Site-selective enhancement of Eu emission in delta-doped GaN

This study demonstrates that delta-doping GaN:Eu with alternating doped and undoped layers can selectively enhance emission from the majority Eu site, enabling both efficient energy transfer for power-efficient LEDs and narrow, homogeneous spectra suitable for quantum technologies.

The Gist

The Big Picture: Tuning a Radio to the Right Station

Imagine you have a giant radio station (the material Gallium Nitride, or GaN) that is supposed to broadcast a specific, beautiful red song (light) using a special singer named Europium (Eu).

In a standard factory setup, when you try to get this singer to perform, you run into a problem: the singer gets confused. Instead of standing in one perfect spot on the stage, the singer ends up scattered in many different, slightly messy spots.

• The Problem: Because the singer is in different spots, they sing the song slightly differently in each spot. Some spots are loud, some are quiet, and some are off-key. When you listen to the whole stage, it sounds like a messy, jumbled choir rather than a single, clear voice. This is bad for making bright, efficient lights (LEDs) and terrible for quantum computers, which need a single, pure note.

The Solution: The "Delta-Doping" Sandwich

The researchers in this paper came up with a clever way to fix this. Instead of letting the singer wander around a huge, open stage, they built a sandwich.

They created a structure with alternating layers:

1. A layer of plain GaN (empty stage).
2. A thin layer of GaN with the Europium singer (the stage).
3. Repeat this 40 times.

This is called Delta-doping. Think of it like building a multi-story apartment building where the singer is only allowed to live on specific, very thin floors, separated by empty hallways.

What They Discovered: The Magic of Thickness

The team built four different versions of this "sandwich" with different thicknesses for the singer's floors and tested them. Here is what they found, using our analogies:

1. The "Crowded Room" (Uniform Doping)

• The Setup: The singer is allowed to stand anywhere in a giant, 300-meter tall room.
• The Result: The singer is everywhere. Most of the time, the singer is in the "wrong" spots (the minority sites), making the light dim and messy. The energy from the light source gets lost trying to find the singer.

2. The "Perfect Floor" (10:1 Sandwich)

• The Setup: They made the singer's floor extremely thin (1 nanometer thick—about the width of a few atoms).
• The Result: This was the most surprising discovery. Because the floor was so thin, the singer had to stand in the perfect spot. No other spots were available.
• The Outcome: The light became pure and homogeneous. It was like the singer finally found the perfect microphone stand. Even though the total amount of light wasn't the brightest, the quality was perfect. This is exactly what quantum computers need: a single, pure note with no background noise.

3. The "Bright Stage" (10:2 Sandwich)

• The Setup: They made the singer's floor slightly thicker (2 nanometers).
• The Result: This was the "Goldilocks" zone. It was thick enough to hold a lot of singers, but thin enough to force them all into the "majority" (best) spot.
• The Outcome: This sample was much brighter than the standard room. The energy transfer was super efficient. This is great for making super-bright, energy-saving LED lights for displays.

Why Does This Work? (The "Trap" Analogy)

Why does making thin layers help?

Imagine the Europium atoms are like fish and the light energy is like food.

• In the big, open room (standard sample), the food floats around, and the fish in the "bad spots" eat it first. The fish in the "good spots" go hungry.
• In the sandwich layers, the thin layers act like traps. When the food (energy) falls from the ceiling, it gets caught in the shallow "wells" of the thin layers before it can drift away. This forces the energy to stay right where the "good spot" fish are waiting.
• The researchers found that the 10:2 trap was deep enough to catch the energy efficiently, making the light very bright. The 10:1 trap was so shallow that it only caught the "perfect" fish, filtering out the messy ones.

Why Should We Care?

This research is a big deal for two main reasons:

1. Better Lights (Classical Tech): By using the "10:2" sandwich, we can make LEDs that are much brighter and use less electricity. This could lead to better screens and more efficient lighting.
2. Quantum Computers (Future Tech): By using the "10:1" sandwich, we can create a material that emits a single, perfect color of light. This is crucial for quantum technologies, which rely on precise, identical particles to store and process information.

The Bottom Line

The researchers didn't need to invent a new material or add new chemicals. They simply changed the architecture of the existing material, stacking it like a precise sandwich. By tuning the thickness of the layers, they could choose whether they wanted a super-bright light or a perfectly pure quantum signal. It's a simple, elegant trick that could revolutionize how we build optical devices.

Technical Summary

1. Problem Statement

Europium-doped gallium nitride (GaN:Eu) is a promising material for both classical red LEDs and quantum technologies due to its bright, narrow-bandwidth emission at ~622 nm ($^5D_0 \rightarrow ^7F_2$ transition). However, the development of high-efficiency devices is hindered by the formation of non-equivalent incorporation sites (labeled OMVPE1–8) during organometallic vapor-phase epitaxy (OMVPE).

