Visible Meta‐Displays for Anti‐Counterfeiting with Printable Dielectric Metasurfaces

Abstract Metasurfaces, 2D arrays of nanostructures, have gained significant attention in recent years due to their ability to manipulate light at the subwavelength scale. Their diverse applications range from advanced optical devices to sensing and imaging technologies. However, the mass production of dielectric metasurfaces with tailored properties for visible light has remained a challenge. Therefore, the demand for efficient and cost‐effective fabrication methods for metasurfaces has driven the continuing development of various techniques. In this research article, a high‐throughput production method is presented for multifunctional dielectric metasurfaces in the visible light range using one‐step high‐index TiO 2‐polymer composite (TPC) printing, which is a variant of nanoprinting lithography (NIL) for the direct replication of patterned multifunctional dielectric metasurfaces using a TPC material as the printing ink. The batch fabrication of dielectric metasurfaces is demonstrated with controlled geometry and excellent optical response, enabling high‐performance light‐matter interactions for potential applications of visible meta‐displays.


SI-1. Process flowchart for high-throughput fabrication of meta-displays
In view of the production cost, the second-generation soft mold and the final meta-display replicas are fabricated in batch mode using low-cost high-throughput ultra-violet nanoimprinting lithography (UV-NIL).Specifically, the whole production process can be categorized into three distinct steps: first, the creation of the silicon (Si) master mold, also known as the 1 st -generation rigid template; second, the replication of the soft polystamp (PS) mold, referred to as the 2 nd -generation soft template; and third, the production of the final metadisplay items, as depicted in Figure S1.The Si master mold fabrication procedure for crafting the meta-display duplicates involves the utilization of electron beam writing (Elionix F125) to generate a pattern that is an exact match to the target TPC meta-display.To initiate this process, a layer of photoresist (AR-6200) is applied to the silicon wafer, with an inverse pattern being transferred to the photoresist through electron beam exposure.Following this, a 50 nm thick chromium (Cr) layer is applied to the sample via the electron beam evaporation system (ULVAC ei-5z).Then, a lift-off process is carried out in acetone to define a Cr hard mask on the sample.Subsequently, the sample undergoes etching along the Cr mask using inductivelycoupled plasma reactive ion etching (LEUVEN INSTRUMENTS).The Si master mold is considered complete after the removal of any residue from the Cr hard mask, which is achieved by employing a Cr etchant (nitrate wet etching).
In the next phase, liquid photosensitive PS resin is evenly spread over the Si master mold, and a PET substrate is placed on top of the coated PS film.The synthetic soft mold, composed of the PET substrate and the PS structures, is then solidified under ultraviolet light irradiation while maintaining a contact pressure of 5000 Pa.Subsequently, the soft mold and the Si template are carefully separated at a controlled, gradual pace.Finally, the high-throughput UV-NIL manufacturing process for TPC-based meta-display replicas is automatically executed using a commercial NIL equipment (GL8/12 CLIV Gen2, GermanLitho GmbH).The scanning electron microscope (SEM) images and the corresponding inset optical microscope images amplified from the process flowchart depict the first-generation Si rigid template, the secondgeneration PS soft mold, and the desired meta-display replica (i.e., final product), respectively.

SI-2. The additional three-channel meta-display sample 3 (MDS3)
To assess the imaging performance of MDS3, our evaluation process commences with the capture of the nanoprinting image of the 1 st replica of MDS3, as illustrated in Figure S2(j), utilizing an optical polarizing microscope (Caikon XP-550c).MDS3 is seamlessly integrated into the optical pathway of the microscope for orthogonal polarization, accompanied by a white light source from a halogen lamp.Employing a 10× objective and a micro digital camera (Model: CKC2000), we document the light intensity distribution near the MDS3 surface in transmission mode.The results distinctly reveal an image of Isaac Newton's picture, confirming MDS3's effective performance in amplitude modulation in the near field.The presence of minor noise speckles can be attributed to the near-field interaction between adjacent nanofins, a concern that can be alleviated through the implementation of a more intricate super-cell design.
Furthermore, both the 10 th replica and the 20 th replica exhibit similar near-field nanoprinting imaging behavior, as depicted in Figure S2(k, l).It is worth noting that the experimental observations for all three replicas (replica 1, replica 10, and replica 20) are consistent with the simulation results for near-field greyscale nanoprinting, as presented in Figure S2(i).
Subsequently, we employ an alternative optical system to observe the resulting far-field holographic images of these three replicas (replica 1, replica 10, and replica 20), showcased in Figure S2(b-d, f-h).For illuminating MDS3, we employ a tunable super-continuum laser source (NKT-SuperK EXTREME) operating at a specific wavelength of 532 nm.To control the polarization state of the incident beam and achieve circular polarization, we utilize a polarizer and a quarter waveplate (QWP).By rotating the QWP by 90°, we can switch between two independent information channels, leading to the appearance of two distinct far-field holographic images with different QWP configurations.
As presented in Figure S2, the numerous replicas (in this work: replica 1, replica 10, and replica 20) of MDS3 exhibit near-field nanoprinting images that are exceedingly similar and nearly indistinguishable from each other.Moreover, the far-field holographic images they produce are almost identical as well.Notably, when we compare the near-field nanoprinting images of the Isaac Newton portrait on the MDS3 replicas to those of the Niels Bohr portrait, they demonstrate nearly identical grayscale imaging performance.However, it is important to acknowledge that due to the spatial separation of the designed holographic images and the restricted selection of eight unit cell structures (i.e., meta-atoms) for independent control of leftcircularly polarized (LCP) and right-circularly polarized (RCP) light incidence, our experiments reveal that when LCP light is incident, the resulting holographic image includes some slight noise from the RCP-generated holographic image, and vice versa.This spatial separation of holographic images, with the LCP pattern occupying one half and the RCP pattern occupying the other half, implies that half of the meta-atoms for phase control are unavailable, leading to the introduction of unwanted noise from the ineffective half of the unit structures.It is worth noting that these experimental results are consistent with the outcomes of simulations based on vector diffraction theory, as the accompanying holographic pattern (i.e., noise) stemming from the opposite circular polarization can be theoretically predicted.Additionally, it is important to mention that the holographic images we generate are visually clear to the naked eye and enlarge in size as the observation distance increases (i.e., the holographic images diverge).To capture these images, we utilize a white paper as a screen positioned 30 cm away from the sample, and then take photographs using an ordinary smartphone commonly used in daily life.In general, the recorded LCP and RCP holograms closely correspond to the simulated LCP and RCP holographic images, as depicted in Figure S2(a) and S2(e), respectively.

