Molecular beam epitaxy of wafer-scale O-band InAs/InGaAs quantum dots on GaAs for quantum photonics

This paper presents a scalable molecular beam epitaxy strategy utilizing gradient sub-monolayer InAs deposition and strain-reducing capping to achieve wafer-scale, low-density, electrically tunable InAs/InGaAs quantum dots on GaAs that emit single photons in the O-band for quantum photonics applications.

The Gist

The Big Picture: Building Tiny Light Bulbs for the Future Internet

Imagine you are trying to build a super-fast, unhackable internet (Quantum Internet). To do this, you need tiny, perfect light bulbs that can flash exactly one photon (a single particle of light) at a time. These are called Single-Photon Sources (SPSs).

The problem? Most of these "light bulbs" are made of materials that flash light at a wavelength (color) that gets absorbed by the air or doesn't travel well through fiber-optic cables. We need them to flash at a specific color called the "O-band" (around 1.3 micrometers), which is the "golden highway" for fiber-optic communication.

This paper is a recipe for growing these perfect, single-photon light bulbs on a standard silicon-like wafer, but with a few clever tricks to make them the right color and the right size.

---

The Challenge: The "Goldilocks" Problem

The scientists are growing Quantum Dots (QDs). Think of these as microscopic islands of material (Indium Arsenide) sitting on a flat plain (Gallium Arsenide).

• Too small: They don't emit the right color of light.
• Too big: They emit the wrong color or stop working.
• Too crowded: If you have too many islands, they bump into each other, and you can't pick just one to use.

The goal is to grow islands that are:

1. The right size to emit the "O-band" color.
2. Very far apart (low density) so we can pick individual ones.
3. Uniform so they all behave the same way.

The Solution: A Three-Part Recipe

The team developed a new way to grow these dots using a technique called Molecular Beam Epitaxy (MBE). Think of MBE as a very precise, high-tech spray painting machine that builds materials atom by atom.

1. The "Rough Carpet" Trick (Surface Roughness Modulation)

Usually, when you spray paint a smooth floor, the paint spreads out evenly. But the scientists wanted the paint to clump up in specific spots.

• The Analogy: Imagine trying to build sandcastles on a perfectly flat beach. The sand just spreads out. But if you first lay down a rug with a bumpy texture, the sand will naturally pile up in the valleys and on the bumps.
• The Science: They grew a special "pattern-defining layer" (a slightly rough carpet) before growing the dots. This roughness acts like a magnet, forcing the quantum dots to form only in specific "rough" spots, keeping them far apart and organized.

2. The "Stop-and-Go" Dance (Gradient Deposition)

They needed to control exactly how much "paint" (Indium) was sprayed. If they sprayed too much, the dots would merge into a big blob. If too little, no dots would form.

• The Analogy: Imagine a chef sprinkling salt on a giant pizza. Instead of shaking the shaker all at once, they sprinkle a tiny pinch, wait, sprinkle another pinch, and rotate the pizza.
• The Science: They used a "sub-monolayer" technique. They sprayed Indium for 3 seconds, then stopped for 15 seconds (letting the atoms settle), then sprayed again. By synchronizing this "stop-and-go" with the rotation of the spinning wafer, they created a gradient.
  • One side of the wafer got a little bit of material (few dots).
  • The other side got more (many dots).
  • This allowed them to find the "sweet spot" on the wafer where the dots were perfectly spaced out.

3. The "Red-Hat" Trick (Strain-Reducing Layer)

Even with the right spacing, the dots were still flashing light that was too "blue" (too short a wavelength) for fiber optics. They needed to shift the color to "red" (the O-band).

• The Analogy: Imagine a spring. If you stretch it, it changes shape. The scientists put a special "hat" (a layer of Indium Gallium Arsenide) on top of the dots. This hat is slightly different in size than the dot underneath, which gently stretches the dot.
• The Science: This stretching changes the energy inside the dot, shifting the color of the light it emits from 1.2 micrometers to the target 1.3 micrometers (O-band). Crucially, they grew this "hat" at a lower temperature to prevent the materials from mixing too much, which would ruin the dot's shape.

