Kitaev-Heisenberg Model Overview

• The Kitaev-Heisenberg model is a spin Hamiltonian combining bond-dependent Kitaev and isotropic Heisenberg exchanges on the honeycomb lattice, underpinning a spectrum of magnetic orders and quantum phases.
• It interpolates between exactly solvable limits and frustrated regimes, exhibiting Neel, stripy, zigzag, and quantum spin-liquid phases as identified through methods like exact diagonalization and spin-wave analysis.
• The model further predicts novel states in doped or field-driven systems, offering insights into unconventional superconductivity and topological excitations in layered iridate materials.

Searching arXiv for the cited Kitaev–Heisenberg papers to ground the article in published work. The Kitaev-Heisenberg model is a class of spin Hamiltonians that combines bond-directional Kitaev exchange with isotropic Heisenberg exchange, originally formulated on the honeycomb lattice as an effective description of the low-energy magnetism of layered iridates $A_2$IrO$_3$ with $A=$Li, Na. In that setting, strong spin-orbit coupling acting on $\mathrm{Ir}^{4+}$ ions in edge-sharing octahedra produces effective $j_{\mathrm{eff}}=1/2$ moments and highly anisotropic nearest-neighbor exchange. The model interpolates between the antiferromagnetic Heisenberg limit and the exactly solvable Kitaev limit, while also exhibiting an intermediate exact stripy antiferromagnet at a special point of the interpolation (Chaloupka et al., 2010).

1. Microscopic origin and canonical formulation

In the iridate construction, each $\mathrm{Ir}^{4+}$ ion sits in an octahedral environment, so the $t_{2g}$ manifold carries an effective orbital angular momentum $l=1$. Strong spin-orbit coupling then produces a Kramers doublet with total angular momentum $j_{\mathrm{eff}}=1/2$, treated as an effective spin-$\tfrac12$ degree of freedom. Because the octahedra share edges, neighboring ions exchange through bond-dependent $_3$0 Ir–O–Ir hopping paths and also direct $_3$1 overlap, generating highly anisotropic exchange interactions (Chaloupka et al., 2010).

On the honeycomb lattice, the nearest-neighbor Hamiltonian on a bond of type $_3$2 is

$_3$3

The first term is a bond-dependent Ising interaction of Kitaev type, while the second is the ordinary Heisenberg exchange. The three bond types correspond to the three cubic components $_3$4 in the local axes of the IrO$_3$5 octahedra. In the microscopic derivation,

$_3$6

with $_3$7 from Hund’s coupling on the Ir ion, $_3$8 from oxygen-mediated processes involving $_3$9, and $A=$0 from direct Ir–Ir $A=$1 hopping. The energy unit is $A=$2, where $A=$3 (Chaloupka et al., 2010).

A commonly used one-parameter form sets

$A=$4

so that $A=$5 is the antiferromagnetic Heisenberg limit and $A=$6 is the pure Kitaev limit. Later studies often use angular parametrizations such as $A=$7 or $A=$8, which sweep the full nearest-neighbor coupling space including ferromagnetic and antiferromagnetic sectors (Gotfryd et al., 2016, Dong et al., 2019).

2. Solvable limits, hidden symmetry, and duality

The defining structural feature of the model is the coexistence of exactly solvable or symmetry-enhanced limits with intervening frustrated regimes. At $A=$9, the honeycomb model is the antiferromagnetic Heisenberg model; at $\mathrm{Ir}^{4+}$0, it is the exactly solvable Kitaev model. The original honeycomb study also identified a special solvable point at $\mathrm{Ir}^{4+}$1, where a four-sublattice spin rotation maps the Hamiltonian to an isotropic ferromagnet in rotated variables,

$\mathrm{Ir}^{4+}$2

so the exact ground state is a fully polarized ferromagnet in the rotated basis and a stripy antiferromagnetic pattern in the original spins (Chaloupka et al., 2010).

This hidden ferromagnetic structure became the basis for a broader duality framework. On the honeycomb and on other lattices built from edge-sharing $\mathrm{Ir}^{4+}$3 octahedra, site-dependent $\mathrm{Ir}^{4+}$4-rotations about spin axes define a Klein four-group structure $\mathrm{Ir}^{4+}$5. In the conventional exchange language, the duality acts as

$\mathrm{Ir}^{4+}$6

and it predicts fluctuation-free ordered states that are analogs of honeycomb stripy order on triangular, kagome, hyperkagome, fcc, and pyrochlore lattices (Kimchi et al., 2013). On the three-dimensional hyperhoneycomb lattice, a hidden four-sublattice symmetry similarly generates hidden SU(2) points at $\mathrm{Ir}^{4+}$7 and $\mathrm{Ir}^{4+}$8, in addition to the ordinary Heisenberg points at $\mathrm{Ir}^{4+}$9 and $j_{\mathrm{eff}}=1/2$0 (Fukui et al., 2023).

