Jensen and Luttinger (1952) showed that the Thomas-Fermi (TF) statistical model yields smooth orbital-angular-momentum distributions for atoms and nuclei -- a mean squared angular momentum of 0.262 Z2/3ℏ20.262\,Z^{2/3}\hbar^{2} for neutral atoms and 0.885 Z2/3ℏ20.885\,Z^{2/3}\hbar^{2} for a sharp-edged nuclear Fermi sphere -- but acknowledged that it cannot locate closed shells, rests on an order-of-magnitude nuclear surface-thickness parameter, and admits no stable mapping from its continuous angular momentum to the discrete quantum number ll. The nuclear self-consistency and surface-diffuseness threads were largely resolved by extended Thomas-Fermi theory with effective interactions and by the Strutinsky smooth-plus-oscillatory shell decomposition, and the atomic asymptotic thread was resolved rigorously in 2025, when the TF angular-momentum distribution was proven (as a liminf statement) to be the correct large-ZZ limit of quantum occupation numbers (so that the Madelung filling rule is not). Two cores remain open with direct 2016-2026 evidence: discrete shell closures and first-appearance numbers are still beyond orbital-free or semiclassical prediction -- described in 2025 as a longstanding textbook challenge and only recently touched by machine-learned functionals for two light nuclei -- and the nuclear angular-momentum (spin) distribution, characterized today by the spin cutoff parameter of nuclear level densities, the direct descendant of the 1952 calculation, still disagrees with measured spin distributions below the neutron separation energy, with pairing and symmetry-imposed corrections only partially understood.

Background

The Thomas-Fermi (TF) model is the original semiclassical ("statistical") description of many-fermion systems such as atoms and nuclei: instead of solving for individual quantum orbitals, it fills position-momentum phase space uniformly up to a local maximum momentum P(r)P(r) related to the local particle density by ρ(r)=P3(r)/(3π2ℏ3)\rho(r)=P^{3}(r)/(3\pi^{2}\hbar^{3}) (the spin factor of 2 is included). Because position and momentum are treated as continuous classical variables, the orbital angular momentum L=∣r×p∣L=|\mathbf{r}\times\mathbf{p}| has a continuous distribution in TF theory, whereas in quantum mechanics it takes the discrete values l(l+1) ℏ\sqrt{l(l+1)}\,\hbar with integer ll.

The question raised by Jensen and Luttinger (1952) was motivated by the nuclear shell model, established only a few years earlier: nuclei are exceptionally stable at the empirical "magic" neutron or proton numbers 2, 8, 20, 28, 50, 82, 126, analogous to the noble-gas atomic numbers 2, 10, 18, 36, 54, 86, and the shell model explains these as closures of groups of single-particle levels. Several authors had tried to interpret the shell closures in TF terms, identifying the closing of a shell with the first appearance of a particle with a higher orbital angular momentum. Jensen and Luttinger showed this identification fails already for atoms (closed shells at Z=2,10,18,…Z=2,10,18,\ldots while electrons with l=1,2,3l=1,2,3 first appear at Z=5,21,58Z=5,21,58), argued that the TF model "can give no information about the location of the closed shells" because all its properties vary smoothly with particle number, and proposed as the well-defined TF observable the mean squared orbital angular momentum per particle, ⟨L2⟩Av\langle L^{2}\rangle_{Av}. For atoms, TF theory gives the universal scaling law ⟨L2⟩Av=0.262 Z2/3\langle L^{2}\rangle_{Av}=0.262\,Z^{2/3} (in ℏ2\hbar^{2}); for nuclei, treating protons and neutrons as separate Fermi gases of fixed interior density and radius R∝A1/3R\propto A^{1/3}, a sharp-edged sphere gives, for the proton gas, ⟨L2⟩Av=0.885 Z2/3\langle L^{2}\rangle_{Av}=0.885\,Z^{2/3} (with ZZ the proton number), and a Gaussian density falloff of fixed surface thickness 1.8×10−131.8\times10^{-13} cm (1.8 fm) brings the TF prediction into excellent agreement with shell-model values. They stressed, however, that this thickness is an order-of-magnitude fit rather than a consequence of any nuclear force, that the nuclear TF density itself is undetermined because the nuclear effective interaction was unknown, and that any TF prescription for the "first appearance" of a given ll is numerically unstable under reasonable changes of the continuous-to-discrete mapping. Their conclusion was that smooth averages such as ⟨L2⟩Av\langle L^{2}\rangle_{Av} are the only robust TF observables.

