Dependence of the Mn sticking coefficient on Ga-rich, N-rich, and Ga/N-flux-free conditions in GaN grown by plasma-assisted molecular beam epitaxy

This study demonstrates that the sticking coefficient of Mn in GaN grown by plasma-assisted molecular beam epitaxy is highly dependent on flux conditions, being highest under N-rich environments and lowest under Ga-rich conditions.

The Gist

Imagine you are trying to build a very specific type of LEGO castle (Gallium Nitride, or GaN) that needs to be magnetic. To make it magnetic, you need to sneak in some special "magnetic bricks" (Manganese, or Mn) while you are building the wall.

This paper is like a report from a master builder who discovered that how you build the wall changes how many of those special magnetic bricks actually stick.

Here is the story of their experiment, broken down into simple concepts:

1. The Setup: The Three Building Modes

The scientists were building layers of this material using a high-tech oven called Molecular Beam Epitaxy (MBE). They wanted to see how the "atmosphere" of the oven affected the magnetic bricks. They tried three different scenarios:

• The "Nitrogen Party" (N-rich): Imagine the room is packed with Nitrogen guests, but there are only a few Gallium builders. The builders are rushing to grab the Nitrogen.
• The "Gallium Flood" (Ga-rich): Imagine the room is flooded with Gallium builders, but there are very few Nitrogen guests. The builders are crowded and bumping into each other.
• The "Silent Pause" (No-flux): The builders and guests stop coming in completely. The room is empty, except for the magnetic bricks being dropped in one by one.

2. The Experiment: Dropping the Magnetic Bricks

In all three scenarios, the scientists dropped the same amount of Manganese (the magnetic bricks) onto the surface. They wanted to see: How many of these bricks actually stuck to the wall, and how many bounced off?

3. The Results: The "Sticky" Factor

The results were dramatic, like a game of musical chairs where the rules change every round:

• In the "Nitrogen Party" (N-rich): The magnetic bricks stuck like glue. This was the best condition. The scientists found that almost all the magnetic bricks they dropped were incorporated into the wall.
  • Analogy: Think of the Nitrogen-rich surface as a clean, empty parking lot. When you drop a car (Mn) in, it has plenty of empty spots to park in.
• In the "Silent Pause" (No-flux): The bricks stuck moderately well. It wasn't as good as the Nitrogen party, but it was better than the flood.
  • Analogy: This is like a parking lot that is empty but slightly dusty. Some cars park, but a few might slip off.
• In the "Gallium Flood" (Ga-rich): The magnetic bricks barely stuck at all. Most of them bounced right off the surface. The amount of magnetic material in the wall dropped by a factor of 100 compared to the Nitrogen condition!
  • Analogy: Imagine a parking lot that is already completely packed with regular cars (Gallium). When you try to drop a new car (Mn) in, there is no room! The new car has nowhere to go, so it bounces off the roof. The Gallium atoms are "hogging" all the parking spots.

4. Why Does This Happen?

The scientists explain that the Manganese atoms want to sit in the same seats as the Gallium atoms (they are "cations").

• When there is a Nitrogen rush, the Gallium atoms are busy bonding with Nitrogen, leaving empty seats for the Manganese to take.
• When there is a Gallium flood, the Gallium atoms are everywhere, fighting for the seats. The Manganese atoms get pushed out because they can't compete with the sheer number of Gallium atoms.

5. The Big Takeaway

If you want to make magnetic Gallium Nitride (which could be used for future super-fast computers or spintronic devices), you must build it in a Nitrogen-rich environment.

If you try to build it when there is too much Gallium, your magnetic material will just bounce off the surface, and your device won't work. The paper gives us a "sticking coefficient" (a score of how well it sticks):

• Nitrogen-rich: Score of 1.0 (Perfect).
• No-flux: Score of 0.31 (Okay).
• Gallium-rich: Score of 0.01 (Terrible).

In short: To get the magnetic properties you want, don't crowd the surface with Gallium. Let the Nitrogen do the work, and the magnetic bricks will happily stick around.

Technical Summary

Here is a detailed technical summary of the paper "Dependence of the Mn sticking coefficient on Ga-rich, N-rich, and Ga/N-flux-free conditions in GaN grown by plasma-assisted molecular beam epitaxy."

1. Problem Statement

Manganese (Mn) doping in Gallium Nitride (GaN) is critical for developing spintronic devices (e.g., ferromagnetic GaMnN) and semi-insulating layers. While it is generally known that Nitrogen-rich (N-rich) growth conditions enhance Mn incorporation compared to Gallium-rich (Ga-rich) conditions, the specific quantitative influence of Ga/N flux ratios on Mn sticking coefficients during Plasma-Assisted Molecular Beam Epitaxy (PA-MBE) has not been systematically investigated.

Furthermore, N-rich conditions in PA-MBE often lead to unfavorable 3D growth modes and surface roughness. Understanding the trade-off between Mn incorporation efficiency and surface morphology under different flux regimes is essential for optimizing the growth of high-quality magnetic nitride semiconductors.

2. Methodology

The researchers employed Plasma-Assisted Molecular Beam Epitaxy (PA-MBE) to grow a multilayer GaN:Mn/GaN structure on a Ga-polar GaN(0001) template.

