Chirality refers to the property of asymmetric objects, by which they cannot be superimposed on their mirror image with translations and rotations. This property is widespread in nature, with important consequences in very different areas, such as chemistry, biology, or photonics [1]. Well-known examples of structures that exhibit inherent chirality include proteins, sugars, or DNA sequences. A chiral molecule (enantiomer) can be a poison, while its mirror enantiomer can act as a medicine, even though both molecules have the same chemical composition. Due to several reasons, such as random orientation, interaction with the environment, and the influence of thermal motion, most natural materials exhibit weak intrinsic chiral responses, which limits their application in, e.g., displays, optical recording, or optical communication. To overcome these limitations and explore more possibilities for chiral materials, controlling the chiral response of natural materials becomes crucial. One approach to achieve this control in optics/photonics is by coupling photonic or plasmonic metasurfaces with strong chiral responses and supporting modes with high quality factors (Q-factors) and large electric field (E-field) enhancements, such as surface lattice resonances [2–13], to weakly chiral or even non-chiral emitters [14–23].

Particularly interesting are bound states in the continuum (BICs), which are optical modes with infinite lifetimes and Q-factors that originate from destructive interference of radiative channels or from the symmetry of the electromagnetic field. These modes remain localized to the structure even though they coexist with the continuum of radiative modes, since they cannot couple to this continuum [24]. BICs offer many promising possibilities to modify light–matter interactions due to their divergent Q-factor and the strong enhancement of the E-field [25–27]. In particular, symmetry-protected BICs in periodic two-dimensional metasurfaces with inversion symmetry have recently been proposed and investigated, both fundamentally and in relation to applications [26,28–30]. By slightly breaking the inversion symmetry of the metasurface, BICs evolve into quasi-BICs (Q-BICs), with finite but still very large Q-factors and strong E-field enhancements. In addition, if the structure with broken inversion symmetry is mirror asymmetric, it will also introduce optical chirality, resulting in chiral Q-BICs [31,32].

Intrinsic chiral BICs (chiral structures for light incident in the normal direction) with large optical dichroism are difficult to achieve experimentally, as they require breaking the out-of-plane symmetry [33,34]. However, it is relatively easy to obtain intrinsic and extrinsic chiral Q-BICs (chiral at an oblique angle) only by breaking the symmetry in the plane [35,36], which extends the applications of chiral Q-BICs [37–39]. Recent studies show that chiral emission can be achieved through chiral Q-BICs [40–42]. However, to the best of our knowledge, studies on directional chiral emission based on chiral Q-BICs have not yet been reported. Directional and polarized emission from metasurfaces will have an impact in different application areas, such as communication systems, imaging systems, radar/laser systems, and light-emitting diodes [43–48].

Motivated by these recent advances, we present here an investigation of the directional chiral emission from achiral molecules coupled to extrinsic chiral Q-BICs in a silicon metasurface. We choose achiral molecules to be able to attribute any change in their polarized emission to the effect of the metasurface. The metasurface is formed by dimers of Si nanodisks. Detuning of geometric parameters, including diameters and relative distances between nanodisks, is used to break the inversion symmetry of the metasurface and introduce optical chirality, resulting in four extrinsic chiral Q-BICs in the visible range. The nature of the chiral Q-BICs was confirmed and investigated experimentally and theoretically by the circular dichroism of the optical extinction. The directional emission of achiral dye molecules coupled to extrinsic chiral Q-BICs has also been explored, showing several directional chiral emitting modes emerging from the dispersion of the lattice. With our design of symmetry-broken metasurface, we demonstrate a very large degree of circular polarization (DCP as high as 0.8) in defined directions. In addition, there is an up to 13-fold enhancement of the photoluminescence due to the coupling of the emission to the array and the subsequent decoupling of polarized emission in defined directions. The directionality is controlled by the lattice constant, and the divergence angle of the far-field directional chiral emission is extremely small (∼2deg). These results demonstrate the possibility of achieving strongly chiral Q-BIC emission with good control of directionality and promising perspectives for applications in chiral optical devices.

