First-hand information for everyone
This paper reports a first‑principles, or ab initio, calculation that reproduces the famous two‑neutron halo of the nucleus lithium‑11 (11Li) directly from nuclear forces and the full many‑body dynamics. The authors use a neural‑network variational Monte Carlo method to build a flexible quantum wave function and show that the extended size and weak binding of 11Li come out of the calculation rather than being put in by hand.
The team represents the many‑nucleon wave function with a neural‑network ansatz called FeynmanNet, which combines known shell structure with flexible correlation pieces (a Slater‑Jastrow‑backflow form). They optimize that wave function by stochastic (Monte Carlo) methods. The nuclear Hamiltonian they use includes two‑nucleon (NN) and three‑nucleon (3N) forces that are fixed only by very light systems (A ≤ 4, meaning up to helium‑4). Because that minimal Hamiltonian underbinds some nuclei, they add a short‑range spin‑orbit interaction (a term that couples a nucleon’s motion to its spin) and fix its strength by reproducing the splitting of P‑wave phase shifts in neutron–alpha (neutron–helium‑4) scattering. The spin‑orbit parameter they find is Cso = −2.8 fm4.
With this setup the calculations reproduce the binding energies and the neutron separation energies along the lithium isotopic chain, including the small but positive two‑neutron separation energy of 11Li. The root‑mean‑square deviation for the separation energies is reported as about 0.36 MeV. The matter radii calculated for 6–9Li agree with experiment, and the radius shows a sharp increase at 11Li, signaling the halo. The computed 11Li radius is somewhat smaller than the experimental value. The authors report that numerical uncertainties from choices in the wave‑function representation are small (below ~0.1 MeV for energies and ~0.02 fm for radii) and that independent training runs give consistent results.
The study gives a physical connection between a few‑body observable and the many‑body halo: the size of the 11Li halo correlates with the splitting of P‑wave neutron–alpha phase shifts, which highlights the importance of neutron–alpha spin‑orbit effects in making the halo. The calculations also show that dineutron correlations (two neutrons tending to pair up spatially) appear naturally from the full many‑body wave function. In other words, the halo and the paired‑neutron behavior emerge without assuming a preformed core‑plus‑two‑neutron model.
This result matters because 11Li has been a long‑standing challenge for microscopic nuclear theory: it combines strong short‑range correlations, weak binding, and an extended density tail that couples to the continuum. Demonstrating that a halo can emerge from an interaction calibrated only in very light systems—and linking it to measurable neutron–alpha scattering—strengthens the bridge between few‑body scattering data and complex emergent nuclear structure.
There are important caveats. The authors use a minimal “essential” Hamiltonian and model the missing spin‑dependent physics with an added short‑range spin‑orbit term fit to neutron–alpha scattering rather than derived from higher‑order interactions. They also note that the absence of explicit pion‑exchange physics in their Hamiltonian may explain the somewhat smaller 11Li radius. Finally, parts of the three‑nucleon interaction (the T = 3/2 component) are not constrained by the light‑nuclei data and were projected out in this study, so further refinements of the interaction could change the quantitative results.