Multilayer Resonant Metasurfaces

• Multilayer resonant metasurfaces are ultrathin, stacked photonic architectures that use engineered resonant modes and precise interlayer coupling to achieve high-Q and broadband spectral performance.
• They employ controlled geometric design and decoupled spectral channels to facilitate advanced optical filtering, wavefront shaping, and pulse control across multiple frequency bands.
• Innovative fabrication techniques like direct-write lithography and self-rolling enable scalable on-chip integration, paving the way for next-generation nanophotonic devices.

Multilayer resonant metasurfaces are ultrathin, planar photonic structures comprising multiple stacked layers of subwavelength-patterned films, each supporting one or more engineered resonant modes. By combining resonators in sequence with controlled spacing and symmetry perturbations, these metasurfaces exhibit enhanced spectral, spatial, and temporal control of electromagnetic waves, overcoming limitations of single-layer systems. The multilayer paradigm provides access to complex transfer functions, high-Q resonances, decoupled spectral channels, broadband dispersion engineering, and scalable on-chip integration for applications ranging from spectral filtering to wavefront shaping and nanophotonic pulse control.

1. Theoretical Foundations: Resonant Modes, Stacking, and Coupling

Resonant metasurfaces employ spatially and spectrally engineered eigenmodes. In the multilayer context, each patterned layer supports localized or collective resonances—such as Mie, plasmonic, or quasi-bound states in the continuum (q-BIC)—with properties determined by layer geometry, symmetry, and material composition.

Modal Picture

• Single-layer modes: Symmetry-protected BICs in periodic slabs become q-BICs under controlled symmetry breaking (e.g., post tilt angle $\theta$), yielding radiatively coupled resonances with linewidth $\Gamma$ scaling as $\Gamma\propto \theta^2$. The Q-factor is $Q\approx\omega_0/(2\gamma_{\rm r}(\delta))$ in the perturbative regime, providing geometric tunability (Baspinar et al., 19 Jan 2026).
• Multilayer stacks: For $N$ decoupled layers separated by low-index spacers ($d_{\rm spacer}\gg\lambda$), the composite transfer matrix is $$\mathbf{M}_{\rm tot} = \prod_{n=1}^N \left[\mathbf{M}^{(n)}_{\rm layer}(\omega)\mathbf{M}_{\rm spacer}\right]$$ (Baspinar et al., 19 Jan 2026, Zhou et al., 2011).
• Interlayer coupling: Weak modal hybridization introduces off-diagonal coupling in coupled-mode theory: $$\frac{d a_n}{dt} = (i\omega_{0,n}-\gamma_n)a_n + \sum_{m\neq n}\kappa_{nm}a_m + \kappa_n s_+$$. This results in collectively shifted resonance wavelengths and linewidths (Baspinar et al., 19 Jan 2026).
• Supercavity limit: Stacking identical dielectric metasurfaces (e.g., TiO$_2$ Mie arrays) at integer half-wavelength separations, $d=n\lambda/2$, produces bound states closely analogous to Fabry–Pérot supercavity modes, supporting $Q>10^5$ and field enhancement by $>30\times$ (Zhang et al., 2023).

Stacked Resonance Design Principles

• Independent spectral channels: Provided the interlayer interaction is suppressed, each metasurface can be tuned for an independent spectral resonance and spatial phase response (Baspinar et al., 19 Jan 2026, Malek et al., 2020).
• Transfer-matrix approach: The stack response is modeled by concatenating individual metasurface and spacer matrices, enabling reflection, transmission, and absorption to be calculated as a function of frequency, polarization, and incidence angle (Zhou et al., 2011).

2. Multilayer Resonant Metasurface Design and Architectures

Material Platforms and Geometries

• Dielectric and semiconductor platforms: Sb$_2$S$_3$ elliptical posts (n=2.1), TiO$_2$ nanocubes, amorphous silicon, Au/SiO$_2$ or Au/Si nanostructures (Baspinar et al., 19 Jan 2026, Zhang et al., 2023, Kumagai et al., 2020, Bermúdez-Ureña et al., 2019).
• Layer morphologies:
  • Rectangular or square meta-atom lattices, with period and aspect ratio controlling $\lambda_{\rm res}$ (Baspinar et al., 19 Jan 2026, Malek et al., 2020).
  • Stacks of electric- and magnetic-dipole resonators, e.g., interleaved split-ring or nanorod arrays (Tsilipakos et al., 2019, Tsilipakos et al., 2023).
  • Self-rolled architectures: planar nanohole/nanorod arrays transformed into multilayer tubes via strain engineering (Bermúdez-Ureña et al., 2019).

