High precision wavefront control in point spread function engineering for single emitter localization

TL;DR

This paper presents a calibration protocol for fluorescence microscopes with SLMs that achieves wavefront control below 20 mλ, enabling highly precise 3D and spectral emitter localization with sub-10 nm accuracy.

Contribution

The authors develop a calibration and alignment method for SLM-equipped microscopes that significantly reduces wavefront aberrations, improving PSF engineering precision for single emitter localization.

Findings

Wavefront error below 20 mλ achieved after calibration.

Localization precision below 10 nm in x, y, and wavelength.

Residual wavefront error comparable to SLM calibration error.

Abstract

Point Spread Function (PSF) engineering is used in single emitter localization to measure the emitter position in 3D and possibly other parameters such as the emission color or dipole orientation as well. Advanced PSF models such as spline fits to experimental PSFs or the vectorial PSF model can be used in the corresponding localization algorithms in order to model the intricate spot shape and deformations correctly. The complexity of the optical architecture and fit model makes PSF engineering approaches particularly sensitive to optical aberrations. Here, we present a calibration and alignment protocol for fluorescence microscopes equipped with a spatial light modulator (SLM) with the goal of establishing a wavefront error well below the diffraction limit for optimum application of complex engineered PSFs. We achieve high-precision wavefront control, to a level below 20 m$\lambda$ wavefront aberration over a 30 minute time window after the calibration procedure, using a separate light path for calibrating the pixel-to-pixel variations of the SLM, and alignment of the SLM with respect to the optical axis and Fourier plane within 3 $\mu$m ($x/y$) and 100 $\mu$m ($z$) error. Aberrations are retrieved from a fit of the vectorial PSF model to a bead $z$-stack and compensated with a residual wavefront error comparable to the error of the SLM calibration step. This well-calibrated and corrected setup makes it possible to create complex `3D+$\lambda$' PSFs that fit very well to the vectorial PSF model. Proof-of-principle bead experiments show precisions below 10~nm in $x$, $y$, and $\lambda$, and below 20~nm in $z$ over an axial range of 1 $\mu$m with 2000 signal photons and 12 background photons.