First Plasma Atomic Layer Etching of Diamond via O$_2$/Kr Chemistry

This paper reports the first demonstration of plasma atomic layer etching (ALE) for diamond using a cyclic O$_2$/Kr chemistry process that achieves atomic-scale precision with a 6.85 Å etch depth per cycle while preserving the diamond bonding structure and reducing surface roughness.

The Gist

Imagine you have a diamond, the hardest substance on Earth. It's so tough that trying to carve it with standard tools is like trying to shave a rock with a butter knife. Usually, when scientists try to cut or shape diamond using plasma (a super-hot, electrically charged gas), they have to use such high energy that they end up smashing the surface, leaving it rough and damaged. It's like using a sledgehammer to fix a watch; you get the job done, but you break the delicate parts inside.

This paper reports a breakthrough: the first time scientists successfully used a technique called Atomic Layer Etching (ALE) to carve diamond with the precision of a surgeon's scalpel, without causing any damage.

Here is how they did it, explained through a simple analogy:

The "Sandpaper and Paint" Analogy

Think of the diamond surface as a very hard, smooth floor. The scientists wanted to remove just a tiny, invisible layer of it—thinner than a single strand of hair.

The Problem:
If you try to sand the floor directly, you have to push hard. If you push too hard, you gouge the floor. If you push too soft, nothing happens.

The Solution (The ALE Process):
Instead of just sanding, they used a two-step "sandwich" process that repeats over and over.

1. Step 1: The "Paint" (Oxygen Modification)
   First, they expose the diamond to an Oxygen plasma. Imagine this as spraying a special, invisible "paint" or "softener" onto the very top layer of the diamond. This paint doesn't just sit there; it chemically bonds to the surface, making that top layer of carbon atoms weak and "loose." It's like putting a layer of soft clay on top of a hard rock. The rock underneath is still hard, but the clay on top is easy to move.

2. Step 2: The "Gentle Sweep" (Krypton Removal)
   Next, they turn off the oxygen and blast the surface with Krypton ions (heavy, inert gas atoms). Think of this as a gentle breeze or a soft broom. Because the top layer was "softened" by the oxygen paint, this gentle breeze is strong enough to blow only that top layer away. However, because the breeze is too weak to move the hard rock underneath, the diamond below remains perfectly untouched.

3. The Cycle
   They repeat this "Paint, then Sweep" cycle thousands of times. With every cycle, they remove exactly one atomic layer.

Why is this a Big Deal?

• Precision: They managed to remove exactly 6.85 Angstroms per cycle. To put that in perspective, an Angstrom is one-tenth of a billionth of a meter. They are removing layers thinner than a single atom's width, one by one.
• No Damage: Usually, carving diamond leaves it looking like a craggy mountain range. With this method, the surface actually became smoother than it was before! It's like polishing a gem while you cut it.
• No "Graphite" Scars: When you damage diamond, it turns into graphite (the stuff in pencils). The scientists checked the chemical makeup and found the diamond stayed pure diamond, with no pencil-like scars.

The "Sweet Spot"

The researchers found a very narrow "Goldilocks zone" for the energy of the Krypton ions.

• If the ions are too weak, they can't blow away the softened layer.
• If they are too strong, they smash through the soft layer and start chipping the hard diamond underneath.
• They found the perfect middle ground (a window of just 1.5 electron-volts) where the process works perfectly.

Why Do We Care?

Diamond isn't just for jewelry. It's a super-material for the future of technology:

• Super-fast computers: It handles heat better than anything else.
• Quantum computers: It can hold "quantum bits" (qubits) that are the basis of future computing.
• Sensors: It can detect magnetic fields with incredible sensitivity.

To build these devices, engineers need to carve diamond into tiny, perfect shapes. If they damage the surface, the device stops working. This new "Oxygen-Krypton" method is like giving engineers a pair of tweezers instead of a sledgehammer, opening the door to building the next generation of ultra-powerful, ultra-fast, and quantum-powered technology.

In short: They figured out how to gently peel off diamond, one atom at a time, without scratching the surface. It's the difference between using a jackhammer and a scalpel.

Technical Summary

1. Problem Statement

Diamond is a premier material for high-power electronics, photonics, quantum sensing, and quantum computing due to its ultra-wide bandgap, high carrier mobility, high breakdown field, and superior thermal conductivity. However, the fabrication of diamond devices is severely hindered by the difficulty of achieving precise, low-damage etching.