The quantities in this problem live on in several later frameworks that the reader needs. The nuclear level density -- the number of nuclear excited states per unit energy, central to reaction theory -- must be resolved by spin JJ for applications; this is done with a Gaussian spin distribution whose width is the spin cutoff parameter σ2=⟨Jz2⟩\sigma^{2}=\langle J_{z}^{2}\rangle, in the independent-particle Fermi gas fixed by exactly the single-particle angular-momentum content that Jensen and Luttinger computed (their 2/32/3-power scaling survives in the standard Fermi-gas spin-cutoff formula σ2≈0.088 A2/3aU\sigma^{2}\approx0.088\,A^{2/3}\sqrt{aU}, with aa the level-density parameter and UU the effective excitation energy; Gilbert and Cameron's underlying single-particle moment ⟨m2⟩=0.146 A2/3\langle m^{2}\rangle=0.146\,A^{2/3} was deduced from the 1952 angular-momentum analysis). The Strutinsky shell-correction method (1967) splits every nuclear observable into a smooth TF-like part plus an oscillatory shell part built from quantum single-particle levels, and is the standard bridge between semiclassics and shell structure. Extended Thomas-Fermi (ETF) theory adds gradient (Weizsäcker-type) corrections to the kinetic energy and, combined with effective interactions of Skyrme type (a standard family of density-dependent nuclear forces), determines nuclear densities and their surface diffuseness self-consistently. Semiclassical Bohr-Sommerfeld (Einstein-Brillouin-Keller, EBK) quantization of the TF potential supplies discrete orbital energies asymptotically, with the Langer replacement l→l+1/2l\to l+1/2 near turning points. Orbital-free density functional theory (OFDFT), the modern descendant of TF used for very large electronic and nuclear simulations, expresses the energy through the density alone and is by construction blind to shell effects -- precisely the 1952 limitation. The Madelung (Aufbau) rule is the empirical ordering by which electrons fill subshells (n,l)(n,l) across the periodic table. Hauser-Feshbach compound-nucleus reaction calculations -- the statistical reaction framework used in nuclear astrophysics and nuclear-data evaluation -- are a principal consumer of spin-resolved level densities and hence of σ2\sigma^{2}.

The modern status against which the open question is judged: for atoms, the TF angular-momentum distribution was proven in 2025 (as a liminf statement) to be the correct large-ZZ limit of the quantum occupation distribution (and the Madelung rule was proven not to be that limit), settling the leading-order reconciliation of smooth TF behavior with discrete filling; for nuclei, ETF with effective interactions supplies self-consistent densities and surface properties, but the measured spin distributions of nuclear levels still deviate systematically from the phenomenological spin-cutoff formulas descended from the 1952 calculation, and no orbital-free or purely semiclassical scheme predicts discrete shell closures and first appearances with quantum-level accuracy.

Problem Statement

Can semiclassical, Thomas-Fermi-type (orbital-free) theory -- corrected by a controlled quantization scheme and built on realistic effective interactions with self-consistent densities -- predict the discrete shell structure that Jensen and Luttinger (1952) showed the smooth Thomas-Fermi model cannot supply? Concretely: predict the closed-shell (magic) numbers of atoms and nuclei and the particle numbers at which each orbital angular momentum ll first appears -- the quantity ν(l)\nu(l), which the 1952 paper computed only for atoms and deemed 'so poorly defined in a model with continuous LL' that it declined to use first appearances for nuclei -- for atoms across and beyond the periodic table (including superheavy elements) and for nuclei, from an orbital-free (density-only) scheme, validated against quantum-mechanical shell-model values and, where these exist, empirical data; at issue is whether any such scheme recovers the discrete shell structure, not whether one particular numerical search fails to. The nuclear spin-cutoff lineage -- the spin cutoff parameter σ2(E,A)\sigma^{2}(E,A) of nuclear level densities, fixed by the same single-particle angular-momentum content behind the 1952 mean squared orbital angular momentum ⟨L2⟩\langle L^{2}\rangle -- serves as motivation and context for this question rather than as a jointly required deliverable; its own documented open problems, including the failures of the standard phenomenological formulas below the neutron separation energy SnS_{n} (the threshold energy for neutron emission), remain related open questions. Partial resolutions already achieved (the rigorous large-ZZ atomic asymptotics of Bjerg et al. 2025; ETF self-consistency with Skyrme-type forces; Strutinsky-type shell corrections) are part of the accepted foundation and define what remains: the discrete, shell-resolved level of the same question.

