Expand this Topic clickable element to expand a topic
Skip to content
Optica Publishing Group

Mathematical synthesis and analysis of a second-order magneto-electrically nonlinear metasurface

Open Access Open Access

Abstract

We propose a discussion on the synthesis and scattering analysis of nonlinear metasurfaces. For simplicity, we investigate the case of a second-order nonlinear isotropic metasurface possessing both electric and magnetic linear and nonlinear susceptibility components. We next find the synthesis expressions relating the susceptibilities to the specified fields, which leads to the definition of the nonlinear metasurface conditions for no reflection, themselves revealing the nonreciprocal nature of such structures. Finally, we provide the approximate expressions of the scattered fields based on perturbation theory and compare the corresponding results to finite-difference time-domain simulations.

© 2017 Optical Society of America

1. Introduction

Over the past few years, metasurfaces, the two-dimensional counterparts of three-dimensional metamaterials, have proven to be particularly effective at controlling electromagnetic waves. However, most studies on metasurfaces have been restricted to purely linear structures and only few studies, as for instance [1–4], have investigated the synthesis and/or scattering from nonlinear metasurfaces, but without providing extensive discussion on the topic. Since nonlinearity may potentially bring about a wealth of new applications to the realm of metasurface-based effects, such as for instance nonreciprocity, second-harmonic generation and wave-mixing [5], we propose here a rigorous discussion on the synthesis and scattering analysis from second-order nonlinear metasurfaces.

In the following, we will use the generalized sheet transition conditions (GSTCs) to obtain the metasurface susceptibilities (linear and nonlinear components) in terms of specified incident, reflected and transmitted waves. Based on that, the conditions for no reflection for second-order nonlinear metasurfaces will be derived, which will be next used to analyze the scattering from such structures. The scattered field will be computed using perturbation analysis and the results will be compared with FDTD simulations.

2. Synthesis of second-order nonlinear metasurfaces

The GSTCs are boundary conditions that apply to zero-thickness discontinuities such as metasurfaces [6–8]. These conditions relate the discontinuities of the electric and magnetic fields to the presence of excitable surface polarization densities. We hereafter assume the harmonic time dependence ejωt. In the case of a metasurface lying in the xy-plane at z = 0 and assuming only transverse polarizations, the GSTCs read

z^×ΔH=jωP,
ΔE×z^=jωμ0M,
where Δ indicates the difference of the fields between both sides of the metasurface.

Let us now investigate the case of a metasurface with nonzero second-order nonlinear electric and magnetic susceptibility tensors. The presence of non-negligible electric nonlinearities may be found, for instance, in optical nonlinear crystals [5] while magnetic nonlinearities may be found in ferrofluids [9]. We restrict our attention to the case of monoanisotropic metasurfaces [8], whose electric and magnetic polarization densities are

P=0χ̿ee(1)Eav+0χ̿ee(2)Eav2,
M=χ̿mm(1)Hav+χ̿mm(2)Hav2,
where χ̿(1) and χ̿(2) correspond to the first-order (linear) and second-order (nonlinear) susceptibility tensors, respectively, and the subscript “av” denotes the average of the field between both sides of the metasurface. For simplicity, we assume that the only nonzero components of these tensors are χ̿aa,(1) and χ̿aaa,(2), where a = {x, y}, which generallly corresponds to the case of a birefringent nonlinear metasurface. Since nonlinear media generate new frequencies [5], the frequency-domain GSTCs in (1) are not appropriate to investigate the response of nonlinear metasurfaces since they relate electromagnetic fields with the same frequency. To overcome this issue, we express the GSTCs in the time-domain instead of the frequency-domain. To further simplify the discussion, we consider, without loss of generality, only the case of x-polarized waves which, upon insertion of (2) into the time-domain version of (1), reduces the time-domain GSTCs to
ΔH=0χee(1)tEav+0χee(2)tEav2,
ΔE=μ0χmm(1)tHav+μ0χmm(2)tHav2,
where E and H are, respectively, the x-component of the electric field and the y-component of the magnetic field, and where the susceptibility components are those corresponding to x-polarized excitation. Equations (3) assume susceptibility time-invariance. They also seem to assume no dispersion, since dispersion would generally imply time convolutions in (2) preventing the susceptibilities to be factored out of the time derivative operators. In reality, since the metasurface will be specified to respond only at harmonics of the excitation, the actual condition on dispersion is dramatically relaxed to the requirement χ(ω0) = χ(2ω0) = χ(3ω0) = . . ., which may be much easier to satisfy in practice. Note that this condition is equivalent to the phase matching condition conventionally used in nonlinear optics [5].

