3D printed sub-terahertz photonic crystal for wireless passive biosensing

The work addresses the need for pre-infection, stand-off detection of airborne pathogens highlighted by the COVID-19 pandemic. Existing rapid tests detect only post-infection, and RFID-based wireless sensors require active circuitry for energy harvesting and communication, increasing complexity and limiting read range. Passive refractive index sensors using electromagnetic waves can provide label-free, specific detection via analyte-induced refractive index changes, but most current approaches require proximity or have short read distances. The authors propose a passive, wireless biosensor using a high-Q sub-terahertz photonic crystal (PhC) slot resonator operating in the 100 GHz–1 THz range, a compromise that balances antenna size, EM wavelength for communication range, and interaction scale for sensing. The research goal is to demonstrate a 3D-printed alumina PhC resonator integrated with a dielectric rod antenna (DRA) enabling remote readout (up to ~0.9 m) and sensitive detection of thin biomolecular films, validating feasibility for distributed wireless biosensing networks.

The paper reviews several EM-based sensing platforms: metasurfaces with split-ring resonators, surface plasmon resonance (SPR) sensors, and whispering-gallery-mode (WGM) resonators. Metasurfaces and SPR suffer from relatively low quality factors, while WGMs have limited field concentration at the analyte location, yielding low figures of merit (FOMs). Optical PhC resonators are highlighted as exceptions, offering design flexibility, ultra-high Q, and small size, with demonstrated ultralow limits of detection for biomarkers. However, most passive concepts require near-field readout. Prior research has explored THz PhCs for wireless communication and biosensing, and the authors’ earlier work showed a sub-THz PhC slot resonator with high FOM compared to other THz sensors. These insights motivate developing a sub-THz PhC resonator optimized for both sensitivity (via high Q and slot field enhancement) and wireless readability.

Design and simulation: The sensor comprises a PhC slot resonator for sensing, a PhC waveguide for coupling, and an integrated dielectric rod antenna (DRA) for wireless backscatter. The PhC slab (periodic holes) creates a bandgap; line defects form the waveguide and cavity. The resonator is designed in CST (frequency-domain solver) and, for optimization, coupled to two PhC waveguides to observe S21. Alumina (Al2O3) is used with relative permittivity ~9 and loss tangent ~0.00022 in W-band, enabling low-loss high-Q operation and compatibility with lithography-based ceramic manufacturing (LCM) 3D printing. Geometric parameters include lattice period p ≈ 1100 µm and slab thickness t = 0.5 p for W-band operation (75–110 GHz). The resonator length is 6 periods to approximate a Gaussian field distribution and reduce radiation losses. A narrow slot is introduced at the cavity center to concentrate the electric field at the analyte location; an equivalent thin-film analyte is modeled on the slot walls to evaluate sensitivity. Adjacent holes around the slot are laterally shifted (e.g., along the resonator and transverse directions with displacements on the order of 100–200 µm) to maximize Q by tailoring the mode envelope. The optimized slot width is about 244.5 µm. Coupling trade-offs are explored by varying the number of hole rows between waveguide and cavity and hole radius r. Three rows are used to maintain sufficient transmitted signal for wireless readout. Two hole radii are compared: r = 0.30 p and r = 0.27 p. Simulated S-parameters show resonances at ~97.440 GHz (Q ≈ 2830) and ~95.144 GHz (Q ≈ 1740) for r = 0.30 p and r = 0.27 p, respectively. Sensitivity simulations: A thin analyte film is modeled with varying refractive index (n from 1 to 2 at fixed 0.5 µm thickness) and varying thickness (0.1–1.0 µm at fixed n ≈ 1.8). Resonance shifts vary approximately reciprocally with n (for very thin films) and linearly with thickness. The higher-Q design (r = 0.30 p) yields larger shifts and higher FOM than the lower-Q (r = 0.27 p). Antenna and wireless design: The DRA is a tapered dielectric rod integrated with the PhC waveguide (base cross-section ~1.25 mm width × 0.55 mm height) and terminated in a tip. Taper lengths of 7, 14, and 21 mm are studied. When excited by a PhC waveguide alone, simulated gains are ~9.3, 11.5, and 13.0 dBi with a main lobe along the rod axis. When excited by the resonator mode (which differs from the fundamental waveguide mode), the radiation pattern becomes multi-lobed, and peak gains in the rod-tip direction reduce to approximately −2.7, 5.1, and 4.5 dBi for 7, 14, and 21 mm rods, respectively, while still covering a large angular range. Fabrication: Sensors are 3D printed from alumina using an LCM process (CeraFab 7500 printer; ~25 µm resolution). Green bodies are cleaned and sintered (up to ~1600 °C, 2 h) to yield dense ceramic parts. Six variants were fabricated: S027T7, S027T14, S027T21, S030T7, S030T14, S030T21 (S027/S030 denote r = 0.27 p/0.30 p; T7/T14/T21 denote DRA lengths). The PhC body footprint excluding the DRA is ~12 × 15 mm. Dimensional checks show slot width ~245.8 µm and hole radius ~330.1 µm, close to design; long DRAs may show slight warpage due to sintering temperature gradients. Measurement and signal processing: A VNA (ZVA67 with ZC110 W-band extender) and a 26 dB gain horn antenna serve as the reader. Frequency resolution is 1 MHz. The sensor’s DRA tip is aligned to the horn using a laser; the distance L is the horn-edge to DRA-tip separation. Frequency-domain S11 is transformed to time-domain via inverse Fourier transform. The received signal contains: (i) instrument and setup reflections (clusters), (ii) sensor reflections, and (iii) a decaying oscillatory component from the high-Q resonator. A Tukey time gate isolates the oscillation (e.g., starting ~1 ns after the second cluster and ending before a later internal reflection), and short-time Fourier transform (STFT) visualizes time–frequency content. The resonance is then read from the time-gated spectrum. Wireless read range and angle tests: L is swept from 0 to 0.9 m (0.1 m steps) with a fixed 10 ns time gate to assess backscattered resonance peak magnitude for all six samples. Time-window length is varied (start at 5 ns; end 10–20 ns) at L = 0.5 m to study Q-factor estimation and peak evolution. Angular acceptance is measured by rotating the sensor ±45° around axes parallel and perpendicular to the PhC at L = 0.3 m in 2.5° steps. Wireless biomolecule sensing: Bovine serum albumin (BSA) solutions at 2.4, 4.8, 7.2, and 9.6 µg/µL are pipetted (1.5 µL) into the slot; after drying, protein films of estimated thickness ~0.5, 1.0, 1.5, and 2.0 µm form on slot walls. For each condition, n = 5 independent measurements are made; resonance shifts are quantified from time-gated spectra.

