Metasurface-enabled three-in-one nanoprints by multifunctional manipulations of light

Summary In metasurface-based ultra-compact image display, color-nanoprints, gray-imaging elements, and binary-pattern-imaging elements are three different types of nanoprints, implemented with different mechanisms of light manipulation. Here, we show the three functional elements can be integrated together to form a “three-in-one” nanoprint with negligible crosstalk, merely with a single-cell nanostructured design approach. Specifically, by decoupling spectrum and polarization-assisted intensity manipulations of incident light, the proposed metasurface appears as a dual-color nanoprint under a broadband unpolarized light source illumination, while simultaneously displaying an independent continuous gray image and another binary-pattern in an orthogonal-polarization optical setup with different polarization controls. Our approach can increase the system integration and security of metasurfaces, which can be of interest to many advanced applications such as data storage, optical information encoding, high-end optical anti-counterfeiting, and optical information hiding.

Optical pattern can be encoded not only into color profiles but also into spatially varying intensity, i.e. grayscale modulation. Inspired by the Malus law, researchers have proposed the polarization-controllable image display technique, with which one can utilize nanostructures acting as half-wave plates (Yue et al., 2018;Shan et al., 2020) or polarizers (Li et al., 2021c;Deng et al., 2019;Deng et al., 2020b;Dai et al., 2020b;Guo et al., 2019) to construct gray-imaging elements with ultra-high resolution and extraordinary ability of continuous grayscale modulation. In addition, in some applications such as quick response (QR) code for information recognition and watermark for anti-counterfeiting, binary patterns are more suitable for information encoding. Recently, multiplexing grayscale or binary-pattern nanoprints have been proposed by finely setting the size or orientation of nanostructures (Dai et al., 2020a(Dai et al., , 2020cDeng et al., 2020c;Fan et al., 2020;Liu et al., 2021), which further improve the density of information storage.
Merging a color-nanoprint, a gray-imaging element and a binary-pattern-imaging element into a single metasurface are an artful approach to increase the information security and system integration, which can also provide a new information multiplexing method. However, different types of nanoprints always correspond to different light control mechanisms. The difficulty of realizing multifunctional manipulations of light hinders the development of ''three-in-one'' nanoprints. In this paper, we show a route of integrating color and grayscale manipulations into a single metasurface and control them separately to form different information channels, which enables the concept of ''three-in-one'' nanoprint, simply by a single-cell design approach. Specifically, based on the spectral differences of two dielectric nanobricks with different dimensions, a dual-color nanoprinting image can be recorded right at the metasurface plane. At the same time, the two different nanobricks have equal polarization conversion efficiency (PCE) near the designing wavelength of 610 nm, which ensures that they can produce an equal intensity governed by Malus law. Based on this characteristic, a continuous grayscale image can be encoded into the dual-color nanoprint. Interestingly, inspired by the orientation degeneracy of anisotropic nanostructures, the same metasurface can simultaneously record an additional binary-pattern, merely with polarization controls. Figure 1 shows the basis concept of the proposed metasurface. Apparently, the metasurface is a dual-color nanoprint observed under a white light illumination without polarization control. Actually, two additional information channels have been hidden into the metasurface, and the corresponding images can be decoded by utilizing specific optical keys. Specifically, when we put the metasurface into an orthogonal-polarization optical path consisting of two bulk-optic polarizers and a narrow-band filter, a continuous Figure 1. Schematic illustration of the ''three-in-one'' nanoprints with a single-cell-nanostructure design approach and some application prospects The metasurface is composed of two types of nanobricks with different dimensions but each unit-cell contains only one nanobrick (i.e. single-cell-nanostructure). Under the white light illumination, a dual-color image appears right at the metasurface plane (channel 1). An orthogonal-polarization optical path consisting of two bulk-optic polarizers and a narrow-band filter is taken as an optical key to decode the hidden continuous grayscale image (channel 2) and binarypattern (channel 3). iScience Article grayscale image can be decoded. If we rotate the metasuface around its optical axis by 22.5 , a binarypattern appears (the two images can also be switched by rotating the two bulk-optic polarizers, as shown in Figure S1). Therefore, three different types of nanoprinting images can be recorded with a piece of metasurface. With aforementioned unique characteristics, the ''three-in-one'' nanoprints have potential applications in multi-folded anti-counterfeiting, optical storage, information encoding and hiding, etc.

