One-Dimensional THz Photonic-Crystal Cavity

• The paper demonstrates that defect-based THz cavities use engineered periodic structures to induce localized resonant modes with high quality factors and deep subwavelength confinement.
• It highlights diverse implementations—including wafer stacks, suspended nanobeams, and Tamm structures—that optimize mode profiles and polarization selectivity for applications like fingerprint detection and nonlinear generation.
• The research elucidates key design trade-offs between mirror reflectivity, field localization, and fabrication tolerances, paving the way for sensitive spectroscopy and chiral light–matter coupling.

A one-dimensional terahertz photonic-crystal cavity is an electromagnetic resonator in which confinement in the terahertz spectral range is produced by a one-dimensional photonic crystal, typically through a defect state inside a distributed Bragg reflector stopband, a localized mode in a tapered nanobeam, or a Tamm state at a metal–Bragg-reflector interface. Across reported implementations, the platform spans free-space Si/air defect cavities, suspended semiconductor nanobeams, metal-terminated distributed Bragg reflectors, and gyrotropic magnetoplasma structures that support helicity-selective resonances. These devices have been used for cavity-enhanced spectroscopy, extreme electric-field concentration, nonlinear frequency conversion, polarization-selective transport, and broken-time-reversal-symmetry cavity electrodynamics in the THz regime (Shi et al., 2017, Lu et al., 2018, 0908.0463, Messelot et al., 2021, Kulkarni et al., 17 Sep 2025).

1. Structural archetypes

The most direct realization is the defect cavity formed by placing a central cavity layer between two one-dimensional distributed Bragg reflector mirrors. In the Si/air implementation used for terahertz fingerprint detection, each Bragg mirror consists of four silicon layers separated by air gaps, with silicon thickness $t = 100\,\mu\mathrm{m}$, air-gap thickness $w = 233\,\mu\mathrm{m}$, lattice constant $a = 333\,\mu\mathrm{m}$, and defect length $d_c = 319\,\mu\mathrm{m}$; the resulting defect mode is centered at $f_0 \approx 0.529\,\mathrm{THz}$ (Shi et al., 2017). A related wafer-stacked architecture replaces isotropic Bragg layers by lightly n-doped InSb under magnetic bias, while retaining a high-resistivity Si central defect layer; in that configuration the side stacks are alternating InSb and air layers, and the defect mode appears near $0.66$–$0.67\,\mathrm{THz}$ for the transmitted helicity (Kulkarni et al., 17 Sep 2025).

A second archetype is the suspended nanobeam cavity. In silicon, the reported one-dimensional nanobeam platform uses a suspended Si slab of thickness $22\,\mu\mathrm{m}$ and width $50\,\mu\mathrm{m}$, with circular air holes of period $a = 33\,\mu\mathrm{m}$ on both sides of the cavity center and a deterministic Gaussian mirror taper defined by a quadratic radius modulation; the regular cavity resonates at $w = 233\,\mu\mathrm{m}$0 with a TE-like mode dominated by $w = 233\,\mu\mathrm{m}$1 (Lu et al., 2018). In GaAs, a THz nanobeam cavity for difference-frequency generation is formed by a ridge waveguide with a periodic array of elliptical holes and an adiabatic taper, supporting an $w = 233\,\mu\mathrm{m}$2-polarized THz mode in the $w = 233\,\mu\mathrm{m}$3–$w = 233\,\mu\mathrm{m}$4 range (0908.0463).

A third archetype is the THz Tamm cavity, which is also one-dimensional but terminates the distributed Bragg reflector with a metal layer rather than a defect slab. The reported implementation uses alternating silicon and vacuum layers with $w = 233\,\mu\mathrm{m}$5 and $w = 233\,\mu\mathrm{m}$6, designed by the quarter-wave condition at $w = 233\,\mu\mathrm{m}$7, and a $w = 233\,\mu\mathrm{m}$8 Au mirror deposited on the final Si layer. Practical samples used Si wafers near $w = 233\,\mu\mathrm{m}$9–$a = 333\,\mu\mathrm{m}$0 and $a = 333\,\mu\mathrm{m}$1 vacuum gaps, for example a “3 Si + 2 vacuum” stack (Messelot et al., 2021).