A recurrent implication is that apparently complicated collinear orders may become simple ferromagnetic or antiferromagnetic states in rotated variables. This suggests that the phase structure of the Kitaev-Heisenberg family is organized not only by competing exchanges, but also by nontrivial transformations of the spin basis that expose hidden isotropic points (Chaloupka et al., 2010, Kimchi et al., 2013).

3. Honeycomb phase diagram and critical behavior

For the spin-$j_{\mathrm{eff}}=1/2$1 honeycomb model in the original $j_{\mathrm{eff}}=1/2$2 interpolation, exact diagonalization on a 24-site cluster and complementary spin-wave analysis found three main zero-temperature regions: a conventional Néel phase near $j_{\mathrm{eff}}=1/2$3, a stripy antiferromagnetic phase centered around $j_{\mathrm{eff}}=1/2$4, and an extended spin-liquid phase near the Kitaev limit. The Néel-to-stripy transition was described as first-order, while the stripy-to-spin-liquid transition was described as second-order or weakly first-order based on the numerical data. The most likely transition points were identified near $j_{\mathrm{eff}}=1/2$5 and $j_{\mathrm{eff}}=1/2$6 (Chaloupka et al., 2010).

Semiclassically, the ordered phases are strongly affected by accidental degeneracies and their lifting by fluctuations. In the original honeycomb study, the classical Néel-stripy boundary sits at $j_{\mathrm{eff}}=1/2$7, where the linear spin-wave spectra develop zero-energy lines; quantum fluctuations shift the boundary upward and open a spin-wave gap, with

$j_{\mathrm{eff}}=1/2$8

near $j_{\mathrm{eff}}=1/2$9. The stripy state at the midpoint is fluctuation-free and has a saturated order parameter despite being antiferromagnetic in the original variables (Chaloupka et al., 2010).

When the full nearest-neighbor coupling circle is considered, the honeycomb phase diagram contains four magnetically ordered phases—Néel, zigzag, ferromagnetic, and stripy—separated by two Kitaev spin-liquid phases around the antiferromagnetic and ferromagnetic Kitaev points. Exact diagonalization, cluster mean-field theory, linear spin-wave theory, and second-order perturbation theory all support the sequence

$\mathrm{Ir}^{4+}$0

with the ferromagnetic-side KSL substantially broader than the antiferromagnetic-side KSL because the neighboring ferromagnetic and stripy phases have very weak quantum fluctuations (Gotfryd et al., 2016).

At finite temperature, the classical honeycomb Kitaev-Heisenberg model exhibits a three-phase structure: a low-temperature magnetically ordered phase with spontaneously broken $\mathrm{Ir}^{4+}$1 symmetry, an intermediate critical Kosterlitz-Thouless phase with emergent $\mathrm{Ir}^{4+}$2 symmetry and algebraic correlations, and a high-temperature disordered phase. Thermal fluctuations select collinear order along cubic axes by order-by-disorder, and finite-size scaling gives exponents near $\mathrm{Ir}^{4+}$3 and $\mathrm{Ir}^{4+}$4 at the lower and upper boundaries of the critical phase, as expected for six-state clock criticality (Price et al., 2012).

The nature of the quantum transition out of the spin liquid remains nontrivial. A slave-particle mean-field treatment of the transition between the gapless $\mathrm{Ir}^{4+}$5 spin liquid and stripy antiferromagnet found a discontinuous transition at mean-field level, but also argued that spinon confinement effects associated with the instability of a gapped $\mathrm{Ir}^{4+}$6 spin liquid in two spatial dimensions may be important, leaving open the possibility of a more exotic continuous transition beyond mean field (Schaffer et al., 2012).

4. Generalizations in spin, dimension, and lattice geometry

The honeycomb model has been generalized in several distinct directions. For spin-1 local moments on the two-dimensional honeycomb lattice, iDMRG on infinite cylinders finds two spin-liquid phases and four symmetry-broken phases. The spin liquids occur near the pure antiferromagnetic and ferromagnetic Kitaev points, are gapless according to finite-entanglement scaling with central charge $\mathrm{Ir}^{4+}$7, and show approximate $\mathrm{Ir}^{4+}$8 local conservations: the plaquette Wilson-loop expectation value stays near $\mathrm{Ir}^{4+}$9, and the static spin-spin correlations remain short-range in the entire spin-liquid phases (Dong et al., 2019).

