2D Materials Device Fabrication: From Graphene and TMDs to Functional Devices

By NineScrolls Engineering · 2026-04-19 · 35 min read · Process Integration

Target Readers: Researchers, process engineers, and graduate students fabricating electronic, optoelectronic, or sensor devices from two-dimensional materials. This guide assumes familiarity with cleanroom processes and provides actionable recipes, equipment parameters, and troubleshooting guidance for each fabrication step.

Introduction: Why 2D Materials Demand a Different Fabrication Approach

Two-dimensional (2D) materials — graphene, transition metal dichalcogenides (TMDs such as MoS₂, WS₂, WSe₂, MoSe₂), and hexagonal boron nitride (h-BN) — have moved from laboratory curiosities to serious candidates for next-generation electronics, photonics, and sensing. Their atomically thin nature enables extreme electrostatic control, while their diverse electronic properties (semimetallic graphene, semiconducting TMDs, insulating h-BN) provide a complete materials palette for heterostructure device engineering.

However, the same properties that make 2D materials technologically exciting also make them extraordinarily challenging to fabricate into devices. Traditional silicon fabrication assumes bulk, mechanically robust substrates that can tolerate aggressive plasma exposure, high-temperature processing, and standard wet chemistries. None of these assumptions hold for materials that are one to a few atoms thick.

This guide addresses the complete fabrication pipeline — from substrate preparation and material sourcing through patterning, metallization, encapsulation, and device integration — with specific emphasis on the plasma processing steps where equipment selection and process tuning most critically determine device performance.

Key Fabrication Challenges Unique to 2D Materials

1) Substrate Preparation and Surface Engineering

The substrate surface directly underlies the 2D material and becomes the dominant source of charge scattering, charge traps, and surface roughness in the finished device. Substrate preparation is not merely a cleaning step — it is the foundation of device performance.

1.1 SiO₂/Si Substrates (Standard Research Platform)

Thermally grown SiO₂ (typically 90 nm or 285 nm for optical visibility of monolayers) on degenerately doped Si remains the most common platform for 2D materials research. The SiO₂ surface quality determines:

• Charge trap density: Dangling bonds and silanol groups at the SiO₂ surface create charge traps that cause hysteresis in device transfer characteristics. Density typically 10¹¹–10¹² cm⁻² eV⁻¹.
• Surface roughness: RMS roughness > 0.3 nm on SiO₂ creates local strain variations in the 2D material, leading to potential fluctuations and inhomogeneous doping.
• Hydrophilicity: The SiO₂ surface adsorbs water layers that act as charge dopants and scattering centers for the overlying 2D material.

Recommended cleaning protocol:

1. Solvent clean: Acetone → IPA → DI water rinse, 5 min each in ultrasonic bath
2. Piranha clean: 3:1 H₂SO₄:H₂O₂ at 90°C for 10 min (removes organic contamination)
3. O₂ plasma activation: 50–100 W, 200 mTorr O₂, 60–120 s in a plasma cleaner. This step removes residual hydrocarbons and creates a hydrophilic surface that promotes adhesion during mechanical exfoliation or transfer.
4. Dehydration bake: 150°C hotplate for 5 min immediately before transfer to minimize surface water

1.2 h-BN Substrates (High-Performance Platform)

For devices requiring the highest carrier mobility, exfoliated hexagonal boron nitride (h-BN) replaces SiO₂ as the substrate/dielectric. h-BN provides an atomically flat, charge-trap-free surface that eliminates the dominant scattering mechanisms present in oxide substrates.

Performance comparison:

h-BN flakes are typically exfoliated from bulk crystals (NIMS or HQ Graphene) and identified by optical contrast on 285 nm SiO₂. Flakes of 10–30 nm thickness provide optimal dielectric properties while maintaining flatness.

1.3 Alternative Substrates

• Sapphire (Al₂O₃): Preferred for CVD growth of TMDs (epitaxial alignment) and for optical devices (transparent, low autofluorescence). Surface preparation: anneal at 1000°C in air for 2 h to form atomically flat terraces.
• SiC: Native substrate for epitaxial graphene growth. Surface preparation: H₂ etching at 1600°C to form Si-terminated step-terrace structure.
• Polymer/flexible substrates: PEN, PET, or polyimide for flexible electronics. Pre-treatment with brief O₂ plasma (< 30 s, < 50 W) improves wettability without degrading the polymer surface.