References

  1. J. H. D. Jensen and J. M. Luttinger, 'Angular Momentum Distributions in the Thomas-Fermi Model', Physical Review 86, 907-910 (1952), DOI: 10.1103/PhysRev.86.907, https://doi.org/10.1103/PhysRev.86.907
  2. N. H. March, 'The Thomas-Fermi approximation in quantum mechanics', Advances in Physics 6, 1-101 (1957), DOI: 10.1080/00018735700101156, https://doi.org/10.1080/00018735700101156
  3. Richard Latter, 'Atomic Energy Levels for the Thomas-Fermi and Thomas-Fermi-Dirac Potential', Physical Review 99, 510-519 (1955), DOI: 10.1103/PhysRev.99.510, https://doi.org/10.1103/PhysRev.99.510
  4. Thomas A. Oliphant, Jr., 'Angular Momentum Distributions in the Thomas-Fermi-Dirac Model of the Atom', Physical Review 104, 954-958 (1956), DOI: 10.1103/PhysRev.104.954, https://doi.org/10.1103/PhysRev.104.954
  5. L. C. R. Alfred, 'Shells and Electron Groups in the Thomas-Fermi Atom', Physical Review 125, 214-219 (1962), DOI: 10.1103/PhysRev.125.214, https://doi.org/10.1103/PhysRev.125.214
  6. Adolf A. Abrahamson, 'Onset of Atomic-Subshell Filling in Ordinary and Superheavy Elements', Physical Review A 4, 454-456 (1971), DOI: 10.1103/PhysRevA.4.454, https://doi.org/10.1103/PhysRevA.4.454
  7. Marten J. Ten Hoor, 'Atomic Thomas-Fermi theory and electron subshell filling', International Journal of Quantum Chemistry 30, 831-844 (1986), DOI: 10.1002/qua.560300612, https://doi.org/10.1002/qua.560300612
  8. Torleif Ericson, 'The statistical model and nuclear level densities', Advances in Physics 9, 425-511 (1960), DOI: 10.1080/00018736000101239, https://doi.org/10.1080/00018736000101239
  9. Donald W. Lang, 'The angular momentum-dependence of the nuclear level density', Nuclear Physics 77, 545-558 (1966), DOI: 10.1016/0029-5582(66)90782-6, https://doi.org/10.1016/0029-5582(66)90782-6
  10. A. Gilbert and A. G. W. Cameron, 'A composite nuclear-level density formula with shell corrections', Canadian Journal of Physics 43, 1446-1496 (1965), DOI: 10.1139/p65-139, https://doi.org/10.1139/p65-139
  11. V. M. Strutinsky, 'Shell effects in nuclear masses and deformation energies', Nuclear Physics A 95, 420-442 (1967), DOI: 10.1016/0375-9474(67)90510-6, https://doi.org/10.1016/0375-9474(67)90510-6
  12. M. Brack, C. Guet and H.-B. Håkansson, 'Selfconsistent semiclassical description of average nuclear properties — a link between microscopic and macroscopic models', Physics Reports 123, 275-364 (1985), DOI: 10.1016/0370-1573(86)90078-5, https://doi.org/10.1016/0370-1573(86)90078-5
  13. Rainer W. Hasse, 'Angular momentum projected semiclassics', Nuclear Physics A 467, 407-421 (1987), DOI: 10.1016/0375-9474(87)90537-9, https://doi.org/10.1016/0375-9474(87)90537-9
  14. G. V. Shpatakovskaya, 'Semiclassical method of analysis and estimation of the orbital binding energies in many-electron atoms and ions', Physics-Uspekhi 62, 186-197 (2019), DOI: 10.3367/ufne.2018.02.038289, https://doi.org/10.3367/ufne.2018.02.038289
  15. S. A. Dyachkov, D. V. Minakov and P. R. Levashov, 'The wide-range model of shell effects in hot plasma with semiclassical approximation for bound electrons', Journal of Physics: Conference Series 1556, 012044 (2020), DOI: 10.1088/1742-6596/1556/1/012044, https://doi.org/10.1088/1742-6596/1556/1/012044
  16. D. V. Gorpinchenko, A. G. Magner, J. Bartel and J. P. Blocki, 'Surface corrections to the moment of inertia and shell structure in finite Fermi systems', Physical Review C 93, 024304 (2016), DOI: 10.1103/PhysRevC.93.024304, https://doi.org/10.1103/PhysRevC.93.024304
  17. S. M. Grimes, A. V. Voinov and T. N. Massey, 'Mass-number and excitation-energy dependence of the spin cutoff parameter', Physical Review C 94, 014308 (2016), DOI: 10.1103/PhysRevC.94.014308, https://doi.org/10.1103/PhysRevC.94.014308
  18. A. V. Voinov, N. Alanazi, S. Akhtar, S. Dhakal, C. R. Brune, S. M. Grimes, T. N. Massey, Z. Meisel, C. E. Parker and A. L. Richard, 'Spin cutoff factor and level density for 59Ni from an analysis of compound nuclear reactions', Physical Review C 108, 034302 (2023), DOI: 10.1103/PhysRevC.108.034302, https://doi.org/10.1103/PhysRevC.108.034302
  19. August Bjerg, Søren Fournais, Peter Hearnshaw and Jan Philip Solovej, 'The angular momentum distribution of electrons in large atoms and its deviation from Madelung's rule' (2025), arXiv:2506.14227, https://arxiv.org/abs/2506.14227
  20. X. H. Wu, Z. X. Ren and P. W. Zhao, 'Machine learning orbital-free density functional theory resolves shell effects in deformed nuclei', Communications Physics 8, 316 (2025), DOI: 10.1038/s42005-025-02234-7, https://doi.org/10.1038/s42005-025-02234-7 (preprint arXiv:2412.20739, https://arxiv.org/abs/2412.20739)
  21. Junchao Guo and Yang Sun, 'Symmetry-imposed correlation in nuclear level statistics: The spin distribution' (2026), arXiv:2602.22669, https://arxiv.org/abs/2602.22669