Reaching the highest efficiency of spin Hall effect of light in the near-infrared using all-dielectric metasurfaces

Abstract

The spin Hall effect of light refers to a spin-dependent transverse splitting of light at a planar interface. Previous demonstrations to enhance the splitting have suffered from exceedingly low efficiency. Achievements of the large splitting with high efficiency have been reported in the microwave, but those in the optical regime remain elusive. Here, an approach to attain the large splitting with high efficiency in the near-infrared is proposed and experimentally demonstrated at 800 nm by using a dielectric metasurface. Modulation of the complex transmission of the metasurface leads to the shifts that reach 10λ along with efficiencies over 70% under two linear polarizations. Our work extends the recent attempts to achieve the large and efficient spin Hall effect of light, which have been limited only to the microwave, to the optical regime.

Fig. 1. Schematic of the spin Hall effect of light (SHEL) and a dielectric metasurface to enhance the shift δ and efficiency ϵ of the SHEL simultaneously.

a, b Illustration of the SHEL in previous demonstrations. The width of the transmitted beams denotes their intensities. a The efficient SHEL is limited by the small shift, b whereas the enhancement of the shift entails a low efficiency. c Large SHEL with high efficiency. d A metasurface that transmits x- and y-polarized incidence with unity amplitudes (∣t_x∣ = ∣t_y∣ = 1) but with the opposite phase (arg(t_x) − arg(t_y) = ± π).

Fig. 2. Design of a dielectric metasurface.

a Schematic of a unit cell. Geometric parameters are given as: height H = 400 nm, length a = 265 nm, width b = 158 nm, periodicities p_x = 378 nm and p_y = 263 nm, and taper angle α. b, c Scanning electron microscopy images of the fabricated sample. b Top view, c perspective view. Inset shows a magnified image.

Fig. 3. Transmittance of the dielectric metasurface under normal incidence.

a Simulated (markers) and measured (solid lines) transmittance under normal incidence. b Phase difference ϕ between the two polarizations obtained by numerical simulation.

Fig. 4. Measurement of the SHEL at 800 nm.

a Schematic of Stokes polarimetry set-up. HWP: half-wave plate, L: lens, QWP: quarter-wave plate, LP: linear polarizer, CCD: charge-coupled device camera. b Phase difference between t_s and t_p, c ${\delta }_H^{+}/\lambda$, d ${\delta }_V^{+}/\lambda$, e ϵ_H, and f ϵ_V. Curve: simulated, markers: measured. Insets in c, d show the magnified view for near-normal incidence where the analytic formula (solid) given by Eq. (1) deviates from the exact shifts (dashed) due to the breakdown of the assumption, i.e., ${\left(2\pi w/\lambda \right)}^2\gg {\cot }^2{\theta }_i$. Error bars of the measured data are not shown because they are all smaller than the markers.