Skip to content

Topological dislocation response in elementary semiconductors

FieldValue
JournalPhysical review. B./Physical review. B
Published2026-07-09
AuthorsAnonymous, Alexandre Chaduteau, Frank Schindler
DOI10.1103/yg53-t56x

Edge dislocations produced a nontrivial topological response in all three modeled 3D semiconductors: silicon, diamond, and germanium. Screw dislocations produced a trivial response. Realistic tight-binding calculations found pairs of inversion-related polarization bands inside the bulk gap for edge dislocations. The response originates in the obstructed atomic insulating phase. It also makes defect orientation a material and device design variable because edge and screw dislocations behave differently.

The four materials are modeled as inversion- and time-reversal-symmetric semiconductors. Silicon, diamond, germanium, and black phosphorene are obstructed atomic insulators in the authors’ models. Their symmetric Wannier functions are exponentially localized, but their electronic centers are shifted from the atomic lattice. For the 3D materials, all four Fu-Kane invariants remain trivial, so the work predicts no strong topological insulator surface states.

The authors instead calculate weak SSH symmetry indicators from occupied inversion eigenvalues at time-reversal invariant momenta. These indicators encode the quantized spatial shift of Wannier centers. A dislocation cut through such a center can bind electronic states when conventional band topology is trivial. The resulting polarization bands differ from the gapless helical dislocation modes discussed for conventional topological insulators.

At neutral filling, states associated with an inversion-related pair of dislocations are half-filled on average. This filling anomaly creates a ground-state degeneracy that can be lifted only by breaking inversion symmetry or closing the bulk gap. The anomaly persists if the states move outside the gap and hybridize with bulk states. A floating band inside the gap is possible, but it is not guaranteed.

Numerical construction of the dislocations

Section titled “Numerical construction of the dislocations”

The calculations use established tight-binding Hamiltonians. Black phosphorene is interpolated between a 90° bond model, which has chiral symmetry, and a realistic model without that extra symmetry. The interpolation parameter λ runs from 0 to 1. The 30 × 30 unit-cell lattice is cut with unit cells intact, and 7 unit cells are removed before the edges are joined.

For silicon, the authors use a 30 × 30 lattice spanned by a1 and a2, with 100 sample points along the surviving momentum κ3. They cut a plane, remove 10 unit cells, and join the lattice under periodic boundary conditions in all three directions. Each simulation contains a pair of inversion-related dislocations.

The authors assess localization with the site-resolved inverse participation ratio. They also plot real-space wavefunction densities or amplitudes to locate the defect states.

Black phosphorene separates a floating band from a filling anomaly

Section titled “Black phosphorene separates a floating band from a filling anomaly”

At λ = 0, the 90° model places two exactly overlapping bound states in the middle of the energy gap. Their degeneracy follows from inversion symmetry because the two defects are far apart. Both models have inversion eigenvalues that yield weak indices of 1 and 1. The bulk gap remains open throughout the interpolation.

The calculated invariant is MνSSH = π(1/a, 1/b). It predicts a nontrivial response for an inversion-related dislocation pair with B = a2 = (0, b), which is an odd combination of lattice basis vectors. The simulated black phosphorene lattice therefore meets the predicted condition.

As λ approaches 1, the polarization states merge with the valence band in the realistic model. The floating mid-gap band is absent at the realistic endpoint, although the imbalance between conduction- and valence-band state counts remains. This imbalance produces a filling anomaly at 5/8 filling. The response therefore survives without a finely tuned in-gap band.

The inverse participation ratio is maximal at λ = 0 and minimal at λ = 1. The authors also relate the localization length to the energy gap, which scales inversely as expected.

Edge dislocations in silicon, diamond, and germanium

Section titled “Edge dislocations in silicon, diamond, and germanium”

The 3D elemental models use lattice constants of 2.698 Å for silicon, 1.798 Å for diamond, and 2.829 Å for germanium. All three exhibit an odd number of band inversions relative to the reference atomic limit along the 1D lines connecting Xi to L.

