Home Physical Sciences Parallel wave-based analog computing using metagratings
Article Open Access

Parallel wave-based analog computing using metagratings

  • Hamid Rajabalipanah , Ali Momeni , Mahdi Rahmanzadeh , Ali Abdolali ORCID logo EMAIL logo and Romain Fleury ORCID logo EMAIL logo
Published/Copyright: March 24, 2022
Download Article (PDF)
Nanophotonics
From the journal Nanophotonics Volume 11 Issue 8

Abstract

Wave-based signal processing has witnessed a significant expansion of interest in a variety of science and engineering disciplines, as it provides new opportunities for achieving high-speed and low-power operations. Although flat optics desires integrable components to perform multiple missions, yet, the current wave-based computational metasurfaces can engineer only the spatial content of the input signal where the processed signal obeys the traditional version of Snell’s law. In this paper, we propose a multi-functional metagrating to modulate both spatial and angular properties of the input signal whereby both symmetric and asymmetric optical transfer functions are realized using high-order space harmonics. The performance of the designed compound metallic grating is validated through several investigations where closed-form expressions are suggested to extract the phase and amplitude information of the diffractive modes. Several illustrative examples are demonstrated to show that the proposed metagrating allows for simultaneous parallel analog computing tasks such as first- and second-order spatial differentiation through a single multichannel structured surface. It is anticipated that the designed platform brings a new twist to the field of optical signal processing and opens up large perspectives for simple integrated image processing systems.

Keywords: analog optical computing; metagrating; metasurface

1 Introduction

For a few decades, digital processors have been widely used to execute computational tasks, as an alternative to analog mechanical and electrical computers. Despite their reliability and high-speed operation, digital processors suffer from high-power consumption, expensive analog-to-digital conversion, and sharp performance degradation at high frequencies, leading to large limitations even for performing simple computing tasks such as differentiation or integration, equation solving, matrix inversion, edge detection, and image processing [1, 2]. With the advent of metamaterials and metasurfaces, spatial analog optical computing resurfaced, finding important applications as compact solutions for high speed, high throughput image processing and parallel computing. Since the seminal proposal of Silva et al., [3] wave-based analog computing has witnessed rapid progress, with the demonstration of optical spatial differentiators [4–13], integrators [2, 8, 12, 14], equation solvers [15–17], spatiotemporal computing [9, 18, 19], and wave-based neuromorphic computing [20–23]. Among them, the Green’s function (GF) method in which a specific-purpose computing operation is directly realized in the real space, without transforming back and forth from the spatial to the spectral domain [14], affords compactness and avoids possible challenges in error propagation and alignment issues. The applicability of the GF method to execute signal processing has been verified in a series of proposals via spin hall effect of light [24], disordered and complex scattering system [7, 25, 26], layered structures [18, 27], topological insulators [28, 29], plasmonic arrays [5], bianisotropic metasurfaces [10, 17], and so on. Nevertheless, prior GF-based studies still face two different challenges: (i) parallel realization of mathematical operators has been only addressed by using bulky structures [7, 16] and array of subwavelength meta-atoms with complex geometries [30, 31] and thus, they are still subject to implementation difficulties arising from high fabrication precision demands; (ii) although reflective optical processing for normal incidences is a good alternative for complex oblique illumination setups, it still needs additional optical components to separate the processed signal from the input one [10]. Further efforts to tackle these barriers must be accompanied with the use of more powerful architectures to implement spatial optical signal processing.

Passive metasurfaces can ensure highly-efficient wavefront molding by leveraging nonlocal effects stemming from the excitation of evanescent [32, 33] or leaky modes [34], but this typically leads to complex design requirements and a need for deeply subwavelength fabrication resolution [35–44]. As an alternative solution, recent efforts have shown that metagratings composed of non-subwavelength periodic patterns enable highly complex diffraction scenarios of significant practical interest, with high efficiency [45, 46]. The core idea of metagratings is based on Floquet–Bloch (FB) theory, remarking that when a plane wave impinges on a periodic structure, a discrete set of diffracted waves can be generated, some of them propagating and others being evanescent [47, 48]. The number of propagating and evanescent waves is determined by the period of the structure and the angle of incidence. The structure is engineered at the scale of the wavelength so as to suppress the unwanted space harmonics and reroute the incident power towards a desired nonspecular channel. Therefore, such design potentially relaxes some of the fabrication challenge of metasurfaces, as it does not require precise lithography techniques [45, 49]. Metagratings have also the ability to provide multiple arbitrarily-oriented space channels for creating multimission surfaces [50]. Although several reports have examined metagratings from different points of view, the potential application of these structures for performing optical analog computation, which requires studying the phase information of the space harmonics, has not been unveiled, yet.

In this study, we reveal that a suitably engineered all-metallic metagrating can open multiple non-specular channels for parallel implementation of symmetric and asymmetric optical transfer functions at both normal and oblique illuminations. Closed-form expressions are presented to predict the reflection phase and amplitude information of the space channels opened by the designed compound metallic grating. The application of the designed metagrating for accomplishing first- and second-order differentiation as well as detecting sharp edges of an input image is investigated through several illustrative demonstrations.

2 Fundamental theory of compound metallic grating

The main idea of this paper is graphically depicted in Figure 1a, where a single metagrating is responsible for realizing parallel mathematical operations on the input signals coming from different directions. The symmetric geometry originated from the presence of a single groove in each period does not allow realizing the asymmetry required for performing odd-order transfer functions. Thus, we consider a two-dimensional (2D) sparse array (∂/∂y = 0) composed of two grooves in each period of a metallic medium filling the half-space z < 0, as shown in Figure 1b and c. The grooves are specified by the widths of w 1, w 2 and center-to-center distance d, heights h 1, h 2, while being filled with dielectrics of permittivity values ϵ r 1 = n 1 2 and ϵ r 2 = n 2 2 , respectively. The whole structure is surrounded by a medium with dielectric constant ε r 0 = n 0 2 . The position and dimensions of the grooves can be set at will and the periodicity along x axis is indicated by L x . The metagrating is illuminated by a general oblique 2D beam profile with the angle θ in and the transverse-magnetic (TM) polarization. By leveraging the superposition principle and spectral decomposition, the input f in and output f out optical signals can be expanded based on an infinite set of plane waves, all traveling in different directions with wavenumbers k x :

