Three-Leg Spin Graphs (3-LSGs)

• Three-leg spin graphs are quasi-one-dimensional quantum systems built from three coupled chains that exhibit geometric frustration, chirality, and diverse quantum phases.
• They are modeled via Hubbard, Heisenberg, and bilinear-biquadratic Hamiltonians to capture phenomena like gapless spin liquids, spontaneous dimerization, and charge ordering.
• Their rich phase behavior is sensitive to boundary conditions and lattice geometry, offering practical insights into deconfined excitations and topological ordering.

Three-leg spin graphs (3-LSGs) are quasi-one-dimensional quantum many-body systems built from three coupled legs, branches, or chains. In the cited literature, the designation is applied to several related but not identical geometries: triangular cylinders of width $L_y=3$, three-leg ladders, Heisenberg tubes with periodic closure around triangular rungs, twisted tubes, and junction graphs in which three open chains meet at a central node. Across Hubbard, Heisenberg, bilinear-biquadratic, Ising-Heisenberg, and double-exchange/superexchange formulations, the common structural ingredients are odd-leg connectivity, geometric frustration, chirality, and strong sensitivity to boundary conditions. These ingredients support a diverse set of phases and phenomena, including gapless spin liquids, spontaneous dimerization, Haldane-like and trivial gapped phases, incommensurate liquids, pair-density waves, Tomonaga-Luttinger liquids, exact deconfined domain walls, and symmetry-enforced degeneracies (Peng et al., 2021, Nishimoto et al., 2010, Charrier et al., 2010, Lajko et al., 2011, Yonaga et al., 2015, Martínez-Carracedo et al., 31 Jul 2025).

1. Geometries and realizations

The geometric content of a 3-LSG depends on the model class. In the Hubbard study on a three-leg triangular cylinder, the lattice is a triangular lattice wrapped onto a cylinder of width $L_y=3$ and length $L_x$ up to $128$, with open boundary conditions along the cylinder axis and periodic boundary conditions around the circumference. The basis vectors are chosen as $e_1=(1,0)$ and $e_2=(\tfrac12,\sqrt3/2)$, so each cross-section is a triangular rung of three sites (Peng et al., 2021). In Heisenberg-tube formulations, the system consists of three spin chains of length $L$, with each rung forming an equilateral triangle and periodicity in the rung direction, so that leg $3$ is coupled back to leg $1$ (Nishimoto et al., 2010, Nishimoto et al., 2011, Charrier et al., 2010).

A second important family is the twisted three-leg spin tube. There the unit cell is again a triangle of three spin-$\tfrac12$ sites, but successive triangles are cyclically shifted, so that each spin has two inter-triangle neighbors rather than one. This twist modifies both the low-energy chirality structure and the magnetization process (Yonaga et al., 2015, Ito et al., 2017). A different usage appears in graphene-based heterospin systems, where a 3-LSG consists of three open chains of lengths $L_y=3$0 joined at a central junction; in the symmetric case $L_y=3$1, and the sites host either spin-$L_y=3$2 or spin-$L_y=3$3 magnetic building blocks such as triangulenes or olympicenes (Martínez-Carracedo et al., 31 Jul 2025).

Three-leg connectivity also arises in ladder descriptions of stripe and oxide materials. In the site-centered stripe analysis of La$L_y=3$4Ba$L_y=3$5CuO$L_y=3$6, an array of three-leg ladders is used as a spin-only model of site-centered stripes (Greiter et al., 2011). In oxyborates, the relevant object is a three-leg ladder of Fe$L_y=3$7 sites with one itinerant electron per rung, where magnetic order, charge order, and lattice dimerization are treated simultaneously (Vallejo et al., 2014).

