Superconductor research has traditionally depended on experiments and theoretical approaches. However, the rapid advancement of data-driven methods and machine learning (ML) has opened avenues for accelerating superconductor discovery. Here, we integrate ML with density functional theory calculations to efficiently screen conventional B-C-N based superconductors and identify potential high-$T_{\mathrm{C}}$ candidates among $R_3{\mathrm{Ni}}_2{\mathrm{O}}_7$-type bilayer nickelates. We identify 13 binary and ternary B-C-N based superconductors with $T_{\mathrm{C}}\ge 10$ K, including 3 with $T_{\mathrm{C}}\ge 25$ K, two structural forms of ${\mathrm{B}}_2\mathrm{CN}$ ($T_{\mathrm{C}}=47.6$ and 43.7 K) and ${\mathrm{TiNbN}}_2$ $(T_{\mathrm{C}}=25.3 \mathrm{K})$. These B-C-N based compounds share a common feature of strong σ bonds, which is key to achieving relatively high $T_{\mathrm{C}}$. Moreover, we propose ${\mathrm{Tb}}_3{\mathrm{Ni}}_2{\mathrm{O}}_7$ $(T_{\mathrm{C}}=61.6 \mathrm{K})$ and ${\mathrm{Ac}}_3{\mathrm{Ni}}_2{\mathrm{O}}_7$ $(T_{\mathrm{C}}=70.3 \mathrm{K})$ as potential high-$T_{\mathrm{C}}$ nickelate superconductors under high pressure. Their electronic structures closely resemble those of ${\mathrm{La}}_3{\mathrm{Ni}}_2{\mathrm{O}}_7$, especially in the hole-type band dominated by Ni $3d_z^2$ orbital character. We also analyze feature importance in the ML results for both conventional and high-$T_{\mathrm{C}}$ superconductors. These results advance the search for new superconductors and enhance the fundamental understanding of superconducting mechanisms.

Close

PrevNext

View figure in article

• Electron-phonon coupling
• Superconductivity
• High-temperature superconductors
• Superconductors
• Density functional theory
• Machine learning