Thermal Metasurfaces: Complete Emission Control by Combining Local and Nonlocal Light-Matter Interactions

Metasurfaces have been enabling the miniaturization and integration of complex optical functionalities within an ultrathin platform by engineering the scattering features of localized modes. However, these efforts have mostly been limited to the manipulation of externally produced coherent light, e.g., from a laser. In parallel, the past two decades have seen the development of structured surfaces that emit partially coherent radiation via thermally populated, spatially extended (nonlocal) modes. However, the control over thermally emitted light is severely limited compared to optical metasurfaces, and even basic functionalities such as unidirectional emission to an arbitrary angle and polarization remain elusive. Here, we derive the necessary conditions to achieve full control over thermally emitted light, pointing to the need for simultaneously tailoring local and nonlocal scattering features across the structure. Based on these findings, we introduce a platform for thermal metasurfaces based on quasibound states in the continuum that satisfies these requirements and completes the program of compactification of optical systems by enabling a full degree of control of partially coherent light emission from structured thin films, including unidirectional emission of circularly polarized light, focusing, and control of spatial and temporal coherence, as well as wave-front control with designer spin and angular orbital momenta.

Popular Summary

Thermal radiation consists of light generated by random fluctuations of matter, and, as such, it is ubiquitous in nature and modern technology. When produced from conventional materials, thermal radiation is incoherent, broadband, and uncollimated. By filtering it, bulky optics may shape its spectral content, polarization, and spatial features but only at a significant cost in terms of efficiency and device footprint. Here, we engineer nanostructured thin films that operate as thermal metasurfaces, capable of exquisite command over their own thermal radiation. These devices integrate the effects of bulky filters into a single ultrathin structure without sacrificing efficiency. Based on these findings, we demonstrate the possibility of unidirectional thermal emission of light with chosen spin, efficient focusing, and patterning of emission with wave-front features of choice, including carrying a defined orbital angular momentum.

Our study generally unveils how and why thermal radiation is fundamentally limited in engineered nanostructures, and it introduces a route for passive structures to achieve unprecedented command over thermal radiation based on simultaneously controlling light-matter interactions at subwavelength and superwavelength scales. Correlations at superwavelength scales are necessary to increase the coherence of emitted light; control at the subwavelength length scale is needed to shape the emitted wave front and polarization state; coordination of the response at both scales is required to enable thermal metasurfaces.

Our platform holds the promise to enable custom optical light sources, tunable with intuitive design knobs based on symmetry considerations. Beyond thermal sources and photoluminescence, our design principles can also largely benefit coherent systems by extending the conventional operation of metasurfaces to locally and nonlocally tailor coherent light, which is of great interest for emerging nonlinear and quantum optics applications.

Figure 1

Comparison of conventional optics and thermal metasurfaces producing light with desired SAM and OAM. (a) In the bulk optics approach, the emission from a blackbody is spatially filtered, collimated, spectrally filtered, polarized, retarded, and shaped by separate components, leading to large inefficiencies and bulky setups. (b) With a thermal metasurface, thermal emission is shaped at the source by a subwavelength film.

Figure 2

Comparison of coherent and thermal metasurfaces. (a) A coherent metasurface transforms an incident waveform (blue) into the desired waveform (red) by spatially varying the scattering along the surface, visualized as Huygens’ wavelets (black). Each wavelet is characterized by a combination of amplitude, phase, and polarization state, which in concert produce the desired far-field profile. The coherence of the system implies that the mapping from metasurface to far field is invertible. (b) A thermal metasurface composed of a patterned structure over a ground plane supporting a thermally populated global mode (blue) whose out-of-plane scattering is locally tailored to produce a partially coherent wave form of choice.

Figure 3

Perturbative framework for thermally emissive QBICs. (a) An unperturbed high-symmetry lattice of a lossless material with index $n$ is perturbed (b) both geometrically (with asymmetry parameter $\delta$) and in index (adding a small extinction coefficient $\kappa$). (c) The radiative $Q$ factor (for fixed $k$) varies as $Q_r\propto 1/{\delta }^2$, and (d) the nonradiative $Q$ factor (for fixed $\delta$) varies as $Q_l\propto 1/\kappa$. Dots in (c,d) are calculated by full-wave simulations, while the red lines show linear fits to these data. (e) Absorption spectrum of the QBIC near the resonant frequency for normal incidence and $x$-polarized light. (f) Absorption spectra for a range of incidence angles, showing perfect absorption following a parabolic angular dispersion. The band-edge mode (dashed white line) emits to a narrow range of angles near the normal (inset). Note: The three-dimensional structure used to calculate (c)–(f) is shown in Figs. 6 with the values $\theta =\alpha =0$.

Figure 4

Temporal and spatial coherence of a QBIC thermal metasurface. (a),(b) Reflectance for $x$-polarized light near a critically coupled QBIC calculated based on FDTD [(a), reproduced from Fig. 3] and TCMT (b). (c) Emissivity at normal incidence, with the linewidth inversely related to the temporal coherence. (d) Emissivity profile at the band-edge frequency, with the angular width given by the spatial coherence. (e) Spectral degree of coherence at the band-edge frequency, with the linewidth characterizing the spatial coherence of the emission. (f) Magnitude of the spectral degree of coherence mapped in the space-frequency domain.

