Abstract: We have studied the doping-driven orbital-selective Mott transition in multi-band Hubbard models with equal band width in the presence of crystal field splitting. Crystal field splitting lifts one of the bands while leaving the others degenerate. We use single-site dynamical mean-field theory combined with continuous time quantum Monte Carlo impurity solver to calculate a phase diagram as a function of total electron filling $N$ and crystal field splitting $\Delta$. We find a large region of orbital-selective Mott phase in the phase diagram when the doping is large enough. Further analysis indicates that the large region of orbital-selective Mott phase is driven and stabilized by doping. Such models may account for the orbital-selective Mott transition in some doped realistic strongly correlated materials.
Introduction
In the study of strongly correlated electron systems, the Mott transition is a foundational concept where electron-electron repulsion transforms a metal into an insulator. In multi-orbital (multi-band) systems, this phenomenon can manifest as an Orbital-Selective Mott Transition (OSMT), where specific orbitals become insulating while others remain metallic.
Traditionally, physicists believed that OSMT required orbitals to possess significantly different bandwidths—essentially different "speeds" for electron hopping. This research, published by Yilin Wang and colleagues, challenges that assumption by asking whether an OSMT can occur in bands of equal width, and more surprisingly, whether it can be driven by doping (changing the electron count), which typically increases a material's metallicity.
Theoretical Framework and Methods
The researchers simulated these quantum systems using a multi-band Hubbard model that incorporates Crystal Field Splitting (CFS) to shift the energy levels of different orbitals. To solve the complex lattice equations, they employed Dynamical Mean-Field Theory (DMFT), mapping the lattice problem onto a local "impurity" problem.
High-precision numerical results were obtained using a hybridization expansion continuous-time quantum Monte Carlo (HYB-CTQMC) solver via the iQIST software package. The study focused on the interplay between total electron filling ($N$), crystal field splitting ($\Delta$), and Hund’s coupling ($J_z$), which governs the alignment of electron spins.
The Mechanism: Doping-Induced Half-Filling
The core finding of this study is that doping doesn't always enhance metallicity. Instead, it can stabilize an orbital-selective insulating state. As electrons are added to the system, the "lower" orbitals (those pushed down by CFS) reach a half-filled state and remain there, even as the total electron count $N$ increases.
Strong Hund’s coupling acts as a "band decoupler." When electrons are doped into the "upper" orbital, they interact with electrons in the half-filled "lower" orbital to form stable local magnetic moments. These moments act as scattering centers, effectively increasing the correlation strength and forcing the lower orbitals into a Mott insulating state. This mechanism demonstrates that increasing doping concentration can actually enlarge the region where this selective insulating state exists.
Phase Diagrams and Key Results
The authors produced detailed phase diagrams showing a substantial Orbital-Selective Mott Phase (OSMP) region. In the two-band model, as the total electron filling $N$ moves from 1.0 toward 2.0, the system transitions from a pure metal to an OSMP.
Calculations of the spin-spin correlation function confirmed that in the OSMP, local magnetic moments are "frozen" and well-defined, a hallmark of the proposed mechanism. Furthermore, the effect was observed in both two-band and three-band models, suggesting that doping-driven OSMT is a robust physical phenomenon not limited to specific orbital counts.
Conclusion and Future Outlook
While this study utilizes single-site DMFT—which neglects certain momentum-dependent effects—and an idealized Bethe lattice, it provides a compelling theoretical basis for understanding complex materials. The authors suggest this "doping-driven" logic could explain the behavior of realistic strongly correlated materials, such as certain ruthenates (e.g., $Ca_{2-x}Sr_xRuO_4$). Future work will likely involve direct applications of this model to specific experimental compounds to verify these counter-intuitive findings in the lab.
Summary updated