pith. sign in

arxiv: 2606.21251 · v1 · pith:4WOGUH2Knew · submitted 2026-06-19 · ⚛️ physics.comp-ph · cond-mat.supr-con

AI-accelerated metallized σ-bonding screening for superconductor discovery

Zechen Tang , Wen-Han Dong , Baochun Wu , Jian-Feng Zhang , Yuxiang Wang , Yang Li , Honggeng Tao , Qiyu Zeng
show 6 more authors
Chong Wang Chen Si Zhong-Yi Lu Wenhui Duan Tao Xiang Yong Xu
This is my paper

Pith reviewed 2026-06-26 12:46 UTC · model grok-4.3

classification ⚛️ physics.comp-ph cond-mat.supr-con
keywords superconductor discoveryσDOS descriptormetallized σ-bondingB13SeAI-accelerated screeningDFT electronic structurephonon-mediated superconductorshigh-Tc candidates
0
0 comments X

The pith

The σDOS descriptor, accelerated by deep learning, identifies B13Se from two million materials as an ambient-pressure superconductor candidate with Tc above 40 K.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper develops an efficient way to search for phonon-mediated superconductors by introducing the σ-bonding density of states as a descriptor drawn from the metallized σ-bonding picture. The descriptor works from ordinary DFT electronic structure results and skips the usual expensive phonon calculations. A deep-learning model further speeds up the electronic structure step so that two million materials can be screened. The search turns up B13Se, predicted to have a transition temperature above 40 K at ambient pressure, plus a broader family of B13X candidates. The approach therefore makes large-scale computational discovery of new superconductors feasible.

Core claim

Guided by the metallized σ-bonding picture, the σ-bonding density of states (σDOS) serves as an efficient physical descriptor that identifies high-Tc superconductors directly from DFT-level electronic structure without explicit DFPT phonon calculations. Evaluation of σDOS is accelerated by a deep-learning DFT Hamiltonian method, enabling screening of two million materials. This process identifies B13Se as an ambient-pressure superconductor candidate with predicted Tc > 40 K, together with a family of high-Tc B13X candidates.

What carries the argument

The σ-bonding density of states (σDOS) descriptor, which quantifies electronic states associated with metallized σ-bonds to flag materials likely to show high superconducting transition temperatures from DFT data alone.

If this is right

  • B13Se emerges as a concrete ambient-pressure superconductor candidate with predicted Tc > 40 K.
  • A family of B13X compounds are flagged as additional high-Tc candidates.
  • The σDOS descriptor bypasses DFPT, making large-scale screening computationally tractable.
  • Deep-learning acceleration of the DFT Hamiltonian enables screening of millions of materials in practice.
  • The overall strategy combines a physics-based descriptor with AI to support efficient superconductor discovery.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the σ-bonding picture generalizes, the same descriptor could be applied to other material families beyond B13X without retraining the full workflow.
  • Pairing σDOS screening with targeted synthesis efforts could shorten the time from computation to experimental test.
  • Checking whether σDOS correlates with measured Tc across a wider set of known superconductors would test the descriptor's range.
  • Lowering the cost of screening could make superconductor searches practical on modest computing hardware.

Load-bearing premise

The metallized σ-bonding picture supplies a reliable basis for defining σDOS as a descriptor that identifies high-Tc superconductors directly from DFT electronic structure without explicit DFPT phonon calculations.

What would settle it

Experimental synthesis and measurement of the superconducting transition temperature of B13Se at ambient pressure; a value well below 40 K or absence of superconductivity would falsify the predictive power of the σDOS descriptor.

Figures

Figures reproduced from arXiv: 2606.21251 by Baochun Wu, Chen Si, Chong Wang, Honggeng Tao, Jian-Feng Zhang, Qiyu Zeng, Tao Xiang, Wen-Han Dong, Wenhui Duan, Yang Li, Yong Xu, Yuxiang Wang, Zechen Tang, Zhong-Yi Lu.

