Capturing nuclear quantum effects in high-pressure superconducting hydrides and ice with nuclear-electronic orbital theory

This paper demonstrates that the nuclear-electronic orbital density functional theory (NEO-DFT) method accurately and efficiently captures essential nuclear quantum effects to predict the structures and phase transition pressures of high-pressure superconducting hydrides and ice, offering a scalable alternative to more expensive computational approaches.

The Gist

Imagine you are trying to build a house out of tiny, vibrating marbles. In the world of physics, these marbles are atoms. For most heavy atoms, like lead or gold, they sit pretty still, and we can predict exactly where they will be. But for hydrogen—the lightest, most energetic atom in the universe—it's a different story. Hydrogen is so light that it doesn't just sit there; it vibrates wildly, wiggles, and even "tunnels" through walls, behaving more like a fuzzy cloud of probability than a solid marble.

This paper is about a new, smarter way to predict how these wiggly hydrogen atoms behave when you squeeze them incredibly hard, like deep inside a planet or in a super-conducting material.

The Problem: The "Heavy" Mistake

For a long time, scientists used a standard computer program (called DFT) to model these materials. Think of this program like a camera that takes a picture of a hummingbird. If you use a slow shutter speed, the bird looks like a blur. But if you treat the bird as a solid, stationary statue, you get the picture completely wrong.

Standard programs treat hydrogen atoms as if they were heavy, stationary statues. They ignore the "fuzziness" and the quantum wiggles. Because of this, when scientists tried to predict at what pressure hydrogen turns into a superconductor (a material that conducts electricity with zero resistance) or when ice changes its structure, their predictions were often way off. They were trying to build a house based on a blueprint of a statue, not a vibrating cloud.

The Old Solution: The "Super-Computer" Brute Force

To fix this, scientists developed a method called SSCHA. Imagine trying to figure out where the hummingbird is by taking 1,000 photos a second, averaging them all out, and then doing the math for every single photo. It's incredibly accurate, but it's also exhausting. It requires massive supercomputers and takes days or weeks to run a single simulation. It's like trying to solve a Rubik's cube by trying every single possible move combination one by one.

The New Solution: NEO-DFT (The "Smart Lens")

The authors of this paper are introducing a new method called NEO-DFT.

Think of NEO-DFT as a smart lens that sees the hummingbird while it's flying. Instead of treating the hydrogen atom as a heavy statue, this method treats the hydrogen nucleus (the proton) with the same respect as the electrons. It acknowledges that the proton is a quantum wave, not a solid ball.

Here is the magic trick:

1. It's built-in: Unlike the old method that requires a separate, expensive calculation after the main one, NEO-DFT includes the quantum wiggles during the main calculation.
2. It's fast: Because it's smarter, it doesn't need to take 1,000 photos. It gets the answer in one go. The authors found it to be 100 times faster than the old "brute force" method for similar accuracy.

What Did They Test?

The team tested this new "smart lens" on three famous high-pressure materials:

1. H3S (Superconducting Hydride): This is a material that becomes a superconductor at very high pressures. The old methods said it would change its shape at a certain pressure, but they were wrong. NEO-DFT predicted the exact pressure where the hydrogen atoms "symmetrize" (become perfectly centered), matching real-world experiments perfectly.
2. LaH10 (Lanthanum Hydride): This is an even bigger, more complex molecule. The old methods got confused by the many possible shapes this molecule could take. NEO-DFT cut through the noise and correctly identified the stable, symmetric shape that scientists see in experiments.
3. Ice (Water under pressure): When you squeeze ice hard enough, it changes from "Ice VIII" to "Ice X." In Ice X, the hydrogen atoms sit perfectly in the middle between oxygen atoms. The old methods predicted this would happen at a pressure much higher than reality. NEO-DFT predicted the exact pressure (around 62 GPa for regular water and 71 GPa for "heavy water" or D2O), matching experimental data.

Why Does This Matter?

Imagine you are an architect designing a bridge. If your computer model ignores the wind, your bridge might collapse. Similarly, if we want to design new materials for clean energy, better batteries, or room-temperature superconductors, we need to know exactly how hydrogen behaves under pressure.

Before this paper, getting that right answer was like trying to count every grain of sand on a beach (expensive and slow). Now, with NEO-DFT, it's like having a satellite image that shows you the whole beach instantly.

In short: This paper gives scientists a fast, accurate, and easy-to-use tool to understand the "quantum dance" of hydrogen. This opens the door to discovering new materials that could revolutionize how we store energy and transmit electricity, all without needing to wait months for a supercomputer to finish its homework.

Technical Summary

1. Problem Statement

The accurate prediction of crystal structures and phase transitions in hydrogen-rich materials under high pressure is hindered by the inadequate treatment of Nuclear Quantum Effects (NQEs) and lattice anharmonicity in standard Density Functional Theory (DFT).

• The Challenge: Hydrogen atoms have low mass, leading to significant quantum fluctuations, delocalization, and tunneling. Standard DFT treats nuclei classically, failing to capture these effects. This results in discrepancies between theoretical predictions and experimental observations, particularly regarding:
  • Hydrogen-bond symmetrization: The transition from asymmetric to symmetric hydrogen bonds (e.g., in $H_3S$ and $LaH_{10}$).
  • Phase transitions: Specifically the transition from Ice VIII to Ice X.
  • Superconductivity: NQEs dictate the correct geometry and phonon properties, which directly influence the critical temperature ($T_c$).
• Limitations of Existing Methods:
  • Path-Integral Molecular Dynamics (PIMD) & Quantum Monte Carlo (QMC): Rigorous but computationally prohibitive for large systems.
  • Stochastic Self-Consistent Harmonic Approximation (SSCHA): Considered the "gold standard" for NQEs but requires expensive supercell calculations and ensemble averaging, making it too costly for high-throughput screening or large complex systems.
  • Machine Learning Potentials: Reduce cost but often lack transferability across wide pressure/composition ranges and require extensive training.

