Acoustic and Optical Phonon Frequencies and Acoustic Phonon Velocities in Silicon-Doped Aluminum Nitride Thin Films

This study utilizes Brillouin-Mandelstam and Raman spectroscopy to demonstrate that while Si doping in AlN thin films causes non-monotonic changes in optical phonon frequencies due to strain and dislocation variations, it results in a monotonic decrease in acoustic phonon velocities, a finding critical for optimizing high-power semiconductor devices.

The Gist

Imagine you are trying to build the ultimate high-speed highway for electricity and heat. The material you choose for this highway is Aluminum Nitride (AlN), a super-strong ceramic that acts like a "super-highway" for electrons and heat in next-generation electronics.

However, to make this material useful in real devices (like powerful chargers or UV lights), you have to add a special ingredient: Silicon. Think of Silicon as the "traffic controllers" that help electricity flow. But just like adding too many traffic cones to a highway can slow down cars, adding Silicon atoms changes how the material vibrates.

This paper is a detective story about how adding Silicon changes the "vibrations" (called phonons) inside the AlN material. The scientists used two special "microscopes" made of light to listen to these vibrations.

The Two Types of Vibrations

To understand the story, you need to know there are two types of vibrations in the material, like two different ways a guitar string can move:

1. Optical Phonons (The "Jittery" Dancers): These are fast, high-energy vibrations where atoms in the crystal dance in opposite directions. They are very sensitive to stress, like a tightrope walker who wobbles if the wind changes even slightly.
2. Acoustic Phonons (The "Rolling" Waves): These are slower, smoother waves where atoms move together in a line. These are the main carriers of heat. If these waves slow down, the material gets hotter because the heat can't escape as fast.

The Experiment: Adding Silicon

The scientists grew thin films of AlN and added different amounts of Silicon, ranging from a tiny pinch to a heavy handful. They then used two techniques to listen to the vibrations:

• Raman Spectroscopy: This is like shining a flashlight on the "Jittery Dancers" (Optical Phonons) to see how fast they are dancing.
• Brillouin Scattering: This is like listening to the "Rolling Waves" (Acoustic Phonons) to measure how fast the heat waves are traveling.

What They Found

1. The "Jittery Dancers" (Optical Phonons) were Confused
When they added a little Silicon, the dancers slowed down. But when they added more Silicon, the dancers sped back up!

• The Analogy: Imagine a crowd of people in a room. If you add a few small people (Silicon) to a room of big people (Aluminum), the room feels less crowded, and people can move more freely (slower vibration). But if you keep adding so many small people that they start bumping into each other and pushing the walls, the room gets tense again, and the movement changes direction.
• The Cause: The Silicon atoms are smaller than the Aluminum atoms. At first, they relaxed the stress in the material. But eventually, they squeezed the material so much that it created new stress and tiny cracks (defects) in the crystal structure. This caused the vibration speed to go up and down in a zig-zag pattern.

2. The "Rolling Waves" (Acoustic Phonons) Got Slower, but Steady
The heat-carrying waves behaved differently. As they added more and more Silicon, the waves slowly and steadily got slower.

• The Analogy: Imagine a marching band. If you replace a few heavy drummers with lighter snare drummers, the whole band moves a tiny bit faster. But if you replace all the drummers with very light snare drummers, the whole band becomes lighter and the sound waves travel differently.
• The Result: At the highest level of Silicon, the heat waves slowed down by about 3%. It sounds small, but in the world of super-fast electronics, that's a big deal.

Why Does This Matter?

You might wonder, "So what if the heat waves slow down a little?"

• Heat Management: In high-power devices (like the ones in electric cars or 5G towers), heat is the enemy. If the heat waves slow down too much, the device overheats and breaks. The good news here is that the slowdown in AlN was relatively small (only 3%), which means these materials are still very good at handling heat, even with Silicon added.
• The "Thermal Boundary" Problem: When you stack different materials together (like a sandwich), heat often gets stuck at the interface (the bread). The speed of the sound waves determines how easily heat jumps from one layer to the next. By knowing exactly how Silicon changes the speed of these waves, engineers can design better "sandwiches" where heat flows smoothly without getting stuck.

The Bottom Line

This paper tells us that while adding Silicon to Aluminum Nitride makes the "jittery" vibrations act unpredictably (going up and down), the "heat-carrying" vibrations just get a little slower in a steady way.

This is good news for engineers. It means they can use Silicon to make these materials conduct electricity better without ruining their ability to stay cool. It's like finding a way to add more lanes to a highway without causing a traffic jam that stops the heat from leaving the city.

Technical Summary

1. Problem Statement

Ultra-wide bandgap (UWBG) semiconductors, particularly Aluminum Nitride (AlN), are critical for high-power electronics and UV optoelectronics due to their large bandgap (~6.2 eV) and high thermal conductivity. However, practical device performance relies on controlled doping (specifically Silicon, Si) to achieve necessary carrier concentrations.

• The Core Issue: Dopant atoms act as scattering centers for electrons and phonons. While the impact of doping on electron mobility is well-studied, its effect on acoustic phonon velocities (the primary heat carriers) and optical phonon frequencies in Si-doped AlN remains unclear.
• Key Questions:
  • Does doping reduce acoustic phonon velocity as a general trend?
  • How do strain, dislocation density, and dopant size interact to affect phonon behavior?
  • How do these changes impact thermal transport and thermal boundary resistance (TBR) in heterostructures?
• Gap: No prior studies had reported acoustic phonon data for Si-doped AlN films, creating a knowledge gap in optimizing thermal management for high-power UWBG devices.