Figure 1
Figure 1. Figure 1: FIG. 1. Workflows for searching phonon-mediated superconductors. Traditional approaches rely on density functional theory [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. (a) [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. (a,c) Atomic structures and (d) band structures of undistorted and distorted B [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4. (a) Superconducting properties of the B [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
read the original abstract

The computational discovery of phonon-mediated superconductors is hindered by the prohibitive cost of density functional perturbation theory (DFPT). Here, guided by the metallized $\sigma$-bonding picture, we introduce the $\sigma$-bonding density of states ($\sigma$DOS) as an efficient physical descriptor to identify high-transition-temperature ($T_{\mathrm{c}}$) superconductors from density functional theory (DFT)-level electronic structure without explicit DFPT calculations. The evaluation of $\sigma$DOS can be further accelerated by a deep-learning DFT Hamiltonian method, enabling efficient large-scale screening for superconductors. Screening 2 million materials, we identify B$_{13}$Se as an ambient-pressure superconductor candidate with predicted $T_{\mathrm{c}} > 40$~K, together with a family of high-$T_{\mathrm{c}}$ B$_{13}X$ candidates, supporting the effectiveness of this discovery strategy. By bridging physics priors with AI acceleration, this study delivers an efficient and generalizable route for computational materials discovery in the AI era.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The paper introduces the σ-bonding density of states (σDOS) as a descriptor derived from the metallized σ-bonding picture to identify high-Tc phonon-mediated superconductors directly from DFT electronic structure without explicit DFPT phonon calculations. The evaluation of σDOS is accelerated via a deep-learning DFT Hamiltonian method, enabling screening of 2 million materials. The authors identify B13Se as an ambient-pressure candidate with predicted Tc > 40 K, along with a family of B13X compounds, and claim this supports the effectiveness of the strategy.

Significance. If the σDOS descriptor is shown to reliably separate high-Tc materials from others on independent benchmarks and to correlate quantitatively with Tc, the approach would offer a computationally efficient route for large-scale superconductor discovery by combining a physics-based proxy with AI acceleration, reducing reliance on expensive DFPT.

major comments (1)
  1. [Abstract] Abstract: the central claim that σDOS serves as a reliable descriptor to flag high-Tc candidates from plain DFT bands (without DFPT) is not supported by any reported correlation with DFPT or experimental Tc values on benchmark materials such as MgB2; no quantitative mapping from σDOS to Tc or separation of known high-Tc vs. low-Tc compounds is shown, rendering the screening of 2 million entries and the B13Se identification dependent on an unvalidated proxy.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their careful reading and constructive feedback. We address the major comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that σDOS serves as a reliable descriptor to flag high-Tc candidates from plain DFT bands (without DFPT) is not supported by any reported correlation with DFPT or experimental Tc values on benchmark materials such as MgB2; no quantitative mapping from σDOS to Tc or separation of known high-Tc vs. low-Tc compounds is shown, rendering the screening of 2 million entries and the B13Se identification dependent on an unvalidated proxy.

    Authors: We agree that the abstract does not explicitly reference benchmark validation. The manuscript body motivates σDOS via the metallized σ-bonding picture and reports σDOS values for select known materials, but we acknowledge the absence of a dedicated quantitative correlation analysis or separation plot against DFPT/experimental Tc (including MgB2). In revision we will add a new subsection and figure providing this mapping on a benchmark set to directly support the descriptor's reliability and the subsequent screening results. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The derivation introduces σDOS as a descriptor explicitly guided by the external metallized σ-bonding physics picture and applies it to DFT electronic structure for screening without any fitting to Tc values or self-referential definitions. The large-scale screening and candidate identification (B13Se) follow directly from this independent proxy plus AI acceleration of the Hamiltonian; no equation or step reduces by construction to its own inputs, no load-bearing self-citation chain is invoked, and no ansatz or uniqueness result is smuggled from prior author work. The chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The approach rests on the domain assumption that the metallized σ-bonding picture yields a useful electronic descriptor and on the accuracy of the deep-learning approximation to DFT Hamiltonians.

