Skip to content

Extending the Optical Absorption in a Lumped Element Meander Structure to Far-Infrared Wavelengths

FieldValue
JournalIEEE Transactions on Applied Superconductivity
Published2026-04-22
AuthorsShekhar Chandra Pandey, Shilpam Sharma, Anudeep Singh, Utkarsh Pandey, S. S. Prabhu
DOI10.1109/tasc.2026.3686561

Optimized lumped-element kinetic inductance detector designs reached 95% simulated absorption in the 12 to 50 μm range. Measurements on a fabricated 30-pixel array found up to 75% absorption at selected wavelengths from 14 to 26 μm. The 95% result came from a diamond-supported model, whereas the experiment used a 500 μm SiO2/Si substrate.

An LEKID uses a meander inductor and interdigitated capacitor to set its resonance. Photons above the superconducting energy gap break Cooper pairs. The resulting quasiparticles change kinetic inductance, which appears as a shift in resonance frequency, amplitude, or phase. Absorption is only one part of detection efficiency because the incident field must also couple into the resonator.

The authors matched the meander’s effective sheet impedance to the 377 Ω free-space impedance. Sheet resistance, substrate dielectric constant, line width, line spacing, and filling factor set that match. Their model assumes spacing much less than line width and line width below the photon wavelength. They also kept the meander pitch below one third of the wavelength.

CST Microwave Studio modeled the radiation-sensitive meander across 12 to 50 μm. The side walls used perfect electric conductor boundaries. The upper and lower walls used perfect magnetic conductor boundaries, while ports at both ends used open-space conditions. The authors compared bare substrates with meanders on substrates and with back shorts.

A back short reflects radiation from behind the substrate toward the meander. Its distance was set to odd multiples of one quarter of the wavelength in the dielectric. The dielectric wavelength was λ divided by the square root of relative permittivity. This spacing makes the reflected and incident fields add constructively. Substrate material and thickness also mattered because absorption within the substrate reduces detector performance.

SiO2 showed minimal absorption beyond 25 μm, leading the authors to select a 2.2 μm substrate. With that thickness fixed, they swept sheet resistance from 10 to 100 Ω/sq. Resistance above 50 Ω/sq allowed absorption above 50%, but the response declined above 90 Ω/sq. The geometry study therefore used 60 Ω/sq.

A 0.5 μm line width and 1.5 μm separation produced more than 45% absorption beyond 25 μm. Adjusting width and spacing raised the simulated value to 70 to 80% over a narrow band. These results identify sheet resistance and filling factor as separate controls on impedance matching.

Diamond offered negligible absorption across 12 to 50 μm. The authors also cite a Mohs hardness of at least 10, support for free-standing membranes a few micrometers thick, and established CVD growth on silicon. The best diamond model used a 4 μm meander pitch, with 0.5 μm lines and 3.5 μm spacing. Its substrate and back-short separation was 2.6 μm. The resonance plot was described as near 100% absorption, while the paper reports up to 95% for narrow bandwidths.

Varying the substrate thickness and back-short distance from 1 to 10 μm showed that wavelength-specific tuning can yield at least 95% across 12 to 50 μm. A separation of one quarter of the dielectric wavelength gave at least 50% over a wider band. At 25 μm, that quarter-wavelength separation was 2.6 μm. Higher odd multiples narrowed the useful bandwidth.

The meander was polarization sensitive. Power concentrated in the horizontal lines aligned with the incident electric field, while the vertical sections contributed little.

The layout was modeled in Python with Phidl and checked in KLayout 0.28.7. Fabrication used electron beam lithography followed by reactive ion etching. The absorber was a 20 nm Ti40V60 film on a 500 μm SiO2/Si substrate, arranged as 30 pixels.

The IR-FEL at RRCAT supplied radiation from 14 to 26 μm. The authors measured transmission and reflection simultaneously with WiredSense SPY pyroelectric detectors. A house-built Michelson interferometer provided the wavelength. From these measurements, the array reached up to 75% absorption at selected wavelengths.

The measured curve also contained negative absorption at some wavelengths. The authors suggest surface plasmonic resonance as a possible cause. They link this effect to collective electron oscillations in metallic surfaces and periodic subwavelength apertures.

