High Uniformity GaN Micro-pyramids and Platelets by Selective Area Growth

This study addresses the issue of morphological non-uniformity in GaN micro-pyramids and platelets grown via MOCVD by identifying the limitations of one-step growth and demonstrating that a controlled multi-step strategy combining sequential growth and thermal treatment significantly enhances structural regularity for advanced optoelectronic applications.

The Gist

The Big Picture: Building a Perfect City of Tiny Towers

Imagine you are an architect trying to build a city of millions of tiny, perfect crystal towers (called micro-pyramids) on a single piece of glass. These towers are the building blocks for next-generation screens (like super-bright micro-LEDs) that will one day replace our current displays.

The goal is simple: Every single tower must be exactly the same height and shape. If one tower is too short, or if the top is jagged, the whole city fails to function properly.

However, the scientists in this paper found that building these towers is incredibly tricky. When they tried to grow them all at once, the results were messy. Some towers were giants, some were dwarfs, and many had jagged, broken tops.

This paper is the story of how they figured out why the towers were failing and invented a new "construction method" to build a perfect, uniform city.

---

The Problem: The "Jealous Neighbor" Effect

The scientists started by growing these towers using a process called Selective Area Growth. Think of this like planting seeds in specific holes in a fence. They wanted the crystals to grow only inside the holes.

They quickly noticed two main problems:

1. The "Broken Roof" Problem (V-Pits): Many towers had a jagged hole in the very center of their flat tops, like a crater. This happened because the crystal wasn't growing smoothly; it was collapsing inward.
2. The "Height Inequality" Problem: Some towers grew huge, while their neighbors stayed tiny. Why?

The Discovery: The "Super-Express" Elevator
The scientists discovered the secret behind the height inequality. It turns out that the "soil" they were planting in (the template substrate) had invisible defects called screw dislocations.

• The Analogy: Imagine the crystal growth is like people trying to climb a ladder.
  • Normal Ladder: If the ladder is smooth, people have to walk up step-by-step. It's slow.
  • The "Super-Express" Ladder: If there is a screw dislocation, it acts like a spiral staircase or a magic elevator that never ends. People (atoms) can hop on this elevator and zoom straight to the top.

Towers growing over these "magic elevators" (screw dislocations) grew fast and tall. Towers growing on smooth ground (no dislocations) grew slow and short. Since the "soil" had these defects scattered randomly, the resulting city of towers was a chaotic mix of giants and dwarfs.

The Experiments: Finding the Right Recipe

The team tried different "recipes" to fix the chaos:

1. Changing the Temperature (The Heat)

• Too Cold: The towers grew at the same speed (uniform height), but their tops were jagged and full of craters (V-pits).
• Too Hot: The craters disappeared, but the towers became a chaotic mess of different heights again. The "magic elevators" became too powerful, and the fast growers stole all the building materials from the slow ones.

2. Changing the Air (The Gas)
They tried blowing different gases over the towers.

• Hydrogen Air: Made the tops smooth but kept the height inequality.
• Nitrogen Air: Made the towers grow more evenly in height, but the tops became rough and jagged again.

3. Changing the Soil (The Template)
They tried planting on different types of "soil."

• Dirty Soil (High Defects): Chaos. A mix of giants and dwarfs.
• Clean Soil (Low Defects): Perfect uniformity! Every tower was the same height.
• The Catch: The "Clean Soil" is incredibly expensive to buy, like trying to build a city on a diamond floor. The scientists wanted to use cheaper soil but get the same result.

The Solution: The "Build, Fix, Build" Strategy

Since they couldn't afford the perfect "Clean Soil" for mass production, they invented a clever Multi-Step Strategy.

Instead of trying to build the whole tower in one long, continuous session (which leads to mistakes piling up), they broke the process into small cycles:

1. Grow a little bit: Build the tower for a short time.
2. Stop and "Heal": Turn up the heat and let the tower sit in a special gas bath. This is like a spa treatment for the crystal.
   • What happens here? The jagged holes (V-pits) melt away and smooth out. The crystal reshapes itself into a perfect hexagon.
3. Repeat: Grow a little more, then heal again.

The Analogy: Imagine sculpting a statue out of clay.

• The Old Way: You try to carve the whole statue in one go. If you make a mistake, you can't fix it, and the statue looks lumpy.
• The New Way: You carve a little bit, then stop to smooth out the clay with your hands. Then you carve a little more, and smooth it again. By the end, you have a perfect statue, even if you started with imperfect clay.

The Result: A Perfect Crystal City

By repeating this "Grow-Heal-Grow-Heal" cycle six times, the scientists achieved something amazing:

• They used the cheaper, imperfect soil.
• They eliminated the jagged holes (V-pits).
• They made every single tower the exact same height and shape.

Why Does This Matter?

This discovery is a huge step forward for technology. It means we can mass-produce the tiny crystals needed for super-bright, high-resolution screens (like future VR headsets or smartphone displays) without needing to buy prohibitively expensive materials.

By understanding the "magic elevators" (defects) and inventing a "healing cycle" (annealing), the scientists turned a chaotic construction site into a perfectly organized city, paving the way for the next generation of light-based technology.

Technical Summary

1. Problem Statement

The development of III-nitride micro-light-emitting diodes (micro-LEDs) requires high-quality, uniform arrays of 3D GaN microstructures (such as pyramids and platelets) to overcome efficiency limitations found in planar devices, specifically non-radiative surface recombination and the quantum-confined Stark effect (QCSE).