The Results: A Perfect Light Source

After all this careful engineering, they tested the dots:

• The Map: They scanned the whole wafer and found huge areas where the dots were perfectly spaced out (less than 1 dot per square micrometer).
• The Color: The dots flashed light right in the O-band, ready for fiber-optic cables.
• The Single Photon Test: They fired electricity at a single dot and measured the light. The result was a "g(2)" value of 0.02.
  • What does this mean? In the world of quantum physics, a value of 0 means "perfect single photon." A value of 1 means "random light." Getting 0.02 is like hitting a bullseye from a mile away. It proves the dot is emitting exactly one photon at a time, almost perfectly.

Why This Matters

This paper isn't just about making pretty dots; it's about scalability.

• Old way: Making these dots was like trying to hand-craft a watch; you could do one or two, but not a whole factory full.
• New way: This method works on a whole 2-inch wafer at once. It's like switching from hand-crafting to a 3D printer.

They have shown that you can mass-produce high-quality, single-photon sources that work with existing internet cables. This is a massive step toward building a real, working Quantum Internet that can be used for secure communication and super-fast quantum computers.

Summary in One Sentence

The scientists figured out how to grow millions of perfectly spaced, single-photon light bulbs on a single wafer by using a "rough carpet" to organize them and a "stretchy hat" to tune their color to the perfect frequency for the internet.

Technical Summary

1. Problem Statement

The development of scalable quantum photonic technologies, such as quantum communication networks and quantum computing, requires efficient single-photon sources (SPSs). While InAs/GaAs quantum dots (QDs) are well-established SPSs, they typically emit at wavelengths below 1 µm. However, fiber-based quantum infrastructure operates in the telecommunications O-band (~1.3 µm) and C-band (~1.55 µm).

Achieving O-band emission from InAs QDs grown on GaAs substrates presents two major challenges:

1. Wavelength Redshift: Standard Stranski-Krastanow (SK) growth of InAs on GaAs at typical temperatures (480–520°C) yields emission wavelengths rarely exceeding 1.2 µm. Redshifting to 1.3 µm requires large, Indium-rich dots or specific strain-reducing layers (SRL), which often compromises structural quality.
2. Density Control: Practical integration into photonic circuits (e.g., cavities) requires low-density QDs (< 10⁸ cm⁻²) to allow for the deterministic selection of single emitters. However, achieving low density while maintaining the large size required for O-band emission is difficult due to the steep onset of nucleation and complex interplay of growth parameters (temperature, flux, time).

2. Methodology

The authors developed a scalable Molecular Beam Epitaxy (MBE) strategy combining gradient deposition, surface roughness modulation, and strain engineering.

• Substrate Preparation: Growth was performed on undoped 2-inch GaAs(001) substrates.
• Pattern Defining Layer (PDL): Before QD growth, a 15 nm GaAs layer was deposited without substrate rotation. This created a surface roughness modulation (alternating flat terraces and rough regions with atomic steps) across the wafer, which acts as a template to control nucleation sites.
• Gradient Sub-Monolayer (Sub-ML) Deposition:
  • InAs was deposited in the sub-monolayer regime (1.6–2.8 ML) using a sequence of growth time ($t_g = 3$ s) and pause time ($t_p$).
  • Synchronization: The deposition cycles were synchronized with substrate rotation ($T$). Specifically, $t_p = (2N+1) \cdot t_g$ was used to create a controlled coverage gradient across the wafer, while $t_g = T/2$ ensured uniformity where needed.
  • This approach allowed precise control over the transition from 2D wetting layers to 3D islands, enabling the formation of low-density regions.
• Strain-Reducing Layer (SRL): To achieve the O-band emission, the InAs QDs were capped with a 7 nm In$_{0.29}$Ga$_{0.71}$As SRL.
  • Temperature Optimization: To minimize Indium-Gallium intermixing (which would degrade the dot size and emission), the SRL overgrowth was performed at a reduced temperature of 490°C (compared to the QD formation temperature of 520–525°C).
• Characterization:
  • Optical: Photoluminescence (PL) mapping across the wafer, micro-PL ($\mu$PL) at 4 K, and hyperspectral imaging (HSI) to analyze >500 individual dots.
  • Structural: Atomic Force Microscopy (AFM) and Aberration-corrected Scanning Transmission Electron Microscopy (STEM) with Energy-Dispersive X-ray (EDX) spectroscopy.
  • Device: Fabrication of n-i-Schottky photodiodes to test electrical tunability via the Quantum Confined Stark Effect (QCSE).