axioms (1)
  • domain assumption The metallized σ-bonding picture provides a valid basis for a descriptor of high-Tc superconductivity
    Explicitly stated as guiding the introduction of σDOS in the abstract.
invented entities (1)
  • σDOS no independent evidence
    purpose: Efficient physical descriptor for high-Tc superconductors from DFT electronic structure
    Newly introduced quantity defined from the σ-bonding picture

pith-pipeline@v0.9.1-grok · 5755 in / 1301 out tokens · 45611 ms · 2026-06-26T12:46:47.356089+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

58 extracted references · 1 linked inside Pith

  1. [1]

    Giustino, Electron-phonon interactions from first prin- ciples, Rev

    F. Giustino, Electron-phonon interactions from first prin- ciples, Rev. Mod. Phys.89, 015003 (2017)

    2017

  2. [2]

    Eliashberg, Interactions between electrons and lattice vibrations in a superconductor, Sov

    G. Eliashberg, Interactions between electrons and lattice vibrations in a superconductor, Sov. Phys. JETP11, 696 (1960)

    1960

  3. [3]

    P. B. Allen and R. Dynes, Transition temperature of strong-coupled superconductors reanalyzed, Phys. Rev. B12, 905 (1975)

    1975

  4. [4]

    L. N. d. Oliveira, E. Gross, and W. Kohn, Density- functional theory for superconductors, Phys. Rev. Lett. 60, 2430 (1988)

    1988

  5. [5]

    Dolui, L

    K. Dolui, L. J. Conway, C. Heil, T. A. Strobel, R. P. Prasankumar, and C. J. Pickard, Feasible route to high- temperature ambient-pressure hydride superconductiv- ity, Phys. Rev. Lett.132, 166001 (2024)

    2024

  6. [6]

    D. Duan, Y. Liu, F. Tian, D. Li, X. Huang, Z. Zhao, H. Yu, B. Liu, W. Tian, and T. Cui, Pressure-induced metallization of dense (H 2S)2H2 with high-T c supercon- ductivity, Sci. Rep.4, 6968 (2014)

    2014

  7. [7]

    F. Peng, Y. Sun, C. J. Pickard, R. J. Needs, Q. Wu, and Y. Ma, Hydrogen clathrate structures in rare earth hydrides at high pressures: possible route to room- temperature superconductivity, Phys. Rev. Lett.119, 107001 (2017)

    2017

  8. [8]

    H. Liu, I. I. Naumov, R. Hoffmann, N. Ashcroft, and R. J. Hemley, Potential high-Tc superconducting lanthanum and yttrium hydrides at high pressure, Proc. Nat. Acad. Sci.114, 6990 (2017)

    2017

  9. [9]

    Lilia, R

    B. Lilia, R. Hennig, P. Hirschfeld, G. Profeta, A. Sanna, E. Zurek, W. E. Pickett, M. Amsler, R. Dias, M. I. Eremets,et al., The 2021 room-temperature super- conductivity roadmap, J. Phys.: Condens. Matter34, 183002 (2022)

    2021

  10. [10]

    Stanev, C

    V. Stanev, C. Oses, A. G. Kusne, E. Rodriguez, J. Paglione, S. Curtarolo, and I. Takeuchi, Machine learn- ing modeling of superconducting critical temperature, npj Comput. Mater.4, 29 (2018)

    2018

  11. [11]

    M. J. Hutcheon, A. M. Shipley, and R. J. Needs, Predict- ing novel superconducting hydrides using machine learn- ing approaches, Phys. Rev. B101, 144505 (2020)

    2020

  12. [12]

    X. Wang, C. Zhang, Z. Wang, H. Liu, J. Lv, H. Wang, W. E, and Y. Ma, Computational discovery of high- temperature superconducting ternary hydrides via deep learning, Nat. Sci. Rev. , nwag030 (2026)

    2026

  13. [13]