To synthesize a nonlinear metasurface, one needs to solve (3) so as to express the susceptibilities as functions of the electromagnetic fields on both sides of the metasurface. As it stands in (3), the system has two equations for four unknowns, and is hence under-determined. If we consider two arbitrary transformations instead of just one [8], then the system becomes a full-rank one, reading

[ΔH1ΔH2ΔE1ΔE2]=t[0Eav,10Eav,12000Eav,20Eav,220000μ0Hav,1μ0Hav,1200μ0Hav,2μ0Hav,22][χee(1)χee(2)χmm(1)χmm(2)],
where the subscripts 1 and 2 refer to the fields of two arbitrary transformations. This matrix system is easily solved and yields the following expressions for the susceptibilities
χee(1)=ΔH2tEav,12ΔH1tEav,220(tEav,12tEav,2tEav,1tEav,22),
χmm(1)=ΔE2tHav,12ΔE1tHav,22μ0(tHav,12tHav,2tHav,1tHav,22),
χee(2)=ΔH2tEav,1ΔH1tEav,20(tEav,12tEav,2tEav,1tEav,22),
χmm(2)=ΔE2tHav,1ΔE1tHav,2μ0(tHav,12tHav,2tHav,1tHav,22).

At this stage, one may think that substituting the specified arbitrary incident, reflected and transmitted fields into (5) would lead to well-defined susceptibilities. However, the specification of even “simple” transformations as, for instance, Ei = E0ej(ωitkiz), Er = Rej(ωrtkrz) and Et = Tej(ωttktz), leads to susceptibilities that are time-varying, irrespectively of the values of ωi, ωr and ωt. The fact that the susceptibilities are, in this case, time-varying is inconsistent with the implicit assumption in Eqs. (3) that they are not. Note that if the susceptibilities were time-dependent, insertion of (2) into the time-domain version of (1) would lead, from the derivation chain rule, to additional terms of the form (χuu(k)/t)Eav, Hav, where u =e,m and k = 1, 2. At this point, one may therefore wonder whether the problem has been adequately paused? However, fortunately, we shall see in the next section that this is not the case: the problem is adequately posed but the specified fields must satisfy a specific constraint, while being otherwise still synthesizable. This constraint will be established by considering specified fields satisfying the postulate that the susceptibilities in (5) are not functions of time. For this purpose, we will look at the problem from a different perspective. Instead of trying to synthesize the metasurface, we shall heuristically analyze its scattering with (arbitrary) known susceptibilities in order to understand what kind of reflected and transmitted fields are produced by such nonlinear metasurfaces, and hence deduce a proper way to perform the synthesis.

3. Scattering from second-order nonlinear metasurfaces

The fields scattered from a nonlinear metasurface may be obtained by solving (3). However, Eqs. (3) form a set of nonlinear inhomogeneous first-order coupled differential equations, that is not trivial to solve analytically. The problem may be simplified by assuming that the metasurface is reflectionless, which reduces (3) to a single equation. The conditions for no reflection in a nonlinear metasurface may be obtained by specifying Er = Hr = 0 in (5) and assuming normally incident and transmitted plane waves, i.e. E = ±η0H, where + corresponds to waves propagating in the +z-direction and vice-versa for −. To obtain the susceptibilities in (5), we have to consider the transformation of two sets of independent waves. One may assume that the incident and transmitted waves for the two transformations are either both propagating in the +z-direction, as in Fig. 1(a), or in the −z-direction, as in Fig. 1(b). In these figures, Ψ1 and Ψ2 correspond to two sets of independent x-polarized normally propagating plane waves being fully transmitted by the metasurface. The subscripts i and t refer to the incident and transmitted waves, respectively. One may also consider the case where the waves Ψ1 are propagating in the +z-direction and the waves Ψ2 are propagating in the −z-direction (or vice-versa), but it may be shown that conditions for no reflection do not exist in this case.

 figure: Fig. 1

Fig. 1 Two different approaches to synthesize a nonlinear reflectionless metasurface. (a) Metasurface reflectionlessly transmitting two waves (Ψ1 and Ψ2) in the +z-direction. (b) Metasurface reflectionlessly transmitting two waves (Ψ1 and Ψ2) in the −z-direction.

Download Full Size | PDF

It will next be shown that the conditions for no reflection are not the same in the two cases depicted in Fig. 1. This is because of the presence of the square of the electric and magnetic fields in (5), which introduces an asymmetry in the definition of the susceptibilities in terms of the direction of wave propagation. This asymmetry is due to the different relations between the electric and magnetic fields (E = ±η0H) for forward or backward propagating waves. Solving (5) for the case depicted in Fig. 1(a) leads to the following conditions for no reflection:

χee(1)=χmm(1),
η0χee(2)=χmm(2).
Similarly, the conditions for no reflection in the case depicted in Fig. 1(b) read
χee(1)=χmm(1),
η0χee(2)=χmm(2).
Note the minus sign difference in the relations (6b) and (7b) between the second-order susceptibilities. The fact that different reflectionless metasurface conditions are obtained for different directions of propagation means that the considered nonlinear metasurface inherently exhibits a nonreciprocal response, that we will discuss in more detail later on.

