Three-dimensional Moiré crystallography

Moiré materials, typically confined to stacking atomically thin, two - dimensional (2D) layers such as graphene or transition metal dichalcogenides, have transformed our understanding of strongly correlated and topological quantum phenomena. The lattice mismatch and relative twist angle between 2D layers have shown to result in Moiré patterns associated with widely tunable electronic properties, ranging from Mott and Chern insulators to semi- and super-conductors. Extended to three-dimensional (3D) structures, Moiré materials unlock an entirely new crystallographic space defined by the elements of the 3D rotation group and translational symmetry of the constituent lattices. 3D Moiré crystals exhibit fascinating novel properties, often not found in the individual components, yet the general construction principles of 3D Moiré crystals remain largely unknown. Here we establish fundamental mathematical principles of 3D Moiré crystallography and propose a general method of 3D Moiré crystal construction using Clifford algebras over the field of rational numbers. We illustrate several examples of 3D Moiré structures representing realistic chemical frameworks and highlight their potential applications in condensed matter physics and solid-state chemistry.

💡 Research Summary

**
The manuscript “Three‑dimensional Moiré crystallography” establishes a comprehensive theoretical foundation for constructing periodic Moiré superstructures in three dimensions. Building on the well‑known 2D Moiré phenomenon—where a 2D lattice L is overlaid with a rotated copy rL to produce a new periodic pattern—the authors ask a fundamental question: for an arbitrary 3D prototype lattice L = ℤ³, under what conditions does a rotation r ∈ SO(3,ℝ) exist such that L and rL are commensurate in all three translational directions?

Mathematically, the condition reduces to the existence of a rational matrix h ∈ SL(3,ℚ) satisfying u⁻¹ r u = h, where u is the matrix of basis vectors of L. In other words, the rotation must map the integer lattice onto itself up to a rational change of basis. The set of all such rotations for a given Gram matrix g (which encodes the unit‑cell parameters a, b, c, α, β, γ) is denoted M_g. The authors introduce a homomorphism φ_u : M_g → SL(3,ℚ) and focus on its image H_g, which consists of all rational matrices h that preserve the quadratic form defined by g, i.e. hᵀ g h = g. This equation describes the rational points of the orthogonal group O(ℝ³, Q_g), where Q_g(v)=vᵀ g v.

When g has only rational entries, the problem becomes one of parametrising the rational orthogonal group O(ℚ³, Q). The authors solve this by invoking the Clifford algebra Cℓ(V, Q) associated with the rational quadratic space V = ℚ³. The even sub‑algebra Cℓ⁺ provides a convenient representation: any invertible element p ∈ Cℓ⁺ defines a linear transformation v ↦ p v p⁻¹ that preserves Q, and the corresponding matrix ρ(p) lies in H_g. Consequently, every admissible h can be generated from an element of Cℓ⁺, and the full rotation set M_g is recovered via φ_u⁻¹. This construction extends the earlier result for simple cubic lattices (where H_g ≅ SO(3,ℚ)) to any rational crystal system, including tetragonal, orthorhombic, monoclinic, etc.

For irrational Gram matrices—common in real materials—the authors prove (Supplementary Text S2) that one can approximate g by a rational matrix g′, construct H_{g′} using the Clifford‑algebra method, and then map its elements back to the original irrational setting. This bridge ensures that the theory applies to all realistic lattices, not only the highly symmetric ones.

Having identified a suitable rotation, the paper details how to build the Moiré supercell. The denominators of the rational entries of h define integer scaling factors l₁, l₂, l₃ (least common multiples). Multiplying the rotated basis vectors by these factors yields a provisional supercell {l_i r u_i}. After applying a Niggli reduction to obtain the conventional cell, atomic positions are recomputed using an integer matrix k = l · h⁻¹, which maps the original fractional coordinates of both L and rL into the new cell. An optional translation vector d can be added to rL without breaking commensurability, providing an extra degree of freedom for structural tuning.

To assess chemical plausibility, the authors introduce a bonding criterion based on the minimal interatomic distance D in the prototype lattice and a scaling factor s > 1 (chosen as 1.2). Pairs of atoms whose separation does not exceed s D are considered bonded, forming an infinite graph X. The fundamental graph X₀, obtained by quotienting X by the translational symmetry, captures the topology of the resulting network. Three classes are defined: (i) X₀ is a single connected component → a realistic 3D framework; (ii) X₀ has a few components much fewer than the number of atoms in the cell → layered or cluster‑packed structures; (iii) X₀ has many components comparable to the atom count → unlikely to correspond to a stable solid.

The manuscript presents several illustrative examples. The first reproduces the known cubic‑lattice Moiré crystals, confirming that the allowed rotations belong to SO(3,ℚ) and can be parametrised by five integer parameters, as reported in earlier work. The second example constructs a tetragonal Moiré crystal from a primitive lattice with one atom per cell; after rotation and scaling, the supercell contains eight atoms arranged in a bipartite network where each site is four‑coordinated in a square‑planar geometry. The fundamental graph X₀ is a single component, confirming that the structure is chemically viable. The third example tackles a non‑cubic, non‑rational lattice (e.g., a monoclinic cell). By applying the rational‑approximation bridge, a suitable rotation and translation are found, yielding a Moiré superstructure whose X₀ consists of two components, interpreted as a layered material.

Beyond structural considerations, the authors discuss potential physical applications. Because the rotation angle and relative displacement can be tuned continuously, the resulting 3D Moiré crystals provide a “3D twistronics” platform. They could serve as model systems for ultracold atomic gases trapped in complex optical lattices, host topologically non‑trivial band structures (e.g., 3D Chern or Weyl phases), or realize unconventional superconductivity through engineered flat bands. The ability to generate a full family of commensurate superlattices for any crystal system opens avenues for systematic exploration of emergent phenomena that are inaccessible in conventional crystals.

In summary, the paper delivers a rigorous mathematical framework—grounded in Clifford algebras over the rational field—for the complete classification and construction of three‑dimensional Moiré crystals. It bridges the gap between abstract group‑theoretic conditions and concrete chemical realizability, provides practical algorithms for generating supercells, and outlines a rich landscape of potential condensed‑matter applications. This work constitutes a foundational step toward establishing 3D Moiré crystallography as a distinct sub‑discipline, extending the transformative impact of 2D twistronics into the full three‑dimensional realm.