In three dimensions, the spin-$t_{2g}$0 model on the hyperhoneycomb lattice exhibits a phase diagram closely paralleling the two-dimensional honeycomb case. PFFRG for the isotropic model $t_{2g}$1 finds four magnetically ordered phases—Néel AFM, zigzag AFM, FM, and stripy AFM—and QSL regions around both pristine Kitaev points, with the ferromagnetic-side QSL wider than the antiferromagnetic-side QSL. Introducing anisotropy through $t_{2g}$2 narrows the QSL region and replaces part of it by stripy order, while strong anisotropy can lead to a dimer phase (Fukui et al., 2023).

An earlier hyperhoneycomb study, motivated by $t_{2g}$3-Li$t_{2g}$4IrO$t_{2g}$5, combined semiclassical analysis, the exact solution at the Kitaev point, and slave-fermion mean-field theory. It found four collinear magnetic phases—Néel, polarized ferromagnet, skew-stripy, and skew-zig-zag—together with an extended three-dimensional $t_{2g}$6 spin liquid around the Kitaev point. In that spin liquid, the Majorana spectrum has a gapless line node described as a deformed Fermi-circle of co-dimension $t_{2g}$7 (1308.05333).

The bond-directional mechanism is not restricted to tricoordinated honeycomb-derived systems. Analogous Kitaev-Heisenberg interactions were argued to arise on triangular, kagome, hyperkagome, fcc, and pyrochlore lattices built from edge-sharing $t_{2g}$8 octahedra. On the triangular lattice, 2D DMRG found a fully magnetically ordered phase diagram with $t_{2g}$9 antiferromagnetic, $l=1$0-vortex crystal, nematic, dual $l=1$1-vortex crystal, $l=1$2 ferromagnetic, and dual ferromagnetic phases, with first-order transitions between them (Shinjo et al., 2015). On a honeycomb-triangular interpolation, classical and quantum studies found that known honeycomb and triangular phases can merge through coexistence regions such as HN-$l=1$3VC and extended nematic or stripy regimes, while quantum fluctuations were reported to affect the phase diagram only weakly (Kishimoto et al., 2018).

5. Perturbations and enriched descendants

Doping converts the spin model into a setting for unconventional superconductivity. In an $l=1$4 slave-boson mean-field treatment of the doped honeycomb model, light doping near the Kitaev limit stabilizes a triplet $l=1$5-wave superconducting state $l=1$6SC$l=1$7 that breaks time-reversal symmetry and has Chern number $l=1$8, irrespective of the sign of the Kitaev interaction. At larger doping, the antiferromagnetic Kitaev side favors a singlet $l=1$9 state and eventually an $j_{\mathrm{eff}}=1/2$0-wave state on the AF Heisenberg side, whereas the ferromagnetic Kitaev side favors a distinct time-reversal-symmetric triplet state $j_{\mathrm{eff}}=1/2$1SC$j_{\mathrm{eff}}=1/2$2, which may be topologically trivial or nontrivial depending on doping and interactions (Okamoto, 2012). In an itinerant formulation around quarter filling, strong spin-orbit coupling was argued to generate six finite-momentum inversion symmetry centers of the Fermi surface, leading to spin-triplet FFLO superconductivity with three separated degenerate ground states and finite-momentum Cooper pairing (Liu et al., 2015).

Magnetic fields strongly reorganize the phase structure. A systematic $j_{\mathrm{eff}}=1/2$3 expansion for fields along $j_{\mathrm{eff}}=1/2$4 and $j_{\mathrm{eff}}=1/2$5 showed that quantum corrections substantially modify the classical phase diagram, reduce the stability of the high-field polarized phase, and strongly suppress exotic large-unit-cell phases in a $j_{\mathrm{eff}}=1/2$6 field (Cônsoli et al., 2020). A 24-site exact-diagonalization study of the spin-$j_{\mathrm{eff}}=1/2$7 model in field found overall agreement with nonlinear spin-wave theory and identified an intermediate-field spin-disordered phase, especially robust for $j_{\mathrm{eff}}=1/2$8, where it survives well away from the pure Kitaev point and may end near a quantum tricritical point (Cônsoli et al., 2021).