2) Material Sourcing: Exfoliation, CVD Growth, and Emerging Methods

The choice between exfoliated and CVD-grown materials represents a fundamental trade-off between crystal quality and scalability. Understanding this trade-off is essential for selecting the right sourcing approach for your application.

2.1 Mechanical Exfoliation (Scotch Tape Method)

Mechanical exfoliation from bulk crystals produces the highest-quality monolayers with minimal defects, making it the gold standard for fundamental research and proof-of-concept devices.

Optimized exfoliation protocol:

1. Apply crystal to blue Nitto tape (SPV-224); fold and peel 5–10 times to thin
2. Press tape onto O₂ plasma-treated SiO₂/Si substrate (freshly activated, within 5 min)
3. Bake at 100°C for 2 min with tape in contact (improves adhesion)
4. Peel tape slowly (< 1 mm/s) at low angle (< 30°)
5. Identify monolayers by optical contrast (green channel for MoS₂ on 285 nm SiO₂) or Raman/PL mapping

Typical yield: 5–20 monolayer flakes per cm² substrate, lateral dimensions 5–50 µm. Sufficient for single-device studies but not for statistical characterization or circuit-level integration.

2.2 Chemical Vapor Deposition (CVD)

CVD growth provides wafer-scale material for applications requiring larger areas, higher throughput, or device arrays. The quality gap between CVD and exfoliated materials has narrowed significantly, with CVD graphene now routinely achieving mobilities > 10,000 cm²/V·s and CVD MoS₂ reaching > 30 cm²/V·s.

Graphene CVD

Standard growth on Cu foil (25 µm thick, electropolished) using CH₄/H₂ at 1000–1050°C. Growth conditions control domain size and coverage:

• Low CH₄ flow (< 1 sccm): Large single-crystal domains (> 1 mm) but incomplete coverage
• Higher CH₄ flow (1–5 sccm): Full coverage but smaller domains and more grain boundaries
• Enclosure growth: Cu foil folded into a pocket; produces centimeter-scale single crystals

TMD CVD (MoS₂ as Example)

Powder vaporization (MoO₃ + S) or metal-organic CVD (Mo(CO)₆ + H₂S) on sapphire or SiO₂:

• Powder CVD: Lower equipment cost but poor uniformity and batch-to-batch variation. Typical growth: 700–800°C, Ar carrier gas, 15–30 min.
• MOCVD: Better uniformity and reproducibility, suitable for wafer-scale growth. Demonstrated 2-inch and 4-inch wafer-scale MoS₂ with > 99% monolayer coverage.
• PECVD-assisted: Lower growth temperatures (300–500°C) enabled by plasma activation. A PECVD system with precise RF power control can deposit TMD films at temperatures compatible with back-end-of-line processing.

2.3 Emerging Growth Methods

• Molecular beam epitaxy (MBE): Ultra-high vacuum growth providing the cleanest interfaces; used for exotic TMDs (PtSe₂, NbSe₂, NiTe₂) and heterostructure superlattices.
• ALD-based growth: Sequential exposure to metal and chalcogen precursors at low temperatures (< 400°C). ALD systems with plasma enhancement enable conformal TMD deposition on 3D structures — potentially transformative for integrating 2D channels into FinFET or gate-all-around architectures.
• Liquid-phase exfoliation: Scalable production of nanosheets for thin-film devices, inks, and composites — not suitable for single-device electronics but increasingly important for printed electronics and energy storage.

3) Material Transfer Techniques

Transfer — moving 2D materials from their growth substrate (Cu foil, sapphire) to the target device substrate — is often the most yield-limiting step in the entire fabrication flow. Residues from transfer polymers, wrinkles from mechanical handling, and tears from surface tension forces are the primary failure modes.