3. Key Contributions

• Universal Growth Strategy: Demonstrated a method to control QD density and position on a wafer-scale using standard MBE systems on (001) surfaces, independent of specific flux non-uniformities.
• Density Modulation via PDL: Validated that a non-rotated GaAs PDL creates a roughness landscape that lowers the nucleation barrier in specific regions, allowing for the creation of low-density zones (< 10⁸ cm⁻²) suitable for single-dot selection.
• Synchronization Technique: Introduced a synchronized rotation and sub-ML deposition protocol to create reproducible coverage gradients, enabling the tuning of dot size and density across the substrate.
• Low-Temperature Capping: Established that capping InAs QDs with an InGaAs SRL at 490°C effectively suppresses In-Ga intermixing, preserving the large dot volume necessary for O-band emission while achieving the required redshift.

4. Key Results

• Wavelength Achievement: The combination of large, In-rich InAs QDs and the In$_{0.29}$Ga$_{0.71}$As SRL successfully redshifted the emission to the O-band (~1310 nm).
  • STEM analysis revealed a QD aspect ratio ($H/L$) of ~0.2 and an Indium content in the SRL of $x \approx 0.29$.
  • The SRL contributed ~50–60 meV to the redshift via strain redistribution, while the increased volume and reduced aspect ratio contributed an additional 30–50 meV.
• Density Control:
  • PL mapping and HSI confirmed the ability to create regions with QD densities as low as 0.15 QDs/µm² (approx. $1.5 \times 10^7$ cm⁻²).
  • HSI of 505 individual dots showed a peak emission distribution centered near 1310 nm with an asymmetric tail toward shorter wavelengths (attributed to smaller dots in rougher regions).
• Single-Photon Emission:
  • Micro-PL measurements identified neutral exciton (X), charged exciton (CX), and biexciton (XX) transitions.
  • Second-order autocorrelation measurements ($g^{(2)}(0)$) on the CX transition yielded $0.020 \pm 0.014$, confirming high-purity single-photon emission under continuous wave excitation.
• Electrical Tunability:
  • In a photodiode structure, the QD emission was tuned via the QCSE.
  • The Stark shift followed a parabolic dependence $E(F) = E_0 - p \cdot F - \beta \cdot F^2$.
  • The polarizability ($\beta \approx 0.6\text{--}0.7 \, \mu\text{eV} \cdot \text{kV}^{-2} \cdot \text{cm}^2$) was found to be significantly higher than in standard InAs/GaAs QDs, attributed to the larger dot height and specific strain profile.
  • The static dipole moment ($p/e \approx 0.3\text{--}0.4$ nm) indicated a spatial separation of electron and hole wavefunctions due to the In-composition gradient within the dot.

5. Significance

This work provides a scalable, reproducible pathway for fabricating high-quality, electrically tunable single-photon sources operating in the telecommunications O-band. By overcoming the trade-off between low density and long-wavelength emission, the authors enable the deterministic integration of QDs into nanophotonic circuits (e.g., cavities, waveguides) on standard GaAs platforms. The demonstrated ability to control dot position and density via surface roughness and rotation synchronization makes this approach highly suitable for the mass production of quantum photonic devices required for future quantum networks.