Polarization-insensitive GaN metalenses at visible wavelengths

Abstract

The growth of wide-bandgap materials on patterned substrates has revolutionized the means with which we can improve the light output power of gallium nitride (GaN) light-emitting diodes (LEDs). Conventional patterned structure inspection usually relies on an expensive vacuum-system-required scanning electron microscope (SEM) or optical microscope (OM) with bulky objectives. On the other hand, ultra-thin metasurfaces have been widely used in widespread applications, especially for converging lenses. In this study, we propose newly developed, highly efficient hexagon-resonated elements (HREs) combined with gingerly selected subwavelength periods of the elements for the construction of polarization-insensitive metalenses of high performance. Also, the well-developed fabrication techniques have been employed to realize the high-aspect-ratio metalenses working at three distinct wavelengths of 405, 532, and 633 nm with respective diffraction-limited focusing efficiencies of 93%, 86%, and 92%. The 1951 United States Air Force (USAF) test chart has been chosen to characterize the imaging capability. All of the images formed by the 405-nm-designed metalens show exceptional clear line features, and the smallest resolvable features are lines with widths of 870 nm. To perform the inspection capacity for patterned substrates, for the proof of concept, a commercially available patterned sapphire substrate (PSS) for the growth of the GaN LEDs has been opted and carefully examined by the high-resolution SEM system. With the appropriately chosen metalenses at the desired wavelength, the summits of structures in the PSS can be clearly observed in the images. The PSS imaging qualities taken by the ultra-thin and light-weight metalenses with a numerical aperture (NA) of 0.3 are comparable to those seen by an objective with the NA of 0.4. This work can pioneer semiconductor manufacturing to choose the polarization-insensitive GaN metalenses to inspect the patterned structures instead of using the SEM or the bulky and heavy conventional objectives.