🔬 materials science

Fast Homoepitaxy on (100) \b{eta}-Ga2O3 Substrates with Large Grown-In Offcut

This study demonstrates that scalable (100)-oriented $\beta$ -Ga $_2$ O $_3$ wafers with large grown-in offcuts, produced via Edge-defined Film-fed Growth, enable fast molecular beam epitaxy growth rates and high-quality epilayers that yield Schottky barrier diodes with superior breakdown fields and low unintentional doping, thereby overcoming the historical growth rate limitations of this crystal orientation.

Original authors: M. Brooks Tellekamp, Drew Haven, David Joyce, John Mangum, Henry Garland, Robert Lavelle, Kevin Schulte, Anna Sacchi, Matt Young, Andriy Zakutayev

Published 2026-07-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: M. Brooks Tellekamp, Drew Haven, David Joyce, John Mangum, Henry Garland, Robert Lavelle, Kevin Schulte, Anna Sacchi, Matt Young, Andriy Zakutayev

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

Imagine trying to build a super-fast highway for electricity, but the road you're building on is a bit weird. The material, called beta-gallium oxide (β-Ga2O3), is like a crystal with a lopsided, monoclinic shape. Because of this shape, the direction you choose to build your road matters a lot.

For a long time, scientists thought the best road to build on was the (001) direction because it was easy to make big chunks of the crystal. But there was a catch: if you tried to build on the (100) direction, the road construction crew (the growth process) was incredibly slow. It was so slow that it was practically a snail race, moving at a crawl of 0.1 to 0.15 nanometers per minute. Meanwhile, other directions were zooming along. Because of this speed difference, the (100) direction was mostly abandoned, even though it had a secret superpower: it could theoretically handle a much stronger electrical "traffic jam" (breakdown field) than the others.

The Big Idea: Tilting the Crystal
The team behind this study asked a clever question: What if we don't just lay the crystal flat? What if we tilt it?

In crystal growing, tilting the surface is called an "offcut." Think of it like a staircase. If the floor is perfectly flat, it's hard for new bricks (atoms) to find their place and build a smooth wall; they tend to pile up in messy clumps. But if you tilt the floor slightly, it creates a series of tiny steps. The new bricks can just walk up the steps one by one, building a smooth, fast-growing wall. This is called "step-flow growth."

Previously, scientists tried to make these tilted (100) crystals, but they could only tilt them a tiny bit (up to 6 degrees). They had to grow a giant block of crystal first and then slice and grind it at an angle, which wasted a ton of material and was expensive.

The Magic Trick: Growing the Tilt Directly
This paper introduces a new way to make these tilted crystals without the wasteful grinding. They used a technique called Edge-defined Film-fed Growth (EFG). Imagine pulling a ribbon of crystal out of a melt, like pulling taffy. Usually, you pull it straight up. But here, they rotated the seed crystal (the starting point) before they started pulling.

By rotating the seed, they grew a ribbon that already had the tilt built right into it! They didn't have to grind it down later. They made ribbons with tilts ranging from 3.4° all the way up to a massive 13.4°.

The Results: Speeding Up the Race
When they started growing thin films on these super-tilted ribbons, the results were wild.

The Speed: On the steepest tilt (13.4°), the growth rate skyrocketed to 5.1 nm/min. This is a huge jump from the old slow speed of 0.1 nm/min. In fact, it's just as fast as the fastest-growing crystals on other, more common directions.
The Quality: You might think a steep staircase would be messy, but the surface was incredibly smooth. After polishing, the bumps were less than 0.2 nm high (that's smaller than a single atom!).
The Cleanliness: The new films were very clean, with very few unwanted impurities. They measured the "unintentional doping" (accidental extra electrons) at between 2×10¹⁵ cm⁻³ and 7×10¹⁵ cm⁻³. This is one of the lowest levels ever seen for this type of growth method.

The Proof: Making a Diode
To prove these new, fast-growing, tilted crystals actually work for real electronics, they built a simple diode (a one-way valve for electricity).

They didn't add any fancy extra layers to protect the edges of the device.
The diode worked great, blocking electricity in one direction until it reached a very high voltage.
It held up to a breakdown field of 1.56 MV/cm. This is comparable to the best devices made on other crystal faces, proving that the (100) direction is a serious contender.

What They Didn't Find (and What They Ruled Out)
The paper is careful to note that while the growth is fast, it's not perfect at every temperature.

If they grew the film at 750 °C on the steepest tilt, the surface got bumpy again because the atoms started bunching up in steps (step-bunching).
If they grew at 600 °C, the atoms didn't have enough energy to move around, and the surface looked grainy.
The "sweet spot" for the steepest tilts was actually 650 °C, where the surface stayed smoothest.

The Bottom Line
This study shows that by tilting the (100) crystal face significantly (up to 13.4°) and growing it directly without wasteful grinding, we can finally make these crystals grow fast enough to be useful. They aren't just fast; they are clean, smooth, and capable of making powerful electronic devices. It's like finding a way to build a super-highway on a tricky terrain by simply tilting the road, turning a slow, messy construction project into a smooth, high-speed operation.