However, fabricating these structures via Selective Area Growth (SAG) using Metal-Organic Chemical Vapor Deposition (MOCVD) faces a critical challenge: morphological non-uniformity. Key issues include:

• Height Inconsistency: Significant variations in the height of microstructures within the same array.
• Surface Defects: The presence of V-pits on the top c-plane facets, leading to distorted apexes.
• Irregular Shapes: Truncated tops and inconsistent cross-sections.

These defects hinder the scalability and yield required for next-generation optoelectronic applications. The paper aims to identify the root causes of these non-uniformities and propose a process strategy to achieve high-precision, uniform GaN microstructures.

2. Methodology

The researchers employed a systematic approach combining material characterization, parametric studies, and process innovation:

• Substrate Templates: Three distinct templates were utilized to vary screw dislocation densities:
  • Type A: In-house grown GaN on AlN/sapphire (Low screw dislocation density).
  • Type B: Commercial GaN on sapphire (High screw dislocation density, >10⁸ cm⁻²).
  • Type C: Commercial AlN on sapphire (Extremely low screw dislocation density).
• Growth Conditions: MOCVD growth was performed with varying parameters:
  • Temperature: Ranged from 825°C to 970°C.
  • Carrier Gas: Ratios of H₂ to N₂ were adjusted (Pure H₂, 1:1, 1:3, Pure N₂).
  • Masking: SiNx masks with 2 µm circular openings in a triangular array (3 µm pitch).
• Characterization: Scanning Electron Microscopy (SEM) for morphology, Atomic Force Microscopy (AFM) for surface roughness and height profiling, and X-ray diffraction (XRD) for crystal quality.
• Process Strategies:
  • Single-step growth at various temperatures.
  • Two-step growth (Low-temp nucleation + High-temp growth).
  • Multi-step Growth-then-Annealing: A novel approach dividing total growth time into cycles of short growth followed by thermal annealing in N₂/NH₃.

3. Key Contributions & Mechanisms Identified

A. Mechanism of Height Non-Uniformity

The study identified screw dislocations as the primary driver of height variation.

• Observation: Microstructures containing screw dislocations grew significantly taller than those without.
• Mechanism: Screw dislocations act as continuous sources of atomic steps (spiral growth), serving as efficient sinks for Ga adatoms. In contrast, dislocation-free platelets have atomically smooth surfaces where adatoms must diffuse long distances to find nucleation sites, often migrating to neighboring structures with dislocations.
• Validation: A quantitative correlation was established between the calculated screw dislocation density (derived from XRD FWHM) and the experimentally observed ratio of "tall" vs. "short" platelets. The data matched closely across all template types.

B. Mechanism of V-Pit Formation

V-pits are governed by the ratio of growth rates between the c-plane ($GR_{c-plane}$) and semipolar facets ($GR_{s.p.}$).

• Low Temperature: At 825°C, the growth rate ratio is low, leading to V-pit expansion on the c-plane.
• High Temperature: Higher temperatures increase the ratio, suppressing V-pits but causing extreme height non-uniformity due to the dominance of screw dislocation effects and adatom diffusion competition.

C. Carrier Gas Influence

• H₂-rich Ambients: Promote smooth c-plane surfaces (etching effect, high adatom mobility) but result in high height variability (CV ~30%).
• N₂-rich Ambients: Improve height uniformity significantly (CV ~4%) by suppressing lateral growth and enhancing vertical growth, but degrade surface morphology (increased V-pits) and cause unwanted deposition on the SiNx mask.

4. Results

• Template Dependence:
  • Type B (High Dislocation): Resulted in a mix of tall and short platelets; tall ones were dominant.
  • Type C (Low Dislocation): Produced highly uniform heights but suffered from V-pits due to the lack of dislocation-induced c-plane acceleration.
• Thermal Annealing: Annealing V-pitted samples at 950°C in N₂/NH₃ successfully "healed" the c-plane surface, removing V-pits and reshaping platelets into regular hexagons, albeit with a slight reduction in height.
• Multi-Step Growth-then-Annealing (The Solution):
  • Direct single-step growth failed to produce uniform pyramids without V-pits.
  • A 6-cycle process (120s growth / 600s annealing at 950°C) was developed.
  • Outcome: This approach successfully suppressed V-pit propagation by periodically removing defects during annealing cycles. It resulted in nearly perfect GaN pyramids with:
    • High hexagonal symmetry.
    • Excellent height and size uniformity.
    • Smooth c-plane surfaces free of V-pits.
    • Negligible deposition on the SiNx mask.

5. Significance

This work provides a comprehensive roadmap for the scalable fabrication of high-quality GaN microstructures essential for micro-LEDs and other optoelectronic devices.

• Process Innovation: The proposed "multi-step growth-then-annealing" strategy decouples the conflicting requirements of surface smoothness (requiring low temp/annealing) and height uniformity (requiring specific dislocation management), allowing the use of lower-cost, imperfect substrates (like Type A) to achieve results comparable to high-cost bulk substrates.
• Fundamental Insight: The study clarifies the critical role of screw dislocations in SAG kinetics, offering a predictive model for microstructure morphology based on substrate quality.
• Application Impact: By enabling the production of uniform, defect-free micro-pyramid arrays, this research addresses a major bottleneck in the commercialization of high-efficiency, high-resolution micro-LED displays.