.png){kind=icon}
.png){kind=icon}
Return
Understanding Tc variation in isostructural hydrides: Interpretable machine learning with physical descriptors
.png){kind=icon}
PRE
AI
S
J
B
J
M
Z
Y
Y
S
DOI：10.1016/j.physb.2026.418748.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Prediction of superconducting Tc and identifying key physical descriptors for rational design of high-temperature superconductors. • Develops an interpretable ML framework integrating electronic, compositional, and crystallographic descriptors from CIFs. • Adopts SHAP analysis to quantitatively rank descriptor contributions, bridging statistical learning and physical mechanisms. • Validates on isostructural hydrides Mg2IrH6/Ca2IrH6, accurately predicting their contrasting superconducting behaviors. • Reveals hydrogen fraction as the decisive driver for Mg2IrH6’s high Tc (83.4 K), correlating with strong electron–phonon coupling. • Demonstrates interpretable ML can accelerate superconductor discovery via accurate predictions and deep structural insights.
Keywords:
superconducting Tc
physical descriptors
interpretable machine learning
electron–phonon coupling
hydrides
Journal
P
IF:
0
Papers:
544
Citations:
1
Organization
Cited Papers