3D Chern Vector: Topological Invariants

• Chern vector is a three-dimensional topological invariant defined by integrating Berry curvature over 2D momentum-space slices.
• It underpins the vectorial bulk-boundary correspondence, dictating chiral surface modes and global surface spectral winding patterns.
• It protects unique 3D quantized Hall effects and is implementable in various platforms such as photonic, acoustic, and Floquet systems.

The Chern vector is a three-dimensional topological invariant that generalizes the scalar Chern number of a two-dimensional Chern insulator. In a gapped three-dimensional Bloch band, it is defined from Berry-curvature integrals over two-dimensional momentum-space slices and is written in the literature either as $$\mathbf{C}=(\mathcal{C}_{yz},\mathcal{C}_{xz},\mathcal{C}_{xy})$$ or as $\mathbf{C}=(C_x,C_y,C_z)$. In this role it characterizes 3D Chern insulators, controls chiral surface states and the winding of surface equifrequency or Fermi loops on the surface Brillouin-zone torus, and, in the special case $\mathbf{C}=(0,m,n)$, protects a quantized Hall response whose value depends explicitly on sample dimensions (Zhang et al., 24 Jan 2025).

1. Definition as a three-dimensional bulk invariant

In a three-dimensional periodic system with Bloch eigenstates $\lvert u_n(\mathbf{k})\rangle$, the Chern vector is defined component-wise by integrating Berry curvature over the two-dimensional Brillouin-zone slices perpendicular to each coordinate axis. In the $(C_x,C_y,C_z)$ convention, one writes

$$C_i=\frac{1}{2\pi}\int_{T^2_{jk}}F_{jk}(\mathbf{k})\,dk_j\,dk_k,$$

where $i\in\{x,y,z\}$, $(j,k)$ is its cyclic complement,

$$F_{jk}(\mathbf{k})=\partial_{k_j}A_k(\mathbf{k})-\partial_{k_k}A_j(\mathbf{k}), \qquad A_j(\mathbf{k})=i\langle \psi_n(\mathbf{k})|\partial_{k_j}\psi_n(\mathbf{k})\rangle.$$

Under suitable spectral gaps, each $C_i$ is an integer, so the triplet $\mathbf{C}=(C_x,C_y,C_z)$0 is a quantized topological invariant (Ma et al., 2024).

The same structure is expressed in the $\mathbf{C}=(C_x,C_y,C_z)$1 convention by fixing one momentum coordinate and integrating the corresponding Berry-curvature component over the perpendicular two-torus. For example,

$\mathbf{C}=(C_x,C_y,C_z)$2

with analogous definitions for $\mathbf{C}=(C_x,C_y,C_z)$3 and $\mathbf{C}=(C_x,C_y,C_z)$4. The coexistence of these notations suggests an equivalent slice-based indexing convention: one notation labels a component by the normal direction, the other by the integration plane (Zhang et al., 24 Jan 2025).

This invariant is needed because, in three dimensions, topology is not exhausted by a single scalar. Each fixed-$\mathbf{C}=(C_x,C_y,C_z)$5 two-dimensional slice can carry its own Chern number, and the three slice invariants assemble into the Chern vector. This is the sense in which the Chern vector is a vectorial generalization of the scalar Chern number (Linyun et al., 2024).

2. Vectorial bulk-boundary correspondence

The Chern vector underlies a genuinely three-dimensional bulk-boundary correspondence. In 3D Chern insulators, a surface with normal $\mathbf{C}=(C_x,C_y,C_z)$6 supports $\mathbf{C}=(C_x,C_y,C_z)$7 chiral surface bands crossing the bulk gap, provided the surface normal is not collinear with $\mathbf{C}=(C_x,C_y,C_z)$8. The zero-energy Fermi loop in the two-dimensional surface Brillouin zone is then a closed curve on a torus, and its winding numbers along the two non-contractible cycles are determined by the two components of $\mathbf{C}=(C_x,C_y,C_z)$9 that are not parallel to $\mathbf{C}=(0,m,n)$0 (Linyun et al., 2024).