(1a) f in ( x , z ) = ∫ k 0 x − W k 0 x + W F ̃ inc y k x e x p ( − j k x x + j k z z ) d k x

(1b) f out ( x , z ) = ∫ k 0 x − W k 0 x + W F ̃ inc y k x H ̃ ref y k x × e x p ( − j k x x − j k z z ) d k x

Here, W denotes the beamwidth of the input field and the harmonic time dependency e jωt is omitted. The spatial frequency content of the incident field is represented by the various plane wave amplitudes around k x = k 0 sin(θ), that together form the input signal. The compound metallic grating, depicted in Figure 1c, interacts differently with each of these plane wave components. A Rayleigh expansion can be performed to write the total fields at the upper half-space:

(2a) H ̃ z > 0 y = e j k z 0 z e − j k x 10 x + ∑ m R ̃ m e − j k z m z e − j k x m x

(2b) E ̃ z > 0 x = − Y 0 e j k z 0 z e − j k x 0 x + ∑ m Y m R ̃ m e − j k z m z e − j k x m x

in which, the subscript m corresponds to the order of space harmonics, and

(3a) k x m = k x 0 + 2 m π / L x

(3b) k z m = − j k 0 ( n 0 ⁡ sin ( θ i ) + m λ / L x ) 2 − n 0 2

Here, k x0 = k 0 n 0 sin(θ i ) and Y m = k z m / ω ε 0 n 0 2 denote respectively the x-directed wavenumber the admittance of the mth TM-polarized diffractive mode in the upper half-space. After applying the proper boundary conditions and solving the related equations, the specular and nonspecular reflection coefficients can be expressed as:

(4a) R ̃ 0 = 2 Y 0 A 0 B + 1

(4b) R ̃ m ≠ 0 = 2 Y m A m B

in which, A, A m , and B are complex numbers. A detailed derivation of the mathematical expressions of these coefficients can be found in Supplementary Information A. It should be noted that Eqs. (4a) and (4b) include both phase and amplitude information of the diffractive modes when the metagrating is excited by an oblique plane wave with an arbitrary angle of incidence.

Figure 1: Metagrating-based parallel analog signal processing. (a) Illustration of wave-based analog signal processing and edge detection through the high-order Floquet modes provided by an all-metallic metagrating. (b) Different channels for parallel analog computing in high-order Floquet harmonics. (c) 2D view of the compound metallic grating with two grooves per period and the associated high-order Floquet harmonics.
Figure 1:

Metagrating-based parallel analog signal processing. (a) Illustration of wave-based analog signal processing and edge detection through the high-order Floquet modes provided by an all-metallic metagrating. (b) Different channels for parallel analog computing in high-order Floquet harmonics. (c) 2D view of the compound metallic grating with two grooves per period and the associated high-order Floquet harmonics.

3 Results and discussion

3.1 Unlocked channels

As shown by Eq. (3a), the scattered wave is a discrete superposition of space harmonics, which can be propagative or not, depending on the metagrating design. Upon illuminating by an oblique plane wave with θ in, the reflected wavefronts are oriented along θ out in such a way that [51, 52]:

(5) sin θ out , m = sin θ in + m λ / L x ( m = 0 , ± 1 , ± 2 , … )

In a multichannel configuration like Figure 1b, those channels can serve to perform optical analog computing. We should remark that in our study, the metagrating is excited by oblique beam profiles of beamwidth W, and thus, Eq. (5) must be evaluated for all incident wave angles within θ in ± arcsin(W/k 0). With simple algebraic manipulations on Eq. (5), the unlocking condition for each diffractive mode turns into:

(6) sin θ in 1 − ξ ± cos θ in ξ + m L n x < 1

wherein, ξ = W/k 0 and L nx = L/λ. Given the purpose of design and according to Figure 2a, working with three accessible channels: m = {−1, 0, 1} for θ in = 0, m = {0, −1, −2} channels for θ in < 0, and m = {0, 1, 2} channels for θ in > 0 is sufficient and the higher-order modes must be kept evanescent in each case. Taking ξ = 0.2, Supplementary Figures S1a–f show the solution domain for different FB modes based on Eq. (6), disclosing the best choice for grating periodicity corresponding to the desired angle of incidence. As designed, only three FB modes m = 0, ±1 will be propagating for the input beam profile, and the rest will be evanescent, as long as the periodicity and the incident wave angle satisfy λ < L x < 1.5λ and −50° < θ in < 50°, respectively.

Figure 2: Single-operator metagrating processor. (a) Schematic illustration of the three processing channels and (b) a single-operator metagrating for performing first-order spatial differentiation in a nonspecular reflection mode. The optimum structural parameters are ɛ r1 = ɛ r2 = ɛ r0 = 1, w 1 = 0.288L x , w 2 = 0.145L x , h 1 = 0.17L x , h 2 = 0.087L x , and d = 0.257L x . (c) The angular spectra of the amplitude for different spatial harmonics. (d) The amplitude (dB) of synthesized transfer function for m = −1. (e) The angular spectra of the phase for different spatial harmonics. (f) The input field and the corresponding exact/calculated output signal.
Figure 2:

Single-operator metagrating processor. (a) Schematic illustration of the three processing channels and (b) a single-operator metagrating for performing first-order spatial differentiation in a nonspecular reflection mode. The optimum structural parameters are ɛ r1 = ɛ r2 = ɛ r0 = 1, w 1 = 0.288L x , w 2 = 0.145L x , h 1 = 0.17L x , h 2 = 0.087L x , and d = 0.257L x . (c) The angular spectra of the amplitude for different spatial harmonics. (d) The amplitude (dB) of synthesized transfer function for m = −1. (e) The angular spectra of the phase for different spatial harmonics. (f) The input field and the corresponding exact/calculated output signal.

Hereafter, we intend to show how the multiple channels provided by the metagrating can be exploited for performing analog signal processing and be used for realizing different functionalities at the same time.