ConfigurationGeometry or conditionAbsorption resultResult type
Diamond model4 μm pitch, 2.6 μm substrate and back shortUp to 95% in narrow bandsSimulation
Diamond modelQuarter-wavelength back-short spacingAt least 50% over a wider bandSimulation
SiO2 model2.2 μm substrate, 60 Ω/sq, 0.5 μm width and 1.5 μm spacingMore than 45% beyond 25 μm; 70 to 80% after geometry adjustmentSimulation
30-pixel array20 nm Ti40V60 on 500 μm SiO2/Si, measured from 14 to 26 μmUp to 75% at selected wavelengthsMeasurement

Comparison with prior detectors and limitations

Section titled “Comparison with prior detectors and limitations”

Reference detectors show why this band needs superconducting absorbers. InSb detectors discussed in the paper are sensitive only to about 5.5 μm. HgCdTe detectors are limited to 17 to 20 μm. The paper also cites a 2024 kinetic inductance detector with single-photon sensitivity at 25 μm.

This work extends optical coupling studies to 12 to 50 μm and measures absorption in a fabricated array. It does not demonstrate single-photon sensitivity. The experiment does not report cooling conditions, noise-equivalent power, or detection efficiency. The highest diamond absorption remains a simulation, while the measured result comes from a different substrate and device.

The negative absorption feature is reported but not resolved. The paper gives no uncertainty band for the measured absorption curve. The results therefore describe optical coupling rather than complete detector performance.

Back-short separation controls the bandwidth trade-off. A quarter-wavelength separation favors broadband absorption, while higher odd multiples favor narrower resonances. Incident polarization must align with the horizontal meander lines to place power in the absorber. Diamond can provide a thin, low-loss substrate that can be made into a free-standing membrane. The authors note synthesis and mechanical handling challenges for the thinnest SiO2 layers.

Process control must therefore cover film sheet resistance, line width, line spacing, substrate thickness, and back-short distance. Prior Ti40V60 work cited by the authors showed that deposition pressure and sputtering current tune sheet resistance and transition temperature.

The 95% diamond result and the 75% measured result should not be treated as one prototype. They represent different substrates, thicknesses, and design stages.

The simulated LEKID needs a 2.6 μm diamond substrate with a metal back short. 6CCVD’s diamond materials catalog lists single-crystal and polycrystalline diamond grown by MPCVD from 0.1 to 500 μm thick, plus metallization with Au, Pt, Pd, Ti, W, or Cu. Because the paper used 20 nm Ti40V60, the alloy film would need separate qualification through the linked engineering support page.

View Original Abstract

Superconducting radiation detectors typically exhibit detection and single-photon sensitivity limited to the mid-infrared wavelength range. Extending their detection capabilities into the far-infrared range (>10μm) requires careful selection of substrate materials and detector geometries. The overall detection efficiency of such detectors is inherently linked to their absorption and coupling efficiencies. In this study, the resonator geometry and its absorption efficiency were estimated using electromagnetic simulations in CST Microwave Studio simulation software for a lumped-element meander structure. Simulations were performed over the 12-50 μm wavelength range, which corresponds to the Infrared Free-Electron Laser (IR-FEL) at RRCAT, Indore. The absorption in the meander inductor was significantly influenced by the substrate material and thickness, as well as the impedance matching between the detector structure and the incident photon medium. The results indicate that SiO2and diamond substrates are well-suited for developing lumped element kinetic inductance detectors (LEKID) in the above wavelength range. Optimised meander geometries on diamond substrates demonstrated absorption efficiencies of up to 95% for narrow bandwidths and greater than 50% for wide bandwidths. Based on the simulations, a 30-pixel LEKID structure was fabricated by electron-beam lithography on a 500μm SiO2-coated Si substrate. The resonator was made from a 20 nm-thick Ti40V60alloy thin film. Optical characterization was also performed through simultaneous transmission and reflection measurements of the 30-pixel LEKID. Simulation studies demonstrate that the designed LEKID structure is well-suited for detecting far-infrared wavelengths above 10 μm, offering high absorption efficiency.

  1. **** - A review of the lumped element kinetic inductance detector