A more general statement applies to interfaces between distinct 3D Chern insulators. For an interface normal to $\mathbf{C}=(0,m,n)$1, with Chern-vector discontinuity

$\mathbf{C}=(0,m,n)$2

the vectorial bulk-boundary correspondence asserts that only the components parallel to the interface produce chiral surface modes. The two integers

$\mathbf{C}=(0,m,n)$3

are exactly the winding numbers of the surface equifrequency loop around the two cycles of the surface Brillouin-zone torus. In this formulation, the topologically protected invariant under hybridization is the total winding around each handle, fixed by $\mathbf{C}=(0,m,n)$4 (Devescovi et al., 2022).

Concrete examples make the correspondence explicit. For a $\mathbf{C}=(0,m,n)$5/$\mathbf{C}=(0,m,n)$6 interface, one has $\mathbf{C}=(0,m,n)$7, so the predicted surface-loop winding is $\mathbf{C}=(0,m,n)$8. In the cited photonic numerics, this loop appears tilted in the $\mathbf{C}=(0,m,n)$9 plane and propagates along the $\lvert u_n(\mathbf{k})\rangle$0-direction in real space. For a $\lvert u_n(\mathbf{k})\rangle$1/$\lvert u_n(\mathbf{k})\rangle$2 interface, one finds a single protected chiral surface loop winding once around the $\lvert u_n(\mathbf{k})\rangle$3-handle and zero times around the $\lvert u_n(\mathbf{k})\rangle$4-handle, with group velocity pinned along $\lvert u_n(\mathbf{k})\rangle$5 (Devescovi et al., 2022).

A common oversimplification is to count uncoupled surface channels independently and treat them as permanently distinct. The vectorial correspondence shows that generic surface couplings can hybridize these modes, while the net winding data remain fixed by the Chern-vector discontinuity. This is why the protected content is vectorial rather than a mere sum of decoupled scalar invariants (Devescovi et al., 2022).

3. Surface-state knots, links, and three-dimensional transport geometry

Because the surface Brillouin zone is a two-torus, the closed surface loops controlled by the Chern vector can be topologically nontrivial torus knots or links. This feature was developed explicitly for acoustic 3D Chern insulators with arbitrary Chern vectors. For $\lvert u_n(\mathbf{k})\rangle$6, the loop on the $\lvert u_n(\mathbf{k})\rangle$7 surface winds once along $\lvert u_n(\mathbf{k})\rangle$8 and once in the opposite direction along $\lvert u_n(\mathbf{k})\rangle$9, forming the torus knot $(C_x,C_y,C_z)$0; on the $(C_x,C_y,C_z)$1 surface the knot is again $(C_x,C_y,C_z)$2; and on the $(C_x,C_y,C_z)$3 surface it is $(C_x,C_y,C_z)$4. For the supercell example $(C_x,C_y,C_z)$5, the $(C_x,C_y,C_z)$6 facet carries the torus link $(C_x,C_y,C_z)$7, which decomposes into three interlinked $(C_x,C_y,C_z)$8 knots because $(C_x,C_y,C_z)$9 (Linyun et al., 2024).

The Chern vector also constrains propagation kinematics on crystal surfaces. The group velocity of a surface mode is approximately

$$C_i=\frac{1}{2\pi}\int_{T^2_{jk}}F_{jk}(\mathbf{k})\,dk_j\,dk_k,$$0

When a chiral mode encounters an edge between two facets with normals $$C_i=\frac{1}{2\pi}\int_{T^2_{jk}}F_{jk}(\mathbf{k})\,dk_j\,dk_k,$$1, it can undergo “topological negative refraction” without backscattering according to the iso-frequency contours in the two adjacent surface Brillouin zones. In the acoustic construction with $$C_i=\frac{1}{2\pi}\int_{T^2_{jk}}F_{jk}(\mathbf{k})\,dk_j\,dk_k,$$2, this behavior can occur on all three mutually adjacent surfaces of a single crystal (Linyun et al., 2024).

