Cps532

The experimental setup for the MDNN characterization is presented in Supplementary Fig. S10. A laser diode emitting at 532 nm (Thorlabs CPS532) was utilized as the input light. A linear polarizer was used to create the desired polarizations. The light is then directed onto the metasurface and imaged on a CMOS camera DCC3260C through a 100× objective lens. Videos of the MDNN focusing effect with diffraction distance are obtained by the movement of a stepper motor. Since the metasurface is integrated onto the CMOS imaging sensor, the output images of the experiment in Fig. 5 are collected directly by the image sensor on the CMOS chip (Sony IMX686).

The experimental setup is illustrated in Fig. 2. The light source was a collimated diode-pumped solid-state laser module (532 nm, 4.5 mW, CPS532, Thorlabs, New Jersey, USA). After beam expansion through two achromatic lenses (AC254-030-A-ML; AC254-075-A-ML, Thorlabs, New Jersey, USA), the laser light was spatially modulated using a DMD (DLP7000, Texas Instruments, Texas, USA) and then projected onto the proximal facet of a MMF (105 μm, 0.22 NA, 1 m, M43L01, Thorlabs, New Jersey, USA) via a tube lens (AC254-050-A-ML, Thorlabs, New Jersey, USA) and an objective (20×, 0.4 NA, RMS20X, Thorlabs, New Jersey, USA). The light illuminated area on the DMD included 32 × 32 independent input modes, with each 2 × 2 micromirrors grouped as one mode. An objective (20×, 0.4 NA, RMS20X, Thorlabs, New Jersey, USA) and a tube lens (AC254-0100-A-ML, Thorlabs, New Jersey, USA) were used to magnify the output light beam before it was captured by a complementary metal-oxide-semiconductor (CMOS) camera (C11440-22CU01, Hamamatsu Photonics, Shizuoka, Japan) with a frame rate of 200 frames per second (fps) for MMF characterisation.

For the optical characterisation of the microprints of the 3D-integrated metasurfaces, we used an optical microscope (Carl-Zeiss AXIO-10) with ×5 (0.13NA), ×10 (0.25NA), ×50 (0.75NA), and ×100 (0.85NA) objective lenses. The different magnifications of the devices at each process step were captured with an incoherent white-light source under transmission mode. The microprint display performance was optimised by modifying the exposure time and contrast of the CCD. Then, the fabricated integrated devices were characterised for hologram projection by the experimental setup shown in Fig. S7. The three laser diodes emitting at 450 nm (Thorlabs CPS450), 532 nm (Thorlabs CPS532), and 633 nm (Thorlabs CPS633S) were exploited to generate R, G, and B channels. The polarizers were used to manipulate the source power to achieve a suitable RGB component ratio due to the polarisation-independent property of the proposed metasurfaces. Then, two dichroic lenses (DMLP567T and DMLP490T) were used to combine the three lasers. The hologram images were projected on a screen and captured by an SLR camera. To characterise the diffraction efficiency for each wavelength, we put an optical power metre in position 1 and position 2 as Fig. S7b shows to measure the transmission power and the diffraction power of the hologram image.