3.1 Wet Transfer (PMMA-Assisted)

The most widely used method for CVD graphene and large-area TMDs:

1. Spin-coat PMMA (495K or 950K, A4) onto 2D material on growth substrate
2. Etch growth substrate: FeCl₃ or (NH₄)₂S₂O₈ for Cu foil; HF or buffered HF for SiO₂
3. Scoop PMMA/2D film with target substrate from liquid surface
4. Dry at 50–80°C for 30 min; dissolve PMMA in warm acetone (50°C, 2 h)
5. Critical final clean: Brief Ar/H₂ anneal (350°C, 2 h, 100/100 sccm) or gentle O₂ plasma (10 W, 5 s) in a plasma cleaner to remove PMMA residues. Residue removal is essential — AFM studies show that even 0.5 nm of PMMA residue reduces graphene mobility by 30–50%.

Limitations: Trapped water between layers; polymer residue contamination; wrinkles from surface tension during drying. These issues motivate the development of dry and semi-dry transfer methods.

3.2 Deterministic Dry Transfer (PDMS Stamp)

Essential for building van der Waals heterostructures where interface cleanliness determines device performance:

1. Prepare PDMS stamp (Sylgard 184, 10:1 base:curing agent, cured at 70°C for 2 h)
2. Cover stamp with polycarbonate (PC) or polypropylene carbonate (PPC) film
3. Use micromanipulator stage with optical microscope to align stamp over target flake
4. Contact at controlled temperature (40–60°C) to pick up flake via van der Waals adhesion
5. Transfer to target location; release by heating above polymer glass transition (PC: 180°C; PPC: 80°C)
6. Remove polymer residue: chloroform wash → Ar/H₂ anneal → gentle plasma clean if needed

Key advantages: No liquid contact; atomically clean interfaces between layers; angular alignment to < 0.1° using rotational stage. This method enables the fabrication of twisted bilayer graphene (magic angle ~1.1°) and other moiré superlattice devices.

3.3 Semi-Dry Transfer (Electrochemical Delamination)

Combines the scalability of wet transfer with reduced contamination:

• PMMA-coated graphene on Cu is placed in NaOH electrolyte as cathode
• Hydrogen bubbles at the Cu–graphene interface delaminate the PMMA/graphene film
• Cu foil can be reused (5–10 times), reducing material cost
• Gentler than chemical etching; produces fewer wrinkles and tears

4) Device Patterning: Low-Damage Plasma Etching

Patterning 2D materials into device geometries (channel isolation, contact windows, mesa definition) is where plasma processing equipment selection has the most direct impact on device performance. The challenge is straightforward but severe: remove unwanted 2D material with spatial precision while leaving zero damage in the active channel — in a material where "damage" means displacing even a few atoms per square nanometer.

4.1 Photolithography for 2D Materials

Standard photolithography adapted for the unique requirements of atomically thin materials:

Resist selection considerations:

• PMMA (e-beam resist): Clean removal by warm acetone; minimal residue. Preferred for high-mobility devices where interface cleanliness is critical.
• Conventional photoresists (AZ, Shipley): Higher throughput but leave more residue. Suitable for devices where contact quality matters more than channel mobility (e.g., sensors).
• HSQ (hydrogen silsesquioxane): Converts to SiO₂ upon e-beam exposure; serves as both resist and passivation layer. Eliminates the stripping step but is permanent.

Critical precautions:

• No ultrasonic agitation during development — van der Waals adhesion is insufficient to prevent delamination
• Spin coating at reduced speed (< 3000 rpm) to avoid shear-induced tearing of large flakes
• Edge bead removal by gentle acetone swab rather than aggressive EBR flooding
• Use a coater/developer system with programmable spin profiles to minimize film stress during resist application

4.2 Graphene Patterning

Graphene's zero bandgap and semimetallic nature make it chemically reactive to O₂ plasma, which is both advantageous (clean etching) and dangerous (uncontrolled edge damage).

Recommended O₂ plasma etch process:

The key advantage of ICP-RIE over conventional RIE for graphene patterning is independent control of plasma density (ICP power) and ion energy (platen bias). This decoupling allows high radical density for rapid, uniform etching while maintaining low ion energy to minimize damage at the graphene–resist interface. For labs processing graphene occasionally, a compact RIE with O₂ capability provides a cost-effective alternative for basic patterning.