These geometric features clarify why the Chern vector is not merely a bookkeeping device for stacked 2D Chern layers. The surface spectrum is organized globally on a torus, and the relevant protected objects are winding loops, knots, and links. This suggests that the topological content of a 3D Chern insulator is encoded simultaneously in bulk slice invariants and in the global topology of its surface spectral manifolds (Linyun et al., 2024).

4. $$C_i=\frac{1}{2\pi}\int_{T^2_{jk}}F_{jk}(\mathbf{k})\,dk_j\,dk_k,$$3 and the three-dimensional quantized Hall effect

A central recent result is that the Chern vector $$C_i=\frac{1}{2\pi}\int_{T^2_{jk}}F_{jk}(\mathbf{k})\,dk_j\,dk_k,$$4 protects a quantized Hall effect in three-dimensional systems. Consider a finite-size cubic sample with dimensions $$C_i=\frac{1}{2\pi}\int_{T^2_{jk}}F_{jk}(\mathbf{k})\,dk_j\,dk_k,$$5 and open boundaries, with nonzero components $$C_i=\frac{1}{2\pi}\int_{T^2_{jk}}F_{jk}(\mathbf{k})\,dk_j\,dk_k,$$6 and $$C_i=\frac{1}{2\pi}\int_{T^2_{jk}}F_{jk}(\mathbf{k})\,dk_j\,dk_k,$$7. Naively, each plane of constant $$C_i=\frac{1}{2\pi}\int_{T^2_{jk}}F_{jk}(\mathbf{k})\,dk_j\,dk_k,$$8 carries $$C_i=\frac{1}{2\pi}\int_{T^2_{jk}}F_{jk}(\mathbf{k})\,dk_j\,dk_k,$$9 chiral modes along $i\in\{x,y,z\}$0–$i\in\{x,y,z\}$1 edges, giving $i\in\{x,y,z\}$2, while each plane of constant $i\in\{x,y,z\}$3 carries $i\in\{x,y,z\}$4 modes along $i\in\{x,y,z\}$5–$i\in\{x,y,z\}$6 edges, giving $i\in\{x,y,z\}$7. However, when the two sets of surface modes are coupled, so that backscattering between opposite faces can occur, individual contributions do not remain separately quantized (Zhang et al., 24 Jan 2025).

The protected quantity is the sum. The key result is

$i\in\{x,y,z\}$8

In a two-terminal geometry with current leads on opposite $i\in\{x,y,z\}$9-faces, the Landauer–Büttiker analysis gives the total transmission into lead 1 as

$(j,k)$0

Hence

$(j,k)$1

Similarly, for a Hall bar along $(j,k)$2,

$(j,k)$3

(Zhang et al., 24 Jan 2025).

For the usual six-terminal Hall-bar configurations, the multi-terminal Landauer–Büttiker equations

$(j,k)$4

yield, in units of $(j,k)$5,

$(j,k)$6

$(j,k)$7

These expressions assume that all side faces are attached to ideal leads that couple only to the corresponding chiral surface modes, and that disorder or weak coupling does not close the bulk gap (Zhang et al., 24 Jan 2025).

The physical interpretation is explicitly three-dimensional. Unlike the pure $(j,k)$8 case, where each of the $(j,k)$9 stacked 2D layers independently contributes $$F_{jk}(\mathbf{k})=\partial_{k_j}A_k(\mathbf{k})-\partial_{k_k}A_j(\mathbf{k}), \qquad A_j(\mathbf{k})=i\langle \psi_n(\mathbf{k})|\partial_{k_j}\psi_n(\mathbf{k})\rangle.$$0 chiral modes, the $$F_{jk}(\mathbf{k})=\partial_{k_j}A_k(\mathbf{k})-\partial_{k_k}A_j(\mathbf{k}), \qquad A_j(\mathbf{k})=i\langle \psi_n(\mathbf{k})|\partial_{k_j}\psi_n(\mathbf{k})\rangle.$$1 case produces two families of surface modes that form an interconnecting network. Individual two-terminal conductances along $$F_{jk}(\mathbf{k})=\partial_{k_j}A_k(\mathbf{k})-\partial_{k_k}A_j(\mathbf{k}), \qquad A_j(\mathbf{k})=i\langle \psi_n(\mathbf{k})|\partial_{k_j}\psi_n(\mathbf{k})\rangle.$$2 or $$F_{jk}(\mathbf{k})=\partial_{k_j}A_k(\mathbf{k})-\partial_{k_k}A_j(\mathbf{k}), \qquad A_j(\mathbf{k})=i\langle \psi_n(\mathbf{k})|\partial_{k_j}\psi_n(\mathbf{k})\rangle.$$3 are not robust once inter-surface tunneling is allowed, but their sum is topologically protected by the total Chern vector. The factor $$F_{jk}(\mathbf{k})=\partial_{k_j}A_k(\mathbf{k})-\partial_{k_k}A_j(\mathbf{k}), \qquad A_j(\mathbf{k})=i\langle \psi_n(\mathbf{k})|\partial_{k_j}\psi_n(\mathbf{k})\rangle.$$4 depends on two orthogonal sample dimensions rather than a single area, which the source identifies as the smoking-gun signature of a 3D Chern vector–protected Hall effect (Zhang et al., 24 Jan 2025).

