Type II t-J model and shared antiferromagnetic spin coupling from Hund's rule in superconducting La$_3$Ni$_2$O$_7$

Recently a 80 K superconductor was observed in La$_3$Ni$_2$O$_7$ under high pressure. Density function theory (DFT) calculations identify $d_{x^2-y^2}$ and $d_{z^2}$ as two active orbitals and a bilayer square lattice structure. The averange valence of Ni is $d^{8-x}$ with $x=0.5$ per site. Naively one may expect a description in terms of a two-orbital t-J model. However, there should be significant inter-orbital repulsion $U'$ and Hund's coupling $J_H$ larger than the bare value of $t$ and $J$. Especially the Hund's coupling can share the inter-layer super-exchange $J_\perp$ of $d_{z^2}$ to $d_{x^2-y^2}$, an effect beyond any perturbative and mean field treatment. In the limit that $d_{z^2}$ is Mott localized, we integrate it out and deal with a bialyer t-J model for $d_{x^2-y^2}$ only. We find strong inter-layer pairing due to the transmitted $J_\perp$ which can survive to $50\%$ hole doping relevant to the experiment. In real system we expect that $d_{z^2}$ orbital will also be slightly hole doped and can not be simply ignored. To deal with this situation, we take the $J_H\rightarrow +\infty$ limit and propose a type II t-J model with four singlon ($d^7$) states and three spin-triplet doublon ($d^8$) states. Through a parton mean field treatment of the constrained Hilbert space, we derive the bilayer one-orbital t-J model for an emergent `$d_{x^2-y^2}$' orbital with significant $J_\perp$, justifying our phenomenological treatment. The type II t-J model can also describe the regime where the $d_{z^2}$ orbital is also slightly hole doped through tuning an orbital energy splitting $\Delta$. From our calculation the pairing strength decreases with the hole doping $x$ and $x=0.5$ is likely larger than the optimal doping. We propose future experiments to electron dope the system to further enhance $T_c$.