Figure 5

Reciprocity and spin-selective unidirectionality. (a) Definitions and conventions for the continuous scattering matrices as in (b), which depicts the behavior of a specular mirror (which conserves linear momentum and flips the spin upon reflection). (c) Summary of the behavior of the four devices studied in (d)–(k). (d) Schematic showing two blackbodies, $A$ and $B$, exchanging energy through a locally specular mirror with a nonlocal QBIC, depicted in (e) as a purple dot placeable anywhere along the specular response (solid orange line) via a nonlocal phase gradient. Reciprocity (symmetry about the black dashed line) forbids the existence of unidirectional emission. Similarly, (f) and (g) show the case of a local phase gradient but no nonlocal phase gradient, enabling unidirectional emission but not spin selectivity. To achieve spin selectivity, birefringence must be added, shown in (h) and (i) for the case of a birefringence local response with only a nonlocal phase gradient. However, reciprocity forbids unidirectionality except at normal incidence in this case. Finally, (j) and (k) show the case with a local phase gradient, nonlocal phase gradient, and birefringence, enabling spin-selective unidirectional thermal emission.

Figure 6

Metaunit library of a thermal metasurface. (a) Perspective view of a structured silicon film sitting atop a PEC with a dielectric filler in between. The top silicon elliptical pillars are monomers, and the bottom are dimers. (b) Top and side views of the unit cell in (a), defining the geometric and material parameters. (c) Cross sections of an example quasi-BIC supported by this structure and controlled by the dimer. (d),(e) The latitude and longitude on the Poincaré sphere of the quasi-BIC in (c) of the absorbed polarization state as the geometric parameters $\theta$ and $\alpha$ are varied. (f) Circularly dichroic absorption spectra for the geometry depicted in (c).

Figure 7

Selective and directional emission from a thermal metasurface. (a) The superperiod of a thermal metasurface designed to emit RCP light at non-normal incidence, with low spatial coherence. (b) A superperiod designed to emit LCP light at non-normal incidence, with high spatial coherence. (c),(d) RCP and LCP emission maps for the device in (a) as a function of the wavelength and altitude in the $x$ direction, ${\theta }_x$. (e) Emission patterns for the device in (a) at the wavelength of the quasi-BIC’s band edge [dashed contours in (c) and (d)]. The band is flat near the band edge, meaning the spatial coherence is low and the emitted lobe is broad. (f),(g) RCP and LCP emission maps for the device in (b) as a function of wavelength and altitude in the $y$ direction, ${\theta }_y$. (h) Emission patterns for the device in (b) at the wavelength of the quasi-BIC’s band edge [dashed contours in (f) and (g)]. The band is not flat near the band edge, meaning the spatial coherence is high and the emitted lobe is narrow.

Figure 8

Thermal metasurface focusing narrow-band RCP light emission. (a) Angular parameters of the thermal metasurface producing focused thermal emission. (b) RCP emission map as a function of the wavelength and position $z$ along the optical axis ($x=0$). (c) RCP emission map as a function of the wavelength and position $x$ at the focal plane ($z=40\mu \mathrm{m}$). (d),(e) Profile of emitted light near the focal spot of the device for the band-edge wavelength (d) and nonresonant wavelength (e).

Figure 9

Thermal metasurface generating quasimonochromatic SAM or OAM light. (a) Schematic of a thermal metasurface with an azimuthally varying directivity. (b),(c) Angular parameters defining a $30\times 30\mu \mathrm{m}$ thermal metasurface generating RCP light with an OAM of $\ell =1$ at $\lambda =1.5\mu \mathrm{m}$. (d),(e) Example electric field profile of LCP light launched in full-wave simulations to study the absorption of normally incident OAM light. (f) Absorption spectra for various OAM beams such as that depicted in (d) and (e). (g) Emissivity into the OAM channels at the peak wavelength, $\lambda =1.499\mu \mathrm{m}$, showing preferential emission to the designed OAM order and emission with overall net OAM.

Figure 10

Controlling spectral linewidth and peak emissivity with a doped photonic crystal slab. (a) Library of metaunits varying the dopant concentration (extinction coefficient $\kappa$) to control the nonradiative $Q$ factor $Q_l\propto 1/\kappa$ and perturbation magnitude $\delta$ to control the radiative $Q$ factor $Q_r\propto 1/{\delta }^2$ for $x$-polarized light. (b) Fano resonance $Q$ factor $Q_0$ as a function of $Q_l$ and $Q_r$. (c) Peak emissivity of the quasimonochromatic thermal source, as a function of $Q_l$ and $Q_r$.

Figure 11

Additional example geometries. (a)–(c) Three lattices controlling three different modes with both $Q$ factor and polarization angle control by varying the ellipticity and the in-plane orientation angles of the light-gray ellipses, respectively. The lattices and the targeted modes depicted are named according to the conventions in Ref. [54]. (d) An alternative configuration in the out-of-plane direction, having two elements: Element (1) is a dimerized PCS supporting a chiral QBIC, and element (2) is a monomer local metasurface atop a perfect electric conductor (PEC). (e),(f) RCP and LCP response for the device shown in (d). (g) Angular emissivity profile at the band-edge frequency, showing unidirectional emission of the designed spin.