Ultra‑High‑Q Terahertz Fano Resonance from a Compact Four‑Strip Metamaterial Resonator

Background

Metamaterials are engineered composites that enable electromagnetic responses not found in natural media, such as negative or ultra‑high refractive indices [1,2]. Their properties can be tuned by adjusting the geometry of their periodic metallic units [3], making them attractive for applications ranging from perfect absorption [4,5] to sensing [6–9] and cloaking [10]. One of the most compelling features of metamaterials is the ability to generate Fano resonances—sharp, asymmetric spectral features that arise from interference between a broad (bright) mode and a narrow (dark) mode [11–16]. Compared with Lorentzian resonances, Fano peaks typically exhibit much higher Q‑factors, often an order of magnitude greater [17–31], which is advantageous for sensing and filtering applications.

In the terahertz (THz) regime, Fano resonances have been realized by introducing weak asymmetries into resonator designs [17–20], enabling the excitation of underlying dark modes [21]. Graphene‑based structures have also demonstrated tunable Fano responses [22,23]. However, many reported THz metamaterials exhibit modest Q‑factors [17,32,33], limiting their practical utility. Recent designs employing multilayer or bilayer asymmetry [19,34] have improved Q‑factors but at the cost of fabrication complexity. Thus, a simple, planar structure that delivers ultra‑high Q remains a key challenge.

Here we present a coplanar four‑strip metamaterial that achieves a Q‑factor of 58 at 0.81 THz, while maintaining a compact footprint. By exploiting the bright–dark mode coupling inherent to the asymmetric geometry, we demonstrate that the Fano profile can be tuned and that additional resonances can be introduced by adding more strips.

Methods/Experimental

To break the symmetry of the resonator, strip 2 is displaced by a distance d from the central axis (Fig. 1). The full structure consists of four gold (σ = 4.09×10⁷ S m⁻¹) strips of thickness 0.2 µm patterned on a lossless silica substrate (n = 1.5). Each unit cell has a square period Pₓ = P_y = 180 µm. Strips 1, 2, and 3 are parallel, each 120 µm long and 20 µm wide; strip 4 is perpendicular, 150 µm long and 20 µm wide. The displacement d is set to 30 µm for the baseline design. Finite‑difference time‑domain (FDTD) simulations employ a 1 µm mesh in x and y and 0.02 µm in z, with periodic boundaries along x and y and perfectly matched layers along z. The incident field is normally polarized along y (E‑field) and x (H‑field) as shown in Fig. 1.

Three‑dimensional diagram of the proposed metamaterial (a). Side view (b) and top view (c) of the asymmetric resonator; the equivalent length l is marked by a dotted line.

Results and Discussion

The simulated transmission spectrum (Fig. 2a) shows two dips: a broad Lorentzian at 0.430 THz (10 % transmission) and a sharp Fano dip at 0.809 THz (26.45 % transmission). The Fano dip has a bandwidth of 0.014 THz, yielding a Q‑factor of 58—30 times higher than the Lorentzian mode. The asymmetry originates from the coupling of a bright mode (directly excited by the incident wave) and a dark mode (excited via symmetry breaking) [16,38,39]. Theoretical modeling using the electromagnetic theory of Fano resonance (Eqs. 2–5) reproduces the simulation results, confirming the bright–dark interaction as the underlying mechanism (Fig. 2f).

a Transmittance curve from FDTD. b Transmission of the bright mode only. c Field intensity under a dipole source. d–f Simulated (red) vs. theoretical (black) spectra for d = 10, 20, 30 µm.

Field maps at the two resonant frequencies (Fig. 3) reveal distinct charge distributions: the Lorentzian mode localizes on strips 1 and 3, while the Fano mode concentrates on strips 1 and 2 with strong surface currents flowing between them—signature of bright–dark coupling.

Field distributions: a |E| at 0.430 THz, b H_z with surface‑current arrows, c |E| at 0.809 THz, d H_z. The arrows indicate current direction.

Adjusting the displacement d shifts the Fano resonance: as d increases from 10 to 30 µm, the dip deepens and blue‑shifts due to stronger bright–dark coupling and a reduced equivalent resonator length l (Eq. 6). This tunability enables precise control over resonance frequency.

When a 4‑µm analyte layer of varying refractive index (n = 1–1.6) is placed on top, the Fano dip red‑shifts proportionally (Fig. 5b). The sensor exhibits a sensitivity of 0.105 THz/RIU and a figure of merit (FOM = S/linewidth) of 7.50. Using intensity‑based metrics, the sensor achieves S* = 2.6/RIU and FOM* = 10, indicating strong suitability for high‑resolution refractive‑index detection.

a Sensor cross‑section. b Transmission shift versus refractive index.

Extending the design to five horizontal strips produces two additional high‑Q Fano dips at 0.75 THz (Q = 61) and 0.91 THz (Q = 65) (Fig. 6b), demonstrating the scalability of the concept for multi‑resonance applications.

a Top view of the five‑strip structure. b Simulated transmittance curve.

Conclusion

We have demonstrated a compact four‑strip planar resonator that delivers an ultra‑sharp Fano resonance with a Q‑factor of 58 at 0.81 THz. The asymmetric line shape arises from bright–dark mode coupling, and its position and depth are tunable via the strip displacement d. The structure’s high Q, coupled with its straightforward fabrication and planar geometry, makes it an attractive platform for THz sensing and other applications requiring narrowband filtering. By adding more horizontal strips, the design can support multiple high‑Q resonances, further expanding its functional versatility.

Abbreviations

EIT:

Electromagnetically induced transparency

EM:

Electromagnetic

FOM:

Figure of merit

PIT:

Plasmon‑induced transparency

Q:

Quality factor

RIU:

Refractive index unit