5. Implementations in acoustic, photonic, and Floquet systems

Arbitrary Chern vectors have been proposed and analyzed in several engineered platforms. In acoustic crystals, the construction begins from a 2D acoustic Haldane model formed by hexagonal prism resonators with sites $$F_{jk}(\mathbf{k})=\partial_{k_j}A_k(\mathbf{k})-\partial_{k_k}A_j(\mathbf{k}), \qquad A_j(\mathbf{k})=i\langle \psi_n(\mathbf{k})|\partial_{k_j}\psi_n(\mathbf{k})\rangle.$$5. An on-site potential difference $$F_{jk}(\mathbf{k})=\partial_{k_j}A_k(\mathbf{k})-\partial_{k_k}A_j(\mathbf{k}), \qquad A_j(\mathbf{k})=i\langle \psi_n(\mathbf{k})|\partial_{k_j}\psi_n(\mathbf{k})\rangle.$$6 breaks inversion symmetry, while circulating airflow at speed $$F_{jk}(\mathbf{k})=\partial_{k_j}A_k(\mathbf{k})-\partial_{k_k}A_j(\mathbf{k}), \qquad A_j(\mathbf{k})=i\langle \psi_n(\mathbf{k})|\partial_{k_j}\psi_n(\mathbf{k})\rangle.$$7 implements complex next-nearest-neighbor hopping. Stacking the layers with vertical couplings $$F_{jk}(\mathbf{k})=\partial_{k_j}A_k(\mathbf{k})-\partial_{k_k}A_j(\mathbf{k}), \qquad A_j(\mathbf{k})=i\langle \psi_n(\mathbf{k})|\partial_{k_j}\psi_n(\mathbf{k})\rangle.$$8 and chiral interlayer couplings $$F_{jk}(\mathbf{k})=\partial_{k_j}A_k(\mathbf{k})-\partial_{k_k}A_j(\mathbf{k}), \qquad A_j(\mathbf{k})=i\langle \psi_n(\mathbf{k})|\partial_{k_j}\psi_n(\mathbf{k})\rangle.$$9, realized by air tubes of radii $C_i$0, yields an effective magnetic field $C_i$1 and a 3D bandgap with $C_i$2. Tilting the airflow circulations so that each loop threads three adjacent layers deforms the nearest-neighbor network and yields $C_i$3. A supercell of size $C_i$4 in the tilted geometry carries $C_i$5 (Linyun et al., 2024).

Photonic and acoustic crystals are also the main proposed settings for observing the transport consequences of $C_i$6. Xu et al. and Yang et al. have already demonstrated 3D bandstructures with Chern vectors of this form. Although direct electrical Hall measurements are not possible in wave systems, one can track the backscattering trajectories of wave packets on opposite faces to infer the predicted two-terminal quantization $C_i$7. The same work proposes a condensed-matter route in which a 3D Chern insulator is layered next to a conventional metal in a weak-coupling regime and subjected to a magnetic field $C_i$8 that does not align with the existing Chern vector; candidate materials include vdW magnets such as MnBi$C_i$9Te$\mathbf{C}=(C_x,C_y,C_z)$00 combined with thin metallic films, for example ZrTe$\mathbf{C}=(C_x,C_y,C_z)$01 under field (Zhang et al., 24 Jan 2025).

