Impact of Exchange-Correlation Functionals on Predictions of Phonon Hydrodynamics: A Study of Fluorides, Chlorides, and Hydrides

This study utilizes density functional theory with various exchange-correlation functionals to investigate the thermal and mechanical properties of alkali halides and hydrides, revealing that the choice of functional significantly influences predictions of lattice thermal conductivity and the conditions for observing phonon hydrodynamics, while also identifying new materials exhibiting this phenomenon.

The Gist

Imagine a bustling city where millions of tiny messengers (called phonons) are running back and forth, carrying heat energy from one place to another. Usually, these messengers are chaotic. They bump into each other, hit walls, and get distracted, moving in a messy, random crowd. This is how heat usually travels in most materials: a slow, diffusive shuffle.

But in some special, ultra-pure materials (like certain salts and hydrides), something magical happens at very low temperatures. The messengers stop fighting and start dancing in perfect unison, flowing like a smooth river or a wave. This is called Phonon Hydrodynamics, and the "second sound" is the wave-like ripple of heat you can actually see.

This paper is a detective story about how we use a powerful computer tool called Density Functional Theory (DFT) to predict when and where this magical dancing happens. The twist? The tool has different "settings" (called Exchange-Correlation Functionals), and the choice of setting changes the prediction.

Here is the breakdown of the paper using simple analogies:

1. The Three "Maps" (The Functionals)

To predict how these heat messengers behave, scientists use mathematical models. Think of these models as three different types of maps for the same city:

• LDA (Local Density Approximation): This is like an old, slightly zoomed-in map. It tends to make the city look a bit too crowded and the buildings (atoms) too close together.
• PBE (Perdew-Burke-Ernzerhof): This is a standard, modern map. It's good, but it sometimes makes the city look a bit too spread out.
• PBEsol: This is a "tuned" map, specifically adjusted for solid materials. It tries to find the perfect balance, making the city look just right.

The researchers tested these three maps on eight different "cities" (materials: NaF, LiF, KF, NaCl, KCl, LiH, NaH, and KH).

2. The Traffic Rules (Scattering)

For the heat messengers to flow like a river (hydrodynamics), they need to follow specific traffic rules:

• Normal Scattering (N): Messengers bump into each other but keep their momentum. They just swap places and keep the flow going. This is the "good" traffic that helps the wave form.
• Resistive Scattering (U & Isotopes): Messengers hit a wall, get stuck, or bounce off a pothole (impurities). This stops the wave and turns the flow back into a chaotic shuffle.

Guyer's Criterion is the rulebook: For the wave to exist, the "Good Traffic" (N) must be stronger than the "Bad Traffic" (U), but the "Bad Traffic" must still be stronger than the "Wall Traffic" (Boundary).

3. The Big Discovery: The Map Matters!

The paper found that which map you choose changes the answer.

• If you use the LDA map, it predicts that the "river flow" (hydrodynamics) happens over a wider temperature range. It's like a map that says, "The traffic jam clears up easily."
• If you use the PBE map, it predicts a narrower window. It's like a map that says, "The traffic jam clears up only for a short time."
• PBEsol usually sits right in the middle.

Why does this matter?
Imagine you are trying to build a bridge. If your map says the river is calm, you build a small bridge. If your map says the river is wild, you build a massive one. If you pick the wrong map, your bridge might collapse. Similarly, if scientists pick the wrong functional, they might miss the chance to observe these heat waves in a lab, or they might waste time looking for them in the wrong temperature range.

4. The New Cities

Before this study, we knew about the "dancing heat" in two famous cities: NaF and LiF.
This paper says: "Wait, look at these other cities!"
They predicted that NaH, LiH, KH, KF, NaCl, and KCl also have this special hydrodynamic window. They are like hidden gems waiting to be discovered.

5. The "Impurity" Problem (Isotopes)

Imagine the city has some potholes (isotopes).

• In NaF, the city is very pure (almost no potholes). The river flows smoothly regardless of the map.
• In LiF, the city has a few potholes (mixed isotopes). If the potholes are too big, the river stops flowing. The paper shows that if you have impurities, the "hydrodynamic window" shrinks or disappears entirely. This explains why some experiments work and others fail—it depends on how pure your sample is.