As of now, we continue the evaluation of scattering from the nonlinear metasurface considering the case in Fig. 1(a). Substituting (6) along with the difference of the specified fields, ΔE = EtEi, the average of the specified fields, Eav=12(Et+Ei), and the squared average of the specified fields, Eav2=EavEav*=14(Ei2+EiEt*+EtEi*+Et2), and similarly for the magnetic fields, transforms (3b) into

2χee(1)tEt+χee(2)t(EiEt*+EtEi*+Et2)+4η0μ0Et=4η0μ0Eiχee(2)tEi22χee(1)tEi,
where Ei is a known excitation and Et is an unknown transmitted field. Assuming that Ei = E0 cos (ω0t) and that E0, χee(1) and χee(2) are real quantities, corresponding to a lossless system, the relation (8) becomes
χee(2)tEt2+(2χee(1)+2E0cos(ω0t)χee(2))tEt+(4η0μ02ω0χee(2)E0sin(ω0t))Et=4η0μ0E0cos(ω0t)+ω0χee(2)E0sin(2ω0t)+2ω0χee(1)E0sin(ω0t).
This is a inhomogeneous nonlinear first-order differential equation that allows one to find the transmitted field from a reflectionless birefringent second-order nonlinear metasurface with purely real susceptibilities and assuming a normally incident plane wave excitation in the +z-direction. The large number of assumptions that were required to obtain Eq. (9) reveals the inherent complexity of analyzing nonlinear metasurfaces.

Equation (9) does not admit an analytical solution. To obtain an approximate expression of the transmitted field, we consider that the second-order susceptibilities are much smaller than the first-order ones, which is typically a valid assumption in the absence of second order resonance, χ(2) ≈ 10−12 χ(1) [5]. From this consideration, perturbation analysis [10] may be used to approximate the value of the transmitted field. Perturbation analysis stipulates that the approximate solution may be expressed in terms of a power series of the following form

EtEt,0+Et,1+2Et,2+
where is a small quantity. Truncating the series and solving recursively for Et,0, Et,1 and so on, may help reducing the complexity of the problem. Since
χee(1)χee(2),
it is possible to simplify Eq. (9) using (10) and solving for Et,0 while neglecting all terms containing . This reduces (9) to
2χee(1)tEt,0+4η0μ0Et,0=4η0μ0E0cos(ω0t)+2ω0χee(1)E0sin(ω0t),
which does not contain any nonlinear term and therefore corresponds to a simple reflectionless linear metasurface. The steady-state solution of (12) is, in complex form, given by
Et,0=E02jk0χee(1)2+jk0χee(1)ejω0t,
which exactly corresponds to the expected transmitted field [8], where the frequency of Et,0 is the same as that of the incident wave. Now, Et,1 can be found by inserting EtEt,0 + ∊Et,1 into (9), with Et,0 as given in (13), and neglecting all the terms containing 2 (and higher powers). This leads to the following linear differential equation.
χee(2)tEt,02(2χee(1)+2E0cos(ω0t)χee(2))tEt,0+(4η0μ02ω0χee(2)E0sin(ω0t))Et,0+2χee(1)tEt,1+4η0μ0Et,1=4η0μ0E0cos(ω0t)+ω0χee(2)E0sin(2ω0t)+2ω0χee(1)E0sin(ω0t),
which now contains the nonlinear susceptibility χee(2). This equation, of the same type as (12) in Et,1, is readily solved, and yields the steady-state solution
Et,1=E0k0χee(2)ej2ω0t[4+12E0+χee(1)k0(χee(1)k04j)(E01)4(χee(1)k0j)(χee(1)k02j)2],
which corresponds to second-harmonic generation, i.e. a wave at frequency 2ω0, twice that of the incident wave. The procedure used to obtain Et,1 may now be applied to find Et,2. In this case, the differential equation becomes quite lengthy and is not shown here for the sake of conciseness. The differential equation for Et,2 is also a linear first-order equation and is thus easily solved. The steady-state solution is
Et,2=Ct,2[(2j3χee(1)k0)ejω0t+3(2j+χee(1)k0)ej3ω0t],
where the complex constant Ct,2 reads
Ct,2=j(χee(2)k0E0)22(χee(1)k02j)3[4+12E0+χee(1)k0(χee(1)k04j)(E01)](χee(1)k0j)(χee(1)k0+2j)(3χee(1)k02j).
The expression of Et,2 corresponds to a superposition of two waves, at frequencies ω0 and 3ω0. From (17), we see that the amplitude of Et,2 is directly proportional to the square of χee(2) while the amplitude of Et,1 in (15) is linearly proportional to χee(2). Similarly, Et,2 is proportional to the cube of E0 while Et,1 is proportional to the square of E0. Consequently, relations (15) and (16) remain valid only for values of E0, χee(1) and χee(2) such that Et,0Et,1Et,2 and χee(1)χee(2).

