Tuning magnitude and direction of lattice thermal conductivity in transition metal dichalcogenide heterobilayers

This study utilizes first-principles calculations to reveal how mass contrast and doping in transition metal dichalcogenide heterobilayers govern the magnitude and direction of lattice thermal conductivity through phonon localization and scattering mechanisms, enabling the tunability of thermal transport for novel 2D functional materials.

The Gist

Imagine a sandwich made of two ultra-thin slices of bread, where each slice is a different type of crystal. In the world of nanotechnology, these are called Transition Metal Dichalcogenide (TMD) heterobilayers. They are like microscopic Lego bricks used to build future electronic devices.

The problem? Just like a real sandwich, heat behaves differently depending on how the ingredients are stacked and what they are made of. If a device gets too hot, it breaks. If it's too cold, it doesn't work well. The goal of this research was to figure out exactly how heat travels through these crystal sandwiches and how to control it.

Here is a simple breakdown of what the scientists found:

1. The "Traffic Jam" of Heat

Think of heat not as a warm breeze, but as a crowd of tiny, invisible runners (called phonons) trying to sprint through a stadium.

• In a perfect, clean stadium (Pristine layers): The runners are all wearing the same shoes and running on a smooth track. They can run fast and in any direction equally. The scientists found that in these clean, two-layer sandwiches, heat flows easily and equally in all directions across the surface.
• The "Relaxon" Discovery: Usually, scientists try to track each runner individually. But the researchers found that in these sandwiches, the runners often hold hands and move as a single, coordinated wave. They call these waves "relaxons." It's like a "wave" in a sports stadium; the individual people aren't moving forward, but the wave itself travels. By studying these waves instead of individual runners, the scientists could better understand why heat moves the way it does.

2. The Heavy vs. Light Runner Effect

The scientists noticed a rule about the "weight" of the runners:

• Lighter is usually faster: If the atoms in the crystal are light (like lighter elements), the heat runners can sprint faster.
• The "Heavy" Barrier: However, if you mix heavy atoms with light atoms in the same layer, it creates a "mass contrast." Imagine a track where some lanes have heavy sandbags and others are smooth. This actually helps organize the runners. If the difference in weight between the two layers of the sandwich is big enough, the heat runners get "stuck" in one specific layer, which changes how fast they travel.

3. The "Doping" Experiment: Adding Chaos

Next, the scientists tried "doping" the sandwiches. This means they took one type of crystal and randomly swapped some of its atoms for a different, heavier type (swapping Molybdenum for Tungsten).

• The Result: This is like throwing random obstacles onto the track. The heat runners start bumping into these obstacles (mass disorder).
• The Outcome: The heat flow slowed down significantly. More importantly, it stopped flowing equally in all directions. Now, heat preferred to flow in one specific direction over another, creating a "traffic jam" that was directional.

4. Turning the Heat Flow Like a Dial

The most exciting finding is that by changing how much of the heavy atoms they added (the concentration) and how hot the system was, they could actually rotate the direction of the heat flow.

• Imagine you have a flashlight that shines heat. In a clean sandwich, the beam shines straight out. In a doped sandwich, by tweaking the recipe and the temperature, you can make that beam tilt slightly to the left or right.
• This suggests that in the future, engineers could "tune" these materials to guide heat exactly where they want it to go, or keep it away from sensitive parts of a device.

Summary

The paper is essentially a manual on how to control the "traffic" of heat in microscopic crystal sandwiches.

• Clean sandwiches let heat flow fast and equally in all directions.
• Mixing heavy and light atoms creates a "layered" effect that organizes the heat.
• Adding random heavy atoms (doping) slows the heat down and makes it flow in a specific, tunable direction.

The researchers didn't just guess; they used advanced computer simulations to watch these "heat runners" and "heat waves" in action, proving that by simply changing the ingredients and the temperature, you can steer the flow of heat in new ways. This helps scientists design better, more efficient electronic devices that don't overheat.

Technical Summary

1. Problem Statement

Transition Metal Dichalcogenides (TMDs) are promising 2D materials for electronic, optoelectronic, and tribological applications. However, their thermal management requirements are conflicting: some applications require efficient heat dissipation (high thermal conductivity), while others (like thermoelectrics) require suppressed heat transport (low thermal conductivity).
While the lattice thermal conductivity (LTC) of monolayer TMDs is well-studied, the behavior of heterobilayers (stacks of two different TMD layers) is complex due to interlayer coupling, mass contrast, and symmetry breaking. Furthermore, existing studies often rely on the Single-Mode Relaxation Time (SMRT) approximation, which neglects collective phonon effects. There is a lack of understanding regarding:

• The microscopic mechanisms governing LTC in pristine TMD heterobilayers.
• How doping (specifically W-doping in MoS$_2$) affects both the magnitude and the directional anisotropy of thermal transport.
• The role of collective excitations (relaxons) in describing heat transport in these systems.

2. Methodology

The authors employed a first-principles approach combining Density Functional Theory (DFT) with the exact solution of the Linearized Boltzmann Transport Equation (LBTE).