Disorder changes the competition between spin liquid and magnetic order in a direction-dependent way. Lanczos exact diagonalization with random-coupling and singular-coupling disorder found that, in the nearest-neighbor honeycomb model, disorder shrinks the Kitaev spin-liquid window, broadens sharp transitions into more crossover-like features, replaces long-range zigzag and stripy order by their three domains with different ordering direction, and eventually drives the system toward a spin-glass-like state. Disorder can also close the flux gap, generate vortices in the plaquette-flux arrangement, and produce commensurate flux patterns. When second- and third-neighbor Heisenberg interactions are included, however, singular-coupling disorder can instead suppress long-range magnetic order and expand the spin-liquid regime (Singhania et al., 2022).

Additional interactions and geometry further enrich the phase diagram. Adding the bond-dependent off-diagonal $j_{\mathrm{eff}}=1/2$9 term to obtain the extended Kitaev-Heisenberg model produces a global phase diagram with eight distinct quantum phases—Kitaev spin liquid, FM, AFM, stripy, zigzag, rotated FM, rotated AFM, and a valence bond solid in a quadro-critical region—together with a dual mapping

$\tfrac12$0

that preserves the model form (Lou et al., 2015). In the ferromagnetic phase of the extended model, linear spin-wave theory finds topological magnon bands with chiral zig-zag edge states protected by non-zero Chern numbers, and for $\tfrac12$1 polarization a topological phase transition can be driven by anisotropy of the Kitaev couplings (Joshi, 2018). On the bilayer honeycomb lattice with interlayer Heisenberg exchange, large-scale iPEPS and high-order series expansions find the familiar FM, AFM, zigzag, stripy, and two Kitaev QSL phases, but also a new rung-singlet valence bond solid between AFM and stripy order, adiabatically connected to isolated Heisenberg dimers (Samimi et al., 2024).

6. Materials relevance, interpretation, and open issues

The original physical motivation was that layered iridates may realize a broad swath of the Kitaev-Heisenberg phase diagram because strong spin-orbit coupling converts the orbital structure of the $\tfrac12$2 electrons into bond-dependent magnetic anisotropy. The honeycomb model therefore proposed $\tfrac12$3IrO$\tfrac12$4 as a realistic setting in which Kitaev physics, quantum spin-liquid behavior, and stripy magnetic order can occur in the same microscopic model. At the same time, the original study emphasized that the available experimental information was insufficient to pin down the materials’ exact location in the phase diagram, and that Na/Ir site disorder may complicate interpretation because impurities can induce local moments, including in Kitaev-like systems (Chaloupka et al., 2010).

Subsequent work sharpened both the usefulness and the limitations of the model as a materials description. The finite-temperature classical study argued that the thermal phase structure of the nearest-neighbor honeycomb model is relevant to Na$\tfrac12$5IrO$\tfrac12$6 and likely to Li$\tfrac12$7IrO$\tfrac12$8, but also stated that the model alone is not sufficient to fully describe Na$\tfrac12$9IrO$_3$00, because further-neighbor couplings are needed to capture the zigzag ground state and the detailed excitation spectrum (Price et al., 2012). This suggests that the nearest-neighbor Kitaev-Heisenberg Hamiltonian is best regarded as a minimal or organizing model rather than a complete material-specific Hamiltonian.

In three-dimensional iridates such as $_3$01-Li$_3$02IrO$_3$03, $_3$04-Li$_3$05IrO$_3$06, and $_3$07-ZnIrO$_3$08, the hyperhoneycomb version provides a reference phase diagram with QSL regions near both Kitaev limits and four commensurate ordered phases, but realistic modeling likely requires additional interactions such as the $_3$09 term to account for experimentally observed incommensurate noncoplanar order (Fukui et al., 2023). Disorder studies further indicate that bond randomness can either destabilize or extend proximate spin-liquid regimes depending on whether the dominant ordered state originates from nearest-neighbor or longer-range exchanges, a point highlighted as relevant to $_3$10-RuCl$_3$11 and H$_3$12LiIr$_3$13O$_3$14 (Singhania et al., 2022).

A final interpretive issue is that the “Kitaev-Heisenberg model” now names not one unique Hamiltonian but a family of closely related models: the original spin-$_3$15 honeycomb form, full-circle nearest-neighbor variants, higher-spin and three-dimensional analogs, doped and field-driven descendants, and extended models with $_3$16, further-neighbor, disorder, or interlayer terms. What remains common across this family is the central competition between bond-directional Kitaev exchange and isotropic Heisenberg exchange, together with the hidden symmetry structures and proximate spin-liquid regimes that emerge from that competition (Chaloupka et al., 2010, Gotfryd et al., 2016, Lou et al., 2015).