🔬 materials science

On the importance of Ni-Au-Ga interdiffusion in the formation of a Ni-Au / p-GaN ohmic contact

This study demonstrates that the formation of a Ni-Au/p-GaN ohmic contact is primarily driven by the creation of Ga vacancies associated with an Au-Ga interfacial layer during oxygen-assisted interdiffusion, rather than the presence of Ni or NiOx at the interface.

Original authors: Jules Duraz, Hassen Souissi, Maksym Gromovyi, David Troadec, Teo Baptiste, Nathaniel Findling, Phuong Vuong, Rajat Gujrati, Thi May Tran, Jean Paul Salvestrini, Maria Tchernycheva, Suresh Sundaram, Ab

Published 2026-02-13

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Jules Duraz, Hassen Souissi, Maksym Gromovyi, David Troadec, Teo Baptiste, Nathaniel Findling, Phuong Vuong, Rajat Gujrati, Thi May Tran, Jean Paul Salvestrini, Maria Tchernycheva, Suresh Sundaram, Abdallah Ougazzaden, Gilles Patriarche, Sophie Bouchoule

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: Fixing a "Stuck" Door

Imagine you are trying to build a high-speed highway (an electronic device) using a special type of road material called p-GaN. This material is great for carrying traffic (electricity), but it has a stubborn problem: when you try to connect a metal bridge (the electrical contact) to it, the traffic gets stuck.

In physics terms, this is called a Schottky barrier. It's like a heavy, rusted door that won't open easily. To get the electricity flowing freely (making an "ohmic contact"), scientists need to knock that door down.

For years, the leading theory was that you needed a specific "key" made of Nickel Oxide (NiOx) to unlock the door. The paper you shared challenges this idea. The researchers discovered that the key isn't the nickel oxide at all; it's actually about empty seats in the crowd.

The Experiment: A High-Tech Cooking Class

The scientists set up a kitchen experiment to see what happens when they cook a sandwich made of:

The Bread: p-GaN (the semiconductor).
The Meat: A thin layer of Nickel (Ni).
The Cheese: A thick layer of Gold (Au).

They put this sandwich in a special oven (Rapid Thermal Annealing) filled with Oxygen and heated it up. They wanted to see how the ingredients mixed and what made the "door" open.

The Discovery: It's Not About the Nickel

Here is what they found, broken down into three simple steps:

1. The Great Escape (Diffusion)

When they heated the sandwich in the oxygen-rich oven, a chaotic dance began:

Nickel (Ni) is like a nervous runner. It immediately runs up through the gold cheese to the very top of the sandwich. Once it hits the air (oxygen), it turns into a crust (Nickel Oxide).
Gold (Au) is the opposite. It slides down through the nickel layer, all the way to the bottom, touching the bread (GaN).
Gallium (Ga) is a piece of the bread itself. As the gold pushes down, it pushes the Gallium atoms out of the bread and into the metal layers above.

2. The "Auto-Cleaning" and The "Empty Seats"

The researchers used a super-powerful microscope (TEM) to look at the layers after cooking. They saw two major things happening:

The Clean Sweep: The mixing process seemed to wipe away any dirt or old oxide that was blocking the connection.
The Empty Seats (The Real Hero): Because the Gold pushed the Gallium out of the bread, it left behind empty spots (vacancies) in the crystal structure of the p-GaN.

The Analogy: Imagine a crowded concert hall (the p-GaN). The people (electrons) can't move because the seats are full.

Old Theory: We thought we needed to paint the stage Nickel Oxide to make the lights work.
New Discovery: The real magic happened when Gold pushed some people out of their seats, creating empty seats (Gallium vacancies). Now, the remaining people can move around freely! The electricity flows because there is room to move, not because of the paint on the stage.

3. The Nickel Myth

The most surprising finding was that Nickel doesn't even need to touch the bread to make the contact work.

