## Abstract

Spectral singularities appearing in parity-time (*PT*)-symmetric non-Hermitian optical systems have aroused a growing interest due to their new exhilarating applications, such as bifurcation effects at exceptional points and the coexistence of coherent perfect absorber and laser (so-called CPAL point). We introduce here how the concept of CPAL action provoked in *PT*-symmetric metasurfaces can be translated into practical implementation of a low-loss zero/low-index open channel supporting a nearly undamped fast-wave propagation. Such a *PT*-synthetic metachannel shows the capability to produce a high-directivity leaky radiation, with a tunable beam angle that depends on the gain-loss parameter. The proposed structure may enable new kinds of super-directivity antennas, as well as many applications that demand extreme dielectric properties, such as epsilon-near-zero (ENZ).

© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement

## 1. Introduction

Unusual points in the spectra of non-Hermitian physical systems, such as the exceptional point [1–8] and the merging point of coherent perfect absorber (CPA) and laser (i.e., CPAL point) [9–12], have attracted substantial attention in the past few years. At these points, Taylor series expansion become singular and fail to converge. These isolated singular points are generally missed in a generic situation and may require special engineering of a quantum [13], optical [1–12], acoustic [14] or electronic systems [15,16]. The CPAL systems with parity-time (*PT*) symmetry is of particular interest because they allow realization of laser and CPA within a single component [9–12]. Traditionally, a laser oscillator emits coherent outgoing radiations, whereas a CPA is its time-reversed counterpart that acts as a dark medium absorbing all incoming radiation. At a given CPAL wavelength, the two eigenvalues of the scattering matrix become infinity and zero, which corresponds to laser and CPA modes, respectively. These two modes with dramatically different scattering properties can be switched by altering the initial amplitude/phase offset of two counter-propagating incoming beams. The CPAL action has been theoretically proposed [9,10] and later experimentally demonstrated [11,12] using waveguides and coupled resonators with *PT* symmetry. In the context, *PT*-symmetric metasurfaces have been recently proposed to make a simplified, low-profile alternative to CPAL devices [17], negative-index media [18], optical sensing and imaging devices [19,20], as well as the unidirectional reflectionless channel biased at the exceptional point [21].

In this paper, we will propose new types of electromagnetic medium formed by *PT*-symmetric metasurfaces operating at the CPAL point. This system comprises a pair of active and passive metasurfaces with the spatial dependency of surface impedance given by ${{Z}_{s}}({y}) ={-} {R}\,\delta ({y} - {d}/2) + {R}\,\delta ({y} + {d}/2)$ (see Fig. 1(a)), where ± *R* are the surface resistances, *d* is the spacing between two metasurfaces, and $\delta$ is the Kronecker delta function. The spatially-distributed balanced gain ($- {R}$) and loss ($+ {R}$) form the basis of a *PT*-symmetric optical system [1]. While scattering from *PT*-symmetric metasurfaces has been studied for the above-mentioned applications [17–21], the guided-wave and leakage characteristics of a parallel-plate waveguide formed by PT-symmetric metasurfaces (Fig. 1(a)) is yet to be explored. Understanding basic characteristics and effective medium properties of such a *PT*-synthetic metachannel may lead to new physical phenomena and applications underlying them. In the following, we will show that this low-profile and unsophisticated metachannel can exhibit an extreme (effective) dielectric properties, such as epsilon-near-zero (ENZ) phenomena found in a dispersive lossy medium [6,7] or in a waveguide operating at its cutoff frequency [22,23]. Interestingly, the propagation constant of the guided transverse electric (TE) mode can be continuously varied from nearly zero to that of the background wavenumber (${k} = \omega \sqrt {\mu \varepsilon }$) by changing the dimensionless gain-loss parameter $\gamma = {R}/\eta$ ($\eta = \sqrt {\mu /\varepsilon }$ is the impedance of background medium). From the effective-medium perspective [22,23], the effective permittivity of the *PT*-synthetic metachannel can vary from ENZ to that of the background medium, i.e., $0 \le {\textrm{Re}} [{\varepsilon _{{eff}}}] \le \varepsilon$. However, unlike other ENZ and low-index media [6,7,22,23], the calculated ${\mathop{\rm Im}\nolimits} [{\varepsilon _{{eff}}}]$ related to the power attenuation rate or propagation loss can be vanishingly small. Additionally, the almost “undamped” fast wave propagating in the *PT*-synthetic metachannel can produce the coherent radiation leakage and form a highly directional beam, thanks to the large radiating aperture size of the metachannel. More interestingly, the beam angle can be reconfigured to any direction between broadside and end-fire by altering the gain-loss parameter (namely, the surface impedance profile of metasurfaces).

