The same arrow picture hides two quantum chains
Draw a row of atomic magnets with neighbours pointing in opposite directions. Their magnetic moments measure the strength and direction of these tiny magnets. The drawing looks similar whether each site carries spin one or spin one half. Yet an ideal antiferromagnetic chain can respond very differently at low energy in those two cases. The spin-one chain has a finite threshold for changing the magnetic motion in the chain’s interior, called the Haldane gap. The uniform spin-one-half Heisenberg chain has excitations at arbitrarily small energy in the long-chain limit.
Duncan Haldane predicted the integer-versus-half-integer distinction in 1983 from a continuum description of quantum spin dynamics. The surprise was that increasing the size of a spin did not simply make every property approach a classical result smoothly. The quantum number itself left a qualitative trace. To understand the claim, one must specify how the spins interact and what experiment means by an excitation gap.
Sources: [2][3]
Spin is a quantum degree of freedom, not a little spinning ball
A spin quantum number specifies the allowed angular-momentum states at one site. Spin one has three possible projections along a chosen axis; spin one half has two. These projections do not mean that a tiny rigid ball is turning at three or two speeds. Different components of quantum angular momentum cannot all have definite values at once. Neighbours that favour opposite directions therefore retain quantum fluctuations even at zero temperature.
The reference model has identical nearest-neighbour antiferromagnetic couplings and no preferred spatial direction for spin. The antiferromagnetic exchange constant is positive in the convention used here and has energy units; it sets the preference for opposing neighbours. There is no alternating strong and weak bond, and chains do not couple to one another. These are physical restrictions: bond alternation, anisotropy or additional interactions can change a gap. The integer-versus-half-integer distinction applies to this specified uniform chain; other couplings can give even a half-integer material a gap.
Sources: [2]
The gap concerns a long chain and its interior
An excitation gap is an energy difference above the ground state. For the Haldane question, the relevant difference must stay positive as the chain becomes arbitrarily long. Every finite numerical chain has discrete energy levels. A small gap between its first two levels can therefore reflect the limited length instead of a bulk gap. Open ends can add still lower energy states, so boundary excitations must be separated from excitations of the interior.
Haldane’s field theory kept quantum phase contributions that a classical arrow calculation loses. Their effect differs between integer and half-integer spin. Numerical tests were especially important for spin one because a continuum argument developed from large spins was being applied to a small quantum spin. A successful test needed a sequence of chain lengths, controlled extrapolation and checks of correlations. Following how the lowest bulk excitation changes with length distinguishes the two possible long-chain behaviours.
Sources: [2][3]
Numerics turned a qualitative prediction into an energy scale
In 1993 Steven White and David Huse used the density-matrix renormalization group to study the spin-one chain with high precision. They found a bulk gap about 0.41050 times the nearest-neighbour exchange energy, with exponentially decaying spin correlations. Their open chains also showed effective spin-one-half moments near the ends. Bulk and boundary properties were different parts of one consistent quantum state.
The formula names that bulk energy threshold and the nearest-neighbour exchange energy; both must use the same energy unit. The ratio gives a useful scale estimate. If an ideal model has exchange energy ten millielectronvolts, its bulk gap is about 4.1 millielectronvolts, equivalent to roughly forty-eight kelvin when divided by Boltzmann’s constant. That conversion is an energy scale, not a prediction that the material undergoes a thermal phase transition at forty-eight kelvin. Real directional preferences can split three originally equivalent excitations, and weak coupling between chains can create three-dimensional order. Their effects must be assessed before comparing a material with the isotropic number.
Sources: [3]
Neutrons measure the energy and momentum of magnetic motion
A neutron can give energy and momentum to a magnetic crystal. Measuring its incoming and outgoing motion reveals the transfer to the sample. Inelastic neutron scattering maps magnetic excitation energy against momentum along the chain. The gap is the minimum energy of the relevant bulk mode, not a missing elastic diffraction peak and not the height of a susceptibility curve.
Buyers and colleagues reported neutron evidence in 1986 for the spin-one antiferromagnet CsNiCl3 in its one-dimensional phase. Its lowest spin excitations occurred at finite frequency. Allowing for the known weak interchain coupling brought the gap into agreement with numerical expectations. This qualification was part of their interpretation: the experiment used a real crystal with neighbouring chains, not an isolated mathematical chain. A finite-frequency signal was strong evidence, but identifying its origin still required the exchange geometry and the interchain correction.
Sources: [4]
NENP supplied a separate material test
Another important system was the nickel-chain compound known as NENP. Renard, Verdaguer, Regnault and collaborators reported magnetic susceptibility and inelastic-neutron results in 1987 consistent with a quantum gap. The experimental team’s later ILL account describes work on NENP at the Siloé reactor and at ILL. This was a distinct material investigation after the CsNiCl3 result; merging them into one experiment obscures how the evidence developed.
Susceptibility asks how magnetization changes in a weak applied field. A gapped singlet ground state, whose total spin is zero, has few thermally populated magnetic excitations at sufficiently low temperature, so its intrinsic susceptibility becomes small. Defects and chain ends can add low-temperature magnetic tails. A decrease in bulk susceptibility alone could have several causes. Combining it with a momentum-resolved finite-energy mode, absence or presence of three-dimensional order, and a model of anisotropy makes the gap interpretation much more selective.
For the illustrative gap equivalent to forty-eight kelvin, the simple Boltzmann factor is about eight thousandths at ten kelvin. This estimates suppression of thermal excitation, not a complete susceptibility: degeneracies, dispersion and prefactors still matter. An impurity signal can dominate precisely because the intrinsic contribution has become so small.
Sources: [5][6]
Valence bonds explain why the ends can behave differently
A related exactly constructed model gave a clear picture of a gapped spin-one state. Affleck, Kennedy, Lieb and Tasaki represented each spin one using two auxiliary spin halves and joined halves on neighbouring sites into singlet bonds. A singlet has no preferred spin direction. An open chain leaves an unpaired auxiliary half at each end, which explains how boundary moments can coexist with a nonmagnetic, gapped interior.
Their rigorous valence-bond model has a specially chosen interaction. Its proof does not automatically prove every statement for the nearest-neighbour Heisenberg Hamiltonian. It nevertheless supplied a concrete state with a gap, exponentially decaying correlations and no conventional magnetic order. The edge picture and the numerical Heisenberg results made the phase intelligible beyond a classical Néel cartoon. Modern discussions also ask which symmetries, such as reversing all spins, preserve the distinction from an ordinary nonmagnetic gapped state. The protected distinction concerns how the quantum state is organized, beyond the size of its gap.
Sources: [7][3]
What a simple chain taught us
The theory changed the role of a simple chain. Alternating classical arrows capture the preferred neighbour direction, but quantum phases determine whether the interior has a finite excitation threshold. Numerical extrapolation, neutron energy transfer and boundary moments make different aspects of that state accessible. Spin chains thus became a precise setting in which to connect local interactions with a collective quantum organization, even when the material has little conventional magnetic order.
Sources: [2][1]