• Systems Studied:
  • Pristine Heterobilayers: 15 unique MX$_2$-M'X'$_2$ combinations (where M, M' $\in$ {Mo, W} and X, X' $\in$ {S, Se, Te}).
  • Doped Systems: W-doped Mo$_{1-x}$W$_x$S$_2$ heterobilayers with varying concentrations ($x \in [0.0, 1.0]$), ranging from MoS$_2$ homobilayer to WS$_2$ homobilayer.
• Computational Tools:
  • VASP: For geometry optimization and force constant calculation (using PBE-GGA and DFT-D2 for van der Waals corrections).
  • Phono3py: To calculate second- and third-order force constants and solve the LBTE in the phonon basis (exact solution, not SMRT).
  • Phoebe: To solve the LBTE in the relaxon basis. Relaxons are eigenvectors of the scattering operator, representing collective phonon excitations that account for mode coupling.
  • Machine Learning Potentials (MLP): Used for doped systems with large unit cells (36 atoms) to efficiently generate force constants via finite displacements.
• Key Descriptors & Metrics:
  • Transport-weighted averages: Of relaxon velocities ($\langle V_\alpha \rangle_\kappa$), lifetimes ($\langle \tau_\alpha \rangle_\kappa$), and mean free paths ($\langle \Lambda_\alpha \rangle_\kappa$).
  • Layer Localization: Quantified by the difference in phonon mode participation between layers ($|\Delta P_\lambda|$).
  • Thermal Viscosity ($\mu_{iiii}$): Used as a probe for hydrodynamic (collective) transport regimes.
  • Cophonicity ($C_{ph}$): A descriptor based on the overlap of atom-projected phonon densities of states (DOS) between metal and chalcogen sublattices.
  • Anisotropy ($A$) and Angle ($\Delta\theta$): Measures of the deviation between maximum and minimum conductivity directions relative to crystallographic axes.

3. Key Contributions

1. Relaxon Framework Application: The study provides the first systematic analysis of TMD heterobilayer thermal transport using the relaxon basis, moving beyond independent phonon approximations to capture collective scattering effects.
2. Microscopic Descriptors: Identification of specific descriptors (layer localization, mass contrast, cophonicity) that correlate with LTC, offering a physical basis for predicting thermal properties.
3. Doping-Induced Anisotropy: Discovery that doping not only reduces thermal conductivity but also breaks in-plane symmetry, allowing for the tuning of the direction of maximum heat flow via concentration and temperature.
4. High-Throughput Potential: Demonstration that low-cost descriptors like cophonicity can predict complex transport behaviors, enabling rapid screening of new 2D materials.

4. Key Results

A. Pristine Heterobilayers

• Isotropy: Pristine heterobilayers exhibit isotropic in-plane thermal conductivity, with a preserved ordering of materials across the temperature range (100–500 K).
• Dominant Mechanisms: LTC is dominated by low-frequency acoustic modes (< 2 THz).
• Correlations:
  • Mean Free Path (MFP): Shows the strongest linear correlation with LTC, as it encapsulates both velocity and lifetime.
  • Layer Localization: A large mass contrast between layers induces layer localization of vibrational modes. Counter-intuitively, higher localization (where modes are confined to one layer) correlates with higher LTC because it reduces interlayer scattering channels.
  • Average Mass: Lighter systems generally have higher LTC, but mass contrast is a necessary condition to induce the beneficial layer localization.
• Hydrodynamic Transport: A linear correlation was found between LTC and thermal viscosity. Systems with higher viscosity (indicating dominant Normal scattering over Umklapp scattering) exhibit higher LTC.
• Cophonicity: The relative distribution of vibrational states between metal and chalcogen sublattices (cophonicity) correlates with thermal viscosity, suggesting it controls the balance between momentum-conserving and momentum-dissipating scattering.

B. Doped Heterobilayers (W-doped MoS$_2$)

• Reduced Conductivity: Doping significantly reduces LTC compared to pristine systems due to mass-disorder scattering and reduced group velocities.
• Anisotropy: Unlike pristine systems, doped systems exhibit anisotropic in-plane thermal transport ($\kappa_{max} \neq \kappa_{min}$).
• Directional Tuning: The direction of maximum conductivity ($\Delta\theta$) varies with both temperature and dopant concentration. This implies that the "preferred" heat flow direction can be engineered by adjusting composition and operating temperature.
• Breakdown of Descriptors: In doped systems, the correlation between layer localization/cophonicity and LTC breaks down. The transport is dominated by mass-disorder scattering rather than the intrinsic layer character or collective hydrodynamic effects seen in pristine systems.
• Limitations: The magnitude of anisotropy is modest ($A \approx 0.02 - 0.12$), and the specific dopant configuration (spatial arrangement) plays a critical role, suggesting that configurational disorder in real experiments could affect these trends.

5. Significance and Implications

• Design Rules for Thermal Management: The study establishes that thermal transport in 2D heterostructures can be tuned not just by magnitude (via doping or layer selection) but also by direction. This is crucial for designing thermal circuits and managing hotspots in nanodevices.
• Beyond SMRT: By utilizing the relaxon framework, the authors demonstrate that collective effects are essential for accurately describing transport in layered materials, particularly for understanding the hydrodynamic regime.
• Predictive Screening: The identification of cophonicity as a low-computational-cost descriptor allows for the rapid screening of van der Waals heterostructures for specific thermal applications (e.g., high-conductivity heat spreaders or low-conductivity thermoelectrics) without needing full LBTE calculations.
• Fundamental Insight: The work clarifies the interplay between mass contrast, layer localization, and scattering mechanisms, providing a microscopic explanation for why certain TMD combinations perform better thermally than others.

In conclusion, this paper bridges the gap between microscopic phonon physics and macroscopic thermal transport properties in TMD heterostructures, offering a robust protocol for engineering thermal transport in next-generation 2D functional materials.