In their experiments, the Nickel ran all the way to the top and turned into a crust.
The bottom of the sandwich was now just Gallium and Gold touching the p-GaN.
Even without Nickel touching the bottom, the contact worked perfectly! This proves that the "Nickel Oxide key" theory was wrong. The "Gold-Gallium alloy with empty seats" is the real secret sauce.

Why Does This Matter?

This discovery changes how engineers build better LEDs and lasers.

Simpler Recipes: You don't need to worry about getting the perfect Nickel layer. You just need to ensure the Gold and Gallium mix well to create those "empty seats."
Better Performance: By creating these vacancies, they can lower the resistance, meaning the devices use less energy and shine brighter.
Lower Heat: They found they could get great results even at lower temperatures (350°C instead of 450°C), which saves energy and protects delicate parts of the device.

The Takeaway

Think of the p-GaN contact like a crowded hallway. For a long time, everyone thought you needed a specific security guard (Nickel Oxide) to let people through.

This paper shows that the security guard isn't the important part. The important part is clearing the hallway. By using Gold to push Gallium out of the way, they create space (vacancies) for the electricity to flow freely. It's not about who is standing at the door; it's about making sure the path is clear.

1. Problem Statement

The performance of optoelectronic devices (such as LEDs and laser diodes) based on Gallium Nitride (GaN) is often limited by the high specific contact resistance at the p-GaN/metal interface.

The Challenge: Creating a low-resistance ohmic contact on p-GaN is difficult due to the high activation energy of Mg dopants (resulting in low free hole density) and the lack of metals with a work function matching p-GaN (~6.6 eV).
Current Debate: While thermal annealing of Ni-Au/p-GaN contacts in an oxygen atmosphere is known to improve ohmic behavior, the underlying physical mechanism remains controversial. Previous studies have proposed conflicting theories:
1. Formation of a NiO/p-GaN interface is the primary driver.
2. Removal of surface contamination ("auto-cleaning").
3. Creation of Gallium (Ga) vacancies near the interface.
4. Formation of a (Ni-)Au-Ga alloy.
Objective: This study aims to clarify the exact role of Ni, NiOx, Au, and Ga interdiffusion during Rapid Thermal Annealing (RTA) to determine the true origin of the ohmic behavior.

2. Methodology

The researchers employed a combined approach of electrical characterization and high-resolution structural/chemical analysis.

Sample Preparation:
- Substrate: MOCVD-grown InGaN/GaN blue LED heterostructures with a 270 nm Mg-doped p-GaN layer (terminated by a 12 nm p+-GaN layer).
- Metallization: Ni (20 nm) and Au (200 nm) layers were deposited via electron-beam evaporation. A secondary set of experiments used thinner layers (3 nm Ni / 3 nm Au) with a lower evaporation rate to enhance analysis resolution.
- Patterning: Transmission Line Method (TLM) structures were defined for electrical testing.
Annealing Conditions:
- Samples underwent Rapid Thermal Annealing (RTA) in a pure Oxygen (O2) atmosphere.
- Variables included temperature (350°C, 450°C, 650°C) and duration (2 min to 10 min).
- Control experiments were performed in Nitrogen (N2) to isolate the effect of oxygen.
Characterization Techniques:
- Electrical: TLM measurements to extract specific contact resistance ( $r_c$ ), Schottky barrier height ( $\Phi_B$ ), and maximum current ( $I_{max}$ ).
- Microscopy & Spectroscopy: High-Resolution Transmission Electron Microscopy (HR-TEM) and High-Angle Annular Dark-Field (HAADF) imaging coupled with Energy Dispersive X-ray Spectroscopy (EDX).
- Analysis: 2D elemental mapping and quantitative line scans to determine atomic composition, diffusion profiles, and the presence of vacancies or alloys.

3. Key Contributions and Findings

A. The Role of Oxygen and Ni Diffusion

Oxygen is Critical: Annealing in O2 is essential. Annealing in N2 did not improve contact resistance or induce Ni diffusion.
Ni Transport Mechanism: Oxygen assists the diffusion of Ni upward through the Au layer to the top surface, where it instantly oxidizes to form a NiOx layer.
Au Transport: Simultaneously, Au diffuses downward toward the GaN surface, replacing the Ni layer at the interface.