2. Methodology

The researchers investigated Si-doped AlN thin films grown on c-plane sapphire substrates using Metalorganic Chemical Vapor Deposition (MOCVD).

• Sample Preparation:
  • Template: 200 nm AlN epilayers grown and annealed at 1700 °C to reduce dislocation density.
  • Active Layer: ~1000 nm of Unintentionally Doped (UID) and Si-doped AlN grown at 1100 °C.
  • Doping Range: Si concentrations varied from UID up to $3 \times 10^{19} \text{ cm}^{-3}$ using Silane ($SiH_4$).
• Characterization Techniques:
  1. Raman Spectroscopy: Used to measure optical phonon frequencies ($E_{2}^{low}$, $E_{2}^{high}$, and $A_1$ modes) near the Brillouin zone center ($\Gamma$ point). Measurements were taken in backscattering geometry with a 488 nm laser.
  2. Brillouin-Mandelstam Spectroscopy (BMS): Used to probe bulk acoustic phonon frequencies and calculate group velocities. Experiments utilized a 532 nm laser at a 45° incidence angle with a tandem Fabry–Perot interferometer.
  3. X-Ray Diffraction (HRXRD): Rocking curves of the (0002) peak were measured to calculate dislocation line densities.
  4. Spectroscopic Ellipsometry: Measured the refractive index ($n$) of the films to correct BMS velocity calculations.

3. Key Results

A. Optical Phonons (Raman Spectroscopy)

• Non-Monotonic Behavior: The frequencies of optical phonon modes ($E_{2}^{high}$, $A_1$, and $E_{2}^{low}$) exhibited non-monotonic shifts as Si concentration increased.
  • Low Doping ($\le 5 \times 10^{18} \text{ cm}^{-3}$): Peaks shifted toward bulk AlN values, indicating the relaxation of residual tensile strain (caused by the thermal expansion mismatch between AlN and sapphire).
  • High Doping ($> 5 \times 10^{18} \text{ cm}^{-3}$): Peaks shifted back (blueshift for $E_{2}^{high}$ and $A_1$), suggesting the formation of local compressional strain due to the smaller atomic radius of Si (0.42 Å) compared to Al (0.51 Å).
• Dislocation Correlation: The non-monotonic trend correlated with dislocation density. Dislocation density increased significantly above $1 \times 10^{18} \text{ cm}^{-3}$, peaking for samples with doping $> 5 \times 10^{18} \text{ cm}^{-3}$, which influenced strain relaxation and phonon peak broadening.

B. Acoustic Phonons (Brillouin-Mandelstam Spectroscopy)

• Monotonic Decrease in Velocity: Unlike optical phonons, the longitudinal acoustic (LA) phonon velocity decreased monotonically with increasing Si concentration.
  • Magnitude: The velocity dropped from 10,425 m/s (UID) to 10,160 m/s (at $3 \times 10^{19} \text{ cm}^{-3}$).
  • Percentage Change: This represents a reduction of approximately 2.5% to 3%.
• Comparison to Other Materials: The effect of Si in AlN is weaker than the effect of Boron doping in diamond (where surface acoustic phonon velocity dropped by ~15%). This suggests Si dopants in AlN have a relatively modest impact on lattice stiffness compared to the mass density changes.

C. Refractive Index

• The refractive index at 532 nm showed a sharp increase at low doping concentrations followed by saturation at high levels, which was critical for accurate velocity calculations in BMS.

4. Key Contributions

1. First Acoustic Data for Si-Doped AlN: Provided the first experimental measurements of acoustic phonon velocities in Si-doped AlN films.
2. Differentiation of Phonon Responses: Demonstrated that optical and acoustic phonons respond differently to doping:
   • Optical: Sensitive to local bond strain and dislocation density, leading to complex, non-monotonic shifts.
   • Acoustic: Sensitive to overall lattice stiffness and mass density, leading to a steady, monotonic velocity reduction.
3. Quantification of Velocity Reduction: Established that while doping reduces acoustic velocity, the magnitude (~3%) is small enough that it does not drastically amplify point-defect scattering rates (which depend on $v_g^{-3}$), unlike in materials with larger velocity drops.

5. Significance and Implications

• Thermal Management: The small reduction in acoustic phonon velocity (~3%) is beneficial for thermal management. It implies that the scattering rate of phonons on dopants is defined primarily by concentration rather than being amplified by a drastic drop in group velocity.
• Thermal Boundary Resistance (TBR): The study highlights that phonon velocity changes directly affect the phonon transmission probability at interfaces. Accurate knowledge of these velocities allows for the optimization of UWBG heterostructures to minimize TBR in high-power devices.
• Device Optimization: The findings provide a physical basis for designing Si-doped AlN devices where high carrier concentrations are needed without catastrophic degradation of thermal conductivity or excessive thermal boundary resistance.

In conclusion, the paper establishes that while Si doping introduces complex strain dynamics affecting optical phonons, its impact on acoustic phonon velocity is a predictable, monotonic, but relatively small reduction, offering a favorable trade-off for high-power UWBG semiconductor applications.