A distinct implementation uses temporal-periodic Floquet engineering in a three-dimensional lattice. In a modified stacked kagome lattice with on-site energy $\mathbf{C}=(C_x,C_y,C_z)$02, six bonds carry time-periodic hoppings

$\mathbf{C}=(C_x,C_y,C_z)$03

with $\mathbf{C}=(C_x,C_y,C_z)$04. Because $\mathbf{C}=(C_x,C_y,C_z)$05, time-reversal symmetry is broken. In the high-frequency limit, the Floquet–Magnus or van Vleck expansion yields an effective Hamiltonian whose commutator term generates complex next-nearest-neighbor hoppings with net phase $\mathbf{C}=(C_x,C_y,C_z)$06, corresponding to a gauge flux of $\mathbf{C}=(C_x,C_y,C_z)$07 through each triangular plaquette. Since the triangles project onto all three Cartesian planes, the flux has nonzero components in $\mathbf{C}=(C_x,C_y,C_z)$08, $\mathbf{C}=(C_x,C_y,C_z)$09, and $\mathbf{C}=(C_x,C_y,C_z)$10, and the bands acquire a nonzero Chern vector (Ma et al., 2024).

For the parameter choice

$\mathbf{C}=(C_x,C_y,C_z)$11

the numerical result is

$\mathbf{C}=(C_x,C_y,C_z)$12

The corresponding surface spectral functions on the right, front, and top faces each display a single in-gap branch, with propagation direction fixed by the sign of the relevant component. In a fully open cubic sample with three notch defects, the excited wavepacket propagates chirally around each corner and bypasses defects without backscattering (Ma et al., 2024).

6. Terminological scope and a distinct mathematical usage

In topological band theory, the Chern vector is the three-component Berry-curvature invariant just described. In algebraic geometry and precobordism, however, the phrase can refer to the full collection of Chern classes of a vector bundle. Annala–Yokura’s universal precobordism theory constructs the Chern classes $\mathbf{C}=(C_x,C_y,C_z)$13 of a rank-$\mathbf{C}=(C_x,C_y,C_z)$14 bundle $\mathbf{C}=(C_x,C_y,C_z)$15 via a projective quasi-smooth alteration $\mathbf{C}=(C_x,C_y,C_z)$16, a filtration of the pulled-back bundle by line-bundle quotients, and elementary symmetric polynomials in the corresponding Euler classes. The resulting total Chern class

$\mathbf{C}=(C_x,C_y,C_z)$17

satisfies normalization, the Whitney sum formula, the splitting principle, and the projective bundle formula (Annala, 2019).

A related analytic development concerns Chern forms of singular Hermitian metrics on holomorphic vector bundles. For a singular Griffiths positive or negative metric $\mathbf{C}=(C_x,C_y,C_z)$18, the curvature current $\mathbf{C}=(C_x,C_y,C_z)$19 need not be a current of order $\mathbf{C}=(C_x,C_y,C_z)$20. Nevertheless, under codimension restrictions on the singular or degeneracy set, one can define Chern currents $\mathbf{C}=(C_x,C_y,C_z)$21 as closed currents of order $\mathbf{C}=(C_x,C_y,C_z)$22 with locally finite mass, obtained as limits of smooth regularizations or via Segre currents on $\mathbf{C}=(C_x,C_y,C_z)$23. On compact $\mathbf{C}=(C_x,C_y,C_z)$24, these currents represent the usual topological Chern classes (Lärkäng et al., 2016).

The term Chern vector therefore has at least two technically distinct meanings across current literature: a three-dimensional momentum-space topological invariant in condensed-matter, photonic, acoustic, and Floquet systems, and a shorthand for the collection of characteristic classes of a vector bundle in algebraic-geometric settings. The former governs chiral surface transport and dimension-dependent Hall quantization; the latter belongs to the formalism of characteristic classes, cobordism, and singular metrics (Annala, 2019).