The calculated 3D weak indices are ν2,3SSH = 1, ν3,1SSH = 1, and ν2,1SSH = 1. These values predict a nontrivial response for any pair of edge dislocations when the Burgers vector and line vector are lattice basis vectors. The silicon simulation uses B = a1 = a/2(0, 1, 1) and T = a3 = a/2(1, 1, 0), which satisfy that condition.

Mid-gap dislocation bands remain present at realistic model parameters for silicon, diamond, and germanium. At half-filling, silicon has a filling anomaly at each momentum κ3. The Supplemental Material shows two inversion-related polarization bands in the middle of the gap for diamond and germanium. One band is shifted slightly for visualization, and the two overlap when the gap is large.

For silicon, the inverse participation ratio is smallest at κ3 = 0 and largest at κ3 = 1/2. For diamond and germanium, it is minimal at κ3 = 0 and maximal at κ3 = -1/2.

The screw dislocation result follows from its geometry. Its line vector is parallel to its Burgers vector. The Hamiltonians on the two cut directions are then identical under inversion and time-reversal symmetry, so the diagonal weak indices required for polarization bands vanish.

The results are calculations based on inversion-symmetric tight-binding models. The paper reports no experimental spectrum or transport measurement. The 3D supercells preserve one elemental diamond-type lattice and do not include grain boundaries.

The filling anomaly is the robust part of the prediction. It survives when the defect states leave the gap and mix with bulk states, provided inversion symmetry remains and the bulk gap does not close. The floating mid-gap band is less universal, as black phosphorene demonstrates at its realistic endpoint.

The current scope also excludes several related cases. The authors note that α-Sn has a near-zero band gap, while strain-induced band inversion at Γ does not contribute to the predicted dislocation response. Boron has a more complex tight-binding structure and a large unit cell, which makes its dislocation response more involved. The authors identify extension beyond inversion symmetry as the next step.

Dislocation type and crystallographic orientation determine whether the predicted response can occur. An edge dislocation with suitable Burgers and line vectors can support polarization bands in the diamond model. A screw dislocation has a trivial response under the symmetry assumptions used here. Device and material specifications should therefore distinguish these defect classes.

A model of diamond transport or contacts should include defect-bound bands when those states lie near the Fermi energy. The filling anomaly remains relevant even if hybridization obscures a clean floating band. The paper does not measure how these states affect mobility, contact resistance, or device yield.

The authors suggest scanning tunneling microscopy as a possible test if the polarization states remain close to the Fermi energy. They cite earlier scanning tunneling microscopy work on screw dislocations in bismuth as a precedent. Their proposed diamond measurement would instead target edge dislocations, since the modeled screw response is trivial.

For an experimental scanning tunneling microscopy test of the predicted diamond dislocation response, the relevant material family is bulk single-crystal diamond containing edge dislocations, with defect-bound states accessible near the Fermi energy. 6CCVD offers MPCVD-grown single-crystal plates and wafers up to 125 mm, with thickness from 0.1 to 500 micrometers, through its diamond materials catalog. The test plan must specify the edge-dislocation geometry and surface measurement method, which can be reviewed through engineering support.

View Original Abstract

We study elementary semiconductors and insulators that are symmetric under spatial inversion: silicon, diamond, germanium, and black phosphorene. These materials are ideal candidates for realizing obstructed atomic insulators, which differ from trivial atomic insulators by a quantized spatial shift of their electronic Wannier centers with respect to the atomic lattice. We use symmetry indicator invariants that allow the prediction of nontrivial responses to crystal dislocations with an integer Burgers vector in these materials, and generally, in all inversion-symmetric obstructed atomic insulators. Unlike weak topological insulators where the response only depends on the Burgers vector, the dislocation response of three-dimensional (3D) inversion-symmetric obstructed atomic insulators can also be affected by the line vector of the dislocation. In such insulators, we find that edge dislocations generically exhibit a nontrivial response, while in all 3D obstructed atomic insulators, screw dislocations always display a trivial response. With the aid of numerical simulations of realistic tight-binding models, we confirm the presence of midgap polarization bands localized along dislocations in silicon, diamond, and germanium.

  1. 2023 - Semiconductor Physics [Crossref]