RESULTS
Design of the tri-channel metasurfaces for ''three-in-one'' nanoprints To obtain the tri-channel metasurface for information multiplexing aforementioned, we need to retrieve a pair of nanostructures that have different spectral response but equal PCE at a fixed wavelength, which meets the requirement of forming a dual-color image (channel 1) under white light illumination and two gray images at a fixed wavelength. Because decoding a continuous grayscale image (channel 2) and a binary-pattern (channel 3) requires an orthogonal-polarization optical path, two bulk-optic polarizers acting as a polarizer and an analyzer, respectively, are placed before and after a nanostructure, then we can deduce the intensity after the analyzer as where A and B indicate the complex reflection coefficients when the light waves propagate with polarization along the long and short axes of the nanobricks, q denotes the orientation angle of the nanobrick, a 1 and a 2 are the transmission axis directions of the polarizer and analyzer respectively, and I 0 is the light intensity after the polarizer. In particular, if a 2 = -a 1 = 45 and q = 0 , the ratio of output light intensity to the incident LP light is AÀB 2 2 , which is defined as PCE aforementioned. More details of the formula derivation are presented in theoretical analysis of STAR Methods.
Here, we employ silicon-on-insulator (SOI) materials that are widely employed in integrated circuits, to make a reflective-type all-dielectric metasurfaces. To satisfy the aforementioned conditions, we elaborately design the geometry of nanostructures by using CST Microwave Studio software. Two types of nanobricks with the equal height H = 220 nm and cell size C = 400 nm are employed in our design, named as I and II, respectively. When Nanobrick I is designed with length L 1 = 150 nm and W 1 = 90 nm and Nanobrick II is designed with L 2 = 180 nm and W 2 = 100 nm, the reflection spectra are different enough to produce two different structural colors. At the same time, the PCE of the two types of nanobricks is almost equal at a working wavelength of 610 nm (detailed description about the design and simulation of the nanobricks is provided in the numerical simulations of STAR Methods). Therefore, both of them can be employed to construct a hybrid metasurface for storage of both dual-color image and gray-images.
With the above designed two types of SOI nanobricks, we can now implement the tri-channel metasurface design, as shown in Figure 2 of the design flowchart. Because we use a single-cell design strategy, no supercell is required. In general, the tri-channel metasurface design includes two aspects: (1) spatial distribution of the two types of nanobricks with different dimensions; (2) orientation distribution of nanobricks. Firstly, we can determine the spatial distribution of the two types of nanobrick according to the target image of channel 1. The background and target parts of I c1 are designed with Nanobrick I and II, respectively, as shown in Figure 2B. Next, the target gray-image I c2 and the normalized intensity modulation of channel 2, i.e., I 2 = I 0 cos 2 (2q), are utilized to calculate the initial orientation q, in which all orientations lie in the interval of [0 , 45 ].
The last step is rearranging the orientations to construct channel 2 and 3 simultaneously with the help of polarization multiplexing. Specifically, if one rotates the orthogonal-polarization optical setup (two bulkoptic polarizers) clockwise from the current 0 to an angle such as 22.5 , the new light intensity can be written as I 3 = I 0 cos 2 (2q+45 ). We plot both I 2 and I 3 versus orientation angle, as shown in Figure 2G. And we found that there exists a one-to-four mapping relationship between the light intensity and the orientation of nanobrick in the defined interval of [0 , 180 ], which can be called as the orientation degeneracy of nanobricks. That is, there are four options for the orientation angles, q 1 , q 2 , q 3 , and q 4 , to generate the equal output light intensity corresponding to channel 2. However, in the intensity modulation of I c3 corresponding to channel 3, the four orientation angles possess two different intensity modulations (q 1 and q 3 correspond to a ''low'' intensity value [<0.5]; q 2 and q 4 correspond to a ''high'' intensity value [>0.5]), opening up a new design degree of freedom to create an additional ''binary-pattern'' without complicating the design ll OPEN ACCESS iScience 24, 103510, December 17, 2021 3 iScience Article and fabrication of nanostructures. Therefore, it is promising to search a reasonable orientation distribution that satisfies the requirement of encoding a continuous gray-image and a binary-pattern into channel 2 and channel 3, respectively. Specifically, if the intensity value of I 3 is lower than 0.5, the corresponding initial orientation remains unchanged (= q 1 ) or is changed to q 3 . If the intensity value of I 3 is larger than 0.5, the corresponding initial orientation distribution q is changed to be q 2 or q 4 . Hence, we get the final orientation distribution q f , as shown in Figure 2F. It is worth noting that the intensity value cannot be set to be 0 or 1 in channel 2 (in this case, the intensity value is 0.5 for each pixel in channel 3). Besides, the intensity profile in channel 3 is not a pure binary-intensity (the intensity values are modulated to be exactly 0 or 1) in traditional sense. In our work, the binary image denotes the image has two kinds of intensity value, one is higher than 0.5 and the other is lower than 0.5, so the dark (bright) part on an image is not dark (bright) enough and the contrast is not high enough compared with a traditional binary image. Therefore, there is a trade-off between encoding more images in nanoprint in a single band and generating higher contrast images.