Technical Summary: Fast Homoepitaxy on (100) β-Ga2O3 Substrates with Large Grown-In Offcuts

Problem Statement
The monoclinic β-Ga2O3 semiconductor exhibits significant anisotropy, where crystal orientation critically influences bulk growth, epitaxial synthesis, material properties, and device performance. While (001)-oriented substrates have been the standard for scalable device fabrication due to ease of epitaxy via Halide Vapor Phase Epitaxy (HVPE), the (100) orientation offers superior theoretical advantages, including a higher breakdown field (10.2 MV/cm vs. 7.6 MV/cm for (001)) and a lower impact ionization coefficient. However, (100)-oriented wafers have historically been abandoned for device applications due to extremely slow epitaxial growth rates (0.1–0.15 nm/min) and high densities of twin domains. Previous attempts to utilize (100) faces with offcuts (misorientations) to suppress twins and promote step-flow growth were limited to angles around 6°; larger offcuts were avoided because machining them from standard 3D crystal boules results in prohibitive material waste and loss of crystalline quality.

Methodology
To overcome the trade-off between large offcut angles and material waste, the authors employed Edge-defined Film-fed Growth (EFG) to produce β-Ga2O3 ribbons with "grown-in" offcuts. By rotating the seed crystal during the pulling process (in the [010] direction) toward the [-c] direction, they directly fabricated (100)-oriented ribbons with offcut angles ranging from 3.4° to 13.4°, eliminating the need for post-growth grinding and polishing to achieve these angles.

The ribbons underwent optimized chemical-mechanical polishing (CMP) to achieve surface roughness ≤ 0.2 nm RMS. These substrates were then used for Molecular Beam Epitaxy (MBE) growth. The study systematically varied substrate temperature (600–750 °C) and offcut angle to investigate growth rates, surface morphology, and unintentional doping densities. Characterization techniques included High-Resolution X-Ray Diffraction (HRXRD), Atomic Force Microscopy (AFM), Reflection High-Energy Electron Diffraction (RHEED), Scanning Transmission Electron Microscopy (STEM), and Secondary Ion Mass Spectrometry (SIMS). Electrical performance was evaluated via Electrochemical Capacitance-Voltage (ECV) profiling and the fabrication of planar Schottky Barrier Diodes (SBDs) without edge termination.

Key Contributions and Results

Restoration of Growth Rates: The study demonstrates that large offcuts on (100) substrates can recover growth rates comparable to the fastest-growing (010) orientation. On a 13.4° offcut substrate at 650 °C, a growth rate of 5.1 nm/min was achieved. This matches or exceeds the "ideal" growth rate of 4.0–4.5 nm/min observed on (010) substrates and is significantly faster than previous reports for (100) substrates.
Surface Morphology and Growth Mode: RHEED and AFM analysis confirmed that despite the extreme offcut angles, the surface maintains (100) terraces with (2 $\bar{1}$ 01) step edges, indicative of step-flow growth. However, the optimal temperature for smooth morphology shifts with the offcut angle: 700 °C for moderate offcuts (~6°) and 650 °C for large offcuts (>10°). At higher temperatures (750 °C) on large offcuts, step-bunching occurs due to reduced adatom diffusion length on the narrow terraces.
Interface Quality: Cross-sectional STEM imaging of a 13.4° offcut interface revealed a pristine homoepitaxial boundary with no evidence of twin domains or interface defects, confirming that the step-flow growth mode is maintained even at these extreme angles.
Low Unintentional Doping (UID): ECV profiling of MBE-grown films on 13.4° offcut substrates revealed an unintentional n-type doping density of 2×10¹⁵ cm⁻³ to 7×10¹⁵ cm⁻³. This represents one of the lowest UID concentrations reported for MBE-grown β-Ga2O3 to date, attributed to the use of ultra-pure precursors, reduced plasma power, and the high growth rates which limit impurity incorporation.
Device Performance: Planar Schottky Barrier Diodes fabricated on these unintentionally doped epilayers (without edge termination or field management) demonstrated a rectification ratio of ~10⁵ and an average breakdown field of 1.56 MV/cm. This performance is comparable to similar devices fabricated on (001) and other crystal faces.

Significance
The paper validates the co-design of crystal face and offcut angle for low-symmetry β-Ga2O3. By utilizing EFG to directly grow ribbons with large offcuts, the authors demonstrate that the scalability and performance benefits of (100)-oriented wafers can be realized without the material waste associated with traditional machining. The results confirm that (100) substrates with large offcuts are a viable platform for fast epitaxial growth and high-power electronic device fabrication, offering a pathway to leverage the superior theoretical breakdown field of the [100] direction while maintaining high-quality, scalable manufacturing.

Technical Summary: Fast Homoepitaxy on (100) β-Ga2O3 Substrates with Large Grown-In Offcuts

More like this