According to (13), (15) and (16), the scattered field from the nonlinear metasurfaces may generally be expressed as

Es=n=1Es,nejnω0t,
where Es represents either the reflected or the transmitted electric field and Es,n are complex constants. The form (18) reveals why the synthesis of a nonlinear metasurface is not trivial: all the harmonics playing a significant role in (18) should be included in the specified fields. In other words, if those harmonics were not included in the specified fields, one would not properly describe the physics of the problem. This would in fact precisely lead to the aforementioned contradiction that the susceptibilities in (5) would be found to be depending on time in contraction with the implicit assumption in (3) that they do not.

We shall now validate our theory with full-wave analysis. The metasurface is synthesized so as to satisfy the conditions for no reflection given in (6), and we consider the following arbitrary parameters: E0 = 1.5 V/m, χee(1)=0.1m and χee(2)=0.04m2/V. The simulations are performed with the 1D FDTD zero-thickness metasurface simulation code developed in [11], modified to account for the nonlinear susceptibilities (see Appendix A). In the forthcoming simulations, we use the usual normalized constants c0 = 0 = μ0 = f0 = 1.

In the first simulation, whose results are plotted in Fig. 2(a), the metasurface is illuminated by a plane wave propagating in the +z-direction. In the figure, the metasurface is placed at the center and the simulation area is split into a scattered-field region (SF) and a total-field region (TF). The source is placed at the SF/TF boundary on the left of the metasurface. As expected, the metasurface is reflectionless (E = 0 in the SF region). A time-domain Fourier transform of the steady-state transmitted field is performed and the normalized (to E0) result is plotted in Fig. 2(b). It may be seen that the metasurface generates a transmitted field with several visible harmonics satisfying E(ωn) > E(ωn+1), as expected.

 figure: Fig. 2

Fig. 2 FDTD simulations of a nonlinear metasurface illuminated by a plane wave from the left with the following parameters: E0 = 1.5 V/m, χee(1)=0.1m and χee(2)=0.04m2/V. (a) Electric field waveform. (b) Spectrum.

Download Full Size | PDF

In order to investigate its nonreciprocity, the metasurface is now illuminated from the left by a plane wave propagating in the −z-direction. The corresponding simulated waveform is plotted in Fig. 3(a), where the positions of the SF/TF regions have been changed accordingly. We can see that when this metasurface is illuminated from the right, it is not reflectionless anymore as evidenced by the nonzero electric field in the SF region. This is in agreement with the fact that different conditions for no reflection apply for different directions of propagation, according to the discussion that led to Eqs. (6) and (7).

 figure: Fig. 3

Fig. 3 FDTD simulations of a nonlinear metasurface illuminated by a plane wave from the right with the following parameters: E0 = 1.5 V/m, χee(1)=0.1m and χee(2)=0.04m2/V. (a) Electric field waveform. (b) Spectrum.

Download Full Size | PDF

The time-domain Fourier transform of the transmitted field in Fig. 3(a) is plotted in Fig. 3(b). It reveals that the frequencies that are odd multiples of ω0 are transmitted, while the even multiples of ω0 are strongly suppressed. Since the metasurface is lossless (purely real susceptibilities), we can infer that the frequencies that are even multiples of ω0 are thus mostly reflected. The generation of transmitted waves, with corresponding frequencies that are even multiples of the excitation frequency, is thus a nonreciprocal process since S21ω02nω0S12ω02nω0, where n = 1, 2, 3, ...

Next, we shall compare the theoretical results with FDTD simulations. For this purpose, two metasurfaces satisfying (6) with different susceptibilities are considered. The values of E0 and χee(1) are specified while those of χee(2) are swept. The amplitudes of the first three harmonics (at ω0, 2ω0 and 3ω0) of the transmitted field are obtained from (13), (15) and (16), and compared to the corresponding amplitudes found by FDTD simulation. Note that the amplitude of the harmonic at ω0 is computed from both (13) and (16) since both of these equations include terms contributing to this harmonic. As explained in the Appendix A, the FDTD field update equations are only valid for specific values of E0, χee(1) and χee(2), other specifications leading to a nonphysical behavior. In addition to choosing those parameters so as to follow this constraint, we have ensured χee(1)>χee(2) in order to be consistent with the assumptions of the perturbation analysis method [see Eq. (11)].