These geometries share the same organizing principle: a one-dimensional periodic medium opens a stopband or bandgap along the propagation axis, and a local violation of periodicity creates a spectrally isolated localized mode. What differs is the manner in which that defect is implemented and the physical degrees of freedom that are made available—free-space access in wafer stacks, extreme localization in nanobeams, metal-interface confinement in Tamm structures, or helicity selectivity in gyrotropic stacks.

2. Confinement physics and modal metrics

For defect-based one-dimensional cavities, the Bragg condition is set by the optical thicknesses of the two materials. In the multilayer formulation, the stopband center occurs when $a = 333\,\mu\mathrm{m}$2, and quarter-wave conditions $a = 333\,\mu\mathrm{m}$3 maximize the stopband at the design frequency (Kulkarni et al., 17 Sep 2025). In the Si/air sensing cavity, the resonance satisfies the round-trip phase condition

$a = 333\,\mu\mathrm{m}$4

with $a = 333\,\mu\mathrm{m}$5 the propagation constant in the cavity and $a = 333\,\mu\mathrm{m}$6 the mirror phase (Shi et al., 2017). In the Tamm geometry, the resonance is instead determined by the interface phase condition

$a = 333\,\mu\mathrm{m}$7

so the metal termination itself acts as the defect that localizes the mode at the metal–DBR boundary (Messelot et al., 2021).

The standard spectral metric is the quality factor,

$a = 333\,\mu\mathrm{m}$8

or, in the Lorentzian form used for the chiral cavity,

$a = 333\,\mu\mathrm{m}$9

Representative values span more than two orders of magnitude depending on architecture. The Si/air fingerprint-detection cavity exhibits $d_c = 319\,\mu\mathrm{m}$0 at $d_c = 319\,\mu\mathrm{m}$1, giving $d_c = 319\,\mu\mathrm{m}$2 (Shi et al., 2017). The chiral InSb/Si cavity yields $d_c = 319\,\mu\mathrm{m}$3 from a fit with FWHM $d_c = 319\,\mu\mathrm{m}$4 at $d_c = 319\,\mu\mathrm{m}$5, with $d_c = 319\,\mu\mathrm{m}$6 observed across datasets (Kulkarni et al., 17 Sep 2025). The THz Tamm cavity reaches measured $d_c = 319\,\mu\mathrm{m}$7 at $d_c = 319\,\mu\mathrm{m}$8 with FWHM $d_c = 319\,\mu\mathrm{m}$9 in high-resolution FTIR, while simulations for idealized structures give values up to $f_0 \approx 0.529\,\mathrm{THz}$0 for 3 Si layers and $f_0 \approx 0.529\,\mathrm{THz}$1 for 4 Si layers (Messelot et al., 2021). The nanohole silicon nanobeam cavity maintains $f_0 \approx 0.529\,\mathrm{THz}$2, and horizontally coupled holes with $f_0 \approx 0.529\,\mathrm{THz}$3 yield $f_0 \approx 0.529\,\mathrm{THz}$4 in the reported example (Lu et al., 2018).

Mode volume is equally architecture-dependent. The nanohole cavity defines

$f_0 \approx 0.529\,\mathrm{THz}$5

normalized to $f_0 \approx 0.529\,\mathrm{THz}$6, and reports $f_0 \approx 0.529\,\mathrm{THz}$7 for the type-1 design and approximately a $f_0 \approx 0.529\,\mathrm{THz}$8-fold reduction relative to the regular THz photonic-crystal cavity (Lu et al., 2018). By contrast, the Tamm cavity emphasizes axial localization over roughly $f_0 \approx 0.529\,\mathrm{THz}$9 inside the distributed Bragg reflector when the final Si layer is quarter-wave, while lateral extent remains set by the device footprint and beam (Messelot et al., 2021). This distribution of $0.66$0 and $0.66$1 values underlies much of the functional diversity of one-dimensional THz photonic-crystal cavities.