## 2. Results and discussion

In order to understand singularities in *PT*-symmetric metasurfaces, we first consider scattering of the TE-polarized plane wave from this composite structure [Fig. 1(b)], which can be described by the two-port transmission-line network (TLN) shown in the inset of Fig. 1(b) [17–20]. In the TLN model, the background medium has a tangential wavenumber and a characteristic impedance given by${{k}_{y}} = {k}\cos \alpha$ and ${Z} = \eta {k}/{{k}_{y}},$ where $\alpha$ is angle of incidence. The outgoing scattered waves and the incoming waves can be related by the scattering matrix (see Appendix A). Figure 1(c) presents the evolution of the two eigenvalues of scattering matrix as a function of $\gamma$ and $\alpha ,$ with the electrical length between the two metasurfaces $\Phi { = }{{k}_{y}}{d} = \pi /2.$ From Fig. 1(c), we find that CPAL points can exist when$\gamma = 1/(\sqrt 2 \cos \alpha )$, which makes eigenvalues become zero and infinity. Exceptional points are also observed in Fig. 1(c). Such branch point singularities divide the system into the exact symmetry phase with unimodular eigenvalues and the broken symmetry phases with non-unimodular ones. Moreover, eigenvalues coalesce at exceptional points [10].

Next, we will discuss the use of *PT*-symmetric metasurfaces as a waveguiding channel and will show that the CPAL point found in scattering events can shed light on tailoring effective medium properties of a *PT*-synthetic metachannel. According to the Lorentz reciprocity theorem [24–26], if a current density **J**_{1} placed at point **r**_{1} produces an electric field **E**_{1} at point **r**_{2}, then by switching the position of source and observation, their product remains constant. In our case, we assume that the tangential electric field **E*** _{s}* on

*PT*-symmetric metasurfaces is induced by an incident plane wave sustained by the current density

**J**

*placed in the volume*

_{FF}*V*. Under the lasing condition, a source (

_{FF}**J**

*) of arbitrary input amplitude can produce a huge*

_{s}**E**

*. We can now apply reciprocity considerations to evaluate the radiated field*

_{s}**E**

*(*

_{FF}*x*,

*y*) produced by the equivalent surface current densities ${{\mathbf J}_{s}}({x})\delta ({y} \pm {d}/2)$ on metasurfaces (which is defined the ratio of the tangential electric field to the surface impedance) enclosed by the volume

*V*

_{MTS}. The reciprocity formula yields

*PT*-synthetic metachannel locked to the CPAL point is excited by a source placed at its terminal surface or within it, strong radiated fields can be produced toward specific direction in the far zone, resulting in an angle-selective, directive radiation pattern.

The eigenmodal solutions of this *PT* metachannel can be derived using the transverse-resonance relation that considers the TLN model of the transverse cross section of the waveguide, analogous to the one used for studying scattering of plane waves by *PT*-symmetric metasurfaces [the inset of Fig. 1(b)] [27]. In this case, the line has a transverse propagation constant ${{k}_{y}} = \sqrt {{{k}^2} - {\beta ^2}}$ and a characteristic impedance for the TE mode given by ${Z} = \eta {k}/{{k}_{y}},$ where $\beta$ is the longitudinal propagation constant. If the transverse resonance condition is satisfied, at any point along the *y*-axis, the sum of the input impedances seen looking to either side must be zero, i.e., ${Z}_{{in}}^{( + )} + {Z}_{{in}}^{( - )} = 0,$ where ${Z}_{{in}}^{( + )}$ and ${Z}_{{in}}^{( - )}$ are respectively input impedances seen looking to $+ \hat{{\boldsymbol {y}}}$ and $- \hat{{\boldsymbol {y}}}$ at any point on the resonant line, $- {d}/2 \le {y} \le {d}/2$ (see Appendix B). This yields the following dispersion equation:

*Z*and

*k*are defined differently), if

_{y}*PT*-symmetric metasurfaces can achieve the lasing effect in the scattering event, the same structure also guarantees the transverse resonance relation for waveguiding applications. As a result, if the CPAL condition is satisfied, the tangential wavenumber of incident wave, ${{k}_{x}} = {k}\sin \alpha ,$ in the scattering problem is also the longitudinal propagation constant of the TE guided mode, $\beta ,$ in the

*PT*metachannel. Figure 2 presents the dispersion diagram for

*PT*metachannels with (a) $(\gamma ,{d}) = (1/\sqrt 2 \,,\,\,{\lambda _0}/4)$ and (b) $(\gamma ,{d}) = (1/2\,,\,\,{\lambda _0}/2\sqrt 2 )$, which lead to $\beta = 0$ and $\beta = {k}/\sqrt 2$ at the wavelength (frequency) of operation ${\lambda _0}$ (${{f}_0}$). Such outcomes further confirm the above physical interpretations.

