SSH chain and winding number
Compute a chiral two-band winding invariant while keeping the unit cell, Zak-phase convention, and physical termination explicit.
On this page
The question for today
What you’ll understand
- Classify a fixed-cell SSH model by its planar winding.
- Derive the zero-mode recurrence for a specified open termination.
- State which chiral, gap, and unit-cell assumptions protect the conclusion.
The SSH chain alternates two hopping amplitudes between two sublattices. Its simplest band Hamiltonian traces a closed curve in a plane as momentum crosses the Brillouin zone. Whether that curve encloses the origin defines a winding number when chiral symmetry keeps it in the plane. Turning that integer into an edge-state prediction requires the same unit-cell convention and an actual termination; moving a mathematical cell boundary alone does not change a physical specimen.
Fix the cell and the oriented loop
Take t1 as intracell hopping and t2 as hopping from one cell to the next, both real and positive. Measure momentum in dimensionless inverse-cell units. The two Pauli coefficients trace a circle centred at t1 on the horizontal axis with radius t2. Defining winding using dx plus i dy makes the loop counterclockwise as k increases. It encloses the origin once when t2 exceeds t1 and not at all when t1 exceeds t2.
The Hamiltonian anticommutes with the sublattice Pauli matrix, placing its spectrum symmetrically about zero. Its two bands have energies plus and minus the distance of that circle from the origin. At equal hoppings the loop touches the origin, the gap closes, and the winding is not defined. Adding a staggered sublattice potential moves the Pauli vector out of the plane, allowing interpolation without this chiral winding protection.
The end mode solves a geometric recurrence
Open the chain by ending on an A site and removing the intercell bond that would have connected it to a cell on its left. At zero energy a candidate state supported only on A sites makes each B-site equation a cancellation between neighbouring A amplitudes. The resulting ratio is minus intracell hopping divided by intercell hopping. It produces an exponentially localized semi-infinite state only when its magnitude is below one.
The opposite end supports the complementary sublattice state under the matching termination. In a finite balanced chain these modes can overlap and split into a symmetric positive/negative energy pair, with splitting exponentially small at long length. Chiral symmetry pairs energies but does not magically forbid overlap across a narrow sample. A different end cut can leave no exposed weakly bonded site for the same physical bulk pattern, so the termination must be specified alongside the claimed edge state.
Zak phase remembers an embedding convention
For this two-band chiral convention, the lower-band Zak phase is pi times the winding modulo two pi. The Zak phase is a Brillouin-cycle Berry phase, not automatically a cell-independent absolute polarization. Changing the unit cell or shifting orbital embeddings changes the representation and how boundary charge is assigned. Comparing two written winding numbers therefore requires matching conventions, just as comparing two edge charges requires matching the termination and ionic background.
Disorder that preserves the relevant chiral structure can retain an appropriate real-space topological distinction without a momentum loop. Generic onsite potentials break that protecting symmetry, and sufficiently strong changes can eliminate the simple zero-energy statement. Interacting chains also require a many-body formulation rather than a literal single-particle amplitude recurrence. The clean SSH model teaches a controlled mechanism: a gap, a symmetry constraint, and a localized solution tied to a concrete boundary.
Work it through
Normalize a semi-infinite edge mode
For the stated termination use t1=1.0 meV and t2=2.0 meV. Find winding, bulk band gap, localization length in cells, and probability on the third A site of the normalized zero mode.
The circle has centre 1 and radius 2 in energy units, so it surrounds the origin and w=1. The closest energy magnitude is |t2−t1|=1 meV, giving a full two-band gap of 2 meV.
The recurrence ratio is −1/2 and the normalized first amplitude is √3/2. The decay length is cells, obtained from the exponential envelope rather than its alternating sign.
The third A site has amplitude . Its probability is therefore 3/64≈0.0469. Summing all A-site probabilities gives .
The fixed-cell phase has w=1, a 2 meV full gap, localization length 1.443 cells, and third-A-site probability 3/64 for the semi-infinite zero mode.
The living notebook
Explore the model
Interactive lab · optimization
Test the edge-state recurrence mathematically
The lab recurrence is xₙ=(1−η)ⁿx₀. Reset and set both controls to η=1.5 and x₀=1; one Step gives x₁=−0.5 and a second gives x₂=0.25, matching unnormalized SSH amplitude ratios r=0.5. For a fresh comparison, Reset and set both controls to η=2.5 and x₀=1; one Step gives −1.5, a growing rather than normalizable edge ansatz.
- Position x0
- 2
- Objective x²/2
- 2
- Iteration behavior
- Stable step
Gradient descent on f(x) = x²/2: xₙ = (1 − η)ⁿx₀. Playback advances two discrete iterations per second. Values outside |x| ≤ 4 are explicitly reported. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.
Look forA geometric ratio controls localization. This is a gradient-descent mathematical analogy, not an SSH Hamiltonian solver; its objective readout is not an electron energy or an edge-state probability.
Pause & reason
Check your understanding
Your turn
Try it yourself
Keep the same cell convention and termination but choose t1=3.0 meV and t2=1.0 meV. Find winding and full band gap, and decide whether the stated A-only edge ansatz is normalizable.
Take a moment to make a prediction or work it out on paper. Open a hint when you need one.
Give me a hint
Check whether the hopping circle contains zero, then examine the magnitude of the zero-mode recurrence ratio.
Compare with the solution
The circle centred at 3 with radius 1 excludes zero, giving w=0. The full gap is meV. The candidate A amplitude ratio is −3, whose magnitude exceeds one, so its semi-infinite squared norm diverges unless every amplitude vanishes. This termination therefore has no such normalizable zero mode. Relabelling cells without changing the physical cut does not manufacture a new localized state; the boundary interpretation must transform with the convention.
Keep the idea
Take it with you
- SSH winding requires a fixed oriented unit cell and a chiral gapped Hamiltonian.
- The zero-mode recurrence is normalizable only when its amplitude ratio has magnitude below one.
- Zak phases and edge predictions must retain the orbital embedding and physical termination conventions.
Read further
Original sources and deeper explanations to follow at your own pace.
Optional progress, saved in this browser when available. Revisit any lesson at any time.
Lesson not marked complete.