3. Polarization selectivity, chirality, and broken time-reversal symmetry

A central development is the realization of a one-dimensional terahertz photonic-crystal cavity with broken time-reversal symmetry. In that device, the usual isotropic Si/air Bragg stack is replaced by gyrotropic layers of lightly n-doped InSb in an external magnetic field, while a high-resistivity silicon wafer remains as the central defect layer. The chirality originates from the nonreciprocal magnetoplasma response of InSb in Faraday geometry, with the magnetic field normal to the layer interfaces and parallel to THz propagation (Kulkarni et al., 17 Sep 2025).

For a magnetized plasma with $0.66$2 along $0.66$3, the dielectric tensor is

$0.66$4

and the circularly polarized eigenmodes have

$0.66$5

Here

$0.66$6

In the reported InSb system, $0.66$7, so a modest field already produces THz-scale cyclotron frequencies: $0.66$8 at $0.66$9 and $0.67\,\mathrm{THz}$0 at $0.67\,\mathrm{THz}$1 (Kulkarni et al., 17 Sep 2025). For $0.67\,\mathrm{THz}$2, the cyclotron-resonance-active branch corresponds to LCP and the cyclotron-resonance-inactive branch to RCP. The active branch exhibits strong absorption near $0.67\,\mathrm{THz}$3, while the inactive branch remains low loss and retains an effective refractive index close to Si in the $0.67\,\mathrm{THz}$4–$0.67\,\mathrm{THz}$5 stopband. As a result, the cavity preserves the photonic bandgap and defect resonance for one helicity while suppressing the other.

Experimentally, the field-reversal asymmetry is direct. With RCP incidence, a clear defect-mode transmission peak appears at $0.67\,\mathrm{THz}$6 for $0.67\,\mathrm{THz}$7 and is suppressed for $0.67\,\mathrm{THz}$8. With linearly polarized incidence, the transmitted component is purely circular at the cavity frequency, and its handedness flips with the sign of $0.67\,\mathrm{THz}$9; the ellipticity angle approaches $22\,\mu\mathrm{m}$0 at resonance, where

$22\,\mu\mathrm{m}$1

The mode persists from cryogenic temperature to approximately $22\,\mu\mathrm{m}$2, with optimal transmission near $22\,\mu\mathrm{m}$3 (Kulkarni et al., 17 Sep 2025).

Polarization sensitivity can also arise without nonreciprocity. The THz Tamm cavity uses a subwavelength Au strip grating with period $22\,\mu\mathrm{m}$4 and filling factor $22\,\mu\mathrm{m}$5 to induce opposite frequency shifts for two linear polarizations: the “parallel” polarization shifts to lower frequency and the “orthogonal” polarization shifts to higher frequency, with total tuning exceeding $22\,\mu\mathrm{m}$6 at $22\,\mu\mathrm{m}$7 (Messelot et al., 2021). This distinction is essential: polarization-dependent tuning does not by itself imply broken time-reversal symmetry, whereas the InSb cavity explicitly relies on a nonreciprocal gyrotropic response and field-reversal asymmetry.

4. Field enhancement and deep-subwavelength confinement

The strongest confinement reported for one-dimensional THz photonic-crystal cavities is obtained by embedding nanoholes inside a high-$22\,\mu\mathrm{m}$8 silicon nanobeam cavity. The enhancement mechanism is not merely geometric narrowing but a cascaded boundary-condition effect. At a dielectric interface, continuity of the normal electric displacement imposes

$22\,\mu\mathrm{m}$9

which gives the conventional slot effect in the low-index region. Continuity of the tangential electric field imposes

$50\,\mu\mathrm{m}$0

which produces the anti-slot effect and concentrates electric energy density back into the high-index semiconductor (Lu et al., 2018).

In the Si/air platform, $50\,\mu\mathrm{m}$1, and the reported simulations show that the two effects act sequentially so that the electric energy density in the high-index region is enhanced by

$50\,\mu\mathrm{m}$2

This leads to unusually small mode volume while retaining large $50\,\mu\mathrm{m}$3. The regular THz nanobeam cavity has $50\,\mu\mathrm{m}$4, whereas the type-1 nanohole cavity reaches $50\,\mu\mathrm{m}$5 and the type-2 cavity reaches $50\,\mu\mathrm{m}$6 at optimal geometry. The same work reports $50\,\mu\mathrm{m}$7 using

$50\,\mu\mathrm{m}$8

with an illustrative value $50\,\mu\mathrm{m}$9 for a horizontally coupled multi-hole cavity having $a = 33\,\mu\mathrm{m}$0 and $a = 33\,\mu\mathrm{m}$1 (Lu et al., 2018).