Realization	Geometry	Representative result
Triangular cylinder	$L_y=3$8, OBC along axis, PBC around circumference	Metal, gapless spin liquid, PDW/CDW under doping (Peng et al., 2021)
Heisenberg tube	Three chains with triangular rungs and rung PBC	Dimerized gapped phases and Berry-phase VBS classification (Nishimoto et al., 2010)
Twisted tube	Cyclically shifted triangular unit cells	TLLs, UUD plateau, chirality plateau (Yonaga et al., 2015)
Junction graph	Three open branches joined at a central node	$L_y=3$9-enforced double-$L_x$0 degeneracy in the first excited manifold (Martínez-Carracedo et al., 31 Jul 2025)

A plausible implication is that “3-LSG” functions less as a single microscopic lattice definition than as a structural class of three-branch frustrated spin networks. The detailed phase structure is then controlled by which degrees of freedom are retained: charge, spin, chirality, pseudospin, or lattice distortion.

2. Hamiltonians and effective descriptions

The model space of 3-LSGs is unusually broad. In the three-leg triangular-cylinder Hubbard problem, the microscopic Hamiltonian is the single-band Hubbard model

$L_x$1

with $L_x$2 setting the energy scale and $L_x$3 the on-site repulsion (Peng et al., 2021). This formulation retains itinerancy and allows direct study of Mott physics, spin-liquid behavior, and doped superconducting correlations.

In spin-only settings, the dominant starting point is a Heisenberg Hamiltonian on three coupled legs. For the three-leg $L_x$4 Heisenberg tube,

$L_x$5

with $L_x$6 on the triangular rung and $L_x$7 tunable over ferro- and antiferromagnetic values (Nishimoto et al., 2010). Variants include anisotropic rung couplings $L_x$8 (Fuji et al., 2013), easy-plane XXZ anisotropy $L_x$9 in the twisted tube (Ito et al., 2017), and integer-spin tubes with a rung-anisotropy parameter $128$0 (Charrier et al., 2010).

Several 3-LSGs admit controlled low-energy reductions. In the asymmetric three-leg tube with $128$1, each triangular rung reduces to a low-energy $128$2 doublet carrying an additional parity or chirality label, producing an effective spin-orbital chain in terms of an effective spin-$128$3 $128$4 and a pseudospin-$128$5 $128$6 (Fuji et al., 2013). In the weak-inter-triangle regime of the twisted tube under field, the low-energy theory is again a coupled spin-chirality chain, and at the $128$7 plateau the spin sector is gapped while the chirality sector reduces to an XY chain (Yonaga et al., 2015). In the projection-operator tube, triangle and square-plaquette projectors generate an exact spin-chirality description in the strong-triangle limit and an effective spin-$128$8 chain in the ferromagnetic-triangle limit (Lajko et al., 2011).

The graphene-based 3-LSGs are modeled by an isotropic bilinear-biquadratic Heisenberg Hamiltonian,

$128$9

with nearest-neighbor leg couplings and, when present, a central cross-coupling $e_1=(1,0)$0 (Martínez-Carracedo et al., 31 Jul 2025). The parameters are extracted through a DFT $e_1=(1,0)$1 $e_1=(1,0)$2 LKAG magnetic-force-theorem $e_1=(1,0)$3 exact-diagonalization workflow. By contrast, the spin-$e_1=(1,0)$4 Ising-Heisenberg three-leg tube is exactly reducible because the total spin and total $e_1=(1,0)$5 on each Heisenberg triangle are conserved. The model maps onto a classical composite spin chain and is solved by a transfer matrix (Strecka et al., 2015).

A common misconception is that all 3-LSGs are merely variants of a nearest-neighbor Heisenberg ladder. The literature instead shows a hierarchy of descriptions ranging from itinerant-electron Hubbard cylinders and double-exchange ladders to exact transfer-matrix models and first-principles bilinear-biquadratic nanographene graphs (Peng et al., 2021, Vallejo et al., 2014, Strecka et al., 2015, Martínez-Carracedo et al., 31 Jul 2025).