Experimental demonstration of the ''three-in-one'' nanoprints
To demonstrate the feasibility and flexibility of the ''three-in-one'' nanoprints, we fabricate two different types of samples (labeled with A and B) by using the standard electron beam lithography (see STAR The dual-color images of a Chinese character ''flower'' (sample A) and a picture of sakura (sample B) can be observed under the illumination of a quartz halogen lamp; its color looks orange-red ( Figures 3B-3E). When iScience Article unpolarized white light from light-emitting diode (LED) source is introduced to illuminate the samples, the colors become yellowish (as shown in Figure S2). Due to the spectral difference of the light sources, the dual-color images have different colors. However, all images including the zoom-in views are in clear visual effect under the illumination of a broadband source, which proves the feasibility of encoding a dual-color nanoprinting image.
Next, to decode the information hidden into channel 2 and 3, a red narrow-band filter (the working wavelength is 610 nm with bandwidth of 5 nm), a polarizer, and two analyzers are inserted into the same light path (as shown in Figure 3A). When the transmission axis directions of the polarizer and the analyzer are À45 and 45 , respectively (denoted with white arrows in the upper left corner of Figures 3F and 3H), the reflected nanoprinting images are shown in Figures 3F-3I. The last row presents the experimentally captured nanoprinting images ( Figures 3J-3M) by rotating the orthogonal-polarization optical setup clockwise by 22.5 . The experimental results and the zoom-in views indicate that both continuous grayscale images of a ''rose'' and clear binary-patterns with negligible crosstalk can be observed at the wavelength of 610 nm, which are in good accordance with our design.
In addition, sample A and B are designed to generate the equal continuous gray-images (a ''rose'') in channel 2 and different images in channel 1 and 3, which proves that the three channels are controlled independently. Therefore, we can design the three information channels at will, and the information of the three channels is not related and cannot be inferred with each other.
At last, to explore the spectral response characteristics of the tri-channel metasurfaces, we capture the nanoprinting images under the illumination of green (l = 540 nm) and blue (l = 480 nm) light, respectively, and the obtained experimental results are shown in Figure 4. Figures 4A-4D show the nanoprinting images captured under the illumination of unpolarized green and blue light. It is obvious that the nanoprinting images obtained in green and blue light illumination appear as the target pattern of channel 1 with different brightness. The main reason is that the reflection of Nanobrick I and II is different at two wavelengths of 480 nm and 540 nm (see STAR Methods for the details of numerical simulations). When an orthogonal-polarization optical path consisting of a polarizer and an analyzer is constructed, the experimentally captured results are shown in the second and third rows of Figure 4. Due to the PCE differences between the Nanobrick I and II at 480 nm and 540 nm, the patterns of channel 2 and 3 are always mixed with the pattern of channel 1, which hinder the information identification.

