One-dimensional spin chains and the Haldane gap

Consider dimensionless spins of fixed representation S with positive nearest-neighbor exchange J. The Hamiltonian is rotationally invariant, with no dimerization, single-ion anisotropy, frustration, or applied field. For the uniform spin-half chain, the exact solution gives a gapless spectrum and algebraically decaying ground-state spin correlations, with logarithmic corrections. Its elementary fractional-spin excitations are spinons; describing all its low-energy spectral weight as one conventional noninteracting magnon would miss the continuum structure.

Haldane's field-theory argument predicts a qualitative contrast between integer and half-odd-integer antiferromagnetic Heisenberg chains. The simplest spin-one model has a gapped, short-range-correlated bulk ground state. This is not a blanket claim that every integer-spin Hamiltonian is gapped or every half-integer Hamiltonian must be a unique gapless phase. Changing interactions can produce dimerization, order, or other phases. A translation-symmetric half-integer chain cannot simply acquire a unique fully symmetric short-range gapped state under the usual locality assumptions; degeneracy can provide another possibility.

Numerical studies of the spin-one chain find a bulk gap approximately 0.4105 times J and a finite correlation length. An open chain also carries effective spin-half degrees of freedom near its ends. Their coupling becomes exponentially small as the chain grows. Thus an open-chain calculation can show a very low first excited level that tends toward the ground level while bulk excitations remain separated by a finite energy. The lowest number in a spectrum is not automatically the bulk gap.

Periodic boundaries remove the physical ends, while careful open-chain analysis can identify edge sectors or pin boundary spins. Correlation functions and the spatial profile of an excitation supply further diagnostics. The spin-one Haldane phase has a symmetry-protected characterization: its robust distinction from a trivial product phase requires specified protecting symmetries, such as time reversal or pi rotations about two orthogonal spin axes, and a maintained bulk gap. Edge degeneracy can be split by appropriate boundary perturbations, so an observed boundary splitting should not be treated as an immutable universal number.

A nonzero gap in a finite spin-half chain can arise simply from discrete allowed momenta and disappear when length increases. Conversely, exponentially small edge splittings in a spin-one chain need not close its bulk gap. Extrapolation therefore needs a physically justified finite-size form, boundary control, and a distinction between bulk and edge-sensitive observables. Adding more short-wavelength basis modes at fixed length improves resolution but is not the same operation as increasing the physical chain length.

The Mermin–Wagner restriction concerns spontaneous continuous-symmetry breaking at positive temperature in low dimensions, and does not decide whether a zero-temperature chain is gapped or gapless. At temperatures well below the Haldane gap, bulk excitations are thermally rare, often with an activation factor involving the gap, but dispersion and matrix elements supply additional prefactors. Edge moments and impurities may dominate the measured susceptibility in this regime. A small activated bulk contribution and an observable low-energy edge response can therefore coexist.