Absolute measurement of penetration depth of superconducting thin films using microwave stripline resonators

Source: arXiv:2605.28759 · Published 2026-05-27 · By Arghya Dutta, Ajeet Salunke, Mahesh Poojary, Vivas Bagwe, Sangita Bose, Pratap Raychaudhuri

TL;DR

This paper addresses the long-standing challenge of determining the absolute superconducting penetration depth (λ) in thin films using microwave resonator techniques. While microstrip resonators are highly sensitive to changes in λ through shifts in resonant frequency, extracting the absolute value is complicated by geometric factors, device packaging, and electromagnetic environment. The authors propose a hybrid approach combining precise resonator frequency measurements with detailed finite-element electromagnetic simulations (COMSOL Multiphysics) that model the full experimental geometry. By numerically solving Maxwell's equations with a superconducting surface boundary condition encoding λ, they create lookup tables mapping λ to resonant frequency and invert this to find λ for different films. They validate the approach on NbN films of varying thicknesses and extend it to materials with lower Tc, such as amorphous Re6Zr, by using a novel 'flip-film' geometry which couples the test film capacitively to a characterized NbN microstrip via a thin Mylar spacer.

The technique yields absolute λ values consistent with independent low-frequency mutual inductance methods within experimental uncertainty, and accurately captures their temperature dependence following dirty-limit BCS theory. The flip-film method allows non-destructive characterization without patterning the test film but is most sensitive for thin films (< λ/4). The combined measurement and simulation method overcomes previous calibration and modeling issues, establishing microstrip resonators as a versatile quantitative platform for superconducting electrodynamics characterization at microwave frequencies.

Key findings

• Measured penetration depth λ(0) for 60 nm NbN film is 276 nm, agreeing within 1% with 275 nm from low frequency mutual inductance.
• Uncertainty in λ from film thickness and substrate dielectric constant is estimated at ±8%.
• Higher harmonic resonant frequencies (f2, f4, f5) match simulation using the same λ value, validating modeling accuracy (Fig. 2c).
• Temperature dependence of 1/λ² fits dirty-limit BCS expression well with gap Δ(0) ≈ 2.88 meV consistent with tunneling data (Fig. 2e).
• For thinner NbN films (6 nm), λ(0) increases (by factor ~1.9) compared to thick films, likely due to vortex-antivortex microwave response near BKT transition.
• Flip-film geometry using a-Re6Zr film gives λ(0) = 1039 nm and Δ(0) = 0.96 meV, consistent with two-coil and tunneling measurements, with ±12% error.
• Sensitivity of flip-film frequency shift to λ decreases sharply for films thicker than λ/4, limiting range of this method (Appendix 2).
• Finite-element mesh refinement shows resonant frequency uncertainty below 0.04% with final mesh parameters used.

Threat model

n/a - The paper is focused on physical measurement techniques rather than an adversarial security context. The 'threat' is measurement uncertainty due to device complexity and geometry, not malicious actors.

Methodology — deep read

1. Threat Model & Assumptions: The work is experimental-metrological rather than adversarial. The problem is accurate non-destructive measurement of absolute penetration depth λ, overcoming geometric and environmental uncertainties. Assumptions include superconducting films operating in a regime where kinetic inductance dominates, with real part of conductivity small compared to imaginary part except near Tc.

2. Data: Multiple NbN films of thickness ranging from 6 nm to 100 nm grown by magnetron sputtering on MgO substrates. Additional sample materials include 90 nm Nb3Sn and amorphous Re6Zr films. Film thicknesses and lateral dimensions are independently measured. Resonator devices patterned with laser photolithography and ion milling into microstrip geometries (width 13-19 μm, length ~19 mm). Measurements performed between 2 K and Tc (~15 K for NbN). Transmission S12 measured with vector network analyzer in cryostat, amplitudes and phases recorded from 0.5 GHz to ~16 GHz.

3. Architecture/Algorithm: The microstrip resonator’s resonant modes depend on the total inductance per unit length L = LG + Lk. LG is geometric inductance; Lk is kinetic inductance proportional to λ²/t with geometric factor g. The resonant frequency fn = n/(2l) × 1/√(CL), so shifts in λ change fn. Exact analytic formulas for LG, g, and effective dielectric constant εeff are not sufficiently precise due to device geometry, packaging, and coupling effects.

To address this, the authors model the entire experimental setup including the microstrip (with lithographically measured dimensions), substrate materials, Mylar spacer if present, and package box in COMSOL Multiphysics. They apply appropriate boundary conditions for superconducting films using a transition boundary condition encoding complex conductivity: σ = σ' - iσ'', with σ'' = 1/(μ0 ω λ²).

They solve the full Maxwell eigenvalue problem with finite element mesh comprising very fine mesh (max 5 μm elements) around microstrip and coarser mesh elsewhere. From these eigenfrequencies, they generate lookup tables for fn vs λ.

4. Training Regime: Not a machine learning model but experimental steps include repeated microwave frequency scans to measure resonance peaks at various temperatures. Resonance frequencies extracted by fitting lineshape S12 amplitude to known formula involving loaded Q factor. Resonances measured up to ~0.96 Tc.

5. Evaluation Protocol: Extracted λ by interpolating lookup tables generated by simulations. Validation includes:

• Comparing fundamental and harmonic resonant frequencies.
• Comparing λ(0) to values from independent techniques like 30 kHz mutual inductance.
• Fitting temperature dependence of 1/λ² to dirty-limit BCS formula, extracting superconducting gap.
• Testing flip-film geometry by placing sample films atop NbN microstrip separated by 25 μm Mylar spacer and observing frequency shifts.
• Mesh convergence tests to control simulation error.

