Achieving near-perfect light absorption in atomically thin transition metal dichalcogenides through band nesting

Near-perfect light absorbers (NPLAs) with absorbance A ≥ 99% are important for applications in energy harvesting, sensing, stealth, and secure communications. In simple single-mirror (Salisbury screen) resonators comprising a dielectric spacer and a metal reflector, the absorbance of a monolayer 2D material is far from unity. Two routes commonly pursued to enhance A are: (1) using complex plasmonic nanoparticles or patterned metasurfaces, which can achieve near-unity absorption but require demanding nano-patterning and are limited for large-area applications; and (2) increasing the thickness of 2D layers, e.g., tens of TMD layers, which can yield near-perfect absorption but are not practically controllable to grow and offer limited wavelength tunability. In this work, the authors propose an approach based on preserving band nesting in TMDs to reach NPLAs with only two or three atomic layers. While monolayer TMDs exhibit unusually strong band nesting and high optical conductivity (~1 mS), interlayer electronic coupling in few-layer stacks disrupts nesting and reduces absorption. The central hypothesis is that minimizing interlayer coupling—via twisting two layers or inserting an atomic buffer layer—preserves monolayer-like band nesting in each TMD layer, enabling the optical conductivity to scale nearly linearly with layer count. Combining such decoupled TMD stacks with a Salisbury screen should achieve near-perfect absorption. The study demonstrates this concept theoretically and experimentally and further shows the approach is adaptable across TMD chemistries to cover the visible spectrum.

Prior NPLA efforts using 2D materials have focused on two directions: (i) plasmonic nanoparticles and patterned metasurfaces to enhance absorption to near-unity across wide frequencies, though these require complex, expensive nanofabrication and are constrained to small areas; and (ii) employing thick TMD films (e.g., ~20 layers) with conventional metal reflectors to approach perfect absorption, but controlled growth of such multilayers is not feasible and peak wavelength tunability is limited. The Salisbury screen offers a simple resonator platform, but a monolayer’s absorbance remains below unity without meeting the required optical conductivity. Graphene’s universal optical conductivity (~0.061 mS) in the visible is an order of magnitude too small for NPLA. Monolayer TMDs exhibit strong optical absorption from band nesting with σ′ ~1 mS—much higher than graphene—but still below the 2.17–3.24 mS range required for ideal single-mirror critical coupling, motivating methods to stack layers while preserving monolayer-like nesting.

Theory and modeling:

• Derived absorbance A(ω) for a 2D sheet between dielectrics using the transfer matrix method; for an optimized Salisbury screen, A(ω) = 1 − [1 − σ′(ω)ZVAC]^2 / [1 + σ′(ω)ZVAC]^2, identifying the target σ′ range (2.17–3.24 mS) for A > 99%.
• Performed first-principles DFT calculations (VASP) with plane-wave cutoff 400 eV, PAW pseudopotentials, PBE-GGA XC, and Grimme-D3 vdW corrections. Large vacuum spacing used out-of-plane. Supercells for twisted MoS2 bilayers and MoS2/Gr(hBN)/MoS2 constructed by coincidence lattice method to minimize strain. k-point meshes: 21×21×1 (unit cell), 15×15×1 (MoS2/Gr or hBN/MoS2), 9×9×1 (twisted 2L MoS2).
• Computed frequency-dependent optical conductivity via Kubo-Greenwood formalism using Wannier90; constant broadening η = 100 meV; σ2D(ω) = σ(ω)L, with L the out-of-plane lattice constant.
• Calculated electronic band structures and momentum-resolved band nesting maps to assess interlayer-coupling effects (including supplemental checks with excitonic effects and SOC where noted). Device design and fabrication:
• Salisbury screen cavity: SiO2 dielectric spacer (199 nm) atop a metal reflector stack Ti(3 nm)/Ag(90 nm)/Au(60 nm) on glass; thickness selected to satisfy critical coupling near the MoS2 C-exciton at 2.83 eV. Fabricated via electron-beam evaporation and template stripping to achieve ultraflat surfaces; top Au sacrificial layer removed to expose flat SiO2 over Ag.
• Large-area monolayer MoS2 prepared by Au-assisted exfoliation to yield high-quality, large-area flakes.
• Twisted bilayer MoS2 assembled by dry transfer using a PPC stamp from pre-patterned single-crystal MoS2 on glass; precise twist control achieved (e.g., ~22°, ~32°).
• Buffer-layer heterostructures: MoS2/graphene/MoS2 and MoS2/hBN/MoS2 assembled by sequential transfer; Raman (E2g and A1g) used to assess coupling.
• Additional wafer-scale in situ growth via MBE: WSe2/ZnSe/WSe2 stacks, demonstrating that the decoupling layer need not be 2D/hexagonal. Optical characterization:
• Micro-reflectance/optical contrast spectroscopy with halogen white-light illumination; optical contrast defined as (R − R0)/R0 on transparent substrates proportional to absorbance. Absorbance on cavity evaluated as 1 − R/R0 (R: with heterostructure; R0: cavity substrate only). Raman spectroscopy (532 nm) used for structural assessment.