SI-3. Optical setup for characterizing 2-channel and 3-channel meta-display samples
The experimental setup used to observe the far-field holographic images of the three replicas (replica 1, replica 10, and replica 20) corresponding to the meta-display sample 1 (MDS1) is shown in Figure S3(a).To illuminate MDS1, we utilize a tunable super-continuum laser source (NKT-SuperK EXTREME) with a specific wavelength of 532 nm.In order to manage the polarization state of the incoming beam and achieve circular polarization, we employ both a linear polarizer (LP) and a quarter waveplate (QWP).A 90° rotation of the QWP enables us to alternate between two separate information channels, resulting in the emergence of two discernible far-field holographic images characterized by distinct QWP settings.
To evaluate the imaging capabilities of the massively produced replicas of three-channel meta-display samples (MDS2 and MDS3), our assessment process begins by capturing the nearfield nanoprinting image of the first replica of MDS2, as depicted in the middle section of

SI-4. The operational principle for achieving high-efficiency broadband performance
To scrutinize the underlying mechanism responsible for the high-efficiency broadband performance of meta-display samples, we conduct simulations to evaluate the amplitude (i.e., polarization conversion ratio, PCR) and propagation phase for incident light at various wavelengths, specifically 473 nm (Blue), 532 nm (Green), and 633 nm (Red).The corresponding results can be observed in Figure S4.Notably, the particular data values of PCR and propagation phase for 473 nm (Blue), 532 nm (Green), and 633 nm (Red) wavelengths are displayed in Table S1, Table S2, and Table S3, respectively.
It is evident that the PCR remains consistently above 0.6, although it diminishes as the operating wavelength deviates from the designated 532 nm in our work.Regarding the propagation phase, specifically the phase component of the cross-polarization, it continues to span the complete range from 0 to 2π when the illuminating light's wavelength deviates from the designated 532 nm.In our holographic image simulations, we consider not only the propagation phase but also the geometric phase (wavelength-independent), as well as the amplitude of the cross-polarization conversion (i.e., PCR).As shown in Figure S4(a), due to the fact that green light (532 nm) has the highest PCR, followed by blue light (473 nm) and then red light (633 nm), we can predict that designed green-light holographic images will be the brightest, blue-light holographic images will be somewhat less bright, and red-light holographic images will be the darkest.This predication can also be experimentally verified as observed from Figure 6 in the main text.Generally, Figure 6 depicts the broadband response measurement results, demonstrating that holographic images for both red light and blue light exhibit decent fidelity.Finally, it is worth noting that the PCR fluctuation among the eight metaatoms becomes more pronounced as the wavelength deviation increases, potentially leading to increased noise in the near-field nanoprinting image and far-field holographic images.Table S1.Simulation results of PCR and propagation phase for 633 nm light incidence Table S2.Simulation results of PCR and propagation phase for 532 nm light incidence Table S3.Simulation results of PCR and propagation phase for 473 nm light incidence 11

SI-5. Photos of the corresponding replicas used in the experimental measurements
As shown in Figure S5, the replication process can be executed using an array-based pattern transfer method, which can save both time and space costs.From the massively produced replicas (replica 1, replica 10, and replica 20), we can clearly see that the replicas perform well at the macro scale, with no visible defects to the naked eye, and each replica looks nearly identical.Table S4.Simulation results of PCR and propagation phase for 532 nm light incidence (3D trapezoidal shape with Dspread = 20 nm) 14