3.2 Illustrative examples

3.2.1 Single-operator metasurface

We adjust the periodicity of the metagrating so as to provide three active channels (see Figure 2a). Accordingly, we set the periodicity of the designed metagrating as L x = 1.3λ 0 (f 0 = 1 THz), which orients the channels along θ = ±40° directions at f = 1.2 THz. Once the input and output channels are determined, the width and height of the contributing grooves are optimized so that the angular dispersion of the scattering coefficient between these channels in Eqs. (4a) and (4b), emulates the k x -dependency of desired transfer functions. For instance, we can achieve 1st- and 2nd-order spatial differentiation of the input signal exciting the metasurface from the ith channel and exiting at the jth channel, provided that S ji (θ) = jk 0(sin(θ) − sin(θ inc)) and S ji (θ) = − k 0 2 ( sin ( θ ) − sin ( θ inc ) ) 2 , respectively. Here, our goal is to design a periodic surface that applies 1st-order spatial differentiation operation on the input signal traveling from port 2 to port 1 (see Figure 2b). A comprehensive parametric study based on the theoretical representation of Eqs. (4a) and (4b), has been carried out to find the optimum parameters of the metagrating and minimize the following error function:

(7) E = ∑ q S i j θ q − H ̃ θ q 2 + ∠ S i j θ q − ∠ H ̃ θ q 2

Here, H ̃ ( θ ) indicates the transfer function of choice, i and j refer to the output and input ports, respectively, and the above summation is calculated over q discrete angles in the vicinity of the incident wave angle. The angular spectra of |S 12(θ)| and ∠S 12(θ) are plotted in Figure 2c and d, respectively, and the optimum geometrical parameters are given in the caption of the same figure. From this figure, one can immediately deduce that the nonspecular S 12 channel (m = −1) admits a linear trend and an asymmetric 180° phase jump around θ = 0, which define the transfer function of 1st-order differentiation. To evaluate the performance of our single-operator metasurface differentiator, the Gaussian signal of Figure 2e is utilized to launch the metagrating from port 2 and the calculated output field is shown in the same figure. The result is compared with the exact response indicating that the output field at port 1 is indeed the response expected from a first-order derivative operation, a 2% error, as defined by Eq. (7). The error value of each computing task is given in the caption of the related figure. Indeed, the designed metagrating successfully deflects the first-order derivative of the normally incident signals into θ = − 40° direction without using any additional beam splitting sub-block. Conversely, previous metasurfaces based on the GF method cannot provide a reciprocal solution for asymmetric on-axis transfer functions at the reflection side [10]. Our structure therefore relaxes vexing complexities of previous designs that needed oblique illumination setups and beam splitting devices [10].

3.2.2 Multi-operator metasurface

Up to now, the proposed metagratings realize only a single processing operation on a single channel. Hereafter, we intend to design metagratings which successfully create multiple channels, each of which enables different processing or scattering functionalities. To this aim, the optimization procedure of Eq. (7) is simultaneously accomplished for different scattering parameters, e.g., S ij and S uv . To exemplify how the operation frequency can be adjusted, we decided to change the periodicity of the structure to L x = 1.22λ 0, so that the orientation of m = ±1 channels points in the θ = 40° direction for normal excitations at a higher frequency, namely f = 1.3 THz. Both specular and nonspecular channels can be involved and we carry out the numerical optimizations for different computing scenarios displayed in Figures 3a, 4a, and 5. In the first demonstration, the metagrating is parametrically adjusted so that the S 13(θ) (specular) and S 22(θ) (specular) coefficients implement first- and second-order differentiation operations, respectively (see Figure 3a). The synthesized transfer functions are plotted in Figure 3b and c. A parabolic shape is acquired for |S 22(θ)| around θ = 0° while no phase change is noticed. At the same time, S 13(θ) provides an amplitude null with 180° phase jump around θ = −40°. Two input fields, with profiles varying as exp(−αx 4) and sinc(βx), are employed to excite the designed metagrating from ports 1 and 2, respectively, (α = 0.008 and 1/β = 6 × 108 are two constants). Using Eqs. (1a), (1b) and the achieved scattering coefficients, the output fields leaving the metagrating from ports 2 and 3 are calculated and presented in Figure 3d and e, respectively. As seen, the output signals at ports 2 and 3 are nothing but the first- and second-order derivative of the input fields at ports 1 and 2, respectively. The obtained results indicate that a suitably-designed metagrating successfully provides multi-input multi-output channels, each of which, enables different processing functionalities. Our purpose in the second example is to design a single-input multi-output metagrating manifesting the angular dispersion of jk 0 sin θ (first-order spatial differentiation) in its S 21(θ) (non-specular) and S 31(θ) (specular) scattering coefficients (see Figure 4a). The transfer functions realized by the optimized metagrating are shown in Figure 4b and c from which, one can immediately deduce that both S 21(θ) and S 31(θ) parameters expose a linear trend and an asymmetric 180° phase jump near their corresponding angles, consistent with the transfer function of the 1st-order differentiation operation. The structure has one input field but the output fields at ports 2 and 3 are examined by input signals with profiles of sin(αx) and sinc(βx), respectively. The input and output fields displayed in Figure 4d and e verify that this single-input multi-output processing mission is perfectly accomplished by the designed metagrating. For a quantitative comparison, the exact output responses are also illustrated. The demonstrations presented in Figures 3 and 4 show different aspects of parallel processing ability of the designed metagratings. Another example of parallel computing can be found in Supplementary Information D in which the first-order spatial differentiation can be applied on two distinct input signals at the same time. As a final demonstration example, we form a multi-operator metagrating performing second-order spatial differentiation for the input signals coming from either port 1 or 2 (see Figure 5a). Figure 5b and c demonstrate the angle-dependent amplitude and phase of the transfer function of the designed metagrating. All indicator features of the second-order differentiation operation are observable in S 22(θ) (specular) and S 11(θ) (nonspecular) around θ = 0° and θ = 40°, respectively. Figure 5d and e depicts the input fields, exp(−αx 2) and exp(−αx 4), and the corresponding output signals, confirming that the designed multi-operator metagrating is capable of independently processing two input signals.

