Search for thermodynamically stable ambient-pressure superconducting hydrides in GNoME database

By employing a multi-stage approach combining machine learning and ab initio methods to screen the GNoME database, researchers identified 25 thermodynamically stable cubic hydrides at ambient pressure with superconducting critical temperatures up to 17 K, offering experimentally accessible candidates for high-temperature superconductivity despite their modest $T_c$ values.

The Gist

Imagine you are looking for a magical material that can conduct electricity with zero resistance (superconductivity) without needing to be frozen in liquid helium or crushed by the weight of a mountain. This is the "Holy Grail" of physics: room-temperature, ambient-pressure superconductivity.

For years, scientists found that certain "hydrides" (compounds rich in hydrogen) could do this, but only under extreme pressure—like being buried deep inside the Earth's core. That's great for theory, but useless for building a power grid or a maglev train.

This paper is like a massive, high-tech treasure hunt. The team used a super-smart computer database called GNoME (which contains thousands of crystal structures that are stable at room temperature) to find the "needle in the haystack."

Here is the story of their search, explained simply:

1. The Problem: The "Unstable" Trap

In the past, computer models predicted some hydrides could superconduct at very high temperatures (like 100 K, or -173°C) at normal pressure. But there was a catch: they were thermodynamically unstable.

Think of it like building a house of cards. You can arrange the cards to look like a castle that reaches the sky (high superconductivity), but the moment you touch it, it collapses. Nature hates these unstable structures. They would instantly turn into something else (like a powder or a different crystal) before you could even test them.

2. The Strategy: The "Two-Layer" Filter

The researchers didn't just guess; they used a clever, two-step filtering process to find materials that are both stable (won't collapse) and superconducting.

• Step 1: The Machine Learning Scout (The Fast Filter)
  They used an AI model called ALIGNN (think of it as a super-fast scout with a magnifying glass). It scanned thousands of stable hydrides from the GNoME database. It didn't calculate everything perfectly; it just made a quick guess: "Hey, this one looks promising!"

  • Analogy: Imagine a talent show. The scout is the producer who quickly scans the audience and picks out 50 people who might have a good voice, ignoring the rest.

• Step 2: The Physics Expert (The Deep Dive)
  For the promising candidates, they used heavy-duty, "first-principles" physics calculations (like DFT and DFPT). This is the rigorous, slow, and expensive method that gives the real answer.

  • Analogy: The 50 people picked by the scout now have to sing live on stage in front of a panel of strict judges. Only the ones who actually sing well get the contract.

3. The Discovery: 25 New Contenders

The search was successful! They found 25 cubic hydrides that are stable at normal pressure and can superconduct.

• The "Good" News: They found materials that work at temperatures above 4.2 K (the boiling point of liquid helium). This is a huge deal because it means they are stable and real.
• The "Best" News: One material, LiZrH6Ru (a mix of Lithium, Zirconium, Hydrogen, and Ruthenium), stood out.
  • It has a unique "vacancy-ordered double perovskite" structure. Imagine a Lego castle where some bricks are missing in a very specific, organized pattern. This pattern is key to its superconducting ability.
  • Its predicted superconducting temperature is 17 K (about -256°C).

4. The Reality Check: Why 17 K?

You might be thinking, "17 K? That's still super cold! Why is this a big deal?"

Here is the catch: Stability vs. Performance.

• The unstable "house of cards" materials predicted 100 K, but they don't exist in reality.
• These new stable materials only reach 17 K.

The Analogy:
Imagine you are looking for a car.

• Option A: A Ferrari that goes 200 mph but falls apart if you drive it on a bumpy road (Unstable, high Tc).
• Option B: A reliable Toyota that goes 60 mph and never breaks down (Stable, lower Tc).

This paper found the Toyota. It's not the fastest car in the world, but it's the first one that actually exists and can be built. The fact that it works at all at normal pressure is a massive scientific breakthrough.

5. The Deep Dive on the Winner (LiZrH6Ru)

The authors didn't just stop at the basic calculation for their winner. They dug deeper to make sure the math was right.

• They checked for "magnetic interference" (like static on a radio) that could ruin the superconductivity.
• They checked if the electrons were dancing in multiple bands (like a choir singing in harmony vs. solo).
• The Result: Even with all these complex corrections, the material still holds up at 17 K. This gives scientists high confidence that if they build this in a lab, it will actually work.

The Bottom Line

This paper is a roadmap. It tells us: "Stop looking for unstable, high-speed Ferraris that fall apart. Look here, in these stable, organized structures."

While 17 K isn't "room temperature" yet, finding a stable material that works at all is the first step toward eventually engineering one that works at room temperature. It proves that stable, ambient-pressure superconductors are real, and we just need to keep refining them.

Technical Summary

1. Problem Statement

The discovery of high-temperature superconductivity (HTS) in hydrides (e.g., LaH$_{10}$, YH$_9$) has been a major breakthrough, but these materials require extreme pressures (>170 GPa) to stabilize, rendering them impractical for real-world applications. While theoretical predictions have suggested ambient-pressure hydrides with critical temperatures ($T_c$) near 100 K, these candidates often suffer from thermodynamic instability (lying far above the convex hull), making their synthesis difficult or impossible. Conversely, hydrides that are thermodynamically stable at 0 K are typically semiconductors or insulators with no superconductivity.