SI-7. Elimination of unmodulated light via an additional optical filtering system
In fact, the elimination of holographic image noise can be achieved effectively.As depicted in Figure S7, a viable approach involves filtering the unmodulated co-polarization light through an additional optical system (QWP+LP), resulting in a distinct holographic image.A convex lens is employed to reduce the filtered clear holographic image for capture by a CMOS camera (Prosilica GT2050NIR, Allied Vision).It is noteworthy that the edge distortion (barrel distortion) in the full-frame holographic image (e.g., cartoon Panda, characters "IOE 2023") is attributed to the imaging characteristics of the convex lens and the presence of a specific flange distance between the convex lens and the CMOS camera.Table S5.Measurement results of PCR efficiency for 1 st , 10 th , 20 th replicas from MDS1 under RCP light incidence

Figure S1 .
Figure S1.Schematic of the high-throughput fabrication process.a) SEM image of the firstgeneration rigid template (namely, Si master mold).Inset: the corresponding optical microscope image.b) SEM image of the second-generation flexible template (namely, PS soft mold).Inset: the corresponding optical microscope image.c) SEM image of the final meta-display replica.Inset: the corresponding optical microscope image.The thick white scale bars in SEM images denote a length of 1 μm, and the thin white scale bars in the inset optical microscope images indicate a length of 200 μm.

Figure S2 .
Figure S2.Results of three-channel meta-display sample 3 (MSD3).a) Simulated far-field holographic image (the target image is a sketch of Christmas tree) under LCP light incidence.b-d) Experimentally observed far-field holographic images of the 1 st , the 10 th , and the 20 th replicas, respectively, under LCP light incidence.e) Simulated far-field holographic image (the target image is a sketch of Peter rabbit) under RCP light incidence.f-h) Experimentally observed far-field holographic images of the 1 st , the 10 th , and the 20 th replicas, respectively, under RCP light incidence.i) Simulated near-field greyscale nanoprinting image (the target image is a portrait of Isaac Newton) when inserted in an orthogonal-polarization optical pathway.b-d) Experimentally observed greyscale nanoprinting images of the 1 st , the 10 th , and the 20 th replicas, respectively, when inserted in an orthogonal-polarization optical pathway.Note: The portrait of Isaac Newton is cropped from Charles Jervas' painting of Isaac Newton (1642-1727).

Figure
Figure S3(b).As for the experimental observations of far-field holographic images, the corresponding optical system for LCP (RCP) light incidence is shown in the top (bottom) section of Figure S3(b).

Figure S3 .
Figure S3.Schematic illustrations of the optical setups for characterizing the scalably manufactured replicas of 2-channel and 3-channel meta-display samples.a) Schematic diagram of the experimental setup for evaluating the imaging capabilities of replicas from the twochannel meta-display sample 1 (MDS1).b) Schematic diagram of the optical system for evaluating the imaging capabilities of replicas from the three-channel meta-display samples (MDS2 and MDS3).

Figure S4 .
Figure S4.Simulation results of PCR and propagation phase for R (633 nm), G (532 nm), and B (473 nm) light illuminations.a) Simulated PCR results of eight meta-atoms for RGB light illuminations.b) Simulated propagation phase results of eight meta-atoms under the incidence of RGB light sources.

Figure S5 .
Figure S5.Photos of the 1 st replica, the 10 th replica, and 20 th replica of the designed metadisplay samples (MDS1, MDS2 and MDS3).The MDS1 pattern is situated at the top, MDS2 pattern is located on the left side of the middle layer, and MDS3 pattern is located on the right side of the middle layer.Note: The three remaining larger patterns in the bottom layer are designed and fabricated for other optical testing purposes.

Figure S6 .
Figure S6.Simulation of produced 3D trapezoidal TPC meta-atoms.a) Perspective view of the established 3D trapezoidal TPC meta-atom model in the FDTD simulation.b) Side view of the modeled TPC meta-atom.c) Front view of the modeled TPC meta-atom.Note: Dspread is a structural parameter to indicate the lateral spread of the bottom structure of the 3D trapezoidal shape.Based on SEM characterization results in the main text, Dspread is set as 20 nm.

Figure S7 .
Figure S7.Schematic illustrations of the optical setups for obtaining distinct holographic images.a) Schematic diagram of the experimental setup for characterizing replicas from the two-channel meta-display sample 1 (MDS1).b) Schematic diagram of the optical system for characterizing replicas from the three-channel meta-display samples (MDS2 and MDS3).

Figure S8 .
Figure S8.Schematics for PCR measurements.a) Schematic diagram of the experimental setup for measuring the cross-polarized beam after the optical filter system.b) Schematic diagram of the optical setup for measuring the input beam after the pinhole.c) Size of the pinhole and size of the meta-display sample.