• Passive wireless readout of the PhC sensor is demonstrated up to 0.9 m for all fabricated variants; read range is ultimately limited by instrument noise floor, available transmit power, and antenna gain.
• Time gating effectively removes clutter, revealing a clear resonance peak in the spectrum; STFT shows the decaying oscillation consistent with high-Q cavity behavior.
• DRA length influences backscattered signal: longer tapers increase resonance peak magnitude but narrow the angular beam; slight fabrication warpage affects pattern shape.
• Coupling/Q-factor trade-off: r = 0.30 p (higher Q) shows stronger sensitivity but lower immediate coupling than r = 0.27 p. With longer time gates (>15 ns), the higher-Q design’s measured Q increases while the lower-Q design’s measured Q degrades due to noise dominance.
• Angle of incidence: sensors are readable over a wide angular range; acceptance angles on the order of ~90° (measured by scanning −45° to +45° in two orthogonal planes).
• Biomolecule sensing (BSA): For S030T14, resonance shift increases approximately linearly with concentration (R² ≈ 0.988), slope ≈ 5.2 MHz per (µg/µL) for 1.5 µL deposit; converting to film thickness gives sensitivity ≈ 25 MHz/µm. With 1 MHz frequency resolution, a ~40 nm film is nominally detectable.
• Measurement variability: baseline standard deviation ≈ 2.1 MHz (at 0 µg/µL); standard deviation increases with concentration due to drying and distribution effects. A 3σ criterion implies a minimum reliably detectable mass of ~1.8 µg or ~250 nm film thickness under current setup.
• Simulations: Resonances at ~97.440 GHz (Q ≈ 2830) for r = 0.30 p and ~95.144 GHz (Q ≈ 1740) for r = 0.27 p; resonance shifts scale reciprocally with analyte RI (for very thin films) and linearly with film thickness; higher-Q design achieves higher FOM (e.g., ~1.69 vs ~0.94).

The study demonstrates that a sub-THz PhC slot resonator can simultaneously achieve high surface sensitivity and long-range wireless readability, directly addressing the challenge of stand-off, passive biosensing. By leveraging the high Q-factor to store energy and generate a long-lived oscillation, time-domain gating can isolate the resonance response from environmental clutter without calibration, enabling remote detection. The slot concentrates the electric field at the analyte location to maximize interaction, improving the figure of merit. Empirical results confirm key performance goals: read distances up to 0.9 m, wide acceptance angles, and quantifiable detection of sub-micrometer protein films. The trade-off between coupling and Q-factor is managed to maintain sufficient backscattered signal while retaining sensitivity. Compared with active RFID-based sensors, the passive, chipless, battery-free design promises lower cost, simplicity, and suitability for distributed deployment for environmental pathogen monitoring. The results validate feasibility and outline a path for scaling to networks of passive wireless biosensors.

This work introduces a 3D-printed alumina sub-THz photonic crystal slot resonator integrated with a dielectric rod antenna for passive, wireless refractive index sensing. It achieves up to 0.9 m read range, wide acceptance angles (~90°), and sensitive detection of thin protein films, using time-gated processing to extract resonances amid clutter. The concept offers a compact, chipless, low-cost solution suitable for distributed monitoring of airborne pathogens. Future directions include: improving antenna gain (e.g., planar dielectric antennas targeting ~20 dBi) and DRA–waveguide mode transition for higher coupling efficiency; deploying robust, calibration-free detection algorithms on compact readers to extend range; increasing resonant frequency and enlarging functionalized surface area to approach monolayer biomolecule detection; transitioning to high-throughput, low-cost DRIE microfabrication on silicon for uniformity and scalability; and multiplexing multiple sensors with distinct resonant frequencies using simplified radar readers to locate and interrogate many sensors in a network.

• Current read range is constrained by instrument noise floor, available transmit power, and antenna gain; obstacles in line-of-sight can attenuate signals significantly.
• Fabrication of long DRAs can introduce slight warpage, affecting the radiation pattern and angular response.
• There is an inherent trade-off between coupling strength and Q-factor; too high Q reduces coupled/backscattered energy, while stronger coupling can reduce Q and sensitivity.
• Measurement variability arises from manual liquid handling (volume placement, non-uniform drying), ambient temperature fluctuations, and device noise; baseline standard deviation ~2.1 MHz influenced detection limits.
• Under current conditions, reliable detection threshold using a 3σ criterion corresponds to ~250 nm film thickness; monolayer detection requires further sensitivity improvements.
• Radiation pattern mismatch between resonator-excited mode and ideal DRA mode lowers effective gain and complicates far-field patterns.