Transition metal dichalcogenide metamaterials with atomic precision

Transition metal dichalcogenide (TMD) materials have gained tremendous interest due to the discovery of the direct bandgap in monolayer MoS2 and associated exciton physics, alongside excellent mechanical, electronic, and catalytic properties. Their capability to form van der Waals heterostructures and exhibit Moiré physics enables tailored hybrid systems. Beyond the 2D planes, one-dimensional edges of TMDs possess unique properties distinct from planes/bulk (e.g., metallic/ferromagnetic zigzag edges versus semiconducting/nonmagnetic armchair edges), with implications for catalysis and nonlinear optics. Controlling edges and the edge-to-plane ratio could yield TMD metamaterials with engineered mixed edge–plane–bulk characteristics. Despite various approaches to edge engineering (chemical, surface engineering, thermal annealing, plasma, and other techniques), existing methods often suffer from complexity, exciton damage, limited precision, and harsh conditions. Deterministic, reproducible control of edges and edge-plane ratios in TMDs remains limited. This work presents a simple, scalable, anisotropic wet etching method to achieve nearly atomically sharp, zigzag-terminated edges in TMDs, enabling precise control of nanostructures from bilayers to bulk, and opening routes to study and utilize edge-engineered TMD metamaterials.

Prior efforts to engineer edges in van der Waals materials and monolayers include chemical treatments, surface engineering, thermal annealing, plasma processing, and emerging techniques. Notable anisotropic etching methods for graphene using top-down lithography plus plasma were demonstrated, but their applicability to TMDs required verification. MoS2 etched by XeF2 gas can achieve anisotropy but needs a graphene mask, adding complexity. Many approaches risk damaging excitonic properties, lack precision, or require harsh conditions. Collectively, despite progress, deterministic control over TMD edges and edge-plane ratios has been limited, motivating a simpler, more precise, and scalable method presented here.

Overall approach: A three-step process combines standard top-down nanopatterning with subsequent anisotropic wet etching to produce atomically sharp, zigzag-terminated edges in TMDs. Step (i) Exfoliation and transfer: TMD flakes (e.g., WS2) of selected thickness were mechanically exfoliated from bulk crystals (HQ-graphene or mineral) onto PDMS using scotch tape and then transferred via all-dry viscoelastic stamping onto thermally oxidized SiO2/Si substrates. Thicknesses were measured by profilometry. Step (ii) Nanopatterning: Pre-patterning of circular or arbitrary-shaped holes was done either by EBL + RIE or by FIB.

• EBL: ARP 6200.13 positive resist spin-coated (6000 rpm, 1 min, ~400 nm). Exposure on JEOL JBX 9300FS at 100 kV with alignment marks. Development in n-Amyl acetate (4 min), N2 dry. Dry etching in ICP/RIE using CHF3 (50 sccm) and Ar (40 sccm) at 10 mTorr, 50 W forward power; WS2 etch rate ~10 nm/min (time adjusted to thickness). Post-etch O2 plasma stripping (O2 40 sccm) to remove hardened resist; residual resist removed in acetone at 50 °C (~3 min), rinsed in IPA and DI water.
• FIB: Performed on Tescan GAIA3 FIB-SEM, 40 kV low-energy electron beam, ion beam current tens of pA. Step (iii) Anisotropic wet etching: Immersion of patterned samples in aqueous etchant at ~50 °C. Two chemistries used: H2O2:NH4OH:H2O with 1:1:10 volume ratio (stock solutions 31% H2O2, 25% NH4OH) or H2O2:H2O 1:10. Etch rate tuned by temperature, time, concentration, and composition. Mild heating enables precise control for ultranarrow features; higher concentration/temperature for faster etching of larger structures. Pre-patterned circular holes evolve into hexagonal holes aligned to TMD crystallography with exclusively zigzag edges. Post-etch rinse in DI water, N2 dry, characterization by optical microscopy and SEM. Characterization:
• SEM: Zeiss ULTRA 55 FEG and JEOL 7800F Prime to monitor morphology and edge sharpness.
• TEM/HRTEM and STEM-EELS: Samples transferred to SiN TEM membranes (some windows broken for suspended regions). JEOL Mono NEO ARM 200F at 60 keV with monochromator and aberration correctors. STEM-EELS acquired in dual-EELS mode (probe ~180 pA, few Å size); EDS simultaneously collected. EELS maps generated via background-subtracted integration.
• Optical spectroscopy: Normal-incidence bright-field reflectivity via 20x air objective (NA 0.45) to fiber-coupled spectrometer; normalized to a dielectric-coated silver mirror.
• Electromagnetic simulations: FDTD (Lumerical) using anisotropic WS2 permittivity (in-plane from literature for bulk WS2; out-of-plane fixed at 6.25). Normal incidence plane wave, in-plane periodic boundaries, steep-angle PML in z, mesh accuracy 5. Extracted reflection, transmission, absorption spectra. Experimental parameters/examples: Initial circular holes radius ~1 µm evolve to hexagons within ~8 min at ~50 °C in 1:1:10 H2O2:NH4OH:H2O. Hole sizes from ~50 nm to 10 µm tested. Arrays fabricated over large areas limited by flake size and lithography.

