Ultrathin 2 nm gold as impedance-matched absorber for infrared light

The paper addresses the need for efficient, broadband, and thermally lightweight absorbers for IR/THz thermal detectors. Existing solutions (antenna structures, metamaterials, stacks of plasmonic resonators, and graphene-based approaches) either suffer from narrow bandwidths or low intrinsic absorptivity, often requiring additional structures that limit bandwidth. A classical approach to achieve broadband absorption is impedance matching a metal film’s sheet resistance to half the free-space impedance (~188 Ω), theoretically yielding up to 50% wavelength-independent absorptivity under the condition that the metal’s optical constants satisfy n ≈ κ. Conventional metals typically require film thicknesses of tens of nanometers or face percolation thresholds that make impedance matching difficult at ultralow thickness. Recent advances using oxidized copper as a surfactant allow deposition of ultrathin, continuous Au films below the conventional percolation limit. The study investigates whether a 2 nm Au film on oxidized Cu can serve as a stable, broadband, impedance-matched absorber from 2–20 µm with minimal thermal mass.

• Antenna and metamaterial absorbers can reach high absorptivity but are resonant and thus narrowband; stacking different resonant sizes broadens but still limits bandwidth (e.g., 0.8–1.3 THz).
• Graphene offers ultimate thinness but only ~2.3% absorptivity in near- to mid-IR; plasmonic metastructures or chemical doping are needed, reintroducing bandwidth limits and fabrication complexity.
• Thin metallic/alloy films (Bi, Ag, Pt; TiN, NiCr; ITO) have been used for impedance-matched absorbers, but oxidation, long-term stability, and required thickness (often ~10 nm or more) can be problematic for thermal mass and durability.
• Gold is chemically stable but typically requires thicknesses above the percolation threshold, making sheet resistance too low for impedance matching at ultrathin dimensions.
• Surfactant-mediated growth using oxidized Cu enables ultrathin, continuous Ag and Au films far below the usual percolation threshold, opening a route to ultrathin impedance-matched absorbers.

• Theoretical framework: The impedance-matched absorption model assumes n ≈ κ (imaginary permittivity dominates). Using the Drude model rewritten in terms of plasma frequency ωp and resistivity ρ, the condition ω ≪ ε0 ρ ωp defines the regime where ε2 dominates ε1, enabling wavelength-independent absorption. For ultrathin films, increased ρ broadens this regime toward shorter wavelengths.
• Electrical transport model: Film resistivity is modeled via the scattering hypothesis ρ = ρ0 + ρGB + ρSS + ρSR, with grain-boundary (∝ D−1), surface (∝ d−1), and roughness (∝ d−3) contributions; D is grain size (~d for very thin films, saturating near D∞ ~20 nm). Below percolation, transport is dominated by tunneling/ohmic bridges, outside the scope of the model.
• Sample fabrication: 50 nm LPCVD SiNx membranes (2.5 mm × 2.5 mm) were patterned and backside released in KOH. A Cu layer of 1.2(2) nm was sputtered (1.5 Å s−1) onto Ar plasma-cleaned membranes, then oxidized in air for one day to form the surfactant seed (thickness increases due to oxidation). Au was evaporated from a W boat at 0.3 Å s−1 under 3 × 10−8 mbar; thickness monitored by quartz sensor. AFM indicated smooth, continuous morphology above a percolation threshold of ~1.84 nm.
• Electrical characterization: Four-point probe (Jandel head, Keithley 6221/2182A) measured resistivity and sheet resistance over 10−7–10−3 A with up to 2.2 N contact force. Thickness series for seeded and unseeded Au were measured to determine percolation and fit scattering model.
• Optical characterization: FTIR (Bruker Tensor 27, A 510/Q-T setup, 2 mm aperture) measured transmittance and reflectivity in a single setup to reduce systematic errors. Optical constants of bare LPCVD SiNx were extracted via a generalized matrix method with wavelength-wise nonlinear fitting in overlapping spectral blocks (not Kramers–Kronig constrained due to limited spectral range). Au-on-SiNx data were fitted using the Drude model (with measured ρ and fitted ωp) and the extracted SiNx optical constants to compute reflectance, transmittance, and absorptivity (A = 1 − R − T). Stability tests repeated FTIR measurements after storage under ambient conditions for five months.