We note further that a *PT*-synthetic metachannel locked at the CPAL point exhibits a fast-wave propagation behavior, i.e., $\beta < {k}$, and, thus, has a low effective permittivity given by ${\varepsilon _{{eff}}}/\varepsilon { } = { }si{n^2}\alpha .$ Fast waves propagating in the unbounded channel formed by metasurfaces will result in the radiation leakage and the beam angle measured from broadside $\alpha = {\sin ^{ - 1}}(\beta /{k}).$ The gain-loss parameter that controls the CPAL at a certain angle of incidence (scattering events) also governs the beam angle in the leaky-wave mode. We first consider a metachannel composed of *PT*-symmetric metasurfaces with $\gamma = 1/\sqrt 2$ and a height of one-quarter wavelength, which forms a CPAL for normally-incident waves at frequency *f*_{0}. In accordance with the discussion above, when a *PT*-synthetic metachannel is excited by a waveguide port at *f*_{0}, one can expect that $\beta \approx 0,$ and, thus, an ENZ medium with infinite phase velocity is achieved. Figures 3(a) and 3(b) show the calculated radiation pattern (see Appendix C) and electrical field distributions [28] for this unbounded metachannel at ${f} = {{f}_0} - \delta {f}$ (here, $\delta {f} = { }{10^{ - 4}}{{f}_0}\;$ which leads to $\beta /{k} = 0.005 - {j}0.010$). It is seen from Fig. 3(b) that inside the channel, a nearly constant phase distribution can be obtained due to the ENZ-like characteristics. Moreover, the nearly undamped fast-wave property with $\beta \sim 0$ results in a highly directive broadside radiation, as can be seen in Fig. 3(a). In the far (Fraunhofer) zone, the directivity of 2-D radiative apertures can be defined quantitively as the ratio of the maximum radiation intensity of the main lobe (${{U}_{\max }}$) to the average radiation intensity over all space [29]:

*L*. For example,

*D*is 10.89 (10.37 dB) for ${L} = 2\;\mathrm{\lambda }$, and is increased to 51.86 (17.15 dB) for ${L} = 10\;\mathrm{\lambda }{,}$ and 138.84 (21.43 dB) for an infinitely long structure. Given that ${\mathop{\rm Im}\nolimits} [\beta ]\sim 0$, in light of the contactless gain-loss interaction, the

_{max}*PT*leaky-wave structure can have a very large effective aperture (see Appendix C) and thus an ultrahigh directivity.

Moreover, changing the gain-loss parameter will alter the beam angle, as can be seen in the far-field radiation patterns in Fig. 3(c) and contour plots of electric field distributions in Fig. 3(b). For different targeted beam angles $\alpha =$ 0°, 30°, 45°, and 60°, surface resistances of two metasurfaces ($\gamma = 1/\sqrt 2 \cos \alpha$) and the spacing between them ($\Phi { = kd}\cos \alpha = \pi /2$) must be changed accordingly, in order to lock the system at the CPAL point. The radiation pattern is somehow bidirectional, due to the nature of unidirectional scattering in *PT* systems [21]. Compared with other ENZ medium made of metamaterials or Drude-dispersion materials, the proposed low-index metachannel may not only ease manufacturing complexity, but also greatly reduce the attenuation rate. The proposed structure may therefore facilitate the practice of ENZ-allowed applications (e.g., supercoupling and superluminal effect, energy squeezing, and enhanced nonlinear wave mixing [6,7,22,23], as well as leaky-wave emitters [7,27]). Leaky-wave antennas based on guided-wave devices with periodic grids/slots have been enormously studied in different spectral ranges. However, their effective aperture size is generally limited by the non-negligible attenuation rate, especially for optical applications. Besides, the occurrence of higher-order (Floquet) spatial harmonics could produce unintended grating lobes. These long-standing challenges may be addressed by the *PT*-synthetic leaky-wave structures, with homogeneous non-graded surfaces and contactless gain-loss interactions.

We also analyze radiation from an electric line source ($\overline {\textrm J} = {\boldsymbol {\hat{z}} }{{I}_0}\delta ({x})\delta ({y})\;\;[A/{m^2}]$) placed at the center of a *PT*-synthetic metachannel, as schematically shown in Fig. 4**(**a). Figures 4(b) and 4(c) show the far-field radiation pattern and contours of electric field distributions for the metachannel in Fig. 4(c), under excitation of a line source; here $\delta {f} = { }{10^{ - 1}}{{f}_0}.$ The electric field in the far zone can be obtained as an inverse Fourier transform [29]:

*PT*-symmetric metasurfaces. The positive surface resistance

*R*can be readily achieved by a resistive sheet or passive metasurface made of lossy materials. In the optical region, an active metasurface could be a (patterned) thin layer of material with negative conductivity (e.g., optically-pumped 2D materials [30,31], organic dyes, or some semiconductors), which may be nanopatterned to achieve the effective negative surface resistance and zero net surface reactance. The active metasurface working at microwave frequencies could be a metallic metasurface or frequency-selective surface loaded with negative-resistance elements [16,32].

