RoomCheck: Machine Learning-Based Room Temperature Superconductor Candidate Screening System

Research Background

Room temperature superconductor is the holy grail of materials science. Despite significant progress in high-pressure superconductors (e.g., H3S at 250K under 150GPa, LaH10 at 250K under 170GPa), achieving room temperature superconductivity at ambient pressure remains a huge challenge.

Method Overview

RoomCheck is a machine learning-based screening system designed to identify candidate materials with potential room temperature superconductivity. The system combines:

1. High-throughput first-principles calculation data

2. Graph Neural Network (GNN) prediction models

3. Superconductivity descriptor library

4. Candidate material ranking system

Technical Route

1. Data Collection

• Collect known superconductor data from Materials Project and AFLOW databases

• Compile high-pressure superconductor experimental data (including H3S, LaH10, YH9, etc.)

• Gather electronic structure, phonon spectrum, crystal structure features

2. Feature Engineering

• Electronic Density of States (DOS) features

• Phonon dispersion relation features

• Crystal symmetry and chemical bond features

• Electron-phonon coupling strength estimation

3. Model Training

Use Graph Neural Network (GNN) model to predict superconducting transition temperature (Tc):

import torch
from torch_geometric.nn import GCNConv

class SuperconductivityPredictor(torch.nn.Module):
    def __init__(self, num_features=64, hidden_dim=128):
        super().__init__()
        self.conv1 = GCNConv(num_features, hidden_dim)
        self.conv2 = GCNConv(hidden_dim, hidden_dim)
        self.fc = torch.nn.Linear(hidden_dim, 1)
    
    def forward(self, data):
        x, edge_index = data.x, data.edge_index
        x = torch.relu(self.conv1(x, edge_index))
        x = torch.relu(self.conv2(x, edge_index))
        x = self.fc(x.mean(dim=0))
        return x

4. Candidate Material Screening

Screening criteria:

• Predicted Tc > 300K (above room temperature)

• Structural stability (no imaginary frequencies in phonon spectrum)

• Reasonable synthesis conditions

• Sufficient electron-phonon coupling strength

Preliminary Results

We have identified the following candidate material categories:

1. Hydrogen-Rich Compounds

• Carbon-sulfur-hydrogen compounds (C-S-H system)

• Boron-hydrogen compounds (B-H system)

• Beryllium-hydrogen compounds (Be-H system)

2. Two-Dimensional Materials

• Graphene-based superconductors

• Transition metal dichalcogenides

• Perovskite-type 2D materials

3. Heavy Fermion Systems

• Cerium-based heavy fermion compounds

• Europium-based heavy fermion compounds

Verification Methods

For screened candidate materials, we will use the following methods for verification:

1. Density Functional Theory (DFT) for electronic structure calculations

2. Density Functional Perturbation Theory (DFPT) for phonon spectra and electron-phonon coupling

3. Eliashberg equation for superconducting transition temperature

4. Molecular dynamics simulation for structural stability verification

Next Steps

1. Complete model training and validation

2. Perform first-principles calculations on screened candidate materials

3. Collaborate with experimental teams for synthesis and measurement

4. Optimize candidate material composition and structure

Expected Outcomes

• An open room temperature superconductor screening database

• A validated machine learning prediction model

• A list of promising candidate materials

• Related papers and open-source code

References

1. Drozdov, A. P., et al. Superconductivity at 250 K in lanthanum hydride under high pressures. Nature 569, 528-531 (2019).

2. Somayazulu, M., et al. Superconductivity at 253 K in lanthanum-yttrium compound at high pressure. arXiv:2010.07608 (2020).

3. Snider, E., et al. Room-temperature superconductivity in a carbonaceous sulfur hydride. Nature 586, 373-377 (2020).