Emergent Z2 Gauge Fields, Visons, and Gapped Topological Order

A deconfined Z2 gauge theory is the simplest fully gapped fractionalized phase. Its electric gauge charge $e$, magnetic flux $m$ (vison), and composite $\epsilon=e\times m$ are distinct superselection sectors; $e$ and $m$ have mutual phase $\pi$. In a spin system, $e$ is a spinon only when its symmetry-fractionalization assignment gives it the appropriate projective spin quantum number. The local Gauss law and global loop algebra—not a gauge-variant link expectation value—establish the phase.

Required background. Parton constraints supplies the enlarged Hilbert space; charges and screening supplies charge/flux response; gauge redundancy supplies physical observables.

Helpful background. Higher-form symmetry supplies loop diagnostics in the absence of dynamical endpoints.

Toric-code realization

Put Pauli matrices on square-lattice links and define

$$A_s=\prod_{\ell\ni s}\sigma_\ell^x, \qquad B_p=\prod_{\ell\in\partial p}\sigma_\ell^z, \qquad H=-K_e\sum_sA_s-K_m\sum_pB_p.$$

All $A_s$ and $B_p$ commute. The gauge-invariant ground subspace has $A_s=B_p=+1$. An open string of $\sigma^z$ creates $A_s=-1$ defects $e$ at its ends; an open dual string of $\sigma^x$ creates $B_p=-1$ visons $m$. Crossing the two strings once anticommutes, so taking $e$ around $m$ gives phase $-1$ Kitaev 2003.

On a torus, two independent noncontractible electric loops and their dual magnetic loops obey Pauli-like commutation. The ground space has four sectors in the ideal model. Local perturbations split them only by processes that wind around the system, exponentially small for system size $L\gg\xi$.

From U(1) to Z2

Pairing charge-one partons condenses a gauge-charge-two field and Higgses U(1) down to Z2. The unpaired parton becomes a Z2 charge, while a $\pi$ gauge flux is the vison. “Condensation” here refers to a gauge-fixed description; physical signatures are the surviving Z2 flux, gapped sectors, and long-range entanglement. A charge-one Higgs field would remove the residual Z2 structure.

Wilson loops distinguish deconfined and confined limits sharply only when matter cannot screen their endpoints. With dynamical fundamental charges, string breaking complicates area/perimeter laws; excitation spectra, topological sectors, and braiding remain the safer diagnosis. Vison condensation confines spinons and often produces valence-bond order when the vison carries nontrivial lattice quantum numbers.

Physical tests

A gapped spin-liquid claim should identify spinon and vison gaps, their symmetry fractionalization, topology-dependent sectors, and a gauge-invariant response. Absence of magnetic order alone is compatible with a valence-bond solid or disordered paramagnet. Finite cylinders can select one sector and show exponentially small splittings; circumference extrapolation is required.

Exercise

Why are there four toric-code ground sectors on a torus rather than two?

Solution

There are two independent noncontractible cycles. The Z2 electric flux through each can be $+1$ or $-1$, giving $2\times2=4$ sectors. Dual magnetic loops change these eigenvalues and demonstrate that no local operator connects them in the thermodynamic limit.

References

• Alexei Kitaev, “Fault-Tolerant Quantum Computation by Anyons,” Annals of Physics 303 (2003) 2–30, doi:10.1016/S0003-4916(02)00018-0.