    X.-Q. Han, Z. Ouyang, P.-J. Guo, H. Sun, Z.-F. Gao, and Z.-Y. Lu, InvDesFlow: An AI-driven materials inverse design workflow to explore possible high-temperature su- perconductors, Chin. Phys. Lett.42, 047301 (2025)

    2025

  14. [14]

    Han, P.-J

    X.-Q. Han, P.-J. Guo, Z.-F. Gao, H. Sun, and Z.-Y. Lu, InvDesFlow-AL: active learning-based workflow for in- verse design of functional materials, npj Comput. Mater. (2025)

    2025

  15. [15]

    Ouyang, B.-W

    Z. Ouyang, B.-W. Yao, X.-Q. Han, P.-J. Guo, Z.-F. Gao, and Z.-Y. Lu, High-temperature superconductiv- ity in Li2AuH6 mediated by strong electron-phonon cou- pling under ambient pressure, Phys. Rev. B111, L140501 (2025). 6

    2025

  16. [16]

    B.-W. Yao, Z. Ouyang, X.-Q. Han, C.-J. Wu, P.-J. Guo, Z.-F. Gao, and Z.-Y. Lu, Superconductivity in atom- intercalated quaternary hydrides under ambient pressure, Phys. Rev. B113, 094509 (2026)

    2026

  17. [17]

    Tran and T

    H. Tran and T. N. Vu, Machine-learning approach for discovery of conventional superconductors, Phys. Rev. Mater.7, 054805 (2023)

    2023

  18. [18]

    J. B. Gibson, A. C. Hire, P. M. Dee, O. Barrera, B. Geisler, P. J. Hirschfeld, and R. G. Hennig, Acceler- ating superconductor discovery through tempered deep learning of the electron-phonon spectral function, npj Comput. Mater.11, 7 (2025)

    2025

  19. [19]

    T. F. Cerqueira, A. Sanna, and M. A. Marques, Sam- pling the materials space for conventional superconduct- ing compounds, Adv. Mater.36, 2307085 (2024)

    2024

  20. [20]

    H. Li, Z. Tang, J. Fu, W.-H. Dong, N. Zou, X. Gong, W. Duan, and Y. Xu, Deep-learning density func- tional perturbation theory, Phys. Rev. Lett.132, 096401 (2024)

    2024

  21. [21]

    H. Li, Z. Wang, N. Zou, M. Ye, R. Xu, X. Gong, W. Duan, and Y. Xu, Deep-learning density functional theory Hamiltonian for efficient ab initio electronic- structure calculation, Nat. Comput. Sci.2, 367 (2022)

    2022

  22. [22]

    X. Gong, H. Li, N. Zou, R. Xu, W. Duan, and Y. Xu, General framework for E(3)-equivariant neural network representation of density functional theory Hamiltonian, Nat. Commun.14, 2848 (2023)

    2023

  23. [23]

    Y. Wang, H. Li, Z. Tang, H. Tao, Y. Wang, Z. Yuan, Z. Chen, W. Duan, and Y. Xu, DeepH-2: Enhanc- ing deep-learning electronic structure via an equivariant local-coordinate transformer, arXiv:2401.17015 (2024)

    arXiv 2024

  24. [24]

    Gao, Z.-Y

    M. Gao, Z.-Y. Lu, and T. Xiang, Finding high- temperature superconductors by metallizing theσ- bonding electrons, Physics44, 421 (2015)

    2015

  25. [25]

    An and W

    J. An and W. Pickett, Superconductivity of MgB 2: co- valent bonds driven metallic, Phys. Rev. Lett.86, 4366 (2001)

    2001

  26. [26]

    Rosner, A

    H. Rosner, A. Kitaigorodsky, and W. Pickett, Prediction of high T c superconductivity in hole-doped LiBC, Phys. Rev. Lett.88, 127001 (2002)

    2002

  27. [27]

    J. E. Moussa and M. L. Cohen, Constraints on T c for su- perconductivity in heavily boron-doped diamond, Phys. Rev. B77, 064518 (2008)

    2008

  28. [28]