3. Zero-field quantum phases and criticality

The phase structure of 3-LSGs is highly model-dependent, and odd-leg geometry alone does not determine whether the system is gapless or gapped. In the half-filled Hubbard model on the three-leg triangular cylinder, three regimes are identified as $e_1=(1,0)$6 is increased: a weak-coupling metallic phase for $e_1=(1,0)$7, an intermediate gapless spin liquid for $e_1=(1,0)$8, and a strong-coupling Mott dimer phase for $e_1=(1,0)$9 (Peng et al., 2021). In the intermediate phase, $e_2=(\tfrac12,\sqrt3/2)$0 and $e_2=(\tfrac12,\sqrt3/2)$1, but $e_2=(\tfrac12,\sqrt3/2)$2, the entanglement entropy gives $e_2=(\tfrac12,\sqrt3/2)$3, spin-spin correlations decay algebraically with $e_2=(\tfrac12,\sqrt3/2)$4, and single-particle and scalar chiral correlations decay exponentially with $e_2=(\tfrac12,\sqrt3/2)$5 and $e_2=(\tfrac12,\sqrt3/2)$6. Time-reversal symmetry remains unbroken.

The spin-$e_2=(\tfrac12,\sqrt3/2)$7 triangular $e_2=(\tfrac12,\sqrt3/2)$8-$e_2=(\tfrac12,\sqrt3/2)$9 Heisenberg model on a 3-leg cylinder yields a different sequence. As $L$0 in $L$1, $L$2 is varied, the model supports a quasi-long-range $L$3 phase for $L$4, a gapless columnar phase for $L$5, a Majumdar-Ghosh dimerized phase for $L$6, and partially polarized then fully ferromagnetic phases beyond $L$7. All of these phases are planar and non-chiral, and only the transition at $L$8 is first order (Saadatmand et al., 2015).

In the projection-operator three-leg spin-$L$9 tube, three phases appear as a function of $3$0: a dimerized phase for strongly antiferromagnetic exchange on the triangles, the phase of the spin-$3$1 Heisenberg chain for ferromagnetic exchange on the triangles, and an intermediate gapless incommensurate phase for $3$2 (Lajko et al., 2011). At $3$3, the model admits exact deconfined domain-wall excitations and a gapless quadratic spectrum. This suggests that exact ground-state constructions in 3-LSGs need not rely on a single perfect valence-bond covering; superpositions of imperfect coverings can suffice.

The status of gaplessness is especially sensitive to transverse boundary conditions. In isolated three-leg ladders, the lowest mode is gapless at $3$4, reflecting the odd-leg gapless nature, and in two dimensions any $3$5 immediately generates long-range order; there is no finite critical coupling $3$6 (Greiter et al., 2011). By contrast, in frustrated three-leg tubes with periodic closure around triangular rungs, spontaneous dimerization can open a finite gap throughout the coupling regime. The spin-$3$7 tube has $3$8 for all $3$9 (Nishimoto et al., 2010), and the spin-$1$0 tube also has a finite gap for the whole coupling regime, although it is only a few percent or less of $1$1 in the weak- and intermediate-coupling regimes (Nishimoto et al., 2011).

Integer-spin tubes exhibit an additional topological structure. For integer spins $1$2, the anisotropic three-leg tube has an ensemble of $1$3 phase transitions. For $1$4, DMRG identifies two Haldane-like phases with nonzero string order separated by a trivial singlet phase, with transitions at $1$5 and $1$6 (Charrier et al., 2010). The spin gap and von Neumann entropy suggest a first-order transition but at the close proximity of a tricritical point.

4. Dimerization, Berry phases, chirality, and hidden symmetry

Spontaneous dimerization is one of the most persistent organizing principles in 3-LSGs. In the spin-$1$7 Heisenberg tube, the dimerization order parameter

$1$8

is nonzero for all $1$9, and the spin gap is finite for all $\tfrac12$0 as well (Nishimoto et al., 2010). In the spin-$\tfrac12$1 tube, the same dimerization mechanism operates, but the gap rises very slowly with increasing rung coupling and saturates only in the strong-rung limit, where $\tfrac12$2 and $\tfrac12$3 (Nishimoto et al., 2011). This distinction is quantitative rather than qualitative: both half-integer odd-leg tubes dimerize, but the $\tfrac12$4 gap is substantially harder to resolve experimentally in weak and intermediate coupling.