2. Methodology: Periodic NEO-DFT

The authors utilize Periodic Nuclear–Electronic Orbital Density Functional Theory (NEO-DFT) to address these challenges.

• Core Concept: In NEO-DFT, specific nuclei (protons or deuterons) are treated quantum mechanically on the same footing as electrons.
• Formalism:
  • The wavefunction is a product of an electronic determinant and a protonic (or deuteron) determinant.
  • The method solves coupled electronic and protonic Kohn–Sham equations self-consistently.
  • It includes an electron–proton correlation functional ($E_{epc}$), specifically the epc17-2 functional, to account for interactions between electrons and quantum nuclei.
• Implementation:
  • Calculations were performed using the FHI-aims code.
  • Basis Sets: Numerically tabulated atom-centered orbitals (NAOs) for electrons and Gaussian-type orbitals (GTOs) for quantum protons/deuterons.
  • Efficiency: Unlike SSCHA, which requires supercells and ensembles of nuclear configurations, NEO-DFT performs a single structural optimization using a primitive unit cell. NQEs (delocalization, anharmonic zero-point energy, tunneling) are inherently included in the geometry optimization without post-hoc corrections.
  • Isotope Effects: The method naturally distinguishes between Hydrogen ($H$) and Deuterium ($D$) by scaling basis set exponents by $\sqrt{2}$ to account for mass differences.

3. Key Contributions

• Validation of NEO-DFT: Demonstrated that periodic NEO-DFT accurately predicts hydrogen-bond symmetrization pressures and phase transitions in high-pressure systems, matching experimental data and high-level SSCHA results.
• Computational Efficiency: Established that NEO-DFT offers a massive reduction in computational cost (over 100-fold for $H_3S$) compared to SSCHA while maintaining comparable accuracy.
• Systematic Investigation: Showed that the inclusion of the electron–proton correlation functional ($epc17-2$) is critical for accuracy, particularly in water systems, and allows for systematic tuning of correlation effects.
• Broad Applicability: Successfully applied the method to diverse systems: superconducting hydrides ($H_3S$, $LaH_{10}$) and high-pressure ice phases ($H_2O$, $D_2O$).

4. Key Results

A. Superconducting Hydrides ($H_3S$ and $LaH_{10}$)

• $H_3S$ (Hydrogen Sulfide):
  • Symmetrization Pressure: Conventional DFT overestimates the pressure required for the transition from the asymmetric $R\bar{3}m$ phase to the symmetric $Im\bar{3}m$ phase.
  • NEO-DFT Prediction: Predicted 118 GPa for $H_3S$ and 130 GPa for $D_3S$.
  • Comparison: These values are in qualitative agreement with SSCHA results ($103$ GPa for $H_3S$, $115$ GPa for $D_3S$) and explain the experimental high-$T_c$ regime (observed at 155 GPa) where the symmetric phase dominates.
  • Cost: Achieved this with a primitive unit cell (1 S, 3 H) vs. SSCHA's supercell (24 S, 72 H), resulting in a >100x speedup.
• $LaH_{10}$ (Lanthanum Decahydride):
  • Complex Energy Landscape: Conventional DFT predicts multiple local minima (asymmetric phases). SSCHA collapses these into the symmetric $Fm\bar{3}m$ phase.
  • NEO-DFT Outcome: Starting from asymmetric $R\bar{3}m$ structures at various pressures (125–200 GPa), NEO-DFT spontaneously relaxed into the symmetric $Fm\bar{3}m$ phase, consistent with SSCHA and experimental attribution of the high-$T_c$ phase.
  • Efficiency: The NEO-DFT optimization for $LaH_{10}$ was faster than the SSCHA optimization for the much smaller $H_3S$ system.

B. Ice Phase Transition (Ice VIII to Ice X)

• Transition Mechanism: The transition involves the symmetrization of hydrogen bonds, where the potential energy surface changes from a double-well to a single-well, localizing hydrogen at the midpoint between oxygens.
• Pressure Predictions:
  • Conventional DFT: Predicted a transition at 110 GPa (far too high).
  • NEO-DFT ($H_2O$): Predicted 62 GPa, matching the experimental range (58–62 GPa).
  • NEO-DFT ($D_2O$): Predicted 71 GPa, matching the experimental value of 72 GPa.
  • Comparison: The NEO-DFT result for $H_2O$ is slightly higher than the SSCHA result (51 GPa) but agrees well with experiment. The authors note that removing the $epc17-2$ functional in NEO-DFT brings the prediction closer to SSCHA (50 GPa), suggesting electron-proton correlation treatment is a key variable.

5. Significance and Conclusion

• Paradigm Shift: This work demonstrates that NEO-DFT is a viable, efficient alternative to SSCHA for incorporating NQEs in structural optimization. It removes the need for expensive supercells and ensemble averaging.
• Scalability: The method scales similarly to conventional DFT, making it applicable to larger, more complex systems (like $LaH_{10}$) that are currently intractable for SSCHA.
• Future Impact: The computational savings enable the use of hybrid functionals and facilitate the high-throughput discovery of new high-pressure superconductors and materials with specific geometrical properties.
• Limitations: Current NEO-DFT implementations in this study were performed at 0 K; finite-temperature effects are not yet included, though work is ongoing.

In summary, the paper establishes Periodic NEO-DFT as a robust, accurate, and computationally efficient tool for predicting high-pressure phase transitions in hydrogen-rich materials, successfully capturing nuclear quantum effects that standard DFT misses and expensive methods like SSCHA struggle to compute efficiently.