## 3. Conclusion

In conclusion, we have proposed the concept of a *PT*-synthetic metachannel exhibiting zero or low effective permittivity, for which the CPAL point offers a comprehensive guidance on tailoring the extreme effective permittivity. When this metachannel working at the CPAL point is fed by a waveguide port or line source, the leaky-wave mode can couple the (nearly undamped) guided fast wave into the background medium, resulting in an ultrahigh-directivity radiative leakage. In addition, the beam can be steered from broadside towards end-fire direction by controlling the gain-loss parameter. We envision that the proposed active component may be applied to many applications of interest in different spectra, including ultrahigh-directivity antennas or emitters with tunable radiating angles, as well as low-attenuation ENZ/low-index media with leaky-wave properties.

## Appendix A. Coherent perfect absorber-laser using *PT*-symmetric metasurface

Consider scattering of a plane wave (wave vector ${\mathbf k}{ = }\hat{{\boldsymbol {x}}}{{k}_{x}} + \hat{{\boldsymbol {y}}}{{k}_{y}}$) from *PT*-symmetric metasurfaces, the elements of the scattering matrix can be calculated using the two-port transmission-line network (TLN) model in Fig. 1(b). The background medium has a characteristic impedance *Z*, and the two shunt surface resistances are separated by a portion of transmission line with a characteristic impedance *Z* and an electrical length $\Phi { = }{{k}_{y}}{d}{.}$ The surface resistances have opposite values, ±*R*. In this system, the dimensionless gain-loss parameter (or non-Hermiticity parameter) can be defined as $\gamma = {R}/{Z}{.}$ For the transverse electric (TE) plane wave incident at an arbitrary angle $\alpha$, the wave impedance is given by ${Z = }\eta /\cos \alpha$ and the propagation constant is ${{k}_{y}}{ = k}\cos \alpha ,$ where *η* is the characteristic impedance of background medium. Similar considerations apply to the transverse magnetic (TM) incidence, but with ${Z = }\eta \cos \alpha .$ Using the transfer matrix formalism, and assuming time-harmonic fields *e ^{jωt}*, the scattering parameters, involving transmission (

*t*) and reﬂection (

*r*) coeﬃcients for bottom ($-$) and top (

*+*) incidences are obtained as:

*PT-*symmetry imposes a generalized conservation relation on the scattering matrix: ${{\mathbf S}^\ast }(\omega ) = \mathcal{PT}{\mathbf S}(\omega )\mathcal{PT} = {{\mathbf S}^{ - 1}}(\omega )$ [10,33], where the parity operator $\mathcal{P} = \left( {\begin{array}{cc} 0&1\\ 1&0 \end{array}} \right),$ the time-reversal operator $\mathcal{T} = \left( {\begin{array}{cc} 0&1\\ 1&0 \end{array}} \right)\mathcal{K},$ and $\mathcal{K}$ is the complex conjugation operator. The transition between the exact and broken symmetry can be known from tracing the evolution of eigenvalues of ${\mathbf S}$, given by:

*PT*-symmetry phase, the eigenvalues are nondegenerate and unimodular ($|{\lambda _{1,2}}|= 1$), whereas in the broken symmetry phase, the eigenvalues are non-unimodular (${\lambda _1} = {(\lambda _2^\ast )^{ - 1}}$). In addition to this branch point singularity, the CPAL action is achieved when $\gamma = 1/(\sqrt 2 \cos \alpha )$ and $\Phi = \mathrm{\pi }/2.$ From Fig. 3(c), we find that if $\gamma { = }\sqrt 2 ,$ a self-dual spectral singularity is obtained, with eigenvalue become $0$ and $\infty$; such values stand for the CPA and lasing states, respectively. Electric fields on bottom ($-$) and top (+) sides can be decomposed into forward (

*f*)- and backward (

*b*)-propagating waves, whose relations can be described by the transfer matrix

**M**as: $\left( {\begin{array}{c} {{E}_{f}^ + }\\ {{E}_{b}^ + } \end{array}} \right) = {\mathbf M}\left( {\begin{array}{c} {{E}_{f}^ - }\\ {{E}_{b}^ - } \end{array}} \right).$ The CPAL system based on

*PT*-symmetric metasurfaces can operate in the laser mode when ${E}_{b}^ +{/}{E}_{f}^ - \ne {{M}_{21}},$ or in the CPA mode when ${E}_{b}^ +{/}{E}_{f}^ -{=} {{M}_{21}}.$The lasing oscillator mode provides output fields ${E}_{b}^ - {,E}_{f}^ + \ne 0$ even for zero input fields (${E}_{f}^ - ,{E}_{b}^ +{\approx} 0$), while the CPA mode makes ${E}_{b}^ - { = E}_{f}^ +{=} 0$ even for non-zero input fields (${E}_{f}^ - ,{E}_{b}^ + \ne 0$) [9,17]. In physical systems, the conditions for a laser/oscillator and a CPA are ${{M}_{22}} = 0$ and ${{M}_{11}} = 0$, respectively. Nevertheless, an

*PT*-symmetric system allows ${{M}_{22}}$ and ${M}_{11}^{}$ to be simultaneously zero at the CPAL point. By varying the initial condition, such as the complex amplitude ratio between the two incoming waves, one can switch the operation mode from lasing to CPA, and vice versa.