Berry-phase diagnostics sharpen the internal structure of these dimerized regimes. In the $\tfrac12$5 tube, quantized Berry phases on leg and rung bonds distinguish three valence-bond configurations: the diagonal-singlet regime for $\tfrac12$6, the rung-singlet regime for $\tfrac12$7, and the leg-singlet regime for $\tfrac12$8 (Nishimoto et al., 2010). In the $\tfrac12$9 tube, small-cluster Berry phases identify three topologically distinct valence-bond-solid states with patterns $L_y=3$00, $L_y=3$01, and $L_y=3$02 as $L_y=3$03 is varied (Nishimoto et al., 2011). These are recombination transitions of singlet bonds rather than simple symmetry-breaking transitions.

For integer-spin tubes, the relevant diagnostic is nonlocal string order associated with a hidden $L_y=3$04 symmetry. In the spin-$L_y=3$05 three-leg tube, the nonlocal order parameter built from rung spins $L_y=3$06 is nonzero in the two Haldane-like phases and vanishes in the trivial rung-singlet phase (Charrier et al., 2010). In open chains, the Haldane-like phases exhibit free spin-$L_y=3$07 edge states. This suggests that the topological vocabulary of one-dimensional SPT physics extends naturally into frustrated odd-leg tube geometries, although the intervening transitions can remain weakly first order.

Chirality plays two distinct roles. First, it appears as a microscopic operator, such as the scalar chirality $L_y=3$08 on a triangular plaquette (Peng et al., 2021). Second, it appears as an emergent two-level degree of freedom on a triangle in strong-rung or plateau regimes (Fuji et al., 2013, Yonaga et al., 2015). A common misconception is that triangular frustration necessarily implies chiral long-range order. The available results are more restrictive. In the half-filled Hubbard gapless spin liquid, scalar chiral-chiral correlations decay exponentially and time-reversal symmetry remains unbroken (Peng et al., 2021). In the triangular $L_y=3$09-$L_y=3$10 cylinder, both vector and scalar chirality correlators decay to zero, so the phases are non-chiral and planar (Saadatmand et al., 2015).

5. Doping, magnetic field, and intertwined orders

When charge degrees of freedom are present, 3-LSGs support intertwined orders not available in pure spin models. Light doping of the Hubbard gapless spin liquid on the three-leg triangular cylinder, for $L_y=3$11, produces power-law charge-density-wave correlations and quasi-long-ranged superconducting correlations consistent with a striped pair-density wave (Peng et al., 2021). The rung density oscillates at wavevector $L_y=3$12, with

$L_y=3$13

and the extracted Luttinger exponent is $L_y=3$14. Equal-time singlet pair-field correlations oscillate in sign with wavevector $L_y=3$15, so $L_y=3$16; for $L_y=3$17 and $L_y=3$18, one finds $L_y=3$19 and $L_y=3$20, respectively. The superconducting exponents are $L_y=3$21 for $L_y=3$22 bonds and $L_y=3$23 for $L_y=3$24 bonds, while spin, chiral, and single-particle correlators remain short-ranged. At higher doping, or when the doped state descends from the dimer phase, a pure CDW regime appears with $L_y=3$25 and exponentially decaying non-charge correlators.

Magnetic fields reveal a different hierarchy of gapless and gapped sectors. In the twisted three-leg spin tube, DMRG with sine-square deformation identifies four field-induced regimes: a one-component TLL (TLL1) behaving as a spin-$L_y=3$26 Heisenberg model at strong inter-triangle coupling, a $L_y=3$27-sublattice UUD state with $L_y=3$28 magnetization, a $L_y=3$29-plateau phase that is a TLL of massless chirality in the weak-inter-triangle regime, and a TLL of massless spin mode with or without chirality quasi-long-range order (Yonaga et al., 2015). The boundary between the UUD plateau and the chirality plateau lies near $L_y=3$30. Below and above the weak-coupling plateau, both spin and chirality are gapless, giving a two-component TLL with $L_y=3$31.

The $L_y=3$32 plateau itself is not unconditionally stable. In the easy-plane twisted tube, the Oshikawa-Yamanaka-Affleck condition allows a $L_y=3$33 plateau because $L_y=3$34 with $L_y=3$35, but numerical diagonalization shows a BKT transition between plateau and plateauless regimes as the anisotropy $L_y=3$36 is tuned (Ito et al., 2017). Representative lower-bound critical values are $L_y=3$37 for $L_y=3$38, respectively. This suggests that the plateau is stabilized by a balance of frustration and anisotropy rather than by geometry alone.