Figure 3: Dual-operator metagrating processor. (a) The schematic illustration of a dual-operator metagrating for performing first- and second-order spatial differentiation in specular reflection modes. The optimum structural parameters are ɛ r1 = ɛ r2 = ɛ r0 = 1, w 1 = 0.035L x , w 2 = 0.0385L x , h 1 = 0.467L x , h 2 = 0.45L x , and d = 0.643L x . The angular spectra of the (b) amplitude and (c) phase for different spatial harmonics. (d) The amplitude (dB) of synthesized transfer function around −40° (left) and 0° (right). (e) and (f) The input fields and the corresponding exact/calculated output signals. The error values corresponding to panels (e) and (f) are 3% and 1.5%, respectively.
Figure 3:

Dual-operator metagrating processor. (a) The schematic illustration of a dual-operator metagrating for performing first- and second-order spatial differentiation in specular reflection modes. The optimum structural parameters are ɛ r1 = ɛ r2 = ɛ r0 = 1, w 1 = 0.035L x , w 2 = 0.0385L x , h 1 = 0.467L x , h 2 = 0.45L x , and d = 0.643L x . The angular spectra of the (b) amplitude and (c) phase for different spatial harmonics. (d) The amplitude (dB) of synthesized transfer function around −40° (left) and 0° (right). (e) and (f) The input fields and the corresponding exact/calculated output signals. The error values corresponding to panels (e) and (f) are 3% and 1.5%, respectively.

Figure 4: Single-input multi-output metagrating processor. (a) The schematic illustration of a single-input dual-output metagrating for performing first-order spatial differentiation in both specular and non-specular reflection modes. The optimum structural parameters are ɛ r1 = ɛ r2 = ɛ r0 = 1, w 1 = 0.026L x , w 2 = 0.0476L x , h 1 = 0.43L x , h 2 = 0.427L x , and d = 0.538L x . (b) The angular spectra of the amplitude for different spatial harmonics. (c) The amplitude (dB) of synthesized transfer function. (d) The angular spectra of the phase for different spatial harmonics. (e) and (f) The input fields and the corresponding exact/calculated output signals. The error values corresponding to panels (e) and (f) are 1.5% and 1.8%, respectively.
Figure 4:

Single-input multi-output metagrating processor. (a) The schematic illustration of a single-input dual-output metagrating for performing first-order spatial differentiation in both specular and non-specular reflection modes. The optimum structural parameters are ɛ r1 = ɛ r2 = ɛ r0 = 1, w 1 = 0.026L x , w 2 = 0.0476L x , h 1 = 0.43L x , h 2 = 0.427L x , and d = 0.538L x . (b) The angular spectra of the amplitude for different spatial harmonics. (c) The amplitude (dB) of synthesized transfer function. (d) The angular spectra of the phase for different spatial harmonics. (e) and (f) The input fields and the corresponding exact/calculated output signals. The error values corresponding to panels (e) and (f) are 1.5% and 1.8%, respectively.

Figure 5: Retro-directive processing channels. (a) The schematic illustration of a dual-channel metagrating for performing second-order spatial differentiation in both specular and nonspecular (retro-reflection) reflection modes. The optimum structural parameters are ɛ r1 = ɛ r2 = ɛ r0 = 1, w 1 = 0.0863L x , w 2 = 0.069L x , h 1 = 0.7336L x , h 2 = 0.71L x , and d = 0.6381L x . The angular spectra of the (b) amplitude and (c) phase for different spatial harmonics. (d) The amplitude (dB) of synthesized transfer function around 0° (left) and −37° (right). (e) and (f) The input fields and the corresponding exact/calculated output signals. The error values corresponding to panels (e) and (f) are 2.5% and 3%, respectively.
Figure 5:

Retro-directive processing channels. (a) The schematic illustration of a dual-channel metagrating for performing second-order spatial differentiation in both specular and nonspecular (retro-reflection) reflection modes. The optimum structural parameters are ɛ r1 = ɛ r2 = ɛ r0 = 1, w 1 = 0.0863L x , w 2 = 0.069L x , h 1 = 0.7336L x , h 2 = 0.71L x , and d = 0.6381L x . The angular spectra of the (b) amplitude and (c) phase for different spatial harmonics. (d) The amplitude (dB) of synthesized transfer function around 0° (left) and −37° (right). (e) and (f) The input fields and the corresponding exact/calculated output signals. The error values corresponding to panels (e) and (f) are 2.5% and 3%, respectively.

To further validate the performance of the proposed metagratings, full-wave simulations have been carried out through COMSOL Multiphysics. The configuration is shown in Figure 6g in which two TM-polarized Gaussian-shape beams illuminate the metagrating of Figure 6a from two different directions θ = 0° and θ = 40° (see Figure 6a and d). The medium surrounding the designed metagratings is filled by air and the boundary conditions are selected as perfect match layer (PML). Figure 6b and d demonstrate the scattered fields for each of normal and oblique illuminations. A cut-line of the scattered fields in each case is plotted in Figure 6c and f indicating that the output signals successfully obey the exact version of the second-order spatial derivative of the Gaussian-shape beams. Thus, the full-wave simulations verify well our idea and analytical model, confirming the relevance of high-order Floquet modes to perform analog signal processing. To investigate the possible errors originating from the fabrication tolerances, we made some geometrical variations on the structure and the corresponding results are presented in Supplementary Information D.

Figure 6: Full-wave verification of nonspecular processing channels. Full-wave verification of the proposed signal processing idea for the designed metagrating of (a). (a) and (d) The obliquely- and normally-oriented input Gaussian-shape beams, (b) and (e) the corresponding reflected fields, and (c) and (f) cut-line of the reflected fields along with the exact second-order derivative of the input fields. (g) The 3D schematic view of the studied metagrating.
Figure 6:

Full-wave verification of nonspecular processing channels. Full-wave verification of the proposed signal processing idea for the designed metagrating of (a). (a) and (d) The obliquely- and normally-oriented input Gaussian-shape beams, (b) and (e) the corresponding reflected fields, and (c) and (f) cut-line of the reflected fields along with the exact second-order derivative of the input fields. (g) The 3D schematic view of the studied metagrating.