## Appendix B. Eigenmodes in a *PT*-synthetic channel

Consider first the eigenmodes of the *PT*-synthetic channel in Fig. 1(a), a guided wave propagates along the *x*-axis with a factor${{e}^{ - {j}\mathrm{\beta }{x}}}.$ Electromagnetic ﬁelds can be separated into transverse electric (TE) and transverse magnetic (TM) ﬁelds with respect to a lateral coordinate. The TE mode has the following electric field distributions:

**E**,

**H**) for the TE mode in each region can be obtained from source-free Maxwell’s equations. The complex coefﬁcients ${c}_{i}^{TE}$ may be determined by matching the boundary conditions enforced on the metasurface: ${{\mathbf J}_{\mathbf s}} = \hat{{\mathbf n}} \times ({{{\mathbf H}^ + } - {{\mathbf H}^ - }} )= {{\mathbf E}_{\tan }}{\mathbf /}{{Z}_{s}}$ and $\hat{{\mathbf n}} \times ({{{\mathbf E}^ + } - {{\mathbf E}^ - }} )= 0,$ where $\hat{{\mathbf n}}$ is the surface normal vector, and the surface impedance has a

*PT*-symmetric profile: ${{Z}_{s}}({y}) ={-} {R}\delta ({y} - {d}/2) + {R}\,\delta ({y} + {d}/2),$ ${R}$ and $- {R}$ are the surface resistances for the passive and active metasurfaces, respectively. The resulting dispersion equation for the complex eigenmodal solution$\beta$ is given by:

*PT*-synthetic channel and has a transverse propagation constant ${k}_{t}^{}$ and a characteristic impedance for TE modes given by ${Z} = {k}\eta /{k}_{y}^{}$ and $\eta = \sqrt {\mu /\varepsilon } .$ Due to the fact that the longitudinal propagation constant,

*β*, must be the same in both regions for phase matching of the tangential fields at the interface. For ${y < } - {d}/2$ and ${y > d}/2,$ the transverse line is terminated with an impedance given by ${Z} = {k}\eta /{k}_{y}^{}.$ Applying the transverse resonance condition [Eq. (9)] will lead to the dispersion equation in Eq. (8).

## Appendix C. Radiation from a *PT*-synthetic channel under excitation of a waveguide port

The electric surface current density on a metasurface is induced by discontinuity of magnetic fields. For the *PT*-symmetric metasurface channel sketched in Fig. 5, surface current densities are given by:

**A**due to ${{\mathbf J}_s}$ is given in terms of the Green’s function:

${\mathbf A}(\bar{\rho }) = \int\!\!\!\int\limits_{s^{\prime}} {{{\mathbf J}_s}\frac{\mu }{{4{j}}}} {H}_0^{(2)}({{k|}\bar{\rho } - \bar{\rho }^{\prime}{|}} ){ds}^{\prime},$ where ${g}({\bar{\rho },\bar{\rho }^{\prime}} )= \frac{1}{{4{j}}}{H}_0^{(2)}({{k|}\bar{\rho } - \bar{\rho }^{\prime}{|}} )$ is the two-dimensional Green’s function, $\bar{\rho }^{\prime} = {x}^{\prime}\hat{\textrm x} + {y}^{\prime}\hat{\textrm y}$ and $\bar{\rho }{ = }\rho \hat{\rho } = {x}\hat{\textrm x} + {y}\hat{\textrm y}$ (*ρ* is the radial distance and $\hat{\rho } = \cos \phi {\hat{x}} + \sin \phi {\hat{y}}$) are the position vectors of the source and the observer, respectively, and${H}_0^{(2)}({\cdot} )$ is the Hankel function of the second kind. In the far zone, the electric and magnetic fields produced by sheet currents induced on the metasurfaces only have${\hat{z}}$ and $\hat{\phi }$ components in the cylindrical coordinates. Those constitute a transverse electromagnetic (TEM) wave propagating in the $\hat{\rho } -$ direction, given by:

*PT*metachannel with length

*L*and an infinite width, as sketched in Fig. 5. In the far zone, the distance from any point on the active metasurface ($({x}^{\prime},{d}/2)$ for $- {L}/2 \le {x}^{\prime} \le {L}/2$) to the observation point can be approximately expressed as:

*x*-axis. Similarly, the distance from any point on the passive metasurface ($({x}^{\prime}, - {d}/2)$ for $- {L}/2 \le {x}^{\prime} \le {L}/2$) to the observation point can be written as:

## Appendix D. Radiation from a *PT*-synthetic channels under excitation of an electric line source

The structure considered here is *PT*-symmetric metasurfaces excited by an electric line source along *z* with a time-harmonic dependence, embedded in the middle of the *PT*-symmetric metasurfaces, as sketched in Fig. 5. The transverse-equivalent network as in Fig. 6 can be used to model such an antenna [27,34–36]. The electric field in the background produced by a unit amplitude electric line source can be represented as an inverse Fourier transform [27,36]:

The characteristic impedance, *Z*, for the TE and TM polarizations have the following expressions:

*k*depends on the spherical angle $\theta$ as ${{k}_{y}} = \sqrt {{k}_{}^2 - {k}_{x}^2} = {k}\cos \theta .$ The TLN model can be used for determination of fields radiated by a source through an application of the reciprocity theorem. In this case, The far-zone electric field can readily be obtained through an asymptotic evaluation of Eq. (17) for large distances from the origin (i.e., $\rho \gg {d}$) [27,36]. The result for the upper half-plane is given by: where the normalized far-field pattern is

_{y}Similarly, the far-zone electric field in the lower half-plane is given by:

where*PT*-symmetric metasurfaces that have isotropic and homogeneous surface impedances. The electric field intensity and power density of the radiated wave were monitored in cylindrical coordinates, in order to obtain the radiation patterns in Figs. 3 and 4.

## Appendix E. Practical realization of *PT*-symmetric metasurfaces in the optical regime

According to impedance boundary conditions, a discontinuity on the tangential magnetic field on the metasurface is related to the induced averaged surface current by the surface impedance. For the normally-incident wave on a 2-D array of perfectly-conducting patches, the equivalent surface impedance can be approximately expressed as [37–39]:

*p*and

*g*are the period and gap of patches of ignorable thickness, respectively. When a dispersive medium is used to constitute the metasurface, the surface impedance becomes complex-valued, ${{Z}_{s}} = {{R}_{s}} + {j}{{X}_{s}}\;[\Omega ],$ where the surface resistance ${{R}_{s}}$ and the surface reactance ${{X}_{s}}$ account for the gain/loss magnitude and the net stored energy in the near field, respectively. When considering a complex-valued sheet conductivity, the surface impedance of metasurface should be modified as [20,40,41] As an example, the photopumped graphene monolayer and bilayer could exhibit optical gain [20,42,43]. The non-equilibrium conductivity of a graphene monolayer can be modeled using Green’s functions, taking into account both interband and intraband transitions $\sigma = {\sigma _{{intra}}} + {\sigma _{{inter}}}$, given by:

*q*is the electric charge, $\varepsilon$ is the energy, $\hbar$ is the reduce Planck’s constant, ${{K}_B}$ is the Boltzmann’s constant,

*T*is the temperature,

*ω*is the angular frequency, and

*τ*is the momentum relaxation time of charge carriers. The interband transitions and the cascaded optical-phonon emission could lead to photoexcited electron-hole pairs near the Dirac point, splitting the Fermi level of graphene monolayer into two quasi-Fermi levels${{E}_{{Fn}}},{{E}_{{Fp}}} ={\pm} {\varepsilon _F}.$ Therefore, under optical pumping, ${\textrm{Re}} [{\sigma _{{inter}}}]$ in Eq. (30), could be negative. At sufficiently strong optical pumping, the $\sigma = {\sigma _{{intra}}} + {\sigma _{{inter}}} < 0$ could be achieved in the terahertz and far-infrared regions. Negative conductivity is also observed in other gapped 2D materials and organic dyes [30,44], which pave the way for building

*PT*-symmetric metasurfaces.

## Funding

National Science Foundation (1917678).

## Disclosures

The authors declare no conflicts of interest.

## References

**1. **R. El-Ganainy, K. G. Makris, D. N. Christodoulides, and Z. H. Musslimani, “Theory of coupled optical PT-symmetric structures,” Opt. Lett. **32**(17), 2632–2634 (2007). [CrossRef]

**2. **C. E. Ruter, K. G. Makris, R. El-Ganainy, D. N. Christodoulides, M. Segev, and D. Kip, “Observation of parity-time symmetry in optics,” Nat. Phys. **6**(3), 192–195 (2010). [CrossRef]

**3. **A. Regensburger, C. Bersch, M. A. Miri, G. Onishchukov, D. N. Christodoulides, and U. Peschel, “Parity-time synthetic photonic lattices,” Nature **488**(7410), 167–171 (2012). [CrossRef]