    Gao, Z.-Y

    M. Gao, Z.-Y. Lu, and T. Xiang, Prediction of phonon-mediated high-temperature superconductivity in Li3B4C2, Phys. Rev. B91, 045132 (2015)

    2015

  29. [29]

    Gao, X.-W

    M. Gao, X.-W. Yan, Z.-Y. Lu, and T. Xiang, Phonon- mediated high-temperature superconductivity in the ternary borohydride KB2H8 under pressure near 12 GPa, Phys. Rev. B104, L100504 (2021)

    2021

  30. [30]

    Wang, X.-W

    J.-N. Wang, X.-W. Yan, and M. Gao, High-temperature superconductivity in SrB 3C3 and BaB 3C3 predicted from first-principles anisotropic Migdal-Eliashberg the- ory, Phys. Rev. B103, 144515 (2021)

    2021

  31. [31]

    Di Cataldo, A

    S. Di Cataldo, A. Sanna, and L. Boeri, Ambient-pressure superconductivity from Boron icosahedral superatoms, arXiv:2508.17422 (2025)

    Pith/arXiv arXiv 2025

  32. [32]

    X. Gong, S. G. Louie, W. Duan, and Y. Xu, Generaliz- ing deep learning electronic structure calculation to the plane-wave basis, Nat. Comput. Sci.4, 752 (2024)

    2024

  33. [33]

    See Supplemental Material for calculation methods as well as other detailed information, which includes refer- ences [2, 3, 19, 21, 23, 24, 28–30, 36–40, 42, 44–58]

  34. [34]

    H. Li, Z. Tang, X. Gong, N. Zou, W. Duan, and Y. Xu, Deep-learning electronic-structure calculation of magnetic superstructures, Nat. Comput. Sci.3, 321–327 (2023)

    2023

  35. [35]

    Z. Tang, H. Chen, Y. Li, Y. Qian, Y. Wang, W. Fu, J. Li, C. Si, W. Duan, J. Chen,et al., Deep-learning electronic structure calculations, Nat. Comput. Sci.5, 1133 (2025)

    2025

  36. [36]

    Y. Wang, Y. Li, Z. Tang, H. Li, Z. Yuan, H. Tao, N. Zou, T. Bao, X. Liang, Z. Chen,et al., Universal materials model of deep-learning density functional theory hamil- tonian, Sci. Bull.69, 2514 (2024)

    2024

  37. [37]

    A. Jain, S. P. Ong, G. Hautier, W. Chen, W. D. Richards, S. Dacek, S. Cholia, D. Gunter, D. Skinner, G. Ceder, et al., Commentary: The Materials Project: A materials genome approach to accelerating materials innovation, APL Mater.1(2013)

    2013

  38. [38]

    Schmidt, N

    J. Schmidt, N. Hoffmann, H.-C. Wang, P. Borlido, P. J. Carri¸ co, T. F. Cerqueira, S. Botti, and M. A. Marques, Machine-learning-assisted determination of the global zero-temperature phase diagram of materials, Adv. Mater.35, 2210788 (2023)

    2023

  39. [39]

    Merchant, S

    A. Merchant, S. Batzner, S. S. Schoenholz, M. Aykol, G. Cheon, and E. D. Cubuk, Scaling deep learning for materials discovery, Nature624, 80 (2023)

    2023

  40. [40]

    Li andet al., To be published (2026)

    Y. Li andet al., To be published (2026)

    2026

  41. [41]

    E. R. Margine and F. Giustino, Anisotropic Migdal- Eliashberg theory using Wannier functions, Phys. Rev. B87, 024505 (2013)

    2013

  42. [42]

    Bercx, S

    M. Bercx, S. Ponc´ e, Y. Zhang, G. Trezza, A. G. Ghezel- jehmeidan, L. Bastonero, J. Qiao, F. O. Von Rohr, G. Pizzi, E. Chiavazzo,et al., Charting the landscape of Bardeen-Cooper-Schrieffer superconductors in exper- imentally known compounds, PRX Energy4, 033012 (2025)