DISCUSSION
The proposed ''three-in-one'' nanoprints provide several technical advantages and have potential applications in many interesting fields. In our design, only two types of nanostructure are employed but we don't bring them together to form a supercell. Instead, each unit-cell consists of either Nanobrick I or II. Because our design is based on single-cell design rather than the widely used supercell design for information multiplexing, our approach has a higher resolution and has potential application in high-density optical storage, as each nanostructure has been multiplexed corresponding to three independent channels.
Secondly, it is interesting to see that the encoded information has to be decoded with quite different optical setups, providing a promising application in designing optical anti-counterfeiting labels. In particular, the information of channel 1, i.e. a dual-color nanoprinting image, is retrieved by a broadband light source without polarization control. And the channel 2 and 3 are decoded by an orthogonal-polarization optical setup with different polarization controls. Therefore, the different illumination conditions can be treated as optical keys to decode the hidden information. In addition, only when the PCE of the two types of nanostructures is equal, can the information hidden in the three channels be completely decoded, which further increases the security of the meta-images. The experimentally measured PCE can reach 11% and 10% for Nanobrick I and II. The efficiency could be improved further by applying more precise fabrication procedures or using low-loss dielectric materials. Because security and counterfeiting difficulties are the fundamental requirements of optical anticounterfeiting labels, our approach with three different keys and three independent images at the nanoscale resolution can significantly improve both the security and counterfeiting difficulty of optical anti-counterfeiting labels.
In summary, we propose a new route of multifunctional light manipulation for separately controlling spectrum and polarization-assisted intensity of incident light, which enlightens the concept of ''three-in-one'' ll OPEN ACCESS 6 iScience 24, 103510, December 17, 2021 iScience Article nanoprints with a single-cell-nanostructured metasurface. Specifically, by combining the spectrum manipulation of varied nanostructures, intensity manipulation governed by Malus law, and the orientation degeneracy of anisotropic nanostructures, a multiplexing metasurface capable of simultaneously and independently recording a dual-color image, a continuous grayscale image, and a binary-pattern is proposed. The experimental results are in good accordance with our design: the metasurface apparently acts as a nanoprint presenting a dual-color image under a broadband light source illumination, while displays two hidden information channels when taking an orthogonal-polarization optical setup and a fixed working wavelength as a decoding key. With advantages such as ultracompactness, high resolution, high security, and difficulty in counterfeiting, the proposed tri-channel metasurfaces have potential applications in optical storage, high-end anti-counterfeiting, information hiding, and many other related fields.

Limitations of the study
In this work, the contrast of the observed images is not high enough compared with a traditional printing image. Besides, the efficiency should be improved further by applying more precise fabrication procedures or using low-loss dielectric materials.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

DECLARATION OF INTERESTS
The authors declare no conflicts of interest.

Lead contact
Any further information and requests for resources and materials should be directed to and will be fulfilled by the Lead Contact, Prof. Guoxing Zheng (gxzheng@whu.edu.cn).

Materials availability
This study did not generate new unique reagents.

Date and code availability
Reagents and materials used in the fabrication procedures are listed in the key resources table.
This paper does not report original code.
Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Theoretical analysis
The Jones matrix of an anisotropic nanostructure with an in-plane orientation q can be expressed as where RðqÞ is the rotation matrix, A and B are the complex transmission (or reflection) coefficients of the nanostructure along with the long and short axes, respectively.
If the incident light passes through a polarizer, an anisotropic nanostructure and a bulk-optic analyzer sequentially, the Jones vector of output light can be expressed as J = cos 2 a 2 sina 2 cosa 2 sina 2 cosa 2 sin 2 a 2 :TðqÞ: cosa 1 sina 1 ; (Equation 3) where a 1 and a 2 represent the directions of transmission axis of the polarizer and analyzer, respectively. If the light intensity after the polarizer is I 0 , we can deduce the expression of output light intensity according to Equation 3 as I = I 0 A À B 2 cosð2q À a 2 À a 1 Þ + A + B 2 cosða 2 À a 1 Þ 2 :

(Equation 4)
We find that any anisotropic nanostructure (AsB) can be used for a continuous intensity modulation when the light intensity I 0 , the transmission axes of the polarizer and analyzer are unambiguously given.
When the transmission axis of the polarizer is perpendicular to the transmission axis of the analyzer, we can simplify Equation 4