The central challenge addressed in this work is to identify thermodynamically stable hydrides at ambient pressure that exhibit superconductivity, utilizing the recently released GNoME (Graph Networks for Materials Exploration) database, which contains thousands of 0 K stable crystals.

2. Methodology

The authors employed a multi-stage, machine-learning-accelerated high-throughput screening approach to navigate the vast material space of the GNoME database:

• Initial Filtering: From the GNoME database, they selected 490 cubic, metallic, thermodynamically stable hydrides at 0 K.
• Magnetic Screening: High-throughput spin-polarized Density Functional Theory (DFT) calculations were performed to filter out magnetic materials, leaving 261 non-magnetic metallic hydrides.
• Machine Learning Pre-screening: The Atomistic Line Graph Neural Network (ALIGNN) model was used to predict the superconducting critical temperature ($T_c$). Only candidates with a predicted $T_c \geq 1$ K were retained to reduce computational cost.
• High-Level Ab Initio Validation: The remaining candidates underwent rigorous Density Functional Perturbation Theory (DFPT) calculations coupled with the Allen-Dynes formula to calculate the electron-phonon coupling constant ($\lambda$), logarithmic average phonon frequency ($\omega_{log}$), and $T_c$.
• Advanced Refinement (Case Study): For the top candidate, LiZrH$_6$Ru, the authors performed advanced calculations beyond the standard McMillan-Allen-Dynes approximation:
  • Stochastic Self-Consistent Harmonic Approximation (SSCHA): To account for ionic quantum and anharmonic effects.
  • Ab Initio Eliashberg Theory: Solving the Eliashberg equations including Coulomb interactions (RPA) and spin susceptibility (Kukkonen-Overhauser approach).
  • Multiband Analysis: Investigating anisotropic superconductivity effects using Superconducting Density Functional Theory (SCDFT).

3. Key Contributions

• Systematic Search of GNoME: This is the first comprehensive study to screen the GNoME database specifically for ambient-pressure superconducting hydrides, bridging the gap between thermodynamic stability and superconducting properties.
• Validation of ML Models: The study demonstrates that the ALIGNN model is highly effective for pre-screening, showing a Mean Absolute Error (MAE) of only 2.5 K when compared to high-level DFPT calculations.
• Discovery of Stable Candidates: The identification of 25 cubic hydrides with $T_c > 4.2$ K (the boiling point of liquid helium) that are thermodynamically stable at 0 K.
• Rigorous Characterization of LiZrH$_6$Ru: The paper provides a deep dive into the most promising candidate, moving beyond standard approximations to provide a reliable $T_c$ estimate that accounts for Coulomb repulsion and multiband effects.

4. Key Results

• Top Candidates: The screening identified 25 cubic hydrides. The highest $T_c$ candidates belong to two structural families:
  1. Vacancy-ordered double perovskites (e.g., EuDyReH$_6$, LiZrH$_6$Ru).
  2. Fluorite-like structures (e.g., Ta$_3$NbH$_8$, Ta$_6$NbH$_{16}$).
• Performance Metrics:
  • Most identified hydrides have $T_c$ values between 4.2 K and 10 K.
  • The standout material is LiZrH$_6$Ru (vacancy-ordered double perovskite).
• LiZrH$_6$Ru Analysis:
  • High-Throughput DFPT: Initially predicted $T_c \approx 23.5$ K.
  • Refined Calculations: After including anharmonicity (SSCHA), Coulomb repulsion (RPA), and spin fluctuations, the estimated $T_c$ was refined to 17 K.
  • Mechanism: Unlike high-$T_c$ hydrides where hydrogen states dominate the Fermi level, LiZrH$_6$Ru relies on transition metal states (Zr and Ru) at the Fermi level. The hydrogen contribution is negligible.
  • Pressure Effects: While the material is stable at ambient pressure, applying pressure (up to 80 GPa) increases $T_c$ to ~44 K due to enhanced density of states (DOS) and electron-phonon coupling, though the ambient-pressure stability is the primary novelty.
• Comparison: While 17 K is modest compared to the 100 K+ predictions for unstable hydrides, it is significant because these materials are thermodynamically stable, suggesting they are experimentally accessible.

5. Significance

• Experimental Accessibility: The primary value of this work is the focus on thermodynamic stability. Previous high-$T_c$ predictions often failed because the materials could not be synthesized. These 25 candidates, particularly LiZrH$_6$Ru, offer a realistic pathway for experimental verification.
• Technological Relevance: A $T_c$ of 17 K is above the boiling point of liquid helium (4.2 K), making these materials potentially useful for cryogenic applications without the need for extreme pressure equipment.
• Methodological Benchmark: The study validates the workflow of combining machine learning (ALIGNN) with high-level ab initio methods (DFPT/Eliashberg) as a robust strategy for materials discovery, providing a template for future searches in large databases.
• Insight into Stability vs. $T_c$: The results reinforce the hypothesis that there is a trade-off between thermodynamic stability and ultra-high $T_c$ in hydrides, likely due to the destabilizing effect of delocalized hydrogen electrons in metallic bonding environments.

In conclusion, this paper successfully shifts the paradigm from searching for the highest possible $T_c$ (often in unstable phases) to finding realizable superconductors, identifying LiZrH$_6$Ru as a prime candidate for experimental synthesis and verification.