Spectral Tuning

• Resonance wavelength: Tuned by lattice aspect ratio, post dimension, or layer thickness; e.g., sweeping $\alpha=Py/Px$ in (Baspinar et al., 19 Jan 2026) shifts $\lambda_{\mathrm{res}}$ linearly across 400 nm.
• Q factor: Controlled by symmetry-breaking parameter (e.g., post tilt angle $\theta$), with $Q(\theta)\sim \theta^{-2}$ for small angles [(Baspinar et al., 19 Jan 2026, Malek et al., 2020, Malek et al., 2020) the PB phase rotation].
• Interlayer decoupling: Choice of spacer thickness (e.g., $d_{\rm spacer}\approx 1.4~\mu\mathrm{m}$) ensures independent resonant operation per layer (Baspinar et al., 19 Jan 2026).

Multi-Channel and Broadband Functionalities

• Spectrally decoupled filters: Stacks supporting $N$ programmable sharp resonances in $<10~\mu\mathrm{m}$ thickness (Baspinar et al., 19 Jan 2026).
• Pulse shaping and dispersion engineering: Engineered non-uniform resonance trains enable arbitrary quadratic phase profiles for pulse compression or dispersion compensation, with the number of layers scaling quadratically with the signal bandwidth and desired group-delay dispersion (Tsilipakos et al., 2023, Tsilipakos et al., 2019).

Device Type	Tunable Parameters	Target Functionality
Sb$_2$S$_3$ elliptical stacks	Aspect ratio $\alpha$, $\theta$	Multi-resonant spectral filters
TiO$_2$ Mie supercavity	Cube size, air gap $s$	Ultra-high-Q field enhancement
Self-rolled Au/SiO$_2$ tubes	Roll radius, twist angle $\Delta\theta$	Chiral, plasmonic, multi-resonant
Interleaved SRR stacks	Sequence encoding, PB phase	Beam steering, THz modulation

3. Fabrication Strategies: High-Throughput and Advanced Stacking

Lithography-Driven Multilayer Assembly

• Direct-write EBL with in-situ conversion: Sb-BDCA spin-coating and EBL drive antimony to high-index Sb$_2$S$_3$ without deposition/etch, enabling $>2\times$ reduction in fabrication time and lithography-limited (<10 nm) interlayer alignment (Baspinar et al., 19 Jan 2026).
• Self-rolling technique: A 2D nanopatterned bilayer, upon selective release, rolls into a tube with $N=1$–$4$ well-defined layers, maintaining nanoscale registry of meta-elements (Bermúdez-Ureña et al., 2019).
• Layer-by-layer planarization and alignment: For high-Q supercavities, sub-10 nm stacking using sacrificial spacer layers and wafer bonding is feasible (Zhang et al., 2023).
• Polymer-spacer and lift-off/photolithography: Used in multilayer split-ring THz architectures to achieve near-unity cross-polarized efficiency (Tian et al., 2020).

Process Comparison

Fabrication Approach	# Steps per Layer	Alignment Accuracy	Throughput
Traditional multilayer EBL	$\sim$9	$<10$ nm	Slow (12–15 h/3 layers)
Sb-BDCA direct-write EBL	4	$<10$ nm	Fast ($\sim$4 h/3 layers)
Self-rolled	1 (pattern+release)	Nanoscale	High (batch, wafer-scale)

4. Physical Effects, Modal Interactions, and Performance Metrics

Multi-Resonant and Supercavity Phenomena

• Fabry–Pérot–Mie supercavities: Stacked metasurfaces at $d=n\lambda/2$ generate standing-wave fields with $20$–$35\times$ local field enhancement and $Q>10^5$, outstripping single-layer and symmetry-broken BIC approaches (Zhang et al., 2023).
• Field distributions: Both in-plane (lateral) and out-of-plane (axial) enhancement is observed, with microcavity antinodes at interfacing metasurfaces.
• Programmable spectral response: Multi-layer stacks encode $N$–$2N$ spectral features; decorrelated filter arrays achieve average Pearson correlation coefficients $|r_{xy}|=0.11$–$0.21$, surpassing prior metasurface/photonic-crystal implementations for compressive-sensing applications (Baspinar et al., 19 Jan 2026).