• Growth Conditions:
  • Substrate Temperature: Maintained at 680°C.
  • Flux Regimes: Three distinct conditions were tested:
    1. N-rich: $f_N / f_{Ga} \approx 1.3$ (N flux dominant).
    2. Ga-rich: $f_{Ga} / f_N \approx 1.3$ (Ga flux dominant).
    3. No-flux (Mn $\delta$-doping): Mn shutter open while Ga and N shutters were closed (though N plasma was ignited to maintain background N2).
  • Mn Flux: Fixed beam-equivalent pressure of $1.6 \times 10^{-9}$ Torr for all doped regions.
• Sample Structure:
  • A 225 nm GaN buffer layer (Ga-rich) was grown to ensure a clean surface.
  • Three Mn-doped regions were sandwiched between Ga-rich GaN spacer layers:
    • N-rich GaN:Mn (43 nm).
    • Mn $\delta$-doped layer (34 s shutter time).
    • Ga-rich GaN:Mn (112 nm).
• Characterization:
  • In-situ: Reflection High-Energy Electron Diffraction (RHEED) monitored surface reconstruction and crystal quality.
  • Ex-situ: Atomic Force Microscopy (AFM) measured surface roughness.
  • Depth Profiling: Dynamic Secondary Ion Mass Spectrometry (SIMS) quantified Mn concentration and impurity levels (O, C, Si, H).

3. Key Results

A. Mn Incorporation and Sticking Coefficients

The study quantified the Mn sticking coefficient ($\alpha$) relative to the N-rich condition (normalized to 1.0):

• N-rich Condition: Highest Mn concentration ($\approx 1 \times 10^{20} \text{ cm}^{-3}$).
• No-flux ($\delta$-doping) Condition: Intermediate Mn concentration. The calculated relative sticking coefficient was 0.31.
• Ga-rich Condition: Lowest Mn concentration ($\approx 1 \times 10^{18} \text{ cm}^{-3}$). The calculated relative sticking coefficient was 0.01.

This indicates a two-order-of-magnitude difference in Mn incorporation between N-rich and Ga-rich regimes.

B. Surface Morphology and Crystal Quality

• N-rich Growth: RHEED patterns showed modulation, indicating a rougher surface due to reduced adatom diffusion lengths. However, the wurtzite hexagonal structure was maintained, and no polarity inversion occurred.
• Ga-rich Growth: RHEED patterns remained sharp and streaky, indicating a smooth, 2D growth mode. The surface exhibited a Ga-bilayer reconstruction, confirming no polarity inversion despite Mn doping.
• Impurities: O and C concentrations were significantly higher (by >1 order of magnitude) in N-rich and no-flux regions compared to Ga-rich regions, attributed to the susceptibility of Ga-adlayer-free surfaces to oxygen incorporation.

C. Kinetic Analysis

• The Mn concentration plateaus in SIMS profiles closely matched the shutter opening times, indicating negligible Mn "floating" or delayed incorporation. Unincorporated Mn desorbs immediately from the surface.
• The Mn $\delta$-doped layer showed a peak density of $\approx 8.2 \times 10^{12} \text{ cm}^{-2}$ (areal density), corresponding to a volumetric density of $\approx 3 \times 10^{20} \text{ cm}^{-3}$ if confined to a single monolayer.

4. Key Contributions

1. Quantification of Sticking Coefficients: The paper provides the first systematic determination of relative Mn sticking coefficients under Ga-rich, N-rich, and flux-free conditions in PA-MBE, establishing values of 0.01 (Ga-rich), 0.31 (No-flux), and 1.0 (N-rich).
2. Mechanism Elucidation: It confirms that Mn occupies cation (Ga) sites. Under N-rich conditions, cation sites are readily available for Mn. Under Ga-rich conditions, Mn competes with the abundant Ga flux for these sites, leading to high desorption rates and low sticking coefficients.
3. Morphology vs. Incorporation Trade-off: The study highlights that while N-rich conditions maximize Mn incorporation, they degrade surface morphology (3D growth), whereas Ga-rich conditions preserve smooth surfaces but severely limit Mn doping.
4. Polarity Stability: It demonstrates that at 680°C, Mn doping levels up to $10^{20} \text{ cm}^{-3}$ do not induce polarity inversion on Ga-polar GaN, unlike Mg doping which can cause inversion at lower concentrations.

5. Significance

• Spintronics Optimization: The findings provide a critical guide for growing high-quality GaMnN. Researchers must balance the need for high Mn concentration (favoring N-rich) against the need for smooth interfaces and low defect densities (favoring Ga-rich).
• Growth Kinetics Understanding: The results clarify the surface kinetics of transition metal doping in nitrides, showing that Mn behavior is analogous to Mg (where N-rich also favors incorporation) but with distinct sticking coefficient magnitudes.
• Future Directions: The paper suggests that future studies should investigate the influence of substrate temperature and crystal polarity (N-polar vs. Ga-polar) on Mn incorporation, as these factors significantly alter sticking coefficients and activation energies.

In summary, this work establishes that N-rich conditions are essential for maximizing Mn incorporation in GaN, but this comes at the cost of surface roughness, while Ga-rich conditions suppress Mn incorporation by two orders of magnitude due to site competition, despite offering superior surface quality.