A metasurface with a size of 2mm×2mm was patterned on a 90 nm thick polycrystalline Si on a SiO2 substrate using electron beam lithography and a lift-off process (see Appendix A) [49]. Figure 1(b) shows a scanning electron microscope (SEM) image of the metasurface. The top panel of Fig. 1(c) illustrates a unit cell of the metasurface, indicating the different geometric parameters. A polymethyl methacrylate (PMMA)/dye solution (25% (mass fraction) perylene dye [N,N’-bis(2,6-diisopropylphenyl)-1,7- and -1,6-bis(2,6-diisopropylphenoxy)-perylene-3,4:9,10-tetracarboximide] in PMMA) was spin-coated on top of the metasurface, forming a layer with a thickness of ≈200nm, as illustrated in the bottom panel of Fig. 1(c). Figure 1(d) shows the normalized photoluminescence (PL) spectrum (red curve) and the normalized extinction (1-transmission) (blue curve) of a layer of PMMA/dye molecules. Two peaks are visible in the extinction, corresponding to the electronic transition at 2.24 eV and its first vibronic replica at 2.41 eV, respectively. Similarly, there are also two excitonic peaks (at 2.00 and 2.14 eV) in the PL spectrum.

We notice that PMMA is transparent in the visible and it matches the refractive index of the substrate, yielding a structure with almost perfect xy mirror symmetry, which is the best condition to reduce unwanted radiative losses. Since the dye’s absorption affects the metasurface extinction, we have first studied the optical properties of the bare photonic modes in the metasurface with PMMA alone, before describing the combination of the metasurface with PMMA/dye. The thickness of the PMMA/dye layer is critical, as the interaction of the dye with the optical modes should be significant, while the dye absorption should not mask the contribution from the photonic modes: the chosen thickness T=200nm follows from a trade-off between these two criteria.

To investigate the chiral response of the bare metasurface, we measured and simulated the dispersion of the extinction of circularly polarized light. A layer of PMMA without dye molecules was spin-coated on top of the array to match the refractive index of the SiO2 substrate. Figures 2(a) and 2(b) show the experimental extinction spectra as a function of the incident wavevector (ky) parallel to the surface of the array with right-handed circularly polarized (RCP) and left-handed circularly polarized (LCP) white light, respectively. ky is given by ky=(2π/λ)sinθ, where λ is the wavelength and θ is the incident angle in the yz plane. The experimental setup is described in Appendix A. Four distinct modes, labeled M1 to M4, are visible in the extinction measurements. The RCP extinction decreases substantially for the M1 and M4 modes, along with the increase of the parallel wavevector from −8 to 8rad/μm, limited by the numerical aperture of the objective [Fig. 2(a)]. In contrast, the extinction of the M2 and M3 modes increases from 0 to 0.35 when the wavevector increases from −8 to 8rad/μm. For the extinction of LCP light [Fig. 2(b)], the modes share the same trend as the mirror image of Fig. 2(a). We notice that some of the modes (most notably M3) are spectrally narrow close to k=0. The reason for this narrowing is that they originate from BICs in the size-detuned structure. M4 is considerably broader because it is an electric dipole mode (py) and is radiative even for the size-detuned structure. Figure 2(c) shows the circular dichroism (CD) of extinction, given by CD=LCP−RCPLCP+RCP.(1)

At the Γ point (ky=0), the CD of all modes is zero. Upon increasing ky, the absolute value of the CD of all modes increases. The integrated CD over positive and negative wavevectors is zero, indicating that the metasurface is not intrinsically chiral as can be also deduced from the symmetry.

Simulations of the optical properties were performed using a Bloch-mode scattering matrix method with the software EMUstack [50,51], which yields a solution of Maxwell equations for field propagation in layered media. The structural parameters of the simulated metasurface are those from sample S0. The refractive indices of Si and SiO2 were taken from Ref. [52], and the refractive index of PMMA is considered constant with a value of 1.49. The results of the simulations are shown in Figs. 2(d) and 2(e) for RCP and LCP light, respectively. Figure 2(f) shows the calculated CD in extinction. The simulated results are in excellent agreement with the experiments concerning the dispersion and the intensity of the four modes for the two circular polarizations. In particular, we see that in each of the two wavevector directions, two modes have positive CD while the other two have negative CD. Such features can be understood by the analytic model based on symmetry, as discussed in the next section.