6. The Takeaway

The authors conclude that while these computer maps (functionals) are great at predicting how hard the material is (mechanical properties) or how it conducts heat at room temperature, they are very sensitive when predicting the "dancing" heat waves.

• LDA tends to be too optimistic (predicts the wave exists too easily).
• PBE is a bit too pessimistic.
• PBEsol is often the most reliable for getting the size of the material right, but for the heat waves, you have to be careful.

In a nutshell:
This paper is a warning to scientists: "Don't just pick a computer setting and run. The setting you choose changes the physics you see. If you want to find these magical heat waves in new materials, you need to check your map carefully, or you might miss the wave entirely."

Technical Summary

Here is a detailed technical summary of the paper "Impact of Exchange-Correlation Functionals on Predictions of Phonon Hydrodynamics: A Study of Fluorides, Chlorides, and Hydrides."

1. Problem Statement

Density Functional Theory (DFT) is the standard tool for computational materials science, yet its accuracy relies heavily on the choice of the Exchange-Correlation (XC) functional. While the impact of XC functionals on electronic properties (like band gaps) is well-documented, their influence on phonon hydrodynamics—a thermal transport regime where heat propagates as a wave (second sound) rather than diffusing—is less understood.

Previous experimental observations of phonon hydrodynamics (second sound) have been limited to a few materials like solid Helium, Bismuth, Graphite, and NaF. There is a lack of systematic first-principles studies predicting this phenomenon in other alkali halides and hydrides, and specifically, how different XC approximations (LDA, PBE, PBEsol) alter the predicted "hydrodynamic window" (the temperature and length scale range where hydrodynamics occurs).

2. Methodology

The authors employed a multi-step computational framework to investigate eight rock-salt structured compounds: NaF, LiF, KF, NaCl, KCl, LiH, NaH, and KH.

• DFT Calculations:
  • Functionals: Three primary XC functionals were compared: Local Density Approximation (LDA), Perdew-Burke-Ernzerhof Generalized Gradient Approximation (PBE), and the modified PBEsol (PBEsol).
  • Software: Quantum ESPRESSO was used for structural relaxation and force constants.
  • Parameters: 14×14×14 k-point meshes, 100 Ry kinetic energy cutoff, and convergence thresholds of 1.0E-16 Ry.
• Phonon Properties:
  • Second-order and third-order interatomic force constants were calculated using thirdorder.py.
  • Phonon dispersion relations and densities of states (PDOS) were generated.
• Thermal Transport Modeling:
  • The Phonon Boltzmann Transport Equation (BTE) was solved iteratively using the ShengBTE package.
  • Lattice thermal conductivity ($\kappa$) was computed from 100 K to 1000 K.
  • Scattering Rates: Normal (N), Umklapp (U), Boundary (B), and Isotope (R) scattering rates were calculated.
• Hydrodynamic Criterion:
  • The study applied Guyer's Criterion to determine the hydrodynamic regime. This condition requires that the average Normal scattering rate dominates over resistive scattering (Umklapp + Isotope) and boundary scattering:
    $$\langle \tau_N^{-1} \rangle > \langle \tau_B^{-1} \rangle > \langle \tau_U^{-1} \rangle$$
  • The "hydrodynamic window" was mapped as a function of temperature and characteristic length scale ($L$).
• Validation:
  • Results were benchmarked against experimental data for lattice constants, elastic moduli, dielectric constants, and thermal conductivity.
  • Supplementary tests included Meta-GGA (SCAN) and Hybrid (HSE) functionals to assess band gap accuracy vs. structural accuracy.

