Chemical Vapor Deposition of Nitrides by Carbon-free Brominated Precursors

This study demonstrates that using carbon-free gallium- and aluminum-brominated precursors in an industrial chemical vapor deposition system significantly reduces carbon-related defects in epitaxial GaN and AlN layers compared to conventional metal-organic methods, thereby enhancing the performance of high-frequency and high-power nitride devices.

The Gist

Imagine you are trying to build the ultimate, high-speed race car engine. To make it fast and powerful, you need to construct the engine block out of incredibly pure, flawless metal. In the world of electronics, this "engine" is made of materials called Group 13-nitrides (like Gallium Nitride or Aluminum Nitride). These materials are the superheroes of modern technology, powering everything from fast 5G networks to efficient electric vehicle chargers.

The Problem: The "Grease" in the Engine

For decades, the standard way to build these electronic layers has been a process called MOCVD (Metal-Organic Chemical Vapor Deposition). Think of this process like baking a cake. To get the right ingredients (Gallium or Aluminum) to stick to the baking pan (the wafer), manufacturers use special "mixing bowls" called precursors.

However, the standard mixing bowls are made of metal-organic compounds. The problem? They contain carbon (the "organic" part).

• The Analogy: Imagine you are trying to bake a pristine, white angel food cake. But, the flour you are using is mixed with a little bit of dark chocolate chips. Even if you try to be careful, those chocolate chips (carbon atoms) get baked right into the cake.
• The Consequence: In electronics, these "chocolate chips" (carbon) act like potholes in a highway or grease in a race car engine. They get in the way of electricity, causing the device to overheat, lose power, or fail under high pressure.

The Solution: A New Recipe

The scientists in this paper asked a simple question: "What if we could bake the cake without the chocolate chips?"

They wanted to switch from the "organic" recipe to a Carbon-Free recipe. But there was a catch:

1. Chlorine-based recipes (used for silicon chips) are too aggressive. They are like using a blowtorch to bake the cake; they would melt the oven (corrode the expensive reactor equipment).
2. Iodine-based recipes are too heavy. They are like trying to lift a boulder; they won't turn into gas easily enough to bake the cake.

The Breakthrough: The team found a "Goldilocks" ingredient: Bromine.

• The Analogy: Bromine is like a gentle, precise chef's knife. It's strong enough to do the job but gentle enough not to destroy the kitchen. They used Gallium Bromide (GaBr₃) and Aluminum Bromide (AlBr₃) as their new ingredients.

The Experiment: Baking the Perfect Cake

The researchers took these new bromine-based ingredients and put them into a standard industrial oven (a commercial MOCVD reactor). They heated them up just enough to turn them into a gas, then sprayed them onto the wafer along with ammonia (the nitrogen source).

The Results were surprisingly good:

1. Smooth Surfaces: The layers they grew were as smooth as a polished mirror. No cracks, no bumps.
2. The "Glow" Test: To check for impurities, they shined a special light on the layers.
   • Old Method (TMGa/TMAl): The layers glowed with a dull, messy blue and yellow light. This was the "glow" of the carbon defects (the chocolate chips) messing up the structure.
   • New Method (GaBr₃/AlBr₃): The layers were almost completely dark, except for a sharp, clean glow at the very top. This means the "chocolate chips" were gone! The material was incredibly pure.
3. Electrical Performance: The new layers were highly resistant to electricity (which is exactly what you want for insulating layers in high-power devices), suggesting the "grease" that usually causes short circuits was removed.

Why This Matters

This paper is a proof of concept. It's like the first time someone successfully built a jet engine using a new, untested fuel.

• Before: We were stuck with "dirty" ingredients that limited how fast and powerful our electronics could get.
• Now: We have a recipe that works in existing factories without needing to rebuild the whole plant.

The Bottom Line:
By swapping out the carbon-heavy ingredients for bromine-based ones, the team has shown that we can grow ultra-pure electronic materials. This paves the way for:

• Faster phones and internet (less signal loss).
• More efficient electric cars (less energy wasted as heat).
• Stronger power grids (devices that can handle higher voltages without breaking).

It's a small change in the recipe, but it could lead to a massive upgrade in the technology we use every day. The only thing left to do is for chemical manufacturers to start making these bromine ingredients in "electronic grade" purity, just like they do for the old ingredients. Once that happens, the future of electronics looks very bright—and very clean.

Technical Summary

Here is a detailed technical summary of the paper "Chemical Vapor Deposition of Nitrides by Carbon-free Brominated Precursors."

1. Problem Statement

The epitaxial growth of Group 13-nitride semiconductors (GaN, AlN, AlGaN) for high-frequency and high-power devices is currently dominated by Metal-Organic Chemical Vapor Deposition (MOCVD). This standard process relies on metal-organic precursors (e.g., Trimethylgallium - TMGa, Trimethylaluminum - TMAl) which contain carbon-based ligands.

• The Core Issue: Carbon atoms from these precursors are unavoidably incorporated into the epitaxial layers.
• Consequences: Carbon acts as a deep-level impurity (often substituting nitrogen sites, $C_N$), leading to:
  • Unintentional doping and self-compensation.
  • Reduced free carrier concentration and mobility.
  • Degradation of Two-Dimensional Electron Gas (2DEG) properties in heterostructures.
  • Impaired p-type doping efficiency (Mg activation).
  • Parasitic luminescence (Blue and Yellow bands) indicating defects.
  • Limitations on breakdown voltage and dynamic on-resistance in power devices.
• Current Limitations: While Chlorinated Chemical Vapor Deposition (HCVD/HVPE) offers carbon-free growth, it is highly corrosive (damaging reactor components) and difficult to scale for multi-wafer, complex heterostructure manufacturing. Previous attempts to use brominated precursors (e.g., $GaBr_3 \cdot 4NH_3$) failed due to low vapor pressure, thermal instability, and hygroscopicity.