**4. **L. Feng, Y. L. Xu, W. S. Fegadolli, M. H. Lu, J. E. B. Oliveira, V. R. Almeida, Y. F. Chen, and A. Scherer, “Experimental demonstration of a unidirectional reflectionless parity-time metamaterial at optical frequencies,” Nat. Mater. **12**(2), 108–113 (2013). [CrossRef]

**5. **B. Peng, S. K. Ozdemir, F. C. Lei, F. Monifi, M. Gianfreda, G. L. Long, S. H. Fan, F. Nori, C. M. Bender, and L. Yang, “Parity-time-symmetric whispering-gallery microcavities,” Nat. Phys. **10**(5), 394–398 (2014). [CrossRef]

**6. **S. Savoia, G. Castaldi, V. Galdi, A. Alu, and N. Engheta, “Tunneling of obliquely incident waves through PT-symmetric epsilon-near-zero bilayers,” Phys. Rev. B **89**(8), 085105 (2014). [CrossRef]

**7. **S. Savoia, G. Castaldi, V. Galdi, A. Alu, and N. Engheta, “PT-symmetry-induced wave confinement and guiding in epsilon-near-zero metamaterials,” Phys. Rev. B **91**(11), 115114 (2015). [CrossRef]

**8. **S. K. Ozdemir, S. Rotter, F. Nori, and L. Yang, “Parity-time symmetry and exceptional points in photonics,” Nat. Mater. **18**(8), 783–798 (2019). [CrossRef]

**9. **S. Longhi, “PT-symmetric laser absorber,” Phys. Rev. A **82**(3), 031801 (2010). [CrossRef]

**10. **Y. D. Chong, L. Ge, and A. D. Stone, “PT-symmetry breaking and laser-absorber modes in optical scattering systems,” Phys. Rev. Lett. **106**(9), 093902 (2011). [CrossRef]

**11. **Y. Sun, W. Tan, H. Q. Li, J. S. Li, and H. Chen, “Experimental demonstration of a coherent perfect absorber with PT phase transition,” Phys. Rev. Lett. **112**(14), 143903 (2014). [CrossRef]

**12. **J. Schindler, Z. Lin, J. M. Lee, H. Ramezani, F. M. Ellis, and T. Kottos, “PT-symmetric electronics,” J. Phys. A: Math. Theor. **45**(44), 444029 (2012). [CrossRef]

**13. **C. M. Bender and S. Boettcher, “Real spectra in non-hermitian hamiltonians having PT symmetry,” Phys. Rev. Lett. **80**(24), 5243–5246 (1998). [CrossRef]

**14. **X. F. Zhu, H. Ramezani, C. Z. Shi, J. Zhu, and X. Zhang, “PT-symmetric acoustics,” Phys. Rev. X **4**(3), 031042 (2014). [CrossRef]

**15. **J. Schindler, A. Li, M. C. Zheng, F. M. Ellis, and T. Kottos, “Experimental study of active LRC circuits with PT symmetries,” Phys. Rev. A **84**(4), 040101 (2011). [CrossRef]

**16. **P. Y. Chen, M. Sakhdari, M. Hajizadegan, Q. S. Cui, M. M. C. Cheng, R. El-Ganainy, and A. Alu, “Generalized parity-time symmetry condition for enhanced sensor telemetry,” Nat. Electron. **1**(5), 297–304 (2018). [CrossRef]

**17. **M. Sakhdari, N. M. Estakhri, H. Bagci, and P. Y. Chen, “Low-threshold lasing and coherent perfect absorption in generalized PT-symmetric optical structures,” Phys. Rev. Appl. **10**(2), 024030 (2018). [CrossRef]

**18. **R. Fleury, D. L. Sounas, and A. Alu, “Negative refraction and planar focusing based on parity-time symmetric metasurfaces,” Phys. Rev. Lett. **113**(2), 023903 (2014). [CrossRef]

**19. **F. Monticone, C. A. Valagiannopoulos, and A. Alu, “Parity-time symmetric nonlocal metasurfaces: All-angle negative refraction and volumetric imaging,” Phys. Rev. X **6**(4), 041018 (2016). [CrossRef]

**20. **P. Y. Chen and J. Jung, “PT symmetry and singularity-enhanced sensing based on photoexcited graphene metasurfaces,” Phys. Rev. Appl. **5**(6), 064018 (2016). [CrossRef]

**21. **R. Fleury, D. Sounas, and A. Alu, “An invisible acoustic sensor based on parity-time symmetry,” Nat. Commun. **6**(1), 5905 (2015). [CrossRef]

**22. **A. Alu, M. G. Silveirinha, A. Salandrino, and N. Engheta, “Epsilon-near-zero metamaterials and electromagnetic sources: Tailoring the radiation phase pattern,” Phys. Rev. B **75**(15), 155410 (2007). [CrossRef]