Benchmark Metrics

• Resonance Q: $Q\sim$15–100 (plasmonic, THz); $Q>10^5$ (supercavity).
• Transmission/reflection amplitude: up to 0.8–0.9 for cross-polarized THz devices (Tian et al., 2020); $T_c \leq 0.08$ for narrowband dielectric doublets (Malek et al., 2020).
• Bandwidth: Up to 20% fractional bandwidth for THz splits; pulse compression bandwidth is only limited by number and placement of stack resonances (Tsilipakos et al., 2023).
• Layer-to-layer spectral independence: Demonstrated by shifted $\lambda_{\rm res}$ and $Q$ per layer (Baspinar et al., 19 Jan 2026, Malek et al., 2020).

5. Advanced Multilayer Engineering: Spatio-Temporal and Broadband Designs

Multiresonant and Interleaved Architectures

• Multi-resonant metasurfaces for broadband phase engineering are synthesized by stacking or co-locating Lorentzian electric and magnetic resonances (via engineered meta-atoms), allowing monotonic phase windings and prescribed group delay profiles far exceeding $2\pi$ (Tsilipakos et al., 2019, Tsilipakos et al., 2023).
• Quadratic phase manipulation: By designing a non-uniform train of resonances (positions, strengths, damping rates) to realize a target transfer function $t(\omega) = A e^{i\varphi(\omega)}$ with $\varphi(\omega)$ quadratic, arbitrary pulse chirping and dispersion compensation is achieved, limited only by passivity and causality (Tsilipakos et al., 2023).
• Polarization, chirality, and PB phase stacking: Twisted nanorod layers (Bermúdez-Ureña et al., 2019) and layered SRR/PB phase sequences (Tian et al., 2020) enable control over polarization response, circular dichroism, and spatial beam steering.

Design Guidelines

• The total attainable phase swing in reflection for $N_e$ electric and $N_m$ magnetic resonances: $\Delta\phi \approx 2\pi(N_e + N_m)$ (Tsilipakos et al., 2019).
• Group delay and group-delay dispersion bandwidth are jointly set by the density and arrangement of resonances (Tsilipakos et al., 2023).

6. Applications and Integration Pathways

Spectral Filtering, On-Chip Spectroscopy, and Imaging

• Multi-channel filters: Enable dense spectral coding for compressive hyperspectral imaging and single-shot reconstruction with low inter-filter correlation (Baspinar et al., 19 Jan 2026).
• Integrated optics: Stacks compatible with CMOS processes may be directly bonded to detector arrays, offering monolithic, multi-functional on-chip spectrometers and imaging modules (Baspinar et al., 19 Jan 2026, Zhang et al., 2023).
• Programmable and nonlinear meta-optics: Multilayer structures enable spatially and spectrally distinct wavefront shaping, dynamically reconfigurable beam steering (THz), and nonlinear generation at ultra-low thresholds via field locking in supercavities (Zhang et al., 2023, Malek et al., 2020).

Temporal Signal Engineering

• Ultrathin broadband pulse control: Arbitrarily strong and broadband dispersion compensation, programmable group delays, and pulse chirping for microwave, THz, or optical communications are accessible by this approach (Tsilipakos et al., 2023).

Emerging Frontiers

• Chiral and polarization-selective elements: Self-rolled and multilayer-twisted narrows facilitate circular dichroism and tailored polarization conversion with nanoscale device thickness (Bermúdez-Ureña et al., 2019).
• Thermal emitters and absorbers: Lithography-free multilayer plasmonic stacks act as efficient mid-IR absorbers/emitters for sensing and radiative cooling (Kumagai et al., 2020).

7. Outlook and Scalability

• Wafer-scale and high-throughput fabrication: The transition from multi-step planarization/etching to direct-write or self-rolling approaches massively increases throughput and reduces cost (Baspinar et al., 19 Jan 2026, Bermúdez-Ureña et al., 2019).
• Maximum spectral programmability: By increasing $N$, stacks approach arbitrary response synthesis, only limited by fabrication tolerances, layer alignment, and optical loss.
• Integration with emerging platforms: Compatibility with 2D materials, quantum emitters, or active spacers points toward in situ tunable and hybrid quantum/optical devices (Zhang et al., 2023).
• Quantitative design framework: All key parameters—geometry, materials, mode Q, layer thickness, and spatial alignment—are supplied for direct theoretical and experimental translation (Baspinar et al., 19 Jan 2026, Malek et al., 2020, Tsilipakos et al., 2023).

Multilayer resonant metasurfaces provide a physically transparent, modular, and scalable photonic architecture for advanced light manipulation in spectral, spatial, and temporal domains, setting the foundation for the next generation of programmable, integrated nanophotonic systems (Baspinar et al., 19 Jan 2026, Zhang et al., 2023, Malek et al., 2020, Tsilipakos et al., 2019, Tsilipakos et al., 2023).