2. Methodology

The authors implemented a Carbon-free CVD process using an industrial Aixtron MOCVD reactor equipped with a close-coupled showerhead.

• Precursor Selection:
  • Strategy: Replaced metal-organic precursors with inorganic brominated precursors: Gallium Bromide ($GaBr_3$) and Aluminum Bromide ($AlBr_3$).
  • Rationale: Bromine offers a compromise between the high corrosiveness of chlorine/fluorine and the low vapor pressure of iodine. $GaBr_3$ and $AlBr_3$ are stable, have manageable melting points ($121.5^\circ\text{C}$and$97.5^\circ\text{C}$), and sufficient vapor pressure for controlled flow.
• Vaporization Setup:
  • Custom heating systems were employed to maintain precise temperatures: $GaBr_3$ at $135^\circ\text{C}$** and$AlBr_3$at **$110^\circ\text{C}$.
  • Gas lines and valves were heated to prevent condensation.
  • Carrier Gas: Nitrogen ($N_2$) was used to minimize the formation of HBr byproducts, though some HBr/NH$_4$Br formation is inevitable.
• Growth Conditions:
  • Temperature: $1200^\circ\text{C}$.
  • Pressure: $35\text{ mbar}$(optimized for brominated precursors; standard TMGa runs used$200\text{ mbar}$).
  • Substrates: 4-inch sapphire templates with pre-grown GaN or AlN layers (grown via standard MOCVD) to serve as nucleation layers.
  • Flows: High $NH_3$ flows ($8000\text{ sccm}$for GaN,$1100\text{ sccm}$ for AlN) to ensure high V/III ratios.

3. Key Contributions

• First Demonstration: To the authors' knowledge, this is the first successful epitaxial growth of GaN and AlN in a commercial multi-wafer MOCVD reactor using strictly carbon-free, brominated precursors.
• Process Feasibility: Proved that brominated precursors can be vaporized and controlled with molar flows ($>100\text{ \mu mol/min}$ for Ga, $>240\text{ \mu mol/min}$ for Al) comparable to standard metal-organic precursors, enabling industrial scalability.
• Overcoming Thermodynamic Barriers: Addressed previous thermodynamic concerns regarding the ability to grow alloys with halogenated precursors by demonstrating successful single-crystal growth of GaN and AlN.

4. Key Results

The study compared layers grown with brominated precursors against standard TMGa/TMAl layers using multiple characterization techniques:

• Morphology:
  • Layers were smooth and crack-free.
  • AFM Results: GaN showed an RMS roughness of 0.5 nm ($2\times2\text{ \mu m}^2$); AlN showed 1.2 nm.
• Carbon Contamination (SIMS):
  • Initial Time-of-Flight SIMS (ToF-SIMS) showed no significant difference in carbon signal intensity between brominated and standard layers, likely due to the sensitivity limits of the tool or background carbon in the templates. The authors note that more sensitive tools are required for definitive quantification.
• Optical Properties (Cathodoluminescence - CL & Photoluminescence - PL):
  • GaN: The sample grown with $GaBr_3$ showed a 10-fold reduction in Blue Luminescence (BL) and up to a 1000-fold (3 orders of magnitude) reduction in Yellow Luminescence (YL) compared to TMGa samples. These bands are strongly associated with carbon defects.
  • AlN: The $AlBr_3$ sample showed a 2.5-fold reduction in defect-related bands (Blue and Red luminescence) and a 20-fold reduction in bound-exciton emission at 208 nm compared to TMAl samples.
  • Conclusion: The drastic reduction in defect-related luminescence indicates a significant decrease in electrically and optically active carbon-related defects.
• Crystal Quality (HR-XRD):
  • Full Width at Half Maximum (FWHM) values for $\omega$-scans were comparable to or slightly better than the underlying templates, indicating that the brominated process preserved high crystal quality without introducing significant strain or defects.
• Electrical Properties:
  • Both GaN and AlN layers exhibited very high sheet resistance ($>40\text{ k}\Omega/\text{sq}$ for GaN, $>100\text{ k}\Omega/\text{sq}$ for AlN), consistent with highly resistive, undoped material with minimal unintentional doping from carbon.

5. Significance

• Performance Enhancement: By eliminating carbon incorporation, this method promises to unlock the full potential of Group 13-nitrides, specifically improving:
  • Doping control (especially p-type GaN).
  • Electron mobility and 2DEG properties.
  • Device reliability and breakdown voltage.
  • Reduction of dynamic on-resistance in power devices.
• Industrial Viability: Unlike HVPE (which is limited to thick layers) or MBE (limited by growth rate and scalability), this approach utilizes existing industrial MOCVD infrastructure. This suggests a viable pathway for mass-producing high-performance, carbon-free nitride devices without requiring a complete overhaul of manufacturing facilities.
• Future Outlook: While current precursor purity (specifically Silicon contamination) needs improvement by manufacturers, the study validates the chemical pathway. Future work will focus on optimizing growth rates, alloy composition (AlGaN), and precursor purity to achieve commercial-grade material quality.