Table of Contents
Fetching ...

A simple anisotropic three-dimensional quantum spin liquid with fracton topological order

Olga Petrova, Nicolas Regnault

TL;DR

The authors construct a simple 3D spin model on an $L_x\times L_y\times L_z$ lattice, anisotropic in $z$, that hosts fracton topological order by coupling adjacent toric-code planes into eight-spin cube operators. The Hamiltonian $H=-\sum_c\hat{Q}_c$ yields immobile fractons at cube corners and two-dimensional string-ends that form dimension-2 excitations, along with dimension-1 excitations propagating along $\hat{z}$; the ground-state degeneracy is $\mathcal{D}=2^{L_xL_y-2+2L_z}$, decomposable into a topological part $2^{2L_z}$ and a non-topological part $2^{L_xL_y-2}$. The model exhibits mutual semion statistics between the dimension-2 flavors $e$ and $m$, and the excitations can be combined into mobile composites whose motion is restricted to either straight lines along $\hat{z}$ or the $xy$ plane. Stability analysis via a tunable coupling to decoupled toric-code planes shows fracton order is a fine-tuned point, while open boundaries generate characteristic surface modes: zero-energy modes on $x$/$y$ boundaries and no such modes on $z$ boundaries, explained by the structure of the composite excitations. Together, the work clarifies how a simple, anisotropic 3D lattice model can realize fracton order and illuminates the rich boundary physics that accompanies bulk fracton topological phases.

Abstract

We present a three-dimensional cubic lattice spin model, anisotropic in the $\hat{z}$ direction, that exhibits fracton topological order. The latter is a novel type of topological order characterized by the presence of immobile pointlike excitations, named fractons, residing at the corners of an operator with two-dimensional support. As other recent fracton models, ours exhibits a subextensive ground state degeneracy: On an $L_x\times L_y\times L_z$ three-torus, it has a $2^{2L_z}$ topological degeneracy, and an additional non-topological degeneracy equal to $2^{L_xL_y-2}$. The fractons can be combined into composite excitations that move either in a straight line along the $\hat{z}$ direction, or freely in the $xy$ plane at a given height $z$. While our model draws inspiration from the toric code, we demonstrate that it cannot be adiabatically connected to a layered toric code construction. Additionally, we investigate the effects of imposing open boundary conditions on our system. We find zero energy modes on the surfaces perpendicular to either the $\hat{x}$ or $\hat{y}$ directions, and their absence on the surfaces normal to $\hat{z}$. This result can be explained using the properties of the two kinds of composite two-fracton mobile excitations.

A simple anisotropic three-dimensional quantum spin liquid with fracton topological order

TL;DR

The authors construct a simple 3D spin model on an lattice, anisotropic in , that hosts fracton topological order by coupling adjacent toric-code planes into eight-spin cube operators. The Hamiltonian yields immobile fractons at cube corners and two-dimensional string-ends that form dimension-2 excitations, along with dimension-1 excitations propagating along ; the ground-state degeneracy is , decomposable into a topological part and a non-topological part . The model exhibits mutual semion statistics between the dimension-2 flavors and , and the excitations can be combined into mobile composites whose motion is restricted to either straight lines along or the plane. Stability analysis via a tunable coupling to decoupled toric-code planes shows fracton order is a fine-tuned point, while open boundaries generate characteristic surface modes: zero-energy modes on / boundaries and no such modes on boundaries, explained by the structure of the composite excitations. Together, the work clarifies how a simple, anisotropic 3D lattice model can realize fracton order and illuminates the rich boundary physics that accompanies bulk fracton topological phases.