4.3 TMD Patterning (MoS₂, WS₂, WSe₂)

Transition metal dichalcogenides present a more complex patterning challenge than graphene because:

• Multiple etch chemistries are viable (SF₆, CF₄, Cl₂, O₂), each with different selectivity and damage profiles
• Etch products (MoF_x, WF_x, SOF_x) can redeposit as non-volatile films on chamber walls and device surfaces
• Edge states created during etching can dominate transport in narrow channels (< 100 nm)

Chemistry comparison for MoS₂:

For research labs developing TMD devices, an ICP-RIE system with multi-gas capability (SF₆, Cl₂, O₂, Ar, CF₄) provides the most flexibility. The independent ICP/bias control allows tuning from gentle radical-dominated etching (high ICP, low bias) for layer thinning to more aggressive ion-assisted etching for deep mesa isolation.

4.4 Atomic Layer Etching (ALE) for 2D Materials

ALE represents the ultimate in precision for 2D material patterning — the ability to remove exactly one atomic layer per cycle with minimal damage to the underlying layer. This is particularly valuable for:

• Multilayer → monolayer thinning: Converting a 3–5 layer MoS₂ flake to a monolayer at a precise location (e.g., the channel region of a transistor while keeping thicker material at the contacts)
• h-BN tunnel barrier thickness control: Thinning the tunnel barrier in graphene/h-BN/graphene vertical tunnel junctions to exactly 2–4 layers
• Edge profile engineering: Creating smooth, well-defined edges that minimize edge-state scattering

ALE cycle for TMDs (modification/removal approach):

1. Modification step: Brief, low-energy O₂ plasma exposure (< 20 eV ion energy) oxidizes the topmost layer to MoO₃ (for MoS₂) or WO₃ (for WS₂). Only the top layer is oxidized because the interlayer van der Waals gap blocks radical penetration.
2. Removal step: The oxide layer is volatilized by gentle heating (> 700°C for MoO₃) or removed by wet chemistry (DI water rinse — MoO₃ is water-soluble).
3. Repeat: Each cycle removes exactly one MoS₂ layer (~0.65 nm). Endpoint monitoring by in-situ Raman or PL confirms layer count after each cycle.

This approach requires an ICP-RIE system with precise low-energy ion control — the ability to maintain ion energies below the lattice displacement threshold (~20 eV for most TMDs) while still generating sufficient radical density. Equipment with pulsed plasma capability further improves control by allowing radical generation during the pulse-on phase while ions are quenched during the pulse-off phase. See our detailed ALE guide for process fundamentals.

4.5 h-BN Etching

Hexagonal boron nitride is more chemically resistant than graphene or TMDs, requiring more aggressive etch conditions:

• SF₆/Ar plasma: Primary chemistry. Etch rate: 5–15 nm/min at moderate ICP power (300–500 W, 20–50 W bias). Selectivity to SiO₂: ~1:1, requiring careful mask design.
• CHF₃/O₂: Alternative for better sidewall smoothness. The fluorocarbon polymer deposition provides some passivation of etched surfaces.
• Ar ion milling: For situations requiring purely physical etching (no chemical selectivity needed). An IBE/RIBE system provides precise angle-of-incidence control for taper engineering.

5) Contact Metallization: The Critical Interface

Contact resistance is the single largest performance limiter in most 2D material devices — often dominating channel resistance even in micrometer-scale devices. The contact–2D material interface must be designed at the atomic level, and the deposition process must preserve this interface.

5.1 Contact Geometries

Top contacts (metal deposited on top of 2D material):

• Simplest to fabricate; most widely used
• Contact resistance limited by van der Waals gap between metal and 2D material (typically 3–4 Å)
• Typical contact resistance: 1–10 kΩ·µm for MoS₂; 0.5–5 kΩ·µm for graphene

Edge contacts (metal contacts the 2D material at its exposed edge):

• Metal bonds covalently to dangling bonds at the 2D material edge — eliminates the van der Waals gap
• Contact resistance can be < 100 Ω·µm (approaching theoretical limits)
• Requires encapsulation of the 2D material in h-BN, then selective etching to expose the edge
• More complex fabrication but dramatically better performance

5.2 Metal Selection

5.3 Deposition Process

The metal deposition method critically affects contact quality:

E-beam evaporation: Preferred for most 2D material contacts. Minimal substrate heating, directional deposition for clean liftoff, and low kinetic energy of evaporated atoms (< 0.1 eV) avoids sputtering damage to the 2D surface.