The first comparison, presented in Fig. 4(a), considers E0 = 1.5 V/m, χee(1)=0.1m and χee(2) swept in [0, 0.04] m2/V, while the second comparison, presented in Fig. 4(b), considers E0 = 10 V/m, χee(1)=0.03m and χee(2) swept in [0, 0.01] m2/V. Both comparisons show good agreement between theory and simulation. In both cases, the discrepancies between the results of the two methods increase with increasing χee(2), as expected from the perturbation assumption χee(1)>χee(2). Thus, as χee(2) increases towards χee(1), the error in the approximation of the transmitted field (Eq. (10)) progressively increases. Another source of error is the fact that Eq. (10) is truncated at the third term and that higher-order terms are neglected, which induces additional errors.

 figure: Fig. 4

Fig. 4 Comparisons of the amplitudes of the first three normalized harmonics obtained by theory [Eqs. (13), (15) and (16)] (blue solid lines) and FDTD simulation (red dashed lines) for the following parameters: (a) E0 = 1.5 V/m, χee(1)=0.1m. (b) E0 = 10 V/m, χee(1)=0.03m.

Download Full Size | PDF

We have verified that the FDTD simulations in Figs. 2 and 3, as well as in Figs. 4(a) and 4(b) satisfy the losslessness and passivity power conservation condition n=1|E(ωn)|2=|E0|2. In contrast, this condition does not hold in the case of the theoretical results, due to the truncation of (10), as is clearly apparent in Fig. 4(b), where n=13|E(ωn)|2>|E0|2 for values of χee(2) that are close to 0.01 m2/V.

4. Conclusion

We have investigated a particular case of second-order nonlinear isotropic metasurfaces that possess both electric and magnetic nonlinear susceptibilities. We have found the synthesis expressions relating the susceptibilities to the fields on both sides of the metasurface, which lead to the derivation of the reflectionless metasurface conditions. These conditions reveal the inherent nonreciprocal nature of nonlinear metasurfaces. Then, the scattered field from such metasurfaces was analyzed based on perturbation theory as well as full-wave simulations, and good agreement was found between the two approaches.

A. Finite-difference time-domain scheme for nonlinear metasurfaces

Here, we extend the 1D finite-difference time-domain (FDTD) simulation scheme developed in [11] for the analysis of metasurfaces to the case of nonlinear susceptibility components. This FDTD scheme consists in using traditional FDTD update equations everywhere on the simulation grid except at the nodes positioned before and after the metasurface. For these specific nodes, the update equations are modified, using the GSTCs relations, to take into account the effect of the metasurface. The conventional Yee-grid FDTD 1D equations are given by

Hyn+1(i)=Hyn(i)Δtμ0Δz(Exn+12(i+1)Exn+12(i)),
Exn+1/2(i)=Exn12(i)Δt0Δz(Hyn(i)Hyn(i1)),
where i and n correspond to the cell number and time coordinates and Δz and Δt are their respective position and time steps. The metasurface is placed at a virtual position between cell number i = nd and i = nd + 1, corresponding to a position between an electric node and a magnetic node. To take into account its effect, a virtual electric node is created just before the metasurface (at i = 0) and a virtual magnetic node is created just after the metasurface (at i = 0+).

From (19), the update Eqs. for Hyn+1(nd) and Exn+12(nd+1) are connected to these virtual nodes via the following relations

Hyn+1(nd)=Hyn(nd)+Δtμ0Δz(Exn+12(0)Exn+12(nd)),
Exn+12(nd+1)=Exn12(nd+1)+Δt0Δz(Hyn(nd+1)Hyn(0+)),
where the value of the electric and magnetic fields at the virtual nodes are obtained from the GSTCs relations
ΔHy=0χee(1)tEx,av+0χee(2)tEx,av2,
ΔEx=μ0χmm(1)tHy,av+μ0χmm(2)tHy,av2.
Using (21), the expression of the electric and magnetic fields at the virtual nodes in (20) read
Hyn(0+)=Hyn(nd)0χee(1)Δt(Ex,avn+12Ex,avn12)0χee(2)Δt((Ex,avn+12)2(Ex,avn12)2),
Exn+12(0)=Exn+12(nd+1)+μ0χmm(1)Δt(Hy,avn+1Hy,avn)+μ0χmm(2)Δt((Hy,avn+1)2(Hy,avn)2),
where the average electric field is defined by
Ex,avn+12=Exn+12(nd)+Exn+12(nd+1)2,
and similarly for the average magnetic field. Substituting (22) along with (23) into (20) leads to two quadratic equations that may be independently solved to obtain the final update equations. These two quadratic equations yield, each one of them, two possible solutions but only one of the two correspond to a physical behavior. The two solutions that produce physical results are
Hyn+1(nd)=(2Δzχee(1)Hyn+1(nd+1)χee(2))μ0Δhμ0χee(1),
Exn+12(nd+1)=(2Δzχmm(1)Exn+12(nd)χmm(2))0Δe0χmm(1)
where the discriminant Δh is given by
Δh=μ0{4Δt(Exn+12(nd)Exn+12(nd+1))χee(2)+μ0[4Δz2+(χee(1)+(Hyn(nd)+Hyn+1(nd+1))χee(2))24Δz(χee(1)+(Hyn(nd)+Hyn+1(nd+1))χee(2))]},
and the discriminant Δe is given by
Δe=0{4Δt(Hyn(nd)Hyn(nd+1))χmm(2)+0[4Δz2+(χmm(1)+(Exn12(nd)+Exn12(nd+1))χmm(2))24Δz(χmm(1)+(Exn+12(nd)+Exn12(nd+1))χmm(2))]}.

