Which symmetry conditions let periodic pi-vacancy graphene superlattices open a band gap?

Folding both valleys onto Γ requires 3n supercells

The paper treats a graphene superlattice (GSL) as graphene with a vacancy pattern repeated on a cell larger than the primitive cell. That larger periodicity shrinks the Brillouin zone into a mini-Brillouin zone and back-folds the pristine bands. Whether the two Dirac valleys, K and K', land on the same point after folding is fixed by the supercell size alone: in 3n×3n cells, where n = 1, 2, 3, …, both valleys fold onto Γ, the time-reversal-invariant momentum. In 3n−1 and 3n+1 cells they stay separated, so the Dirac cone survives and no gap can form. This is why the authors call the 3n size a prerequisite rather than a design choice.

The calculation is a nearest-neighbor tight-binding model with one pz orbital per carbon site, hopping t = −2.7 eV, on-site energy zero, and spin and spin-orbit coupling neglected. A π-vacancy is modeled simply as a deleted site: the pz orbital is removed and every hopping matrix element to and from it is set to zero. The authors use this to stand in for functionalized, substituted, or missing carbon sites, and they compare C3-type motifs (C3h(12C), D3h(A≠B), D3h(18C), D6h(6C)) with C2-type motifs (C2h(4C), D2h(2C), D2h(10C), D2h(16C)).

Removing equal A and B sites avoids zero-energy flat bands

Before symmetry can decide anything, the vacancy pattern must remove the same number of sites from the two honeycomb sublattices, A and B. When A ≠ B, the imbalance m = |A − B| produces at least m zero-energy modes, which appear as flat bands at the Fermi level, consistent with Lieb's theorem as the paper cites it. The authors therefore restrict attention to A = B motifs, which they state is required for gap opening.

Equal removal also preserves the bipartite form of the Hamiltonian, so the matrix chirality relation ΓHΓ⁻¹ = −H with Γ² = I holds. The practical consequence is a spectrum symmetric about the Fermi level in every supercell size, which is what allows a conduction-band minimum and valence-band maximum to sit symmetrically around the gap at Γ in the 3n case.

C3 symmetry locks the Dirac cones to high-symmetry points

C3 symmetry makes the two valleys 120° rotationally equivalent, so the cones cannot drift away from the high-symmetry points of the mini-Brillouin zone. In a 3n supercell they therefore stay at Γ, and the gap can be read directly from a standard high-symmetry band path. The paper's clearest example is the D3h(18C) motif at n = 3: a gap of |Δg| = 615 meV opens at Γ, and the fourfold degeneracy at the Fermi level splits into a twofold one, with the global point group of the mini-Brillouin zone remaining D3h.

The authors also report that C3-type vacancies produce larger gaps than C2-type ones at comparable defect concentrations, which they attribute to the twofold degeneracy at Γ; C2-type cases leave nondegenerate bands. Among the motifs studied, D6h(6C) is the most efficient, reaching 314 meV at only 3.7% defect concentration. The abstract further states that gaps in 3n superlattices scale linearly with defect concentration for each motif, and that energy contours near the cone stay isotropic when C3 is preserved.

Two perpendicular mirrors offer a weaker second route

When C3 is absent, the cones are no longer forced onto high-symmetry points. The paper shows that losing C3 shifts the Dirac cone in reciprocal space by Δq, and that the shift magnitude scales with defect concentration, so a sparse C2-type pattern produces only a small displacement. A direct consequence of breaking C3 is the loss of at least two σd and two σv mirror planes, because the product of two mirrors separated by 60° equals the 120° rotation.

A gap can still open in a 3n supercell if the motif reduces the global point group to D2h while preserving a pair of perpendicular mirror planes, σv ⊥ σd. The remaining mirrors restrict the cone displacement along the perpendicular direction, which keeps the cones at Γ in the 3n cell. The authors are explicit that this is a sufficient condition, not a necessary one, and that the resulting gaps are smaller than in C3-preserving cases because the bands are nondegenerate. The D2h(16C) motif illustrates the intermediate situation: the cone shifts slightly off the high-symmetry point, and the little group at K and K' drops to C2v.

Displaced or missing vacancies erode the gap linearly

Real structures are not perfectly periodic, so the authors test what happens when D6h(6C) vacancies are displaced by one or two graphene primitive lattice vectors, or removed entirely. Both perturbations break C3 symmetry, lift the degeneracy, and push the gap back toward the gapless pristine-graphene limit. The suppression is quantified per defect: displaced vacancies cost about 30 meV per defect and the gap nearly closes at 50% defect density, while missing vacancies cost about 20 meV per defect.

This is the comparison that separates the two symmetry routes in practice. A C3-preserving motif is not only the route to a larger gap but also the route whose gap degrades more slowly under the disorder that bottom-up assembly is likely to leave behind. The paper frames this as the practical argument for choosing a C3-type motif when perfect periodicity is hard to achieve.

The tight-binding vacancy model sets the scope

The results are tight-binding results, not measured band gaps. Spin polarization and spin-orbit coupling are neglected, on-site energies are uniform at zero, and only nearest-neighbor hopping is included. The vacancy is a deleted pz site, which the authors argue captures the low-energy π physics of hydrogenation-driven sp3 functionalization or BN substitution while remaining cheap enough for large superlattices. They note that such chemical changes can cause local deformation that shifts the electronic structure quantitatively without changing its overall character, whereas true atomic vacancies bring strong relaxation effects that the model does not represent.

The authors also point out that Dirac points can move to low-symmetry points off the usual high-symmetry path, so a one-dimensional band plot along high-symmetry lines can miss the cone entirely; two-dimensional band structures are needed to see the effect. That is a caution about how such superlattices should be characterized computationally, not a claim about fabricated devices.

Choose C3 first, then check the mirror pair

The paper's decision order is concrete. First, require equal A and B removal with uniform on-site energy, otherwise zero modes appear at the Fermi level. Second, require a 3n supercell, because 3n−1 and 3n+1 leave the valleys separated and the cone alive. Third, prefer a motif that preserves C3, since it pins the cones at Γ, yields the larger gaps, and keeps the gap more robust to displacement. Only if C3 cannot be preserved should the design fall back on the σv ⊥ σd mirror pair, accepting a smaller, nondegenerate gap.

The quantitative anchor for that priority is the comparison within this study: D6h(6C) reaches 314 meV at 3.7% defect concentration, D3h(18C) gives 615 meV at n = 3, and C2-type motifs give smaller gaps, with roughly 30 meV lost per displaced D6h(6C) vacancy. The authors connect the scale to application by noting that a gap above 500 meV is sufficient for stable electronics at room temperature, which places the C3-preserving motifs on the useful side of that threshold and makes symmetry preservation, not defect density alone, the variable to control.

Sources used in this answer