Sputtering: Energetic ions (5–30 eV) from sputter deposition can damage the topmost layer of 2D materials. However, for some metals (Bi, In) that are difficult to evaporate, sputtering with reduced power (DC, < 50 W) and increased target-substrate distance provides acceptable results. RF sputtering enables deposition of high-melting-point metals (W, Mo) that serve as diffusion barriers.

Pre-metallization surface preparation:

• Immediately before metal deposition, clean the contact window with a brief Ar plasma (10 W, 10 s, 50 mTorr) to remove resist residues and native adsorbates
• For TMDs: avoid O₂ plasma pre-clean, which oxidizes the contact surface and increases barrier height
• For graphene: a 5 s O₂ plasma "activation" at very low power (< 10 W) can improve wettability without introducing significant defects
• Transfer to evaporator/sputter within 5 minutes of pre-clean to prevent re-adsorption

5.4 Edge Contact Fabrication (High-Performance Devices)

The process flow for edge contacts to h-BN-encapsulated 2D materials:

1. Assemble h-BN/2D material/h-BN stack by deterministic dry transfer
2. Pattern contact windows by e-beam lithography (PMMA resist)
3. Etch through top h-BN and into 2D material edge using SF₆/Ar ICP-RIE (300 W ICP, 30 W bias, 10 mTorr)
4. Without breaking vacuum (or with minimal air exposure), deposit contact metals by e-beam evaporation
5. Liftoff in warm acetone

The critical requirement is minimizing the time between edge exposure (etching) and metallization. Exposed 2D material edges oxidize within seconds in ambient air, degrading contact resistance. Integrated etch-deposition systems or rapid transfer (< 2 min air exposure) are essential.

6) Dielectric Encapsulation and Gate Stack Formation

Depositing gate dielectrics and passivation layers on 2D materials is inherently challenging because the atomically smooth, chemically inert basal plane surfaces resist conventional dielectric nucleation. This section covers strategies for achieving uniform, pinhole-free dielectric films without damaging the underlying 2D material.

6.1 The Nucleation Problem

On bulk silicon, ALD precursors (trimethylaluminum for Al₂O₃, TDMA-Hf for HfO₂) chemisorb on abundant surface –OH groups. The basal planes of graphene, MoS₂, and h-BN lack these reactive sites — precursors physisorb weakly and desorb during purge steps, resulting in island growth, pinholes, and thickness non-uniformity.

Solutions (in order of increasing complexity):

1. Seed layer approach: Deposit a thin (1–2 nm) metal oxide seed layer by e-beam evaporation (Al₂O₃ or Y₂O₃) before ALD. This provides nucleation sites but introduces an uncontrolled interface.
2. Functionalization: Brief O₃ or remote O₂ plasma treatment (< 5 s) creates a controlled density of oxygen-containing defect sites that serve as ALD nucleation points. Must be carefully calibrated — too much treatment damages the 2D material.
3. Low-temperature ALD: Reducing deposition temperature to 80–120°C on an ALD system increases physisorption residence time, improving nucleation on pristine surfaces. Trade-off: lower film quality (higher impurity content, lower density).
4. h-BN capping: Transfer a thin (2–5 nm) h-BN flake as the dielectric layer. Provides an atomically clean, perfectly uniform van der Waals interface — the highest-quality gate dielectric for 2D material FETs, though limited to exfoliated-flake dimensions.

6.2 ALD Gate Dielectrics

Atomic layer deposition is the primary method for depositing gate dielectrics on 2D material FETs due to its thickness control (Å-level), conformality, and relatively gentle processing conditions.

Common ALD gate dielectric processes:

Plasma-enhanced ALD (PEALD) consideration: PEALD uses O₂ or N₂ plasma as the oxidant instead of H₂O, producing denser films at lower temperatures. However, the plasma step can damage the 2D channel underneath. PEALD is acceptable for top-gate stacks where a seed layer or h-BN buffer separates the plasma from the active channel, but should be avoided for direct deposition on exposed 2D material surfaces.