Finally, a concrete 1D edge detection application is evaluated by using the metagrating differentiator of Figure 2b, where the logo of ”EPFL” university is utilized as the input image (see Figure 7a). Due to the 1D nature of the proposed system, the reflected image displayed in Figure 7b highlights all edges of the incident image along the vertical direction. Supplementary Information E depicts another interesting application in the field of image processing in which the metagrating applies a denoising filter on the input image. It should be noted that the idea can be extended to 2D scenarios, including differentiation and edge detection, the reflection coefficients of two-dimensional metagratings should be extracted, allowing us to manipulate the information of both spatial frequencies kx and ky [53–55]. Besides, the proposed multi-operator metagrating can also be an important key in parallel image processing applications, including THz imaging. With the ability to penetrate many materials, terahertz waves have intriguing potential in revealing hidden features [56]. In order to extract the important geometric features of objects, the edge detection method can be directly employed through different THz imaging systems [57]. As a result, THz-based edge detectors like those we proposed in the submitted paper, may be a key component in THz imaging/image processing systems. As a promising outlook, by tuning the interference between the multipoles of all-dielectric metagratings, one can also move this kind of signal processing platforms to optical frequencies.

Figure 7: Edge detection demonstration of the metagrating processor. Illustration of the edge detection application by using the metagrating of Figure 2b. (a) The logo of “EPFL” university is utilized as the input image, (b) the reflected image captured at the first-order space harmonic, demonstrating the vertical edges of the input image.
Figure 7:

Edge detection demonstration of the metagrating processor. Illustration of the edge detection application by using the metagrating of Figure 2b. (a) The logo of “EPFL” university is utilized as the input image, (b) the reflected image captured at the first-order space harmonic, demonstrating the vertical edges of the input image.

4 Conclusions

In summary, we exploited high-order spatial harmonics of a multi-functional angle-multiplexed metagrating to realize various scattering and signal processing functionalities at the same time. Analytical expressions were presented for engineering the phase and amplitude information of each spatial harmonic. We demonstrated metagrating configurations that implement diverse optical computing functions such as first- and second-order differentiation operations, and several scattering functionalities such as retro-reflection when excited from different channels. Full-wave finite-element simulations confirmed our theory. Using the proposed angle-multiplexed metagratings, we completely escape the problem of realizing asymmetric optical transfer functions for normal illuminations, avoiding bulky additional splitting blocks to separate the routes of the incident and reflected signals, and allowing for much more compact configurations compared with previous attempts. Besides, the sparsity of our wavelength-scale array significantly relaxes fabrication tolerance, which is a crucial advantage to move the concept towards optical frequencies. This new class of multifunctional processors may find great potential applications in integrated photonic devices and imaging systems that process optical signals coming from different directions at the speed of light, as they reflect onto structured surfaces.

Corresponding authors: Ali Abdolali, Applied Electromagnetic Laboratory, School of Electrical Engineering, Iran University of Science and Technology, Tehran 1684613114, Iran, E-mail: . ; and Romain Fleury, Laboratory of Wave Engineering, School of Electrical Engineering, Swiss Federal Institute of Technology in Lausanne (EPFL), Lausanne, Switzerland, E-mail:

Funding source: Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung

Award Identifier / Grant number: 181232

  1. Author contributions: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.

  2. Research funding: A. Momeni and R. Fleury acknowledge funding from the Swiss National Science Foundation under the Eccellenza grant number 181232.

  3. Conflict of interest statement: The authors declare no conflicts of interest regarding this article.

References

[1] F. Zangeneh-Nejad, R. Fleury, D. L. Sounas, and R. Fleury, “Topological wave & insulators: a review,” Nat. Rev. Mater., vol. 21, pp. 1–19, 2020. https://doi.org/10.5802/crphys.3.Search in Google Scholar

[2] S. Abdollahramezani, O. Hemmatyar, and A. Adibi, “Meta-optics for spatial optical analog computing,” Nanophotonics, vol. 9, no. 13, p. 4075, 2020. https://doi.org/10.1515/nanoph-2020-0285.Search in Google Scholar

[3] A. Silva, F. Monticone, G. Castaldi, V. Galdi, A. Alù, and N. Engheta, “Performing mathematical operations with metamaterials,” Science, vol. 343, no. 6167, p. 160, 2014. https://doi.org/10.1126/science.1242818.Search in Google Scholar PubMed

[4] Y. Zhou, H. Zheng, I. I. Kravchenko, and J. Valentine, “Flat optics for image differentiation,” Nat. Photonics, vol. 14, no. 5, p. 316, 2020. https://doi.org/10.1038/s41566-020-0591-3.Search in Google Scholar

[5] T. Zhu, Y. Zhou, Y. Lou, et al.., “Plasmonic computing of spatial differentiation,” Nat. Commun., vol. 8, no. 1, p. 1, 2017. https://doi.org/10.1038/ncomms15391.Search in Google Scholar PubMed PubMed Central

[6] J. Zhou, H. Qian, C.-F. Chen, et al.., “Optical edge detection based on high-efficiency dielectric metasurface,” Proc. Natl. Acad. Sci. USA., vol. 116, no. 23, p. 11137, 2019. https://doi.org/10.1073/pnas.1820636116.Search in Google Scholar PubMed PubMed Central

[7] J. Sol, D. R. Smith, and P. del Hougne, arXiv preprint arXiv:2108.06178, 2021.Search in Google Scholar

[8] A. Momeni, H. Rajabalipanah, A. Abdolali, and K. Achouri, “Generalized optical signal processing based on multioperator metasurfaces synthesized by susceptibility tensors,” Phys. Rev. Appl., vol. 11, no. 6, 2019, Art no. 064042. https://doi.org/10.1103/physrevapplied.11.064042.Search in Google Scholar

[9] Y. Zhou, W. Wu, R. Chen, W. Chen, R. Chen, and Y. Ma, “Analog optical spatial differentiators based on dielectric metasurfaces,” Adv. Opt. Mater., vol. 8, no. 4, p. 1901523, 2020. https://doi.org/10.1002/adom.201901523.Search in Google Scholar