Because of the square roots in (24), the update equations may lead to nonphysical behavior depending on the values of the two discriminants. This limits the range of allowable values that the susceptibilities and the amplitude of the incident field may take.

Funding

Natural Sciences and Engineering Research Council of Canada (NSERC) Collaborative Research and Development Project CRDPJ 478303-14 in partnership with the company Metamaterial Technology Inc.

References and links

1. J. Lee, M. Tymchenko, C. Argyropoulos, P.-Y. Chen, F. Lu, F. Demmerle, G. Boehm, M.-C. Amann, A. Alù, and M. A. Belkin, “Giant nonlinear response from plasmonic metasurfaces coupled to intersubband transitions,” Nature 511, 65–69 (2014). [CrossRef]   [PubMed]  

2. H. Wakatsuchi, S. Kim, J. J. Rushton, and D. F. Sievenpiper, “Waveform-dependent absorbing metasurfaces,” Phys. Rev. Lett. 111, 245501 (2013). [CrossRef]  

3. P.-Y. Chen and A. Alù, “Optical nanoantenna arrays loaded with nonlinear materials,” Phys. Rev. B 82, 235405 (2010). [CrossRef]  

4. Y. Yang, W. Wang, A. Boulesbaa, I. I. Kravchenko, D. P. Briggs, A. Puretzky, D. Geohegan, and J. Valentine, “Nonlinear fano-resonant dielectric metasurfaces,” Nano Lett. 15, 7388–7393 (2015). [CrossRef]   [PubMed]  

5. R. W. Boyd, Nonlinear optics (Taylor & Francis, 2003).

6. M. M. Idemen, Discontinuities in the Electromagnetic Field (John Wiley & Sons, 2011). [CrossRef]  

7. E. F. Kuester, M. Mohamed, M. Piket-May, and C. Holloway, “Averaged transition conditions for electromagnetic fields at a metafilm,” IEEE Trans. Antennas Propag. 51, 2641–2651 (2003). [CrossRef]  

8. K. Achouri, M. A. Salem, and C. Caloz, “General metasurface synthesis based on susceptibility tensors,” IEEE Trans. Antennas Propag. 63, 2977–2991 (2015). [CrossRef]  

9. G. Wang and J. Huang, “Nonlinear magnetic susceptibility of ferrofluids,” Chem. Phys. Lett. 421, 544–548 (2006). [CrossRef]  

10. T. Kato, Perturbation theory for linear operators, vol. 132 (Springer Science & Business Media, 2013).

11. Y. Vahabzadeh, N. Chamanara, and C. Caloz, “Space-time varying metasurface simulation in finite difference time domain,” https://arXiv:1701.08760 (2017).

Cited By

Optica participates in Crossref's Cited-By Linking service. Citing articles from Optica Publishing Group journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (4)

Fig. 1
Fig. 1 Two different approaches to synthesize a nonlinear reflectionless metasurface. (a) Metasurface reflectionlessly transmitting two waves (Ψ1 and Ψ2) in the +z-direction. (b) Metasurface reflectionlessly transmitting two waves (Ψ1 and Ψ2) in the −z-direction.
Fig. 2
Fig. 2 FDTD simulations of a nonlinear metasurface illuminated by a plane wave from the left with the following parameters: E0 = 1.5 V/m, χ ee ( 1 ) = 0.1 m and χ ee ( 2 ) = 0.04 m 2 / V. (a) Electric field waveform. (b) Spectrum.
Fig. 3
Fig. 3 FDTD simulations of a nonlinear metasurface illuminated by a plane wave from the right with the following parameters: E0 = 1.5 V/m, χ ee ( 1 ) = 0.1 m and χ ee ( 2 ) = 0.04 m 2 / V. (a) Electric field waveform. (b) Spectrum.
Fig. 4
Fig. 4 Comparisons of the amplitudes of the first three normalized harmonics obtained by theory [Eqs. (13), (15) and (16)] (blue solid lines) and FDTD simulation (red dashed lines) for the following parameters: (a) E0 = 1.5 V/m, χ ee ( 1 ) = 0.1 m. (b) E0 = 10 V/m, χ ee ( 1 ) = 0.03 m.