Abstract

We present a three-dimensional cubic lattice spin model, anisotropic in the direction, that exhibits fracton topological order. The latter is a novel type of topological order characterized by the presence of immobile pointlike excitations, named fractons, residing at the corners of an operator with two-dimensional support. As other recent fracton models, ours exhibits a subextensive ground state degeneracy: On an three-torus, it has a topological degeneracy, and an additional non-topological degeneracy equal to . The fractons can be combined into composite excitations that move either in a straight line along the direction, or freely in the plane at a given height . While our model draws inspiration from the toric code, we demonstrate that it cannot be adiabatically connected to a layered toric code construction. Additionally, we investigate the effects of imposing open boundary conditions on our system. We find zero energy modes on the surfaces perpendicular to either the or directions, and their absence on the surfaces normal to . This result can be explained using the properties of the two kinds of composite two-fracton mobile excitations.

Paper Structure

This paper contains 11 sections, 8 equations, 9 figures.

Figures (9)

  • Figure 1: Operator definitions for the 3D model. The cube operator (a) $\hat{Q}_c$ is the product of the two plaquette operators at adjacent $xy$ planes: (b) $\hat{Q}_{xy}(z)$ and (c) $\hat{Q}_{xy}(z+1)$.
  • Figure 2: Starting with a system in a ground state, acting on the center spin with $\hat{Z}$ creates four $Q_c=-1$ excitations: excited cubes (dark grey and light grey) are diagonal from each other in the $xy$ plane and stacked one on top of the other along the $\hat{z}$ direction.
  • Figure 3: (a) Dimension-2 particles (AB and CD, pairs of cubes with $Q_c=-1$ sharing an $xy$ face) are located at the ends of an open string operator (dotted line), which is realized as a product of $\hat{X}$ and $\hat{Z}$ indicated by filled and empty circles respectively. (b) Pairs of cubes making up dimension-2 particles can be split into individual fractons (A, B, C, and D) by stacking open strings in adjacent $xy$ planes such that their endpoints are directly above one another. The open strings can be freely deformed apart from their endpoints, and are indicated by the dotted lines. (c) Dimension-1 particles are pairs of cubes, diagonal from one another, at the same height $z$: AC and BD.
  • Figure 4: (a) Two dimension-2 particles of different flavors, each consisting of a pair of adjacent excited cubes, are shown. The red ($m$ flavored) and blue ($e$ flavored) strings schematically indicate the presence of a second dimension-2 particle of the same flavor at the other end of the string. (b) The red dotted string indicates that the $m$ flavored dimension-2 particle has been moved around the $e$ particle in a closed loop. The green cross designates the crossing of the $e$ and $m$ strings. (c) Both $m$ (red) and $e$ (blue) dimension-2 particles take the form of rods extended along the $\hat{z}$ direction. (d) The four dotted red cycles stacked on top of one another along $\hat{z}$ depict moving of the $m$-flavored rod in space in a closed loop.This winding process leads to three $e$-$m$ string crossings, indicated by green crosses.
  • Figure 5: Upper panel : Partial lifting of the ground state degeneracy of Eq. (\ref{['eq:Hdecoupled']}) as a function of $\lambda$ for a system with $L_x\times L_y=20$. The energy level shown in red corresponds to the $2^{2L_z}$ ground states of the fracton model (\ref{['eq:H3d']}) with all $Q_{xy}$ quantum numbers equal to $+1$. The gap to the other states indicated in blue closes at the fracton point, $\lambda=0$. Inset shows how the first three excited energy levels of the fracton model are connected to the lowest five excited energy levels of the decoupled toric code planes. Lower panel : Various examples of excited states. Excited cubes are in red and excited plaquettes have a red cross. Red crosses connected by a string lie in the same $xy$ plane. (b) and (c) have four excited cubes (corresponding to $E(\lambda=0)=4$) while having a different number of excited plaquettes corresponding respectively to $E(\lambda=1)=2$ and $E(\lambda=1)=4$. (d) and (e) have six excited cubes (corresponding to $E(\lambda=0)=6$) but have a different number of excited plaquettes corresponding respectively to $E(\lambda=1)=4$ and $E(\lambda=1)=6$.
  • ...and 4 more figures