6.3 PECVD Passivation

For post-device passivation, PECVD SiO₂ or SiNₓ provides robust environmental protection:

• Deposit at lowest possible temperature (100–200°C) and RF power (< 50 W) to minimize ion bombardment damage
• SiNₓ provides better moisture barrier than SiO₂ (important for ambient-sensitive TMDs like WSe₂)
• Typical passivation thickness: 20–50 nm — thick enough for environmental protection, thin enough to avoid stress-induced delamination
• For publication-spotlight-quality devices, sandwich the 2D channel between exfoliated h-BN layers and use PECVD only as an outer environmental seal

7) Van der Waals Heterostructure Assembly

The ability to stack different 2D materials into vertical heterostructures — with atomically clean interfaces and controlled rotational alignment — is the defining capability of 2D materials technology. These heterostructures enable device architectures impossible with conventional epitaxial growth.

7.1 Common Heterostructure Architectures

7.2 Assembly Process Flow

Building a complete heterostructure requires iterating the pick-up/stamp-down sequence described in Section 3.2, with additional considerations:

1. Top h-BN first: The "pick-up" technique starts by contacting the top h-BN flake, which is then used to sequentially pick up underlying layers through van der Waals adhesion. This ensures the most critical interface (h-BN/channel) is never exposed to polymer or ambient contamination.
2. Temperature control: Heating the stamp to 40–60°C during pick-up and cooling to room temperature between steps prevents premature release.
3. Bubble-free stacking: Trapped air or hydrocarbons between layers create "bubbles" that locally separate the layers. Slow contact (< 1 µm/s approach speed) and annealing at 200–350°C in forming gas (Ar/H₂) can eliminate most bubbles.
4. Final release: Heating the assembled stack above the polymer glass transition releases the stamp. Residual polymer is removed by chloroform dissolution followed by Ar/H₂ anneal.

7.3 Post-Assembly Processing

After heterostructure assembly, the stack must be patterned for contacts and device isolation. The critical challenge is etching through multiple layers with different etch chemistries without damaging the active channel buried within the stack.

Typical etch sequence for h-BN/graphene/h-BN stack:

1. Pattern contact windows by e-beam lithography
2. Etch through top h-BN + graphene: SF₆/Ar ICP-RIE (expose graphene edge)
3. Immediate metallization (Cr/Au or Pd/Au) for edge contacts
4. Liftoff in warm acetone
5. Mesa isolation etch: pattern second e-beam lithography, etch through entire stack with SF₆/Ar (higher power, longer time). The ICP etcher provides the selectivity control needed to stop precisely at the bottom h-BN or SiO₂ substrate.

8) Complete Device Fabrication Flows

This section provides end-to-end fabrication flows for three representative 2D material devices, with specific equipment and process parameters at each step.

8.1 Back-Gated MoS₂ FET (Research Standard)

The simplest 2D material transistor — a baseline for material characterization and process development.

1. Substrate: p++ Si with 285 nm thermal SiO₂ (back gate + optical identification)
2. Substrate preparation: Piranha clean → O₂ plasma activation (80 W, 60 s, plasma cleaner) → dehydration bake 150°C
3. Material: Mechanically exfoliate MoS₂ from bulk crystal onto substrate; identify monolayers by optical contrast and confirm by PL
4. Channel patterning: PMMA e-beam resist → e-beam lithography for mesa definition → develop → O₂/Ar plasma etch (30 W, 10 s, RIE) → acetone strip
5. Contact patterning: PMMA bilayer (495K/950K) → e-beam lithography for source/drain → develop → brief Ar plasma pre-clean (10 W, 5 s)
6. Metallization: E-beam evaporate Ti/Au (5/50 nm) at < 10⁻⁶ Torr → liftoff in warm acetone 4+ hours
7. Passivation (optional): 20 nm Al₂O₃ by ALD (TMA/H₂O, 150°C, with 1 nm Al₂O₃ seed layer)
8. Measurement: Probe station under vacuum or N₂ ambient to minimize hysteresis

Expected performance: On/off ratio > 10⁶; field-effect mobility 10–40 cm²/V·s; subthreshold swing 2–10 V/decade (limited by thick back-gate oxide).