• The Challenge: The strong $sp^3$ carbon-carbon bonds make diamond highly resistant to conventional plasma etching.
• Current Limitations: Existing techniques like Ion Beam Etching (IBE) and Reactive Ion Etching (RIE) require high ion energies to remove material. This inevitably causes:
  • Surface roughness.
  • Subsurface structural damage.
  • Partial conversion of the diamond surface into graphitic phases (creating "photonic dead layers").
  • Degradation of device performance (e.g., reduced field-effect mobility in MOSFETs).
• The Gap: While Atomic Layer Etching (ALE) offers atomic-scale precision and damage-free processing for other semiconductors, it had not been experimentally demonstrated for diamond. Previous theoretical concepts (e.g., using $NO_2$) lacked experimental validation and mechanistic clarity.

2. Methodology

The authors developed a cyclic plasma Atomic Layer Etching (ALE) process using an Inductively Coupled Plasma (ICP) reactor. The process relies on the separation of two half-reactions: surface modification and material removal.

• Chemistry Strategy:
  • Modification Step: Exposure to Oxygen ($O_2$) plasma. Reactive oxygen species chemisorb onto the diamond surface, forming oxidized carbon bonds (C–O, C=O). This weakens the surface C–C bonds and lowers the sputtering threshold of the topmost layer.
  • Removal Step: Bombardment with low-energy Krypton (Kr) ions. Kr is chosen as a heavy noble gas to provide directional, controllable physical bombardment.
• Process Cycle:
  1. Oxygen Exposure: Surface modification.
  2. Purge: Prevents overlap between chemical and physical steps.
  3. Kr Ion Bombardment: Removal of the modified layer (RF bias applied only here).
  4. Purge: Prepares for the next cycle.
• Substrates: Single-crystal CVD diamond samples patterned with aluminum masks.
• Characterization: Scanning Electron Microscopy (SEM), Atomic Force Microscopy (AFM), X-ray Photoelectron Spectroscopy (XPS), and optical profilometry were used to analyze etch depth, roughness, and chemical bonding.

3. Key Contributions

• First Experimental Demonstration: This is the first reported experimental realization of plasma ALE for diamond.
• O2/Kr Chemistry Validation: The study validates that the combination of oxygen-based chemical modification and krypton-based physical removal creates a viable ALE pathway for diamond.
• Mechanistic Insight: The paper elucidates the mechanism where oxygen chemisorption lowers the sputtering threshold of the surface layer, creating a narrow energy window for selective removal.
• Process Optimization: Identification of the specific "ALE window" (ion energy range) required to achieve self-limiting behavior.

4. Key Results

• Self-Limiting Behavior & ALE Window:
  • The process exhibits a narrow ALE window of approximately 1.5 eV (corresponding to a DC bias of 17–18.5 V and ion energies of ~37–38.5 eV).
  • Within this window, the Etch Per Cycle (EPC) is constant at 6.85 Å, independent of ion energy fluctuations, confirming self-limiting behavior.
  • Below this window, removal is inefficient; above it, bulk diamond sputtering occurs, destroying the ALE regime.
• Process Synergy:
  • Modification-only: EPC = 0.87 Å (due to weak inherent ion bombardment).
  • Removal-only: EPC = 2.35 Å (due to progressive sputtering/fatigue).
  • Full ALE Cycle: EPC = 6.85 Å.
  • Synergy Factor: The full cycle achieves a 53% synergy, proving that the material removal is driven by the cyclic interaction of oxidation and ion-assisted desorption, not just the sum of individual steps.
• Surface Morphology:
  • Roughness Reduction: The Root-Mean-Square (RMS) roughness decreased from 1.23 nm (pristine) to 1.1 nm (etched). The ALE process acts as a gentle smoothing mechanism, removing pre-existing defects without introducing new damage.
  • Anisotropy: SEM images showed well-defined vertical sidewalls with no micromasking or edge damage.
• Surface Chemistry (Damage Assessment):
  • XPS Analysis:
    • Removal-only and Modification-only processes showed significant graphitization (peaks near 284.2 eV) and structural damage.
    • Full ALE Process: The C 1s spectrum was dominated by the $sp^3$ diamond peak (285.1–285.2 eV) with negligible graphitic contribution. This confirms the preservation of the diamond bonding structure and essentially damage-free etching.

5. Significance

This work establishes a foundational pathway for nanoscale, damage-controlled processing of diamond.

• Device Performance: By eliminating subsurface damage and graphitic conversion, this technique addresses a critical bottleneck in diamond device fabrication. It is expected to significantly improve the performance of diamond-based MOSFETs (potentially increasing mobility by orders of magnitude) and photonic devices.
• Quantum Technologies: The ability to etch with sub-nanometer precision while preserving the lattice structure is crucial for fabricating high-quality quantum sensors and qubits (e.g., Nitrogen-Vacancy centers), where surface states and defects are detrimental.
• Future Outlook: The O2/Kr ALE process provides a scalable, plasma-based alternative to ion beam etching, opening new opportunities for advanced diamond electronics, photonics, and quantum computing hardware.