[论文解读] Spin-orbit excitation energies, anisotropic exchange, and magnetic phases of honeycomb RuCl3

Ravi Yadav, Nikolay A. Bogdanov+4 more|arXiv (Cornell University)|Apr 16, 2016

Advanced Condensed Matter Physics参考文献 47被引用 105

一句话总结

从头算量子化学研究表明，α-RuCl3中的 Kitaev 自旋晶格交换 K 是铁磁的，但比 Ir 氧化物中的要弱，J 较小且反铁磁；更长程的耦合和磁场效应驱动锯齿状有序以及可能的磁场诱导自旋液态。

ABSTRACT

Using quantum chemistry calculations we shed fresh light on the electronic structure and magnetic properties of RuCl3, a proposed realization of the honeycomb Kitaev spin model. It is found that the nearest-neighbor Kitaev exchange K is weaker than in 5d5 Ir oxides but still larger than other effective spin couplings. The electronic-structure computations also indicate a ferromagnetic K in the halide, which is supported by a detailed analysis of the field-dependent magnetization. From exact-diagonalization calculations for extended Kitaev-Heisenberg Hamiltonians we additionally find that a transition from zigzag order to a spin-liquid ground state can be induced in RuCl3 with external magnetic field.

研究动机与目标

• Clarify the electronic structure of α-RuCl3 and the nature of its spin-orbit entangled states.
• Determine the nearest-neighbor magnetic interactions (K, J, Γ) from ab initio calculations.
• Map out the magnetic phase diagram including longer-range interactions and external magnetic field effects.

提出的方法

• Perform embedded-cluster quantum chemistry calculations (CASSCF and MRCI) with spin-orbit coupling to obtain t2g5 configurations and excitations.
• Map the low-energy states of two neighboring RuCl6 octahedra onto an effective S=1/2 spin Hamiltonian with J, K, and symmetric anisotropy Γ (Γxy, Γzx, Γyz).
• Compute NN magnetic couplings for multiple crystal structures and analyze their dependence on Ru–Cl–Ru bond angles.
• Use exact diagonalization of extended Kitaev-Heisenberg models to explore phase diagrams with J2, J3 and external field H.
• Compare calculated g-factors and magnetization with experimental data to validate the model.

实验结果

研究问题

• RQ1 What are the nearest-neighbor magnetic interactions (K, J, Γ) in α-RuCl3 from ab initio calculations?
• RQ2 How does trigonal distortion affect the spin-orbit ground state and g-factor anisotropy?
• RQ3 How do longer-range interactions (J2, J3) and external magnetic field influence the magnetic phases, including possible spin-liquid states?
• RQ4 Can the computed couplings reproduce the observed zigzag order and magnetization curves?
• RQ5 Is there a field-induced transition from zigzag order to a spin-liquid state?

主要发现

Structure	Angle (Ru-Cl-Ru)	K (meV)	J (meV)	Γ_xy (meV)	Γ_zx = -Γ_yz (meV)
C2/m (Cao et al. 2016)	94°	-5.6	1.2	-1.2	-0.7
C2/m (Johnson et al. 2015) – Link 1	94°	-5.3	1.2	-1.1	-0.7
C2/m (Johnson et al. 2015) – Link 2	93°	-4.8	-0.3	-1.5	-0.7
P3_1 12 (Stroganov & Ovchinnikov 1957)	89°	-1.2	-0.5	-1.0	-0.4

• Nearest-neighbor Kitaev exchange K is ferromagnetic and substantially weaker (max |K| ≈ 5.6 meV) than in 4d/5d honeycomb oxides.
• J is antiferromagnetic and relatively small in ab initio results, with J values around 1 meV range, and Γ couplings are of similar magnitude to K.
• Trigonal distortion causes sizable t2g splitting and g-factor anisotropy consistent with experiments.
• Longer-range couplings J2 and J3 are small (≈0.25–0.5 meV) but essential to stabilize zigzag order; a spin-liquid state emerges in ED when field drives zigzag to collapse, indicating a field-induced SL region.
• Model fits to magnetization data require J>0 and K<0, withField-enhanced |K|/J ratio; pure Kitaev-Heisenberg picture with only NN terms is insufficient to capture all features.
• A phase diagram with seven phases (FM, Néel, zigzag, stripy, and three IC plus SL) is obtained, showing zigzag stability for realistic J2, J3 ranges.