3. Key Contributions

1. Systematic XC Functional Comparison: The study provides a rigorous comparison of how LDA, PBE, and PBEsol affect not just static properties, but dynamic thermal transport regimes.
2. Novel Predictions: The authors report the first theoretical predictions of phonon hydrodynamics in NaH, LiH, KH, KF, NaCl, and KCl, expanding the known material space for second sound beyond NaF and LiF.
3. Isotope Scattering Impact: The work quantifies the critical role of natural isotope abundance in suppressing hydrodynamic windows, particularly in LiF and LiH, explaining why high-purity samples are required for experimental observation.
4. Functional Trade-offs: The paper highlights that while LDA often overestimates thermal conductivity due to overbinding (stiffer lattices), it may predict wider hydrodynamic windows. Conversely, PBE often underestimates conductivity but may narrow the hydrodynamic window.

4. Key Results

A. Structural and Mechanical Properties

• Lattice Constants: PBEsol generally provided the closest agreement with experimental lattice constants for fluorides and chlorides (deviations <1%). PBE performed better for hydrides (e.g., LiH).
• Elastic Moduli: LDA consistently predicted the highest bulk moduli (due to overbinding), followed by PBEsol, then PBE. All materials were found to be mechanically stable and brittle (negative Cauchy pressure).
• Band Gaps: As expected, standard functionals significantly underestimated electronic band gaps (e.g., LiF calculated ~8.8 eV vs. experimental ~14 eV). Hybrid (HSE) and mBJ functionals improved band gaps but were less accurate for structural parameters.

B. Thermal Conductivity ($\kappa$)

• Trends: $\kappa$ generally decreased with increasing temperature.
• Functional Dependence:
  • LDA typically yielded the highest $\kappa$ (e.g., NaF: 31.5 W/mK vs. PBE: 18.5 W/mK) because the stiffer lattice increases phonon group velocities and lifetimes.
  • PBE yielded the lowest $\kappa$.
  • PBEsol often provided the best match to experimental room-temperature $\kappa$ values (e.g., KCl: 6.90 W/mK vs. exp 6.70–7.10 W/mK).
• Isotope Effect: Including natural isotope scattering significantly reduced $\kappa$ at low temperatures for materials with mixed isotopes (LiF, LiH) but had negligible effect on monoisotopic materials (NaF, NaCl).

C. Phonon Hydrodynamics Windows

• Regime Mapping: The study mapped ballistic, hydrodynamic, and diffusive regimes for all eight materials across lengths from 1 nm to 100 $\mu$m and temperatures up to 80 K.
• Functional Impact on Windows:
  • LDA generally predicted the widest hydrodynamic windows (highest temperatures) because its stiffer phonon spectrum enhances Normal scattering relative to resistive scattering.
  • PBE predicted narrower windows.
  • LiF exhibited the broadest hydrodynamic window among all studied materials.
  • KCl and NaH showed the narrowest windows.
• Isotope Suppression: For LiF, including isotope scattering (R-scattering) narrowed the hydrodynamic window at $L=1$ mm from ~24.5 K (U-scattering only) to ~9 K. This aligns with experimental difficulties in observing second sound in natural LiF compared to isotopically enriched samples.
• Second Sound Confirmation: Direct solutions to the linearized BTE confirmed oscillatory temperature responses (second sound) in LiF, NaF, KF, KH, and NaH at low temperatures (e.g., 15 K).

5. Significance

• Guidance for Experimentalists: The paper provides a roadmap for experimentalists seeking to observe phonon hydrodynamics in new materials. It suggests that LiH, NaH, and KH are promising candidates, provided high-purity (isotope-controlled) samples are used to minimize resistive scattering.
• Computational Best Practices: It demonstrates that the choice of XC functional is not merely a detail but a critical parameter that can qualitatively change the prediction of transport regimes. While LDA may be "wrong" for band gaps, it can be "useful" for identifying hydrodynamic windows due to its tendency to overestimate phonon stiffness.
• Material Design: The findings suggest that tuning isotopic composition is a viable strategy to engineer hydrodynamic thermal transport in alkali hydrides and halides, potentially useful for thermal management in microelectronics where wave-like heat transport could be advantageous.

In conclusion, this work establishes that while DFT functionals vary in their prediction of absolute thermal conductivity values, they consistently identify the potential for phonon hydrodynamics in a wide range of alkali halides and hydrides, with the specific "window" of observation being highly sensitive to the chosen functional and isotopic purity.