Equations (39)

Equations on this page are rendered with MathJax. Learn more.

z ^ × Δ H = j ω P ,
Δ E × z ^ = j ω μ 0 M ,
P = 0 χ̿ ee ( 1 ) E av + 0 χ̿ ee ( 2 ) E av 2 ,
M = χ̿ mm ( 1 ) H av + χ̿ mm ( 2 ) H av 2 ,
Δ H = 0 χ ee ( 1 ) t E av + 0 χ ee ( 2 ) t E av 2 ,
Δ E = μ 0 χ mm ( 1 ) t H av + μ 0 χ mm ( 2 ) t H av 2 ,
[ Δ H 1 Δ H 2 Δ E 1 Δ E 2 ] = t [ 0 E av , 1 0 E av , 1 2 0 0 0 E av , 2 0 E av , 2 2 0 0 0 0 μ 0 H av , 1 μ 0 H av , 1 2 0 0 μ 0 H av , 2 μ 0 H av , 2 2 ] [ χ ee ( 1 ) χ ee ( 2 ) χ mm ( 1 ) χ mm ( 2 ) ] ,
χ ee ( 1 ) = Δ H 2 t E av , 1 2 Δ H 1 t E av , 2 2 0 ( t E av , 1 2 t E av , 2 t E av , 1 t E av , 2 2 ) ,
χ mm ( 1 ) = Δ E 2 t H av , 1 2 Δ E 1 t H av , 2 2 μ 0 ( t H av , 1 2 t H av , 2 t H av , 1 t H av , 2 2 ) ,
χ ee ( 2 ) = Δ H 2 t E av , 1 Δ H 1 t E av , 2 0 ( t E av , 1 2 t E av , 2 t E av , 1 t E av , 2 2 ) ,
χ mm ( 2 ) = Δ E 2 t H av , 1 Δ E 1 t H av , 2 μ 0 ( t H av , 1 2 t H av , 2 t H av , 1 t H av , 2 2 ) .
χ ee ( 1 ) = χ mm ( 1 ) ,
η 0 χ ee ( 2 ) = χ mm ( 2 ) .
χ ee ( 1 ) = χ mm ( 1 ) ,
η 0 χ ee ( 2 ) = χ mm ( 2 ) .
2 χ ee ( 1 ) t E t + χ ee ( 2 ) t ( E i E t * + E t E i * + E t 2 ) + 4 η 0 μ 0 E t = 4 η 0 μ 0 E i χ ee ( 2 ) t E i 2 2 χ ee ( 1 ) t E i ,
χ ee ( 2 ) t E t 2 + ( 2 χ ee ( 1 ) + 2 E 0 cos ( ω 0 t ) χ ee ( 2 ) ) t E t + ( 4 η 0 μ 0 2 ω 0 χ ee ( 2 ) E 0 sin ( ω 0 t ) ) E t = 4 η 0 μ 0 E 0 cos ( ω 0 t ) + ω 0 χ ee ( 2 ) E 0 sin ( 2 ω 0 t ) + 2 ω 0 χ ee ( 1 ) E 0 sin ( ω 0 t ) .
E t E t , 0 + E t , 1 + 2 E t , 2 +
χ ee ( 1 ) χ ee ( 2 ) ,
2 χ ee ( 1 ) t E t , 0 + 4 η 0 μ 0 E t , 0 = 4 η 0 μ 0 E 0 cos ( ω 0 t ) + 2 ω 0 χ ee ( 1 ) E 0 sin ( ω 0 t ) ,
E t , 0 = E 0 2 j k 0 χ ee ( 1 ) 2 + j k 0 χ ee ( 1 ) e j ω 0 t ,
χ ee ( 2 ) t E t , 0 2 ( 2 χ ee ( 1 ) + 2 E 0 cos ( ω 0 t ) χ ee ( 2 ) ) t E t , 0 + ( 4 η 0 μ 0 2 ω 0 χ ee ( 2 ) E 0 sin ( ω 0 t ) ) E t , 0 + 2 χ ee ( 1 ) t E t , 1 + 4 η 0 μ 0 E t , 1 = 4 η 0 μ 0 E 0 cos ( ω 0 t ) + ω 0 χ ee ( 2 ) E 0 sin ( 2 ω 0 t ) + 2 ω 0 χ ee ( 1 ) E 0 sin ( ω 0 t ) ,
E t , 1 = E 0 k 0 χ ee ( 2 ) e j 2 ω 0 t [ 4 + 12 E 0 + χ ee ( 1 ) k 0 ( χ ee ( 1 ) k 0 4 j ) ( E 0 1 ) 4 ( χ ee ( 1 ) k 0 j ) ( χ ee ( 1 ) k 0 2 j ) 2 ] ,
