Topological Insulator: Bulk-Edge Correspondence

A chain with two different bonds

The Su-Schrieffer-Heeger (SSH) model, introduced in 1979 to explain conduction in the polymer polyacetylene, is a one-dimensional chain of sites with alternating hopping (bond) strengths: a strong bond t1, then a weak bond t2, then strong again, repeating down the chain. It is about as simple a quantum lattice model as exists, and it is the cleanest possible place to see topological physics without the complications of two- or three-dimensional band structure.

live demo · an SSH chain with a protected edge state● LIVE

A gap that can wind around zero, or not

In an infinite periodic chain, the SSH Hamiltonian in momentum space is a 2×2 matrix whose off-diagonal entry traces a small closed loop in the complex plane as the crystal momentum k sweeps through the Brillouin zone:

h(k) = t1 + t2 * e^(i·k)          the off-diagonal element traces a circle in ℂ as k runs 0..2π

winding number w = (1/2π) ∮ d(arg h(k))

t2 < t1  →  the circle does NOT enclose the origin  →  w = 0   (trivial)
t2 > t1  →  the circle DOES enclose the origin      →  w = 1   (topological)

The winding number w counts how many times this loop wraps around the origin, and it can only ever be an integer — 0 or 1 for the SSH model. That integer is the topological invariant: a global property of the entire band, insensitive to the fine details of the Hamiltonian, that can only change if the loop is deformed through the origin itself — which happens exactly when the energy gap closes, at t1 = t2.

Bulk-edge correspondence

Here is the payoff. Take a finite chain that starts weak-bond-first (w = 1, topological) and imagine gluing it to vacuum, or to another finite chain that starts strong-bond-first (w = 0, trivial). The winding number cannot change continuously — it is an integer — so somewhere between the two regions the gap that separates the upper and lower energy bands must close. A closed gap at that boundary means there exist states with energy inside what would otherwise be the forbidden gap, and those states are necessarily localised right at the interface: edge states. This is the bulk-edge correspondence: the difference in a purely bulk, translation-invariant quantity (the winding number on either side) guarantees the existence of localised states at the boundary between them, with no need to solve the boundary problem directly.

Why the edge state is protected

An ordinary surface state — say, a dangling bond at the end of a normal crystal — can be pushed out of the gap or hybridised away by almost any local perturbation: a stray impurity, a slightly different bond length, thermal disorder. The SSH edge state is fundamentally different because its existence is tied to a global topological quantity. To remove it, you would have to close the bulk gap somewhere and change the winding number — and no small, local change at the edge alone can do that, because the winding number is computed from the entire periodic bulk far away from any single defect. This is the sense in which topological edge states are called protected: not immune to everything, but immune specifically to any perturbation that is local and does not close the bulk gap, including moderate structural disorder, missing atoms, or small variations in the individual bond strengths.

From a 1D toy to real topological materials

The SSH chain is the one-dimensional ancestor of the topological insulators discovered in real crystals starting in the mid-2000s — three-dimensional materials like Bi₂Se₃ that are insulating in the bulk but conduct along their surface through states protected by a more elaborate topological invariant built from spin-orbit coupling and time-reversal symmetry rather than a simple winding number. The core logic is identical: a bulk topological invariant that differs between the material and the vacuum outside it forces conducting states to exist at the boundary, and those states survive disorder that would destroy an ordinary metal's surface conductivity. The same winding-number mathematics from the SSH model has also been realised directly and cleanly in photonic waveguide arrays and cold-atom optical lattices, where researchers can literally watch a topological edge mode light up at the boundary between two differently-wound regions.

Frequently asked questions

What makes an insulator topological rather than ordinary?

Both are gapped in the bulk — no low-energy states to conduct through. The difference is a global invariant of the bulk band structure, in the SSH model whether the winding number is 0 or 1. Two insulators with different invariants cannot be smoothly deformed into each other without closing the gap, and that topological distinction is what forces edge states to appear at any boundary between them.

Why are topological edge states called protected?

Because removing them requires closing the bulk energy gap somewhere, which is a global, non-local operation. Local perturbations at the edge — disorder, a slightly wrong hopping value, a missing atom — cannot do this, so the edge state's existence, and often its energy, survives changes that would easily destroy an ordinary surface state. Only something that closes the gap in the bulk (or connects the edge to another edge of matching character) can remove it.

Is the SSH model just a toy, or does it describe real materials?

Both. It was introduced in 1979 to describe soliton excitations in polyacetylene, a real conducting polymer with genuinely alternating bond lengths, so it has direct experimental relevance. It has since become the standard minimal toy model for teaching topological band theory, and its physics has been reproduced directly in photonic waveguide arrays, cold-atom optical lattices and mechanical metamaterials, which is why it remains the first stop for anyone learning the field.

Try it live

Everything above runs in your browser — open Topological Insulator and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.

▶ Open Topological Insulator simulation

Topological Insulators: When Geometry Refuses to Let a Boundary Be Boring