[10] A. Momeni, H. Rajabalipanah, M. Rahmanzadeh, et al.., “Reciprocal metasurfaces for on-Axis reflective optical computing,” IEEE Trans. Antenn. Propag., vol. 69, no. 11, p. 7709, 2021. https://doi.org/10.1109/tap.2021.3078521.Search in Google Scholar

[11] A. Momeni, M. Safari, A. Abdolali, N. P. Kherani, and R. Fleury, “Asymmetric metal-dielectric metacylinders and their potential applications from engineering scattering patterns to spatial optical signal processing,” Phys. Rev. Appl., vol. 15, no. 3, 2021, Art no. 034010. https://doi.org/10.1103/physrevapplied.15.034010.Search in Google Scholar

[12] H. Kwon, D. Sounas, A. Cordaro, A. Polman, and A. Alù, “Nonlocal metasurfaces for optical signal processing,” Phys. Rev. Lett., vol. 121, no. 17, p. 173004, 2018. https://doi.org/10.1103/physrevlett.121.173004.Search in Google Scholar

[13] T. Davis, F. Eftekhari, D. Gómez, and A. Roberts, “Metasurfaces with asymmetric optical transfer functions for optical signal processing,” Phys. Rev. Lett., vol. 123, no. 1, 2019, Art no. 013901. https://doi.org/10.1103/physrevlett.123.013901.Search in Google Scholar

[14] H. Rajabalipanah, A. Abdolali, S. Iqbal, L. Zhang, and T. J. Cui, “Analog signal processing through space-time digital metasurfaces,” Nanophotonics, vol. 10, no. 6, p. 1753, 2021. https://doi.org/10.1515/nanoph-2021-0006.Search in Google Scholar

[15] N. M. Estakhri, B. Edwards, and N. Engheta, “Inverse-designed metastructures that solve equations,” Science, vol. 363, no. 6433, p. 1333, 2019. https://doi.org/10.1126/science.aaw2498.Search in Google Scholar PubMed

[16] M. Camacho, B. Edwards, and N. Engheta, “A single inverse-designed photonic structure that performs parallel computing,” Nat. Commun., vol. 12, no. 1, p. 1, 2021. https://doi.org/10.1038/s41467-021-21664-9.Search in Google Scholar PubMed PubMed Central

[17] A. Abdolali, A. Momeni, H. Rajabalipanah, and K. Achouri, “Parallel integro-differential equation solving via multi-channel reciprocal bianisotropic metasurface augmented by normal susceptibilities,” New J. Phys., vol. 21, no. 11, p. 113048, 2019. https://doi.org/10.1088/1367-2630/ab26f8.Search in Google Scholar

[18] Y. Zhou, J. Zhan, R. Chen, et al.., “Analogue optical spatiotemporal differentiator,” Adv. Opt. Mater., vol. 9, no. 10, p. 2002088, 2021. https://doi.org/10.1002/adom.202002088.Search in Google Scholar

[19] A. Momeni, K. Rouhi, and R. Fleury, arXiv preprint arXiv:2104.10801, 2021.Search in Google Scholar

[20] X. Lin, Y. Rivenson, N. T. Yardimci, et al.., “All-optical machine learning using diffractive deep neural networks,” Science, vol. 361, no. 6406, p. 1004, 2018. https://doi.org/10.1126/science.aat8084.Search in Google Scholar PubMed

[21] T. W. Hughes, I. A. Williamson, M. Minkov, and S. Fan, “Wave physics as an analog recurrent neural network,” Sci. Adv., vol. 5, no. 12, 2019, Art no. eaay6946. https://doi.org/10.1126/sciadv.aay6946.Search in Google Scholar PubMed PubMed Central

[22] Y. Zuo, B. Li, Y. Zhao, et al.., “All-optical neural network with nonlinear activation functions,” Optica, vol. 6, no. 9, p. 1132, 2019. https://doi.org/10.1364/optica.6.001132.Search in Google Scholar

[23] A. Momeni and R. Fleury, arXiv preprint arXiv:2107.08564, 2021.Search in Google Scholar

[24] T. Zhu, Y. Lou, Y. Zhou, et al.., "Generalized spatial differentiation from the spin hall effect of light and its application in image processing of edge detection," Phys. Rev. Appl., vol. 11, no. 3, 2019, Art no. 034043. https://doi.org/10.1103/physrevapplied.11.034043.Search in Google Scholar

[25] M. W. Matthès, P. del Hougne, J. de Rosny, G. Lerosey, and S. M. Popoff, “Optical complex media as universal reconfigurable linear operators,” Optica, vol. 6, no. 4, p. 465, 2019. https://doi.org/10.1364/optica.6.000465.Search in Google Scholar

[26] P. del Hougne and G. Lerosey, “Leveraging chaos for wave-based analog computation: demonstration with indoor wireless communication signals,” Phys. Rev. X, vol. 8, no. 4, 2018, Art no. 041037. https://doi.org/10.1103/physrevx.8.041037.Search in Google Scholar

[27] C. Jin and Y. Yang, “Transmissive nonlocal multilayer thin film optical filter for image differentiation,” Nanophotonics, vol. 10, no. 13, 2021.10.1515/nanoph-2021-0313Search in Google Scholar

[28] F. Zangeneh-Nejad and R. Fleury, “Topological analog signal processing,” Nat. Commun., vol. 10, no. 1, p. 1, 2019. https://doi.org/10.1038/s41467-019-10086-3.Search in Google Scholar PubMed PubMed Central

[29] F. Zangeneh-Nejad and R. Fleury, “Disorder-induced signal filtering with topological metamaterials,” Adv. Mater., vol. 32, no. 28, p. 2001034, 2020.10.1002/adma.202001034Search in Google Scholar PubMed

[30] A. Babaee, A. Momeni, A. Abdolali, and R. Fleury, “Parallel analog computing based on a 2×2 multiple-input multiple-output metasurface processor with asymmetric response,” Phys. Rev. Appl., vol. 15, no. 4, 2021, Art no. 044015. https://doi.org/10.1103/physrevapplied.15.044015.Search in Google Scholar