E t , 2 = C t , 2 [ ( 2 j 3 χ ee ( 1 ) k 0 ) e j ω 0 t + 3 ( 2 j + χ ee ( 1 ) k 0 ) e j 3 ω 0 t ] ,
C t , 2 = j ( χ ee ( 2 ) k 0 E 0 ) 2 2 ( χ ee ( 1 ) k 0 2 j ) 3 [ 4 + 12 E 0 + χ ee ( 1 ) k 0 ( χ ee ( 1 ) k 0 4 j ) ( E 0 1 ) ] ( χ ee ( 1 ) k 0 j ) ( χ ee ( 1 ) k 0 + 2 j ) ( 3 χ ee ( 1 ) k 0 2 j ) .
E s = n = 1 E s , n e j n ω 0 t ,
H y n + 1 ( i ) = H y n ( i ) Δ t μ 0 Δ z ( E x n + 1 2 ( i + 1 ) E x n + 1 2 ( i ) ) ,
E x n + 1 / 2 ( i ) = E x n 1 2 ( i ) Δ t 0 Δ z ( H y n ( i ) H y n ( i 1 ) ) ,
H y n + 1 ( n d ) = H y n ( n d ) + Δ t μ 0 Δ z ( E x n + 1 2 ( 0 ) E x n + 1 2 ( n d ) ) ,
E x n + 1 2 ( n d + 1 ) = E x n 1 2 ( n d + 1 ) + Δ t 0 Δ z ( H y n ( n d + 1 ) H y n ( 0 + ) ) ,
Δ H y = 0 χ ee ( 1 ) t E x , av + 0 χ ee ( 2 ) t E x , av 2 ,
Δ E x = μ 0 χ mm ( 1 ) t H y , av + μ 0 χ mm ( 2 ) t H y , av 2 .
H y n ( 0 + ) = H y n ( n d ) 0 χ ee ( 1 ) Δ t ( E x , av n + 1 2 E x , av n 1 2 ) 0 χ ee ( 2 ) Δ t ( ( E x , av n + 1 2 ) 2 ( E x , av n 1 2 ) 2 ) ,
E x n + 1 2 ( 0 ) = E x n + 1 2 ( n d + 1 ) + μ 0 χ mm ( 1 ) Δ t ( H y , av n + 1 H y , av n ) + μ 0 χ mm ( 2 ) Δ t ( ( H y , av n + 1 ) 2 ( H y , av n ) 2 ) ,
E x , av n + 1 2 = E x n + 1 2 ( n d ) + E x n + 1 2 ( n d + 1 ) 2 ,
H y n + 1 ( n d ) = ( 2 Δ z χ ee ( 1 ) H y n + 1 ( n d + 1 ) χ ee ( 2 ) ) μ 0 Δ h μ 0 χ ee ( 1 ) ,
E x n + 1 2 ( n d + 1 ) = ( 2 Δ z χ mm ( 1 ) E x n + 1 2 ( n d ) χ mm ( 2 ) ) 0 Δ e 0 χ mm ( 1 )
Δ h = μ 0 { 4 Δ t ( E x n + 1 2 ( n d ) E x n + 1 2 ( n d + 1 ) ) χ ee ( 2 ) + μ 0 [ 4 Δ z 2 + ( χ ee ( 1 ) + ( H y n ( n d ) + H y n + 1 ( n d + 1 ) ) χ ee ( 2 ) ) 2 4 Δ z ( χ ee ( 1 ) + ( H y n ( n d ) + H y n + 1 ( n d + 1 ) ) χ ee ( 2 ) ) ] } ,
Δ e = 0 { 4 Δ t ( H y n ( n d ) H y n ( n d + 1 ) ) χ mm ( 2 ) + 0 [ 4 Δ z 2 + ( χ mm ( 1 ) + ( E x n 1 2 ( n d ) + E x n 1 2 ( n d + 1 ) ) χ mm ( 2 ) ) 2 4 Δ z ( χ mm ( 1 ) + ( E x n + 1 2 ( n d ) + E x n 1 2 ( n d + 1 ) ) χ mm ( 2 ) ) ] } .
Select as filters


Select Topics Cancel
© Copyright 2024 | Optica Publishing Group. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.