A notable feature is the location of the field maximum. In many slot or bowtie structures the maximum energy density lies in the low-index gap, but in this nanohole design the maxima reside in the semiconductor. The paper explicitly identifies this as advantageous for quantum-engineered active materials embedded in the high-index region (Lu et al., 2018). A related interface-localization logic appears in the Tamm cavity, where the electric field peaks at the Si–vacuum interface nearest the metal and can be confined over a $a = 33\,\mu\mathrm{m}$2 length within the distributed Bragg reflector (Messelot et al., 2021). Taken together, these results show that one-dimensional THz photonic-crystal cavities are not restricted to diffraction-limited volumetric confinement; some implementations retain propagation-direction localization, while others push into deep-subwavelength modal concentration.

5. Functional modalities

One-dimensional THz photonic-crystal cavities have been exploited in at least three distinct functional regimes: ultrasensitive spectroscopy, nonlinear frequency conversion, and cavity-engineered coupling to quantum materials.

For fingerprint detection, the Si/air defect cavity is tuned so that its defect mode coincides with the absorption line of the target molecule. In the reported example, the cavity resonance is set to the $a = 33\,\mu\mathrm{m}$3-lactose signature at $a = 33\,\mu\mathrm{m}$4. The analyte is placed at the cavity center, where the electric field has an anti-node, and the on-resonance transmission becomes highly sensitive to absorptive loss. The bare cavity has near-unity transmission, $a = 33\,\mu\mathrm{m}$5, and $a = 33\,\mu\mathrm{m}$6. Loading a $a = 33\,\mu\mathrm{m}$7 $a = 33\,\mu\mathrm{m}$8-lactose film reduces the on-resonance transmittance by approximately $a = 33\,\mu\mathrm{m}$9, whereas a bare film of the same thickness produces only about $w = 233\,\mu\mathrm{m}$00 attenuation in regular transmission; the sensitivity improvement factor is approximately $w = 233\,\mu\mathrm{m}$01 at $w = 233\,\mu\mathrm{m}$02 and approaches approximately $w = 233\,\mu\mathrm{m}$03 at $w = 233\,\mu\mathrm{m}$04, which is the calculated detection limit using a conservative $w = 233\,\mu\mathrm{m}$05 transmission-drop criterion (Shi et al., 2017).

For nonlinear generation, the THz cavity can serve as one member of a triply resonant system. The GaAs scheme couples a one-dimensional THz photonic-crystal nanobeam cavity to a doubly resonant near-infrared nanobeam cavity supporting orthogonal TE-like and TM-like modes, so that three mutually orthogonal cavity fields participate in a $w = 233\,\mu\mathrm{m}$06 difference-frequency-generation process. The THz mode is $w = 233\,\mu\mathrm{m}$07-polarized, the NIR modes are dominated by $w = 233\,\mu\mathrm{m}$08 and $w = 233\,\mu\mathrm{m}$09, and the nonlinear overlap exploits the off-diagonal tensor elements of GaAs. The reported overlap parameter is $w = 233\,\mu\mathrm{m}$10, the THz field-overlap metric is $w = 233\,\mu\mathrm{m}$11, and for a representative design with $w = 233\,\mu\mathrm{m}$12, $w = 233\,\mu\mathrm{m}$13, and $w = 233\,\mu\mathrm{m}$14, the estimates are $w = 233\,\mu\mathrm{m}$15, optimal pump power $w = 233\,\mu\mathrm{m}$16, and THz output power $w = 233\,\mu\mathrm{m}$17 at quantum-limited photon conversion (0908.0463).

For cavity-modified quantum materials, the gyrotropic InSb/Si structure is explicitly proposed as a platform for chiral light–matter interactions and vacuum dressed quantum condensed matter in the terahertz regime. Because the confined mode is uniformly circularly polarized and the central Si wafer can serve as a substrate for future thin-film samples, the device is presented as a route toward chiral cavity QED, Berry-curvature engineering, and possible Dirac-gap induction in graphene when embedded in such chiral cavities (Kulkarni et al., 17 Sep 2025). This suggests that one-dimensional THz photonic-crystal cavities are not only passive spectral filters or field concentrators, but also tunable photonic environments for modifying matter response at low energies.