    2025

  43. [43]

    Z. Tang, H. Li, P. Lin, X. Gong, G. Jin, L. He, H. Jiang, X. Ren, W. Duan, and Y. Xu, A deep equivariant neural network approach for efficient hybrid density functional calculations, Nat. Commun.15, 8815 (2024)

    2024

  44. [44]

    Kresse and J

    G. Kresse and J. Furthm¨ uller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Phys. Rev. B54, 11169 (1996)

    1996

  45. [45]

    J. P. Perdew, K. Burke, and M. Ernzerhof, Generalized gradient approximation made simple, Phys. Rev. Lett. 77, 3865 (1996)

    1996

  46. [46]

    Ozaki, Variationally optimized atomic orbitals for large-scale electronic structures, Phys

    T. Ozaki, Variationally optimized atomic orbitals for large-scale electronic structures, Phys. Rev. B67, 155108 (2003)

    2003

  47. [47]

    Giannozzi, S

    P. Giannozzi, S. Baroni, N. Bonini, M. Calandra, R. Car, C. Cavazzoni, D. Ceresoli, G. L. Chiarotti, M. Cococ- cioni, I. Dabo,et al., Quantum ESPRESSO: a modu- lar and open-source software project for quantumsimula- tions of materials, J. Phys.: Condens. Matter21, 395502 (2009)

    2009

  48. [48]

    Pyykk¨ o and M

    P. Pyykk¨ o and M. Atsumi, Molecular single-bond co- valent radii for elements 1–118, Chem. Eur. J.15, 186 (2009)

    2009

  49. [49]

    C. J. Pickard and R. Needs, Ab initio random structure searching, J. Phys.: Condens. Matter23, 053201 (2011)

    2011

  50. [50]

    Schlipf and F

    M. Schlipf and F. Gygi, Optimization algorithm for the generation of ONCV pseudopotentials, Comput. Phys. Commun.196, 36 (2015)

    2015

  51. [51]

    Damle, L

    A. Damle, L. Lin, and L. Ying, Compressed represen- tation of Kohn–Sham orbitals via selected columns of the density matrix, J. Chem. Theory Comput.11, 1463 (2015). 7

    2015

  52. [52]

    Ponc´ e, E

    S. Ponc´ e, E. R. Margine, C. Verdi, and F. Giustino, EPW: Electron–phonon coupling, transport and super- conducting properties using maximally localized Wannier functions, Comput. Phys. Commun.209, 116 (2016)

    2016

  53. [53]

    Aykol, S

    M. Aykol, S. S. Dwaraknath, W. Sun, and K. A. Persson, Thermodynamic limit for synthesis of metastable inor- ganic materials, Sci. Adv.4, eaaq0148 (2018)

    2018

  54. [54]

    Pizzi, V

    G. Pizzi, V. Vitale, R. Arita, S. Bl¨ ugel, F. Freimuth, G. G´ eranton, M. Gibertini, D. Gresch, C. Johnson, T. Koretsune,et al., Wannier90 as a community code: new features and applications, J. Phys.: Condens. Mat- ter32, 165902 (2020)

    2020

  55. [55]

    Vitale, G

    V. Vitale, G. Pizzi, A. Marrazzo, J. R. Yates, N. Marzari, and A. A. Mostofi, Automated high-throughput Wan- nierisation, npj Comput. Mater.6, 66 (2020)

    2020

  56. [56]

    Chen and S

    C. Chen and S. P. Ong, A universal graph deep learning interatomic potential for the periodic table, Nat. Com- put. Sci.2, 718 (2022)

    2022

  57. [57]

    Pellegrini and A

    C. Pellegrini and A. Sanna, Ab initio methods for super- conductivity, Nat. Rev. Phys.6, 509 (2024)

    2024

  58. [58]

    Zhang and T

    F. Zhang and T. Rice, Effective hamiltonian for the su- perconducting cu oxides, Phys. Rev. B37, 3759 (1988)

    1988