8.2 h-BN-Encapsulated Graphene Hall Bar (High-Mobility Platform)

The gold standard for graphene transport studies — maximizes mobility by eliminating substrate-induced scattering.

1. Prepare h-BN flakes: Exfoliate h-BN onto SiO₂/Si; identify top and bottom h-BN flakes (10–30 nm thick each) and graphene monolayer by optical contrast
2. Dry-transfer assembly: Pick up top h-BN → pick up graphene → release onto bottom h-BN using PC/PDMS stamp at 180°C
3. Anneal: 350°C in Ar/H₂ (100/100 sccm) for 3 h to remove polymer residue and collapse bubbles
4. Hall bar patterning: PMMA e-beam resist → e-beam lithography → SF₆/Ar ICP-RIE etch through entire h-BN/graphene/h-BN stack (500 W ICP, 30 W bias, 15 mTorr, 60–90 s depending on h-BN thickness)
5. Edge contact patterning: Second e-beam lithography for contacts → SF₆/Ar etch to expose graphene edges → immediate Cr/Au (1/80 nm) e-beam evaporation → liftoff
6. Top gate (optional): Third e-beam lithography → Cr/Au (5/50 nm) gate → liftoff

Expected performance: Room-temperature mobility > 50,000 cm²/V·s; mean free path > 1 µm at low temperature; Dirac point within ±5 V of zero gate voltage.

8.3 MoS₂/WSe₂ Vertical Photodetector

A van der Waals p-n junction photodetector combining n-type MoS₂ and p-type WSe₂ for broadband light detection.

1. Bottom electrode: Deposit Ti/Au (5/30 nm) on SiO₂/Si by sputtering or e-beam evaporation → pattern by liftoff
2. WSe₂ transfer: Dry-transfer exfoliated WSe₂ (few-layer, 3–5 nm) onto bottom electrode, overlapping electrode edge
3. MoS₂ transfer: Dry-transfer MoS₂ monolayer onto WSe₂ with partial overlap
4. Top contact: E-beam lithography on MoS₂ non-overlap region → Ti/Au (5/50 nm) → liftoff
5. Passivation: 30 nm Al₂O₃ by ALD over entire device to prevent ambient degradation
6. Opening contact pads: Lithography → CF₄/O₂ RIE (50 W, 20 s) to etch through Al₂O₃ at probe pad locations → strip resist

Expected performance: Photoresponsivity > 10 A/W; detectivity > 10¹¹ Jones; response time < 10 ms; photovoltaic mode operation (self-powered).

9) Characterization Integration and Process Monitoring

Inline characterization is even more critical for 2D materials than for conventional semiconductors, because the impact of each processing step on device performance is immediate and severe. Unlike bulk materials where damage can be annealed or buried under subsequent layers, damage to a monolayer is permanent and device-killing.

9.1 Key Characterization Techniques

9.2 Process Monitoring Workflow

A recommended monitoring checkpoint strategy for 2D material device fabrication:

1. Post-exfoliation/growth: Optical microscopy + Raman + PL → confirm material identity, layer number, and crystal quality before investing in further processing
2. Post-transfer: AFM (check for wrinkles, bubbles, residue) + Raman (check for damage from transfer process) → verify no degradation during transfer
3. Post-etch: Raman at channel edges + PL mapping → quantify etch-induced damage. If D/G ratio increased significantly, reduce ion energy and re-optimize etch recipe on test samples before processing real devices
4. Post-metallization: Optical inspection + profilometry → verify liftoff quality. Test contact resistance on TLM structures before proceeding to gate stack
5. Post-encapsulation: C-V measurement → check for interface trap density and flat-band voltage. PL through dielectric → verify no damage from dielectric deposition

10) Equipment Selection Guide for 2D Materials Labs

Setting up or upgrading a 2D materials device fabrication lab requires careful equipment selection that balances versatility, precision, and budget. Below we outline the recommended equipment configuration for three lab tiers.