We consider the metasurface described in Fig. 1(b), with the unit cell in the upper panel of Fig. 1(c). In the approximation of a 2D structure with an effective refractive index neff, the dispersion of photonic modes can be expressed by taking into account diffraction by the lowest-order reciprocal lattice vectors as ω±(k)=cneff(kx+m2πax)2+ky2,m=±1.(2)

There are two pairs of such modes, xy-even and xy-odd, which in principle may have a different effective index neff. The modes are degenerate at k=0, but this degeneracy is removed by the periodic structure of the metasurface. To get a better description of the photonic dispersion, we show in Fig. 3 the dispersion of photonic modes in the symmetry-broken metasurface of Fig. 1(b), calculated by the guided-mode expansion method [53–55]. Notice that xy-even modes are often called TE-like in photonic crystal terminology, while xy-odd modes are often called TM-like: see Ref. [55] for a full discussion of polarization properties. At the Γ-point, there are four modes between 2 and 2.3 eV, which result from diffraction by the wavevectors G=(±G,0) along the x direction. As shown by the group-theory treatment in Appendix B, these four modes, in order of increasing energy, have symmetries of electric and magnetic dipole moments along the y and z directions, py, pz, mz, my, as indicated in the inset. The dispersion along the Γ-X direction was investigated in Ref. [49] for TE polarization, which probes the xy-even modes. The dispersion along Γ-Y can be compared with the results of Section 3, with the modes my, mz, pz, py corresponding to the modes M1, M2, M3, M4 in Fig. 2.

From this analysis, we can derive the main chiral properties of the photonic modes and compare them with those of the experiments. At any wavevector ky≠0, all mirror symmetries are broken and the properties of the metasurface become extrinsically chiral. Indeed, at any ky≠0 the py and mz modes are coupled together, as are the pz, my modes. We can relate the chiral optical response to the far-field polarization vector and the Stokes parameter S3, following the approach of Ref. [56]. A non-zero Stokes parameter S3 indicates a finite ellipticity of the eigenmodes, which results in a circular dichroism. The detailed analysis is reported in Appendix B, and only the main results are summarized here. Basically, the two coupled py and mz modes give rise to two chiral eigenmodes with opposite ellipticity, and the same applies for the two coupled pz, my modes. Thus, for any wavevector ky>0, we expect that two of the four eigenmodes display positive CD, while the other two have negative CD. In addition, the eigenmodes for ky<0 can be obtained from the corresponding eigenmodes for ky>0 by mirror reflection, which reverses the ellipticity and the resulting CD in extinction and photoluminescence, i.e., CD(−θ)=−CD(θ). Thus, we can explain the main features observed in the experiments and the numerical simulations.

To investigate the chiral emission of the achiral dye molecules coupled to the different modes, and to understand the tunability of the chirality in emission, the achiral dye mixed with PMMA (n=1.59, T≈200nm) has been spin-coated at the top of the metasurface. Figures 5(a) and 5(b) show the dispersion of the PL from the bare dye layer on the substrate (off the metasurface) for right-handed (RCPL) and left-handed (LCPL) circularly polarized emission, respectively. A description of how the measurements have been performed can be found in Appendix A. The non-dispersive emission band at around 2.16 eV corresponds to direct exciton emission. There is almost no difference in the intensity of the PL between the RCPL and the LCPL, as can be seen in the degree of circular polarization of the photoluminescence (DCP), shown in Fig. 5(c) and defined as DCP=LCPL−RCPLLCPL+RCPL.(3)

The absence of DCP for all wavevectors confirms the achiral nature of the dye. The results are quite different for the emission of the dye on top of the metasurface, as shown in Figs. 5(d) and 5(e) for RCPL and LCPL, respectively. The PL spectra exhibit a pronounced dispersion and a strong chiral response, similar to the extinction of the bare metasurface shown in Fig. 2. Therefore, the emission depends mainly on the chiral properties of the metasurface. Note that the M1 mode in Figs. 5(d) and 5(e) is missing compared to the extinction of the bare metasurface. The reason why M1 is not visible in the PL spectra is that the energy of this mode is higher than the emission range of the dye. Figure 5(f) shows the DCP of the dye on top of the metasurface. Similarly to the CD of the extinction of the bare metasurface, the DCP at the Γ point is 0. The value of the DCP increases along with the increase of the modulus of ky, reaching its maximum (0.5) for ky around ±7rad/μm. Since the nature of the M4 mode is different from that of M2 and M3, the CD of the modes shows the opposite sign when ky is negative or positive.