No cross-validation or adversarial tests since empirical physics measurement.

6. Reproducibility: COMSOL model files and experimental data available on request. Film growth and patterning detailed sufficiently for replication. Some reliance on unpublished or proprietary parameters such as dielectric constants at low temperatures, but efforts made to calibrate (e.g. εr Mylar). Overall, reproducible with standard thin film and microwave measurement setups.

Concrete Example: For a 60 nm NbN film microstrip with measured f1 = 4.08515 GHz at 2.2 K, the simulation lookup curve f(λ) yields λ = 276 nm. Higher harmonic frequencies predicted with this λ closely match measurement. Measurement repeated at increasing T to trace out λ(T) which follows expected BCS dependence. In flip-film setup, an a-Re6Zr film atop 60 nm NbN resonator with Mylar spacer shows frequency shift allowing extraction of λ for unpatterned a-Re6Zr.

Overall, the detailed electromagnetic modeling combined with precise microwave spectroscopy enables reliable absolute λ extraction from resonator frequency data in complicated real-world geometries.

Technical innovations

• Combining finite-element electromagnetic simulations of full resonator geometry with microwave resonant frequency measurements to extract absolute penetration depth without relying on simplified analytical models.
• Use of transition boundary condition in COMSOL for superconducting films characterized by complex conductivity to efficiently simulate microwave response without volumetric mesh of film.
• Development of flip-film measurement geometry coupling a test superconducting film capacitively via a thin dielectric (Mylar) spacer to a calibrated NbN microstrip resonator enabling measurement of λ in unpatterned low Tc films.
• Systematic mesh optimization in finite-element analysis balancing accuracy and computational time, allowing practical λ extraction in ~10-15 minutes per simulation.

Datasets

• NbN thin films — thickness 6-100 nm — grown in-house by DC magnetron sputtering on MgO substrates
• Nb3Sn thin film — 90 nm thickness — grown in-house on MgO substrate
• Amorphous Re6Zr thin films — 28 nm thickness — pulsed laser deposition on MgO substrates

Baselines vs proposed

• Low-frequency mutual inductance technique: λ(0) = 275 nm for 50 nm NbN film vs proposed COMSOL-based resonator method: λ(0) = 276 nm for 60 nm NbN film
• Fitting temperature dependence of 1/λ² to dirty-limit BCS gap model yields Δ(0) = 2.88 meV consistent with tunneling measurements vs literature
• Flip-film method: λ(0) = 1039 nm for a-Re6Zr film consistent with two-coil mutual inductance and tunneling reported values

Figures from the paper

Figures are reproduced from the source paper for academic discussion. Original copyright: the paper authors. See arXiv:2605.28759.

Fig 1: (a) Schematic of the microstrip resonator device and (b) optical image of a device made from a 60 nm thick

Fig 2: (a) Measured transmission coefficient || as a function of frequency for a microstrip resonator made from

Limitations

• Accuracy depends critically on precise knowledge of film thickness and substrate/spacer dielectric constants; ±4% thickness uncertainty leads to ±8% uncertainty in λ.
• Flip-film method less sensitive for films thicker than λ/4 due to geometric effects dominating kinetic inductance-related frequency shifts.
• Real device environment including box resonances and parasitic modes impose complexity; some resonances (e.g. 3rd harmonic) cannot always be resolved clearly.
• Assumes constant dissipative conductivity σ' in simulation which is valid except very close to Tc; near Tc effects may bias results.
• Computational cost: each COMSOL eigenfrequency simulation requires 10-15 minutes on high-end desktop; finer mesh increases time significantly.
• Q-factor modeling for surface resistance extraction remains numerically challenging and not fully addressed in this work.

Open questions / follow-ons

• Develop improved numerical methods to reliably extract surface resistance Rs from Q-factor simulations of superconducting microstrip resonators.
• Extend measurement technique to broader range of superconductors including unconventional order parameters and strongly disordered films.
• Investigate microwave kinetic inductance contributions from vortex-antivortex pairs near Berezinskii-Kosterlitz-Thouless transitions in ultrathin films.
• Adapt flip-film and direct microstrip resonator methods for in-situ and real-time monitoring of superconducting film properties during growth or device operation.

Why it matters for bot defense

While this paper focuses on microwave characterization of superconducting thin films rather than bot defense or CAPTCHA technologies, the presented methodology offers a rigorous approach to combining high-sensitivity experimental data with precise numerical modeling to extract intrinsic physical parameters. For CAPTCHA and bot-defense practitioners exploring novel sensor technologies or hardware-based verification methods leveraging superconducting devices, this study provides a detailed blueprint for overcoming challenges in absolute parameter calibration amidst complex system geometries. The hybrid experimental-simulation approach exemplifies a pathway to transform sensitive resonant phenomena into quantitatively reliable metrics, a principle potentially transferable to physical-layer bot defense mechanisms involving RF or quantum sensors. However, direct application to CAPTCHA or bot detection is limited; the main value lies in methodology for high-accuracy device characterization under complicated boundary conditions.

Cite

@article{arxiv2605_28759,
  title={ Absolute measurement of penetration depth of superconducting thin films using microwave stripline resonators },
  author={ Arghya Dutta and Ajeet Salunke and Mahesh Poojary and Vivas Bagwe and Sangita Bose and Pratap Raychaudhuri },
  journal={arXiv preprint arXiv:2605.28759},
  year={ 2026 },
  url={https://arxiv.org/abs/2605.28759}
}

Read the full paper

• arXiv:2605.28759 (abstract)
• arXiv:2605.28759 (PDF)