[31] A. Babaee, A. Momeni, M. M. Moeini, R. Fleury, and A. Abdolali, Fourteenth International Congress on Artificial Materials for Novel Wave Phenomena (Metamaterials), IEEE, 2020, pp. 195–197.10.1109/Metamaterials49557.2020.9285089Search in Google Scholar

[32] X. Wang, A. Díaz-Rubio, V. S. Asadchy, et al.., “Extreme asymmetry in metasurfaces via evanescent fields engineering: angular-asymmetric absorption,” Phys. Rev. Lett., vol. 121, no. 25, p. 256802, 2018. https://doi.org/10.1103/physrevlett.121.256802.Search in Google Scholar

[33] L. Li, K. Yao, Z. Wang, and Y. Liu, “Harnessing evanescent waves by bianisotropic metasurfaces,” Laser Photon. Rev., vol. 14, no. 12, p. 1900244, 2020. https://doi.org/10.1002/lpor.201900244.Search in Google Scholar

[34] E. Abdo-Sánchez, M. Chen, A. Epstein, and G. V. Eleftheriades, “A leaky-wave antenna with controlled radiation using a bianisotropic Huygens’ metasurface,” IEEE Trans. Antenn. Propag., vol. 67, no. 1, p. 108, 2018.10.1109/TAP.2018.2878082Search in Google Scholar

[35] Y. Yang, I. I. Kravchenko, D. P. Briggs, and J. Valentine, “All-dielectric metasurface analogue of electromagnetically induced transparency,” Nat. Commun., vol. 5, no. 1, p. 1, 2014. https://doi.org/10.1038/ncomms6753.Search in Google Scholar PubMed

[36] Y. F. Yu, A. Y. Zhu, R. Paniagua-Domínguez, Y. H. Fu, B. Luk’yanchuk, and A. I. Kuznetsov, “High-transmission dielectric metasurface with 2π phase control at visible wavelengths,” Laser Photon. Rev., vol. 9, no. 4, p. 412, 2015. https://doi.org/10.1002/lpor.201500041.Search in Google Scholar

[37] S. M. Kamali, E. Arbabi, A. Arbabi, and A. Faraon, “A review of dielectric optical metasurfaces for wavefront control,” Nanophotonics, vol. 7, no. 6, p. 1041, 2018. https://doi.org/10.1515/nanoph-2017-0129.Search in Google Scholar

[38] M. Kiani, M. Tayarani, A. Momeni, H. Rajabalipanah, and A. Abdolali, “Self-biased tri-state power-multiplexed digital metasurface operating at microwave frequencies,” Opt. Express, vol. 28, no. 4, p. 5410, 2020. https://doi.org/10.1364/oe.385524.Search in Google Scholar PubMed

[39] M. Kiani, A. Momeni, M. Tayarani, and C. Ding, “Spatial wave control using a self-biased nonlinear metasurface at microwave frequencies,” Opt. Express, vol. 28, no. 23, p. 35128, 2020. https://doi.org/10.1364/oe.408622.Search in Google Scholar

[40] H. Rajabalipanah, A. Abdolali, J. Shabanpour, A. Momeni, and A. Cheldavi, “Asymmetric spatial power dividers using phase-amplitude metasurfaces driven by huygens principle,” ACS Omega, vol. 4, no. 10, p. 14340, 2019. https://doi.org/10.1021/acsomega.9b02195.Search in Google Scholar PubMed PubMed Central

[41] H. Rajabalipanah, A. Abdolali, J. Shabanpour, A. Momeni, and A. Cheldavi, arXiv preprint arXiv:1901.04063, 2019.Search in Google Scholar

[42] A. Momeni, K. Rouhi, H. Rajabalipanah, and A. Abdolali, “An information theory-inspired strategy for design of re-programmable encrypted graphene-based coding metasurfaces at terahertz frequencies,” Sci. Rep., vol. 8, no. 1, p. 1, 2018. https://doi.org/10.1038/s41598-018-24553-2.Search in Google Scholar PubMed PubMed Central

[43] H. Rajabalipanah, K. Rouhi, A. Abdolali, S. Iqbal, L. Zhang, and S. Liu, “Real-time terahertz meta-cryptography using polarization-multiplexed graphene-based computer-generated holograms,” Nanophotonics, vol. 9, no. 9, p. 2861, 2020. https://doi.org/10.1515/nanoph-2020-0110.Search in Google Scholar

[44] K. Rouhi, H. Rajabalipanah, and A. Abdolali, “Multi-bit graphene-based bias-encoded metasurfaces for real-time terahertz wavefront shaping: from controllable orbital angular momentum generation toward arbitrary beam tailoring,” Carbon, vol. 149, p. 125, 2019. https://doi.org/10.1016/j.carbon.2019.04.034.Search in Google Scholar

[45] Y. Ra’di, D. L. Sounas, and A. Alù, “Metagratings: beyond the limits of graded metasurfaces for wave front control,” Phys. Rev. Lett., vol. 119, no. 6, 2017, Art no. 067404.10.1103/PhysRevLett.119.067404Search in Google Scholar PubMed

[46] Y. Ra’di and A. Alù, “Reconfigurable metagratings,” ACS Photonics, vol. 5, no. 5, p. 1779, 2018.10.1021/acsphotonics.7b01528Search in Google Scholar

[47] M. Rahmanzadeh and A. Khavasi, “Perfect anomalous reflection using a compound metallic metagrating,” Opt. Express, vol. 28, no. 11, p. 16439, 2020. https://doi.org/10.1364/oe.393137.Search in Google Scholar PubMed

[48] M. Rahmanzadeh, A. Khavasi, and B. Rejaei, “Analytical method for diffraction analysis and design of perfect-electric-conductor backed graphene ribbon metagratings,” Opt. Express, vol. 29, no. 18, p. 28935, 2021. https://doi.org/10.1364/oe.436241.Search in Google Scholar