6. Fabrication, characterization, and design trade-offs

Fabrication strategies differ sharply across implementations. Free-space multilayer cavities are assembled from bulk wafers and controlled air gaps. The sensing cavity uses silicon plates separated by air and held with double-sided adhesive, with the sample-loading region at the cavity center (Shi et al., 2017). The chiral cavity uses alternating InSb wafers and air gaps, with paper spacers maintaining the air layers and thicknesses verified by micrometer (Kulkarni et al., 17 Sep 2025). The THz Tamm cavity uses manual stacking of double-side-polished high-resistivity silicon wafers and metallic spacers in a custom holder with grooves and clamps, followed by thermal evaporation of a $w = 233\,\mu\mathrm{m}$18 Au mirror or laser-lithographic definition of a strip grating (Messelot et al., 2021). Nanobeam implementations are monolithic: the silicon nanohole cavity is compatible with photolithography and deep reactive-ion etching, while the finer nanoholes and sub-$w = 233\,\mu\mathrm{m}$19 gaps are associated with electron-beam lithography, focused-ion-beam milling, or atomic layer lithography (Lu et al., 2018). The GaAs triply resonant scheme adds a multi-scale alignment problem, because the THz nanobeam is combined with an NIR nanobeam separated by an air gap of about $w = 233\,\mu\mathrm{m}$20 (0908.0463).

Characterization methods are similarly diverse. THz time-domain spectroscopy is the default for free-space multilayer cavities. The chiral cavity uses THz time-domain magneto-spectroscopy in Faraday geometry, optical rectification in ZnTe driven by an $w = 233\,\mu\mathrm{m}$21 ultrafast amplifier, electro-optic sampling in ZnTe, and a spectral range of $w = 233\,\mu\mathrm{m}$22–$w = 233\,\mu\mathrm{m}$23; Stokes parameters are extracted from the complex transmitted fields to quantify ellipticity (Kulkarni et al., 17 Sep 2025). The fingerprint-detection work establishes trends through finite-element simulations under normal incidence and notes that typical experimental realization uses free-space THz-TDS (Shi et al., 2017). The Tamm cavity is studied with transfer-matrix calculations, COMSOL finite-element simulations, standard FTIR, angle-resolved reflectivity, and synchrotron-based high-resolution FTIR with $w = 233\,\mu\mathrm{m}$24 resolution (Messelot et al., 2021). The nanohole cavity uses 3D and 2D FEM in COMSOL Multiphysics 4.3 to extract eigenmodes, $w = 233\,\mu\mathrm{m}$25 factors, and field maps (Lu et al., 2018). The DFG cavity combines 3D-FDTD field extraction with temporal coupled-mode theory (0908.0463).

Across all variants, the key trade-offs are between mirror reflectivity and dynamic range, confinement and radiation leakage, material loss and external coupling, and spectral selectivity and fabrication tolerance. Increasing the number of Bragg periods raises reflectivity and $w = 233\,\mu\mathrm{m}$26 in defect cavities, but narrows linewidth and can reduce usable dynamic range in sensing (Shi et al., 2017). In the chiral cavity, strong dichroism requires lightly doped, high-mobility InSb, and performance degrades at high temperature because intrinsic thermal carriers increase absorption; room-temperature operation with the reported wafers is limited (Kulkarni et al., 17 Sep 2025). In nanohole cavities, larger perturbations can lower radiation-limited $w = 233\,\mu\mathrm{m}$27 even as mode volume shrinks (Lu et al., 2018). In Tamm cavities, the balance between DBR radiative leakage and Au dissipation determines whether the system approaches critical coupling, and for grating mirrors an additional full Au reflector at $w = 233\,\mu\mathrm{m}$28 above the grating is required to preserve high $w = 233\,\mu\mathrm{m}$29 for the orthogonal polarization (Messelot et al., 2021). These trade-offs define the practical design space of one-dimensional terahertz photonic-crystal cavities: highly adaptable, but always conditioned by the interplay of stopband engineering, field localization, loss channels, and coupling geometry.