Researchers used a large optical lattice of ultracold atoms to create and probe extended regions that behave like a U(1) quantum spin liquid. A quantum spin liquid is an unusual state of matter that does not show ordinary order. Instead it is a highly entangled superposition of many different particle configurations. The team reports a simulator spanning more than 3,000 lattice sites and finds direct signatures of the emergent U(1) gauge structure and long-range coherence over roughly 100 lattice sites.

The experiment starts from a simple, ordered “all‑monomer” state where certain lattice sites hold single atoms (monomers). By tuning the lattice and interactions they suppress ordinary single‑particle motion and make two‑particle correlated hops the main process. Two neighboring monomers can jointly hop and form a doubly occupied site on a link; these doublons act like dimers in an effective two‑dimensional monomer‑dimer model. The effective dynamics can be written in terms of a small tunnelling energy Jeff ≈ 4√2 J^2/U and a tunable mass m set by the superlattice, with experimental parameters around J/h ≈ 130 Hz, ∆/J ≈ 12 and U/J ≈ 24. The effective tunnelling time is Teff = ħ/Jeff = 5.13(7) ms.

A key test of the gauge theory is Gauss’s law, a local constraint on allowed patterns of monomers and dimers. To read out that constraint the team developed doublon‑resolved imaging that converts doubly occupied sites into singly occupied sites before fluorescence detection. This avoids the usual parity projection that hides doublons. Using a roughly 47×47 site region of interest (more than 500 vertices and 1,000 links) they directly classified the vertex patterns and showed most vertices follow the Gauss’s‑law constraint. The doublon conversion reached a peak fidelity of about 98(1)%, allowing quantitative comparison of constrained and unconstrained dynamics. The fraction of Gauss’s‑law satisfying vertices decays only slowly in time and is further stabilized by a diagonal tilt (δ/J ≈ 2) that suppresses competing processes.