[49] D. Sell, J. Yang, S. Doshay, and J. A. Fan, “Periodic dielectric metasurfaces with high-efficiency, multiwavelength functionalities,” Adv. Opt. Mater., vol. 5, no. 23, p. 1700645, 2017. https://doi.org/10.1002/adom.201700645.Search in Google Scholar

[50] D. Sell, J. Yang, S. Doshay, R. Yang, and J. A. Fan, “Large-angle, multifunctional metagratings based on freeform multimode geometries,” Nano Lett., vol. 17, no. 6, p. 3752, 2017. https://doi.org/10.1021/acs.nanolett.7b01082.Search in Google Scholar PubMed

[51] S. Behroozinia, H. Rajabalipanah, and A. Abdolali, “Real-time terahertz wave channeling via multifunctional metagratings: a sparse array of all-graphene scatterers,” Opt. Lett., vol. 45, no. 4, p. 795, 2020. https://doi.org/10.1364/ol.383001.Search in Google Scholar PubMed

[52] H. Rajabalipanah and A. Abdolali, “Analytical design for full-space spatial power dividers using metagratings,” J. Opt. Soc. Am. B, vol. 38, no. 10, p. 2915, 2021. https://doi.org/10.1364/josab.437379.Search in Google Scholar

[53] O. Rabinovich and A. Epstein, “Reflection-contrast limit of fiber-optic image guides,” Phys. Rev. Appl., vol. 14, no. 6, 2020, Art no. 064028. https://doi.org/10.1103/physrevapplied.14.064028.Search in Google Scholar

[54] B. Zhou, W. Jia, P. Sun, J. Wang, W. Liu, and C. Zhou, “Polarization-independent high diffraction efficiency two-dimensional grating based on cylindrical hole nano arrays,” Opt. Express, vol. 28, no. 20, p. 28810, 2020. https://doi.org/10.1364/oe.402131.Search in Google Scholar

[55] M. Rahmanzadeh and A. Khavasi, arXiv preprint arXiv:2110.13912, 2021.Search in Google Scholar

[56] I. Duling and D. Zimdars, “Revealing hidden defects,” Nat. Photonics, vol. 3, no. 11, p. 630, 2009. https://doi.org/10.1038/nphoton.2009.206.Search in Google Scholar

[57] T.-J. Huang, J. Zhao, L.-Z. Yin, and P.-K. Liu, “Terahertz subwavelength edge detection based on dispersion-induced plasmons,” Opt. Lett., vol. 46, no. 11, p. 2746, 2021. https://doi.org/10.1364/ol.421992.Search in Google Scholar

Supplementary Material

The online version of this article offers supplementary material (https://doi.org/10.1515/nanoph-2021-0710).

Received: 2021-11-12
Accepted: 2022-03-09
Published Online: 2022-03-24

© 2022 Hamid Rajabalipanah et al., published by De Gruyter, Berlin/Boston

This work is licensed under the Creative Commons Attribution 4.0 International License.

Articles in the same Issue

  1. Frontmatter
  2. Reviews
  3. Photoactive nanomaterials enabled integrated photo-rechargeable batteries
  4. Recent progress in terahertz metamaterial modulators
  5. Research Articles
  6. Droplet epitaxy symmetric InAs/InP quantum dots for quantum emission in the third telecom window: morphology, optical and electronic properties
  7. Monolithic dual-wedge prism-based spectroscopic single-molecule localization microscopy
  8. Four-dimensional multi-particle tracking in living cells based on lifetime imaging
  9. Orthogonal gap-enhanced Raman tags for interference-free and ultrastable surface-enhanced Raman scattering
  10. Parallel wave-based analog computing using metagratings
  11. Gauge-independent emission spectra and quantum correlations in the ultrastrong coupling regime of open system cavity-QED
  12. Combining one and two photon polymerization for accelerated high performance (3 + 1)D photonic integration
  13. High power downconversion deep-red emission from Ho3+-doped fiber lasers
  14. Topological optical parametric oscillation
  15. Optically tunable split-ring resonators controlled lead sulfide quantum dots modulator for wide THz radiation
  16. Ultrasensitive nanoscale optomechanical electrometer using photonic crystal cavities
  17. Negative index metamaterial at ultraviolet range for subwavelength photolithography
  18. Erratum
  19. Erratum to: Physics and applications of quantum dot lasers for silicon photonics
https://doi.org/10.1515/nanoph-2021-0710
Keywords for this article
analog optical computing; metagrating; metasurface
Creative Commons
BY 4.0

Articles in the same Issue

  1. Frontmatter
  2. Reviews
  3. Photoactive nanomaterials enabled integrated photo-rechargeable batteries
  4. Recent progress in terahertz metamaterial modulators
  5. Research Articles
  6. Droplet epitaxy symmetric InAs/InP quantum dots for quantum emission in the third telecom window: morphology, optical and electronic properties
  7. Monolithic dual-wedge prism-based spectroscopic single-molecule localization microscopy
  8. Four-dimensional multi-particle tracking in living cells based on lifetime imaging
  9. Orthogonal gap-enhanced Raman tags for interference-free and ultrastable surface-enhanced Raman scattering
  10. Parallel wave-based analog computing using metagratings
  11. Gauge-independent emission spectra and quantum correlations in the ultrastrong coupling regime of open system cavity-QED
  12. Combining one and two photon polymerization for accelerated high performance (3 + 1)D photonic integration
  13. High power downconversion deep-red emission from Ho3+-doped fiber lasers
  14. Topological optical parametric oscillation
  15. Optically tunable split-ring resonators controlled lead sulfide quantum dots modulator for wide THz radiation
  16. Ultrasensitive nanoscale optomechanical electrometer using photonic crystal cavities
  17. Negative index metamaterial at ultraviolet range for subwavelength photolithography
  18. Erratum
  19. Erratum to: Physics and applications of quantum dot lasers for silicon photonics
Sign up now to receive a 20% welcome discount
Institutional Access
How does access work?
Have an idea on how to improve our website?
Please write us.
© 2026 De Gruyter Brill
Downloaded on 29.1.2026 from https://www.degruyterbrill.com/document/doi/10.1515/nanoph-2021-0710/html
Scroll to top button