To tune the dispersion of the chiral emission, different metasurfaces (S1–S5) with varying geometric parameters, including diameters, periods, and asymmetry parameters, were prepared. The geometric parameters of all samples are listed in Table 1, where S0 refers to the previously described sample (Fig. 5). The DCP of the metasurfaces is presented in Fig. 6. An increase in period and diameter leads to the redshift of the Q-BIC modes [Figs. 6(a)–6(e)], in agreement with Eq. (2). Similarly to S0, S1 to S5 exhibit extrinsic chirality. When ky=0, the DCP is 0 and increases with the modulus of ky. The DCP reaches its maximum when ky is close to ±5.5rad/μm. The size asymmetry parameter, which defines the normalized difference in radius between nanodisks (αsize=Δr/r), remains nearly constant at around 0.25 from S1 to S5 (see Table 1). In contrast, the position asymmetry parameter, defined as αposition=Δx/x, where x denotes the distance from the middle of the unit cell, increases from 0.102 to 0.246. This increase enhances the DCP from S1 to S5. The maximum absolute value of DCP is approximately 0.8 for S5 when ky is around ±5.46rad/μm.

The dispersive nature of extrinsic chiral modes leads to directional chiral optical sources. To visualize this directional emission more effectively, polar plots of the emission at 620±10nm (2 eV) are presented in Fig. 7. For S1 [Fig. 7(a)], the modes are at higher energies and the PL intensity peaks at 0°, leading to similar RCPL and LCPL emissions. As the diameters of the nanodisks and periods of the metasurfaces increase (S2 to S5), the modes redshift, reaching the 620 nm wavelength and revealing the directional chiral emission [Figs. 7(b)–7(e)]. Figure 7 also illustrates that the ideal angle for chiral emission ranges from 30° to 60°, with a divergence as low as 2°. Simulations of directional emission are discussed in Appendix E. The richness of modes in Si metasurfaces makes multidirectional chiral emission feasible and motivates further exploration for potential applications.

We emphasize that the coupling mechanism leading to the directional chiral emission is the spontaneous decay of the excited dye molecules spatially located in the near field of the metasurface, as illustrated in Fig. 4. The molecules decay to the modes supported by the metasurface and these modes leak to the continuum outcoupling the chiral emission due to the broken symmetries discussed in Section 4. Due to the dispersion introduced by the metasurface, the emission outcoupling is highly directional. Although the emission from the molecules is non-chiral, the decay of the emission to a chiral optical mode and subsequent leak of this mode produce a strongly chiral signal. Thus, the symmetry-broken metasurface has a crucial role in producing a strongly directional chiral emission.

In summary, we have demonstrated the chiral directional emission of achiral molecules coupled to metasurfaces formed by asymmetric Si nanodisk dimers. Narrow and dispersive quasi-BICs are supported by the metasurface. The origin of the chiral response of the metasurface and the symmetries of the different optical modes are explained with the guided-mode expansion method and are supported by finite-element numerical simulations. These symmetries are responsible for the large modification of the spectrum, polarization, and direction of emission of the molecules. A large emission enhancement, up to 13 times (shown in Appendix D) over a wide angular range, a small divergence (≃2°), and a DCP as high as 0.8 have been reached. Our study yields a thorough understanding of extrinsic chirality in the investigated metasurfaces, and the results are promising for applications to optical devices that take advantage of strong chiral and directional emission.

Acknowledgment. Funded by the European Union (SCOLED Grant Agreement Number 101098813). Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union or the European Innovation Council and SMEs Executive Agency (EISMEA). Neither the European Union nor the granting authority can be held responsible for them. Funded by the Dutch Organization for Scientific Research (NWO) through the Gravitation Grant Research Centre for Integrated Nanophotonics. LCA acknowledges financial support from Italian MUR through PNRR project PE0000023-NQSTI. JASG and JLP acknowledge financial support from Spanish Ministerio de Ciencia, Innovación y Universidades and Agencia Estatal de Investigación (MCIN/AEI/10.13039/501100011033), ERDF A way of making Europe, and NextGenerationEU/PRTR (PID2022-137569NB-C41, TED2021-131417B-I00, TED2021-130786B-I00, PID2021-126046OB-C22, PID2020-113533RB-C33, and PRTR-C17.I1). JLP also acknowledges the financial support from NextGenerationEU (CONVREC-2021-23). SM acknowledges financial support from the Bilateral Joint Research Project (JPJSBP120239921) from Japan Society for the Promotion of Science. HD acknowledges financial support from the China Scholarship Council (CSC202206320284).