In‑situ calibration cuts systematic localization bias to 5.3 nm for quantum‑dot nanofabrication
This paper tackles a practical problem in making quantum photonic devices: optical microscopes used to find single quantum emitters can give very precise positions but still be systematically wrong. The authors show an in‑place calibration method that measures and corrects spatial distortions in a cryogenic wide‑field photoluminescence imaging system. After correction the remaining systematic bias across the analyzed field of view is 5.3 nm, with a two‑dimensional scatter of 24.6 nm.
The team used lithographically defined gold nanodisk arrays as reference marks. They image both the markers and quantum dot emission in the same optical coordinate frame, then map the field‑dependent offsets between measured and known marker positions. Those offsets are fitted with a Zernike vector‑field model — a compact mathematical description that uses familiar optical basis functions (Zernike polynomials) to represent the smooth, position‑dependent distortions of the imaging system. The model was validated on held‑out marker patterns that were not used for calibration.
The experiments were done on quantum dots emitting near 785 nm, cooled to 4 K under a microscope objective with numerical aperture 0.82. The wide field of view on the sample was about 76.5 µm, giving an effective pixel size of about 70.8 nm. Typical statistical fitting uncertainties for a combined quantum‑dot and marker localization were small (mean 6.6 nm), showing the imaging data are precise; the calibration acts to turn that precision into accurate positions.
Why this matters: many photonic devices require placing nanostructures within a few to a few tens of nanometers of a chosen emitter. Marker‑based localization methods can achieve sub‑10 nm statistical uncertainty, but uncorrected optical distortion can ruin device registration. As a practical demonstration, the authors applied the calibrated coordinates when fabricating circular mesa structures around semiconductor quantum dots. After correction, the device set showed a 49% reduction in variance of emission polarization, which the authors interpret as evidence of improved alignment between dots and fabricated structures.