• Continuous, metallic-like Au films achieved down to ~2 nm on oxidized Cu seed; an offset of 1.84 nm accounts for percolation threshold/effective thickness in resistivity fits.
• Resistivity vs thickness is well described down to 2 nm by the scattering hypothesis, dominated by grain-boundary and surface scattering (∝ d−1). A 13 nm Au film shows ~3× the bulk Au resistivity (bulk ρ ≈ 2.2 × 10−8 Ω·m). Layers <1.5 nm exhibit an anomalous resistivity trend attributed to island growth and additional scattering below percolation.
• Sheet resistance indicates an optimal impedance match to half free-space impedance (~188 Ω) at ~2.5 nm Au thickness.
• FTIR shows near wavelength-independent absorptivity of 40–50% for 2–4 nm Au; peak near 12 µm for sub-2 nm layers is due to SiNx substrate features.
• Best-performing films reach 47(3)% absorptivity from 2 to 20 µm, with slight decrease toward shorter wavelengths; calculated absorptivity agrees with measurements with a ~0.35 nm thickness offset (within quartz sensor uncertainty).
• Extracted plasma frequency remains approximately constant across metallic-like thicknesses and is slightly higher than bulk; Drude fits are valid down to ~2 nm but degrade below percolation (R² decreases).
• Dielectric analysis shows Im[ε] > Re[ε] over the entire 2–20 µm range for 2 nm Au, satisfying the criterion for wavelength-independent absorption; thicker films require longer wavelengths to satisfy Im[ε] dominance.
• Long-term stability: 2 nm Au absorptivity unchanged over five months in ambient; oxidized Cu alone loses conductivity/absorptivity over time, underscoring the role of the Au film.
• Sub-percolation Au (<~1.5–2 nm) exhibits antireflection behavior, consistent with literature.

The study verifies that ultrathin Au films (≈2 nm) grown on an oxidized Cu surfactant achieve impedance matching (sheet resistance ~188 Ω) and consequently near wavelength-independent absorptivity close to the theoretical 50% limit across 2–20 µm. The enhanced resistivity of ultrathin films, governed by surface and grain-boundary scattering, shifts the regime where the imaginary permittivity dominates, allowing broadband absorption to extend to shorter wavelengths than predicted by bulk Au properties. Empirical FTIR results align with Drude-based calculations using measured resistivity and fitted plasma frequency, with small thickness calibration offsets. The absorber combines high, flat spectral response with negligible thermal mass and demonstrated stability, making it highly relevant for thermal IR/THz detectors and other broadband absorption applications. The dielectric analysis (Im[ε] ≫ Re[ε]) clarifies the physical basis of the impedance-match in ultrathin metallic films and suggests material strategies to further extend bandwidth.

The work demonstrates a practical, robust, and ultralight broadband absorber: a 2 nm Au film on oxidized Cu that delivers 47(3)% absorptivity from 2 to 20 µm, approaching the 50% theoretical maximum for impedance-matched thin films. Electrical and optical analyses confirm increased resistivity and Drude behavior consistent with broadband absorption criteria (Im[ε] dominance), and calculated spectra agree with measurements. The films exhibit long-term ambient stability and minimal thermal mass, ideal for thermal IR/THz detectors. Potential future directions include exploring materials with higher plasma frequency and resistivity (e.g., Al) or engineered materials (doped semiconductors) to push the wavelength-independent regime further into shorter wavelengths, and leveraging sub-percolation Au as antireflection coatings. Additional studies could extend direct measurements into the far-IR/THz to verify predicted performance.

• Optical modeling of SiNx and Au films was not Kramers–Kronig constrained due to limited spectral range, potentially introducing uncertainties in extracted optical constants.
• Drude-model fits break down below the percolation threshold (<~2 nm), where transport involves tunneling and island growth; optical/electrical behavior in this regime is not fully captured.
• Far-IR/THz absorptivity is inferred from theory; measurements were limited to 2–20 µm.
• Thickness calibration relies on a quartz crystal sensor; a ~0.35 nm offset indicates possible systematic uncertainty in nominal thickness.
• Results are specific to Au on oxidized Cu seed on 50 nm LPCVD SiNx; generality to other substrates/process conditions requires verification. Bare oxidized Cu is unstable over time, necessitating the Au overlayer for stable performance.