Title: Nagaoka Ferromagnetism in $3 \times 3$ Arrays and Beyond

Abstract: Nagaoka ferromagnetism (NF) is a long-predicted example of itinerant ferromagnetism (IF) in the Hubbard model that has been studied theoretically for many years. The condition for NF, an infinite onsite Coulomb repulsion and a single hole in a half-filled band, does not arise naturally in materials, and was only realized recently for the first time in experiments on a 2 \times 2 array of gated quantum dots. Atomically precise fabrication of dopant arrays in Si allows for engineering highly controllable systems with complex geometries. This makes dopant arrays a good candidate to study NF in different array geometries through analog quantum simulation. Here we present theoretical simulations done for $3 \times 3$ arrays and larger $N \times N $ arrays of quantum dots, and predict the emergence of different forms of ferromagnetism in different geometries. We find NF in perfect $3 \times 3$ arrays, as well as in $N \times N $ arrays for one hole less than half-filling. The ratio of the Hubbard on-site repulsion U to hopping t that defines the onset of NF increases as N increases, approaching the bulk limit of infinite U for large N. Additional simulations are done for geometries made by removing sites from $N \times N $ arrays. Different forms of ferromagnetism are found for different geometries. Loops show ferromagnetism, but only for three electrons. For loops, the critical U/t for the onset of ferromagnetism decreases as the loop length increases. We show that this different dependence on size for loops and $N \times N $ arrays can be understood by a scaling argument. Our results show how analog quantum simulation with small arrays can elucidate the role of effects including wavefunction connectivity; system geometry, size and symmetry; bulk and edge sites; and kinetic energy in determining the quantum magnetism of small systems.