10.1 Essential Equipment (Tier 1 — Startup Lab)

10.2 Advanced Equipment (Tier 2 — Research Group)

10.3 Specialized Equipment (Tier 3 — Advanced Lab)

11) Troubleshooting Guide

12) Future Outlook: Scaling 2D Materials to Production

The transition from laboratory demonstrations to manufacturable 2D material devices requires solving several interconnected challenges:

12.1 Wafer-Scale Growth and Transfer

• Epitaxial growth on wafer substrates: MOCVD of MoS₂ and WS₂ on 2-inch to 8-inch sapphire wafers has been demonstrated with > 99% monolayer coverage. The remaining challenge is grain boundary density and its impact on device-to-device variability.
• Transfer-free integration: Direct growth on device substrates eliminates the transfer step entirely. Low-temperature PECVD and ALD-based growth methods are key enablers, making compatible plasma deposition equipment increasingly important.
• Roll-to-roll transfer: For graphene, continuous roll-to-roll transfer from Cu foil to flexible substrates has reached pilot scale (> 30 cm width). Similar approaches for TMDs are under development.

12.2 Integration with Existing Fab Infrastructure

• BEOL-compatible temperatures: All 2D material processing must occur below ~400°C to avoid damaging underlying CMOS. This constrains growth, annealing, and dielectric deposition options.
• Standard etch platforms: The same ICP-RIE and RIE platforms used for conventional semiconductor etching can process 2D materials with recipe modifications — no new capital equipment categories required, only process development.
• Contamination management: 2D materials introduce new elements (Mo, W, Se, Te) into the fab. Dedicated chambers or thorough between-run cleaning prevents cross-contamination.

12.3 Device Architecture Evolution

• Gate-all-around (GAA) 2D FETs: The atomically thin channel of TMDs is ideally suited to GAA architectures, providing superior electrostatic control compared to Si nanosheets at the same gate length.
• 3D heterogeneous integration: Stacking 2D material layers vertically (with interlayer dielectrics) could provide the ultimate compute density improvement, since each "floor" of the 3D stack adds negligible thickness (< 5 nm per active layer).
• Neuromorphic and analog computing: The tunable memristive behavior of TMD-based junctions and the gate-tunable Schottky barriers of 2D contacts enable analog synapse devices for brain-inspired computing.

Conclusion

Two-dimensional materials device fabrication demands a fundamental rethinking of semiconductor processing assumptions. Every step — from substrate cleaning to final passivation — must be re-optimized for materials where the active layer is one atom thick and every surface interaction matters.

The good news is that the essential processing steps (plasma etching, thin-film deposition, surface cleaning) use the same equipment platforms as conventional semiconductor fabrication, requiring recipe optimization rather than entirely new tool categories. An ICP-RIE with proper low-energy capability, an ALD system with low-temperature operation, and a plasma cleaner with gentle process modes form the core of a 2D materials processing capability.

As the field moves from single-device demonstrations toward circuit-level integration and manufacturing, the precision and reproducibility of plasma processing equipment will become the key differentiator between labs that publish proof-of-concept papers and those that demonstrate wafer-scale, manufacturing-compatible 2D material device technology.

References and Further Reading

• Liu, Y., et al. "Promises and prospects of two-dimensional transistors." Nature 591, 43–53 (2021).
• Wang, L., et al. "One-dimensional electrical contact to a two-dimensional material." Science 342, 614–617 (2013). (Edge contact breakthrough)
• Shen, P.-C., et al. "Ultralow contact resistance between semimetal and monolayer semiconductors." Nature 593, 211–217 (2021). (Bismuth contact breakthrough)
• Cao, Y., et al. "Unconventional superconductivity in magic-angle graphene superlattices." Nature 556, 43–50 (2018).
• Chhowalla, M., et al. "Two-dimensional semiconductors for transistors." Nature Reviews Materials 1, 16052 (2016).
• Akinwande, D., et al. "Graphene and two-dimensional materials for silicon technology." Nature 573, 507–518 (2019).
• Illarionov, Y. Y., et al. "Insulators for 2D nanoelectronics: the gap to bridge." Nature Communications 11, 3385 (2020). (ALD nucleation challenge)
• NineScrolls. "The Complete Guide to Reactive Ion Etching (RIE)"
• NineScrolls. "Atomic Layer Etching (ALE) Precision Guide"
• NineScrolls. "Etching Beyond Silicon: Emerging Materials"
• NineScrolls. "ALD Thin Film Deposition Guide"