• Anisotropic wet etching of multilayer TMDs (WS2, MoS2, MoSe2) in aqueous H2O2 (with/without NH4OH) at ~50 °C converts pre-patterned circular or arbitrary-shaped holes into hexagons with nearly atomically sharp edges.
• Crystallographic alignment: All hexagonal holes share a fixed orientation synced with the TMD lattice; etched surfaces are perpendicular to the <11-00> direction, indicating exclusively zigzag-terminated edges (supported by SAED and HRTEM).
• Edge quality: HRTEM shows atomically sharp sidewalls with roughness of only a few atomic planes. STEM-EELS at the edge detects strong A-, B-, and C-exciton signals, confirming preserved excitonic properties up to the edge. A ~13.8 eV EELS feature consistent with zigzag edges is observed. Occasional 0–3 nm amorphous residue layers may be present on edges.
• Size and shape convergence: Initial holes from ~50 nm to 10 µm convert to hexagons; small holes rapidly reach perfect hexagons and then etching largely self-limits; larger holes approach hexagons more slowly and can appear incomplete if stopped early. Arbitrary initial shapes (ellipses, triangles, squares, pentagons, trapezoids) all converge to (regular/irregular) hexagons, indicating defect-driven regions can be converted into zigzag edges.
• Layer dependence: Monolayer WS2 survives etching (retains bright A-exciton PL) but circular holes transform into irregular triangles rather than hexagons; adjacent bilayers and thicker multilayers etch into hexagons, implying anisotropy requires multilayers and does not etch along the basal plane. Bilayers remain bilayers throughout etching (no thinning), suggesting distinct chemical behavior between mono- and multi-layers.
• Heterostructures: In stacked WS2/WS2 heterostructures with ~30° rotational misalignment, etched hexagons rotate accordingly, enabling crystallographic orientation mapping between layers.
• Arrays and scalability: Regular arrays (honeycomb, bow-tie, vortex patterns) fabricated over hundreds of microns with edges as sharp as single holes; geometry controlled by array orientation relative to crystal and hole size. True-color optical images of ~70 nm WS2 arrays show structural colors dependent on geometry.
• Optical response: Reflectivity spectra of honeycomb arrays exhibit resonances dispersing linearly with hole size in NIR (<1.9 eV) and an avoided crossing near the WS2 A-exciton (~2.0 eV), suggesting mode–exciton hybridization; FDTD reproduces key spectral features and avoided crossing. A specific array (#7) shows low reflectivity (~15%) across visible (dark-blue color) and high absorption.
• Ultranarrow features: Controlled over-etching of patterned arrays yields nanoribbons and nanojunctions with edge-to-edge distances ~3 nm, approaching WS2 multilayer exciton Bohr radius; features are limited by initial lithography resolution, suggesting even smaller gaps may be achievable with improved masks.
• Process practicality: Method uses abundant, inexpensive chemicals (H2O2, NH4OH), operates at ambient conditions with mild heating, and is broadly applicable to several group-VI TMDs.

The work addresses the challenge of deterministic edge engineering in TMDs by introducing an anisotropic wet etch that self-terminates on crystallographic planes, producing zigzag-terminated, nearly atomically sharp edges. The resulting structures exhibit preserved excitonic features at the edges and well-defined crystallographic orientation, validating that the method enables precise control of edge-plane ratios central to forming TMD metamaterials. The observed monolayer versus multilayer behavior supports a mechanistic hypothesis: differing chemical stabilities of M- and X-terminated zigzag edges lead monolayers to favor a single zigzag type (yielding triangles), while in 2H-stacked multilayers alternating M/X zigzags vertically enable hexagonal symmetry. The ability to pattern arrays with controlled orientation and size allows systematic tuning of photonic resonances and observation of avoided crossing with the A-exciton, demonstrating potential for tailored light–matter interactions. Ultranarrow nanoribbons and nanojunctions with nanometer-scale gaps offer routes to quantum confinement, enhanced catalytic activity, and nonlinear optics, directly leveraging edge-dominated properties. Overall, the method’s scalability and simplicity overcome limitations of previous techniques and open a pathway for comprehensive studies and applications of edge-engineered TMD metamaterials.

A simple, scalable anisotropic wet etching process was demonstrated to fabricate complex hexagonal nanostructures with nearly atomically sharp, exclusively zigzag edges in multilayer TMDs (WS2, MoS2, MoSe2) using aqueous H2O2 (with/without NH4OH) at mild temperatures. The method reliably converts arbitrary pre-patterned holes into crystallographically aligned hexagons from bilayers to bulk, preserves excitonic properties at the edges, enables large-area arrays with tunable optical resonances (including signatures of exciton–photon hybridization), and produces ultranarrow nanoribbons and nanojunctions down to ~3 nm gaps. These advances define a class of TMD metamaterials with engineered edge–plane–bulk characteristics and high surface-to-volume ratios, promising impacts in catalysis, sensing, photodetection, quantum transport, nonlinear optics, nanophotonics, and optomechanics. Future research directions include: elucidating the exact etching mechanism and edge-termination chemistry; extending the approach to other 2D materials (graphene, hBN, MXenes); optimizing lithographic pre-patterns for sub-nanometer features; and comprehensive studies of quantum-confined and edge-driven optical/electronic phenomena in the fabricated structures.

• The exact chemical mechanism of anisotropic etching remains a hypothesis (relative stability of M- vs X-terminated zigzag edges); further studies are needed for confirmation.
• Monolayer TMDs do not directly form hexagonal holes under the reported conditions; they tend to form triangular holes, so the method’s hexagonal precision applies from bilayers upward.
• Feature sizes are currently limited by the resolution and quality of pre-patterning (EBL/RIE or FIB); ultranarrow gaps could be further reduced with improved masks.
• Occasional 0–3 nm amorphous residue layers observed on edges may arise from processing/transfer, potentially affecting certain measurements or applications.
• Demonstrated thickness range is from bilayers up to a few hundred nanometers; behavior for much thicker samples was not tested.
• Arrays’ uniformity over macroscopic scales is ultimately limited by flake size and lithography alignment/quality.