled light field, not the later detection statistic. The 2.5-fold value refers to the simulated intensity at the focus, while the reported 20% and four-times comparison refer to measured single-atom signal detection under a particular transport condition.
For the laboratory test, the system prepared cold 87 Rb atoms in a magneto-optic trap about 550 µm from the chip. An optical conveyor belt then transported the atoms toward the waveguide, allowing the collection path to be examined as the atoms moved.
The researchers recorded background-subtracted fluorescence during the first 10 ms of each measurement and used those counts to form a ratio for comparing the collection conditions.
What the experiment measured
The counts-ratio curve fell faster when the collection path crossed the waveguide than when the path was 7.5 µm away. That difference was the direct experimental signature used to examine whether the waveguide changed the spatial profile of the collected fluorescence.
The team fitted the curves with a Rayleigh length, a measure of how quickly a focused beam spreads along its direction of travel. The reported fit was 6.76 ± 10.48 µm with the nanolens and 17.97 ± 2.32 µm without it. The first estimate carries a large uncertainty, which makes the fitted comparison less precise than the central values alone suggest.
The analysis then looked for individual-atom events. Signals were identified using threshold criteria for the step height, the post-step baseline and the pre-step count level. The detection-probability analysis used 600 measurements and reported binomial standard errors.
At a transport displacement of 0 µm, the reported probability of detecting a single-atom signal with nanolensing was 20%. The paper described that result as four times greater than the probability without nanolensing. The supplied analysis gives the relative comparison but does not provide absolute uncertainty intervals for both conditions.
A narrow zone for the effect
The reported signals were concentrated close to the waveguide surface. Across transport displacements from 0 µm to 2.13 µm, 50% of single-atom signals were reported to occur within about 2 µm of the surface.
The supplemental characterization also found nanolensing only when the alignment error between the collection path and the waveguide was less than 2 µm. That makes positioning a central part of the reported measurement, rather than a minor experimental detail.
Repeated measurements estimated that the collection and auxiliary beam centers were less than 0.3 µm apart at a 95.6% confidence level. This indicates that the experiment could assess alignment with submicrometre precision, while not showing that the nanolensing effect is insensitive to misalignment.
What remains untested
The authors interpret the observations as evidence that an on-chip waveguide can serve as a nanoscale lens, localize the collection field and raise the efficiency of single-atom fluorescence detection. They also suggest that the approach could eventually assist atom trapping and atom–photon coupling, but those possibilities were not directly demonstrated in the reported experiment.
The evidence is limited to one GaN waveguide geometry and a specific arrangement of cold 87 Rb atoms and optical collection hardware. It does not establish that the same performance would hold across other nanostructure designs, materials, wavelengths, atom species or experimental platforms.
The comparison was not randomized, and the exact number of atoms and their allocation across the two collection conditions were not reported. The work therefore describes what happened in a defined laboratory arrangement, rather than providing a causal effect estimate for a broader population.
The authors attributed the higher detection probability to collection efficiency because the fluorescence counts during the first 10 ms were described as almost the same. The underlying numerical comparison for that explanation was not provided in the supplied analysis, leaving the interpretation less complete than the headline detection result.
Further tests would need to examine how reproducible the enhancement is across waveguide dimensions, materials, wavelengths and alignment conditions. They would also need clearer absolute detection probabilities and uncertainty intervals, along with experiments using independently controlled atom positions and larger samples.
A preprint-stage result
The manuscript is an arXiv preprint, version 1. It reports support from the National Key R&D Program, the National Natural Science Foundation of China, the Fundamental Research Funds for the Central Universities and USTC Research Funds of the Double First-Class Initiative.
The supplied document also includes supplemental material with an appendix on the numerical simulation of the nanolensing effect. That material adds detail to the model, but the central evidence remains the combination of the simulated focus, the measured counts-ratio change and the single-atom detection analysis.
Paper data and sources
Original title: Lensing and enhanced single atom detection via a single-pixel nanostructure
Authors: Ling-Xiao Wang, Lei Xu, Ai-Ping Liu et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-20
DOI: Not available
Original paper · Full text