• Analytical modeling shows a Salisbury screen can reach A → 100% when σ′ ≈ ε0c; target σ′ range 2.17–3.24 mS yields A > 99%.
• Monolayer MoS2 (freestanding) exhibits a strong C-exciton near 2.81 eV with σ′ ≈ 0.96 mS and A ≈ 25.5%.
• Pristine 2L MoS2 (freestanding) suffers degraded band nesting due to interlayer coupling; peak splits, yielding σ′ ≈ 1.47 mS and A ≈ 34% (below target).
• Twisted 2L MoS2 (freestanding) reduces interlayer coupling, restoring a single C-exciton peak and enhancing absorbance: Amax ≈ 39% (21.81°) and 40% (32.22°).
• Buffer-layer stacks preserve monolayer-like nesting and approximately double σ′: MoS2/hBN/MoS2 and MoS2/Gr/MoS2 (freestanding) reach A ≈ 41.4% and 42.0%, exceeding the ~41.2% threshold compatible with A > 99% in a single-mirror cavity.
• Salisbury screen (SiO2 199 nm/Ag) calculations near 2.83 eV: 2L MoS2 Amax ≈ 91.0%; twisted 2L MoS2 Amax ≈ 98.7% (32.22°); MoS2/hBN/MoS2 ≈ 98.8%; MoS2/Gr/MoS2 ≈ 99.0%.
• Experimental absorbance on the cavity: pristine 2L MoS2 A ≈ 78% (slight redshift); twisted 2L MoS2 reaches A ≈ 89.8% at 28° (and ≈86.3% at 26°, ≈84.5% at 23°), peaking near 30° twist; MoS2/Gr/MoS2 achieves A ≈ 94.8% with thickness <2 nm.
• Optical contrast of the C exciton in MoS2/Gr/MoS2 is enhanced by ~116% vs 1L and ~48% vs 2L MoS2; C-exciton blueshift observed with increased twist (2.71 eV → 2.82 eV from 0° to 32°), consistent with reduced coupling.
• Extension to 29 visible-range 2D materials (cavity-optimized): among 1L materials, A up to 82.5% (MoS2), 79.9% (2H-CrTe2), 77.7% (2H-WS2); with 2L + buffer, strong candidates include 2H-CrTe2 (97.9%), 2H-WS2 (96.4%), 2H-ZrTe2 (95.8%), indicating broad spectral coverage with NPLAs.
• Demonstrated all in situ, wafer-scale growth route (WSe2/ZnSe/WSe2 via MBE) confirming the generality of buffer-layer decoupling (buffer need not be 2D/hexagonal).

The study addresses the challenge of achieving near-unity absorption in atomically thin films without complex nanostructuring by identifying and preserving monolayer band nesting in multilayer TMD stacks. Interlayer electronic coupling in conventional bilayers degrades band nesting and reduces optical conductivity; by twisting layers or inserting an atomic buffer, the coupling is minimized, restoring monolayer-like nesting and enabling optical conductivity to scale nearly linearly with layer number. When combined with a simple single-mirror (Salisbury screen) cavity tuned to the C-exciton, these decoupled stacks yield near-perfect absorption in theory and high experimental absorbance approaching 95%. The dependence on twist angle and the consistent blueshift with increasing twist reinforce the coupling-nesting link. The strategy is robust across TMD chemistries and can be implemented via scalable growth (e.g., MBE WSe2/ZnSe/WSe2), suggesting a practical route to wafer-scale, ultra-thin optoelectronic absorbers spanning the visible. This advances atomically thin photonic device design, offering a simple, tunable, and material-diverse platform for applications where ultrathin, high-absorbance layers are desired.

By harnessing band nesting in TMDs and suppressing interlayer coupling via twisting or buffer-layer insertion, the authors realize an ultimate Salisbury screen absorber using only two or three atomic layers. First-principles and transfer-matrix analyses predict A ≈ 99% at the C-exciton with optimized cavities, and experiments validate strong absorption enhancement, achieving up to 94.8% with MoS2/graphene/MoS2 on a simple SiO2/Ag mirror. The approach generalizes to a broad set of 2D materials, enabling near-perfect light absorbers across the visible spectrum and offering a scalable pathway (including all in situ growth) for atomically thin optoelectronics. Potential future directions include further optimization of interlayer decoupling and excitonic linewidths to bridge the gap between theoretical and experimental A, integration with large-area growth and device processing, and exploration of additional material combinations and buffer layers for spectral tailoring and multifunctionality.

The experimental absorbance values (up to ~94.8%) are below the theoretical near-99% predictions, indicating practical limitations such as residual interlayer coupling, cavity loss/mismatch, or imperfect spacer thickness control. Primary conductivity/absorption calculations largely use single-particle approximations (with SOC neglected in most cases) and a constant broadening parameter (η = 100 meV); while excitonic features were examined in supplemental calculations (showing C-exciton redshift and linewidth sensitivity), these approximations may affect quantitative accuracy. Achieving and maintaining precise twist angles near ~30° is experimentally sensitive; deviations reduce absorption. Graphene buffer layers add small intrinsic absorption that slightly modifies spectra. Results can be sensitive to material quality, excitonic lifetimes, and sample-to-sample variations.