**23. **C. Argyropoulos, P. Y. Chen, G. D’Aguanno, N. Engheta, and A. Alu, “Boosting optical nonlinearities in epsilon-near-zero plasmonic channels,” Phys. Rev. B **85**(4), 045129 (2012). [CrossRef]

**24. **D. M. Pozar, * Microwave engineering, 4nd edition* (Wiley, 2012).

**25. **C. A. Balanis, * Advanced engineering electromagnetics, 2nd edition* (Wiley, 2012).

**26. **J. Lee, M. Tymchenko, C. Argyropoulos, P. Y. Chen, F. Lu, F. Demmerle, G. Boehm, M. C. Amann, A. Alu, and M. A. Belkin, “Giant nonlinear response from plasmonic metasurfaces coupled to intersubband transitions,” Nature **511**(7507), 65–69 (2014). [CrossRef]

**27. **G. Lovat, P. Burghignoli, F. Capolino, D. R. Jackson, and D. R. Wilton, “Analysis of directive radiation from a line source in a metamaterial slab with low permittivity,” IEEE Trans. Antennas Propag. **54**(3), 1017–1030 (2006). [CrossRef]

**28. **Comsol multiphysics, COMSOL, Inc, http://www.comsol.com.

**29. **C. A. Balanis, * Antenna theory: Analysis and design* (Wiley, 2016).

**30. **T. Low, P. Y. Chen, and D. N. Basov, “Superluminal plasmons with resonant gain in population inverted bilayer graphene,” Phys. Rev. B **98**(4), 041403 (2018). [CrossRef]

**31. **T. J. Guo, L. Zhu, P. Y. Chen, and C. Argyropoulos, “Tunable terahertz amplification based on photoexcited active graphene hyperbolic metamaterials,” Opt. Mater. Express **8**(12), 3941–3952 (2018). [CrossRef]

**32. **P. Y. Chen, C. Argyropoulos, and A. Alu, “Broadening the cloaking bandwidth with non-foster metasurfaces,” Phys. Rev. Lett. **111**(23), 233001 (2013). [CrossRef]

**33. **L. Ge, Y. D. Chong, and A. D. Stone, “Conservation relations and anisotropic transmission resonances in one-dimensional PT-symmetric photonic heterostructures,” Phys. Rev. A **85**(2), 023802 (2012). [CrossRef]

**34. **R. E. Collin, * Field theory of guided waves* (Wiley, 1990).

**35. **L. B. Felsen and N. Marcuvitz, * Radiation and scattering of waves* (Wiley, 1994).

**36. **D. R. Jackson and A. A. Oliner, “A leaky-wave analysis of the high-gain printed antenna configuration,” IEEE Trans. Antennas Propag. **36**(7), 905–910 (1988). [CrossRef]

**37. **S. Tretyakov, * Analytical modeling in applied electromagnetics* (Artech House, 2003).

**38. **O. Luukkonen, C. Simovski, G. Granet, G. Goussetis, D. Lioubtchenko, A. V. Raisanen, and S. A. Tretyakov, “Simple and accurate analytical model of planar grids and high-impedance surfaces, comprising metal strips or patches,” IEEE Trans. Antennas Propag. **56**(6), 1624–1632 (2008). [CrossRef]

**39. **F. Costa, S. Genovesi, A. Monorchio, and G. Manara, “A circuit-based model for the interpretation of perfect metamaterial absorbers,” IEEE Trans. Antennas Propag. **61**(3), 1201–1209 (2013). [CrossRef]

**40. **P. Y. Chen, M. Farhat, and H. Bagci, “Graphene metascreen for designing compact infrared absorbers with enhanced bandwidth,” Nanotechnology **26**(16), 164002 (2015). [CrossRef]

**41. **P. Y. Chen and A. Alu, “Terahertz metamaterial devices based on graphene nanostructures,” IEEE Trans. Terahertz Sci. Technol. **3**(6), 748–756 (2013). [CrossRef]

**42. **T. Watanabe, T. Fukushima, Y. Yabe, S. A. B. Tombet, A. Satou, A. A. Dubinov, V. Y. Aleshkin, V. Mitin, V. Ryzhii, and T. Otsuji, “The gain enhancement effect of surface plasmon polaritons on terahertz stimulated emission in optically pumped monolayer graphene,” New J. Phys. **15**(7), 075003 (2013). [CrossRef]

**43. **D. Svintsov, V. Ryzhii, A. Satou, T. Otsuji, and V. Vyurkov, “Carrier-carrier scattering and negative dynamic conductivity in pumped graphene,” Opt. Express **22**(17), 19873–19886 (2014). [CrossRef]

**44. **Z. Wang, H. Sun, Q. Y. Zhang, J. B. Feng, J. X. Zhang, Y. Z. Li, and C. Z. Ning, “Excitonic complexes and optical gain in two-dimensional molybdenum ditelluride well below the mott transition,” Light: Sci. Appl. **9**(1), 39 (2020). [CrossRef]