B. Refutation of the NiOx Interface Theory

Key Finding: The presence of Ni or NiOx at the interface with p-GaN is not the primary cause of the ohmic behavior.
Evidence: After 10 minutes of annealing, Ni and NiOx completely diffused to the top surface, leaving a clean interface between p-GaN and an Au-rich layer. Despite the absence of NiOx at the interface, the contact remained highly ohmic with low resistance. This contradicts previous claims that a NiO/p-GaN interface is necessary.

C. The Crucial Role of Ga Vacancies and Au-Ga Alloy

Ga Out-Diffusion: The study provides direct evidence that Ga atoms diffuse out of the p-GaN lattice into the metal layer during annealing.
Vacancy Creation: This Ga out-diffusion creates a high concentration of Ga vacancies in the top p-GaN layer.
Alloy Formation: The diffused Ga reacts with Au to form a Ga-Au interfacial alloy (sometimes with Ni nano-crystallites).
Mechanism: The creation of Ga vacancies significantly reduces the Schottky barrier height, facilitating hole injection and maximizing current flow.

D. Impact of Evaporation Rate and Layer Thickness

Thin Layers & Low Rate: Samples with very thin layers (3 nm) deposited at a low rate (0.1 nm/s) showed improved ohmic behavior even before annealing.
Reason: The low evaporation rate and thinness allowed spontaneous Ga out-diffusion and interdiffusion to begin immediately upon deposition, creating vacancies and an Au-Ga interface without external thermal treatment.

4. Results Summary

Electrical Performance:
- Non-annealed (Thick layers): High barrier (~0.70 eV), high resistance ( $r_c \sim 1.5 \times 10^{-1} \Omega cm^2$ ).
- Annealed (450°C, 2 min): Barrier reduced, $r_c$ dropped to $7.5 \times 10^{-3} \Omega cm^2$ , and $I_{max}$ increased significantly.
- Thin Layers (3 nm): Even without annealing, $r_c$ was reduced by a factor of 10 compared to thick non-annealed layers, proving that the diffusion mechanism can be triggered by deposition conditions alone.
Structural Observations:
- 2 min Anneal: Ni is partially oxidized at the top; Au has reached the interface; Ga is detected in the metal layer (indicating vacancy creation).
- 10 min Anneal: Ni is fully oxidized at the top; the interface is purely Au-Ga rich; the contact remains ohmic, confirming NiOx is not required at the interface.
- 650°C Anneal: The Au-Ga layer becomes discontinuous with large voids, degrading electrical contact stability.

5. Significance and Conclusion

This study fundamentally shifts the understanding of Ni-Au/p-GaN ohmic contacts:

Paradigm Shift: It refutes the long-held belief that a NiO/p-GaN interface is the primary mechanism for ohmic behavior. Instead, it establishes that Ga vacancies created by Ga out-diffusion are the dominant factor.
Mechanism Clarification: The process involves a "self-cleaning" interdiffusion where Ni moves up (oxidizing) and Au moves down, dragging Ga out of the semiconductor to form a conductive Au-Ga interface.
Practical Implications:
- Optimization: A moderate annealing temperature (350°C–450°C) for short durations (2–5 min) is optimal. Higher temperatures (650°C) degrade the interface.
- Process Control: Controlling evaporation rates and layer thickness can induce the necessary diffusion mechanisms even without high-temperature annealing.
- Universality: The findings suggest that creating Ga vacancies is a universal mechanism for forming ohmic contacts on p-GaN, potentially applicable to metals other than Ni-Au.

In summary, the paper demonstrates that the creation of Ga vacancies via interdiffusion, rather than the presence of a specific oxide layer, is the key to achieving high-performance ohmic contacts on p-GaN.