Three-leg ladder models of materials introduce an additional spin-charge-lattice channel. In the Fe-based oxyborate ladder with one itinerant electron per rung, the competition between double exchange and antiferromagnetic superexchange produces a ferromagnetic phase, a phase with ferromagnetic rungs ordered antiferromagnetically, and a zig-zag canted phase (Vallejo et al., 2014). Each magnetic pattern induces a distinct charge distribution. The $L_y=3$39 phase gives exactly $L_y=3$40, $L_y=3$41, while the zig-zag $L_y=3$42 phase yields $L_y=3$43. Coupling to a rung distortion $L_y=3$44 with elastic energy $L_y=3$45 shows that the zig-zag phase strongly softens the lattice and can drive spontaneous distortion.

6. Exact solvability, entanglement structure, and discrete symmetry

Some of the sharpest 3-LSG results arise in models where frustration coexists with exact constraints. In the spin-$L_y=3$46 Ising-Heisenberg three-leg tube, the conservation of each triangle’s total spin $L_y=3$47 and $L_y=3$48 maps the quantum problem to a classical composite spin chain with an $L_y=3$49 transfer matrix (Strecka et al., 2015). The model has three zero-temperature phases: classical antiferromagnetic (CAF), symmetric quantum trimerized (SQT), and chiral antiferromagnetic (DCA). Thermal entanglement between two spins within a triangle, measured by the concurrence, appears only in the frustrated region of parameter space. At the same time, the Bell function satisfies $L_y=3$50 at all temperatures, so pairwise entanglement does not imply Bell nonlocality. The specific heat can show up to three peaks, including a sharp low-temperature peak originating from massive excitations into the highly degenerate chiral DCA manifold.

Boundary conditions have a similarly strong effect in entanglement-oriented Heisenberg ladders. In the antiferromagnetic spin-$L_y=3$51 three-leg Heisenberg ladder with alternating intraleg coupling $L_y=3$52, open and cylinder boundary conditions lead to qualitatively different concurrence patterns (Li et al., 2024). Under OBC and $L_y=3$53, intraleg concurrence shows odd-even alternation; increasing $L_y=3$54 reverses the concurrence distribution, and a phase transition point is predicted near $L_y=3$55. The same $L_y=3$56 induces two types of long-distance entanglement under OBC, intraleg and inter-leg, which saturate to a constant for sufficiently large system size. Under CBC, however, frustration inhibits long-distance entanglement altogether, while only edge-localized interleg concurrence pockets may appear. This provides a direct entanglement-level demonstration that transverse periodic closure is not a minor technicality in 3-LSGs.

The projection-operator tube supplies an exact example of deconfined excitations. At $L_y=3$57, two kinds of domain walls become exact, deconfined zero-energy excitations above the dimerized background, and the one-domain-wall dispersion

$L_y=3$58

has a lower branch that vanishes at $L_y=3$59 (Lajko et al., 2011). On even-length tubes, a third exact ground state at $L_y=3$60 can be built from two deconfined domain walls. This is a particularly explicit realization of fractionalized low-energy objects in a short-range frustrated three-leg geometry.

In graphene-based quantum heterospin graphs, the decisive structure is discrete swapping symmetry rather than dimerization. For fully symmetric couplings $L_y=3$61 and identical spin species on all legs, the first excited manifold is two-fold degenerate in total spin $L_y=3$62 for all six 3-LSGs with $L_y=3$63 and spin-$L_y=3$64 or spin-$L_y=3$65 building blocks (Martínez-Carracedo et al., 31 Jul 2025). The commuting leg-swapping operators generate a group $L_y=3$66, and the numerically observed first-excited states furnish its two-dimensional irreducible representation $L_y=3$67. The resulting “double-$L_y=3$68” degeneracy is distinct from the usual $L_y=3$69 spin-multiplet degeneracy. A plausible implication is that 3-LSGs occupy an intermediate position between conventional ladders and finite molecular spin graphs: local frustration can coexist with a finite non-Abelian spatial symmetry that leaves an imprint directly on the many-body spectrum.