• Inhomogeneity: Uniformly doped samples exhibit an inhomogeneous photoluminescence (PL) spectrum dominated by "minority" sites (e.g., OMVPE1/2, OMVPE7) when excited above the GaN bandgap.
• Inefficient Energy Transfer: The energy transfer from the GaN host bandgap to the "majority" site (OMVPE4), which constitutes >90% of Eu ions, is inefficient. This results in poor external quantum efficiency for LEDs and broad, inhomogeneous spectra unsuitable for quantum applications requiring narrow linewidths.
• Current Limitations: Previous attempts to improve efficiency focused on increasing minority sites or using complex codoping (e.g., with Mg), which often introduces new defects or requires additional materials.

2. Methodology

The authors investigated the effect of delta-doping (alternating doped and undoped layers) on the optical properties of GaN:Eu.

• Sample Fabrication: Four samples were grown via OMVPE at 960°C:
  1. Uniformly Doped (UD): A single 300-nm thick GaN:Eu active layer.
  2. Delta-Doped (DD) Series: Structures with 40 alternating pairs of undoped GaN and doped GaN:Eu layers. The thicknesses of the doped layers were varied while keeping the undoped spacer constant (10 nm):
     • 10:10 DD: 10 nm doped / 10 nm undoped.
     • 10:2 DD: 2 nm doped / 10 nm undoped.
     • 10:1 DD: 1 nm doped / 10 nm undoped.
• Characterization Techniques:
  • Combined Excitation-Emission Spectroscopy (CEES): Performed at 10 K to map distinct incorporation sites and their relative abundances.
  • Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS): Used to measure Eu concentration profiles and normalize PL intensity against actual dopant density.
  • Temperature-Dependent PL: Measured from 16 K to 295 K under both above-bandgap (351 nm) and below-bandgap (364 nm) excitation to study carrier dynamics and energy transfer efficiency.
  • Spectral Decomposition: Non-negative matrix factorization (NMF) was used to deconvolve overlapping emission spectra into contributions from specific sites (OMVPE1/2, OMVPE4, OMVPE7).

3. Key Contributions

• Demonstration of Site-Selectivity via Delta-Doping: The study proves that reducing the thickness of the doped layer in a superlattice structure can selectively suppress minority sites and enhance the majority site (OMVPE4).
• Concentration-Normalized Brightness: By combining PL data with SIMS measurements, the authors quantified the efficiency of energy transfer per Eu atom, revealing that thinner doped layers significantly improve the optical activity of the majority site.
• Mechanism Elucidation: The work provides evidence that the enhancement is driven by carrier confinement in shallow quantum wells formed by the bandgap reduction in doped layers, rather than a change in the intrinsic radiative lifetime of the Eu ions.

4. Key Results

• Spectral Homogeneity:
  • The 10:1 DD sample (1 nm doped layers) exhibited emission exclusively from the majority site (OMVPE4). Under all excitation conditions (resonant, above-bandgap, and below-bandgap), the spectrum was identical, narrow, and homogeneous. This is ideal for quantum technologies.
  • The 10:2 DD sample showed a dramatic increase in the relative contribution of OMVPE4 compared to the UD sample, effectively suppressing minority sites.
• Enhanced Efficiency:
  • Despite having significantly less total Eu volume, the 10:2 DD and 10:1 DD samples showed a ~3x enhancement in concentration-normalized PL intensity ($I_{PL}/\sigma_{SIMS}$) for OMVPE4 compared to the UD sample.
  • The 10:2 DD sample was the brightest overall, suggesting an optimal balance between Eu concentration and carrier confinement.
• Carrier Dynamics and Temperature Dependence:
  • Thermal Quenching: The 10:10 and 10:2 DD samples maintained intensity better than the UD sample up to ~100 K, indicating that the shallow wells (induced by bandgap reduction) effectively trap carriers at higher temperatures.
  • 10:1 DD Anomaly: This sample showed a sharp drop in intensity between 20–50 K and a secondary increase around 140 K. This behavior is attributed to the extremely shallow wells (due to thin layers) failing to trap carriers at low thermal energies, followed by the release of carriers from shallow trap states at higher temperatures.
• Lifetime Measurements: The excited-state lifetime of OMVPE4 remained nearly constant (272–274 $\mu$s) across UD, 10:10, and 10:2 samples, confirming that the intensity enhancement is due to improved excitation/energy transfer efficiency, not a change in radiative decay rates. The 10:1 sample showed a slightly longer lifetime (306 $\mu$s).

5. Significance

• For Quantum Technologies: The ability to generate a pure, homogeneous emission spectrum from a single Eu site (OMVPE4) without complex codoping is a major breakthrough. This enables the creation of GaN:Eu ensembles suitable for quantum memories and single-photon sources with narrow linewidths.
• For Classical Optoelectronics: The 10:2 DD structure demonstrates a pathway to highly efficient red LEDs. By enhancing energy transfer to the majority site, the external quantum efficiency can be significantly improved, addressing a long-standing bottleneck in GaN:Eu LED performance.
• General Applicability: The delta-doping technique is simple, tunable (by adjusting layer thickness), and does not require additional dopants. The authors suggest this approach could be broadly applicable to engineering defect properties in other rare-earth doped semiconductors (e.g., Si:Er), offering a versatile tool for material design.