Hybrid Straight-Curved Waveguide Array

• Hybrid straight-curved waveguide arrays are two-dimensional Floquet photonic lattices that combine fixed honeycomb sites with dynamically modulated Kagome sites.
• The design enables controlled modulation of inter-site couplings, yielding distinct topological phases such as Floquet Chern insulators and Floquet anomalous topological insulators.
• Bulk diagnostics via Bloch oscillations allow for practical measurement of topological invariants, distinguishing between nonzero Chern numbers and loop winding effects.

Searching arXiv for recent and related papers on hybrid straight–curved waveguide arrays and Floquet photonic lattices. A hybrid straight–curved waveguide array is a photonic Floquet lattice in which two different two-dimensional lattices are interlaced and driven in very different ways: a straight honeycomb waveguide array, whose transverse positions are fixed along the propagation coordinate $z$, and a curved Kagome waveguide array, whose sites are periodically modulated along $z$ in the transverse plane. In the realization introduced for diagnosing Floquet Chern and anomalous topological insulators, the straight honeycomb sites act as a static backbone, while the curved Kagome sites mediate time-modulated couplings around hexagonal loops, yielding a Floquet honeycomb–Kagome photonic lattice (Zuo et al., 1 Aug 2025). The structure is embedded in a three-dimensional sample but implements a two-dimensional Floquet system because the transverse $(x,y)$ coordinates define the lattice and the physical propagation coordinate $z$ plays the role of time.

1. Architecture, geometry, and implementation

In the specific construction, each unit cell $\{m,n\}$ contains two straight honeycomb sites, labeled $A$ and $B$, and three curved Kagome sites, labeled $j=1,2,3$. The honeycomb lattice constant is

$a=\sqrt{3}\,d,$

with projections

$a_x=\frac{3d}{2}, \qquad a_y=\frac{\sqrt{3}d}{2}.$

The curved Kagome sites are placed so that their positions evolve as

$z$0

with

$z$1

Here $z$2 is the driving amplitude and $z$3 is the Floquet frequency. The relative phases $z$4, shifted by $z$5, make the motion asynchronous: the three Kagome sites do not reach closest approach simultaneously. Instead, the strong honeycomb–Kagome coupling circulates around a hexagon in time, explicitly breaking time-reversal symmetry and producing a chiral hopping sequence (Zuo et al., 1 Aug 2025).

The relevant separations are

$z$6

so the coupling hotspot moves around each hexagonal ring as $z$7 increases. This arrangement differs from fully helical three-dimensional waveguide arrays because it uses only one-dimensional periodic curving in the transverse plane rather than full three-dimensional helices, a design choice stated to simplify fabrication and modeling while still realizing nontrivial Floquet topology (Zuo et al., 1 Aug 2025).

The proposed platform is standard femtosecond laser–written waveguides in glass. The modeling uses a background index $z$8 and probe wavelength $z$9. A single waveguide is represented by

$(x,y)$0

with $(x,y)$1 and effective radius $(x,y)$2. Coupled-mode calibration gives

$(x,y)$3

Typical parameters are $(x,y)$4, base honeycomb coupling $(x,y)$5, and driving choices $(x,y)$6 for an FCI example and $(x,y)$7 for a FATI example, both at $(x,y)$8. Propagation distances extend to hundreds of driving periods, i.e. tens of centimeters, and these values are stated to be compatible with current femtosecond laser–written waveguide technology (Zuo et al., 1 Aug 2025).

2. Paraxial and Floquet formulation

Light propagation is governed by the paraxial Helmholtz equation

$(x,y)$9

with

$z$0

In the tight-binding limit, the field is expanded in localized Gaussian-like modes centered on the five sites per unit cell, producing coupled-mode equations of the form

$z$1

Only nearest-neighbor couplings between honeycomb and Kagome sites are retained. The time-periodic couplings are

$z$2

with $z$3 and $z$4 determined by the asynchronous motion.

Under periodic boundary conditions, the system becomes a five-band time-periodic Bloch Hamiltonian $z$5 satisfying $z$6. The evolution operator is

$z$7

and over one modulation period,

$z$8

The effective Floquet Hamiltonian is defined stroboscopically by

$z$9

whose eigenvalues give five Floquet–Bloch quasienergy bands within a $\{m,n\}$0 quasienergy Brillouin zone.

This formulation places the hybrid straight–curved array within the broader theory of curved photonic lattices, where curvature enters as an effective linear potential in a Schrödinger-like propagation equation (Longhi, 2010). In the present structure, however, curvature does not simply supply a uniform force: it generates an asynchronous, loop-resolved modulation of inter-sublattice couplings.

3. Floquet Chern and anomalous phases

Scanning the driving amplitude $\{m,n\}$1 and frequency $\{m,n\}$2 yields three regimes: a trivial insulator at $\{m,n\}$3, a Floquet Chern insulator (FCI) for moderate $\{m,n\}$4, and a Floquet anomalous topological insulator (FATI) for larger $\{m,n\}$5 after closure and reopening of the gap between the bottom and top quasienergy bands (Zuo et al., 1 Aug 2025). In the trivial phase there are no chiral edge states and all band Chern numbers are zero. In the FCI regime, the bottom and top bands acquire Chern numbers $\{m,n\}$6 and $\{m,n\}$7, respectively, while the middle bands remain at zero. In the FATI regime, all five band Chern numbers are zero, yet chiral edge states persist in every gap.

For the FCI, the topological characterization uses the Berry curvature

$\{m,n\}$8

and the Chern number

$\{m,n\}$9

Bulk–edge correspondence takes the standard form: the number of chiral edge-state pairs crossing a gap equals the sum of Chern numbers of all bands below that gap.

For the FATI, the relevant invariant is not a band Chern number but a gap winding number built from the full micromotion $A$0: $A$1 The band Chern numbers and gap windings are related by

$A$2

Accordingly, when all $A$3, all gaps share the same winding number; in this model the winding number is $A$4 for every gap in the FATI regime, explaining why each quasienergy gap carries one pair of chiral edge states despite vanishing band Chern numbers (Zuo et al., 1 Aug 2025).

A central consequence is that edge transport is not sufficient to distinguish the two phases. Both FCI and FATI exhibit chiral boundary channels, but only the FCI carries nonzero band Chern numbers in the stroboscopic spectrum.

4. Bloch oscillations as a bulk diagnostic

The proposed diagnostic is a two-dimensional tilted potential created by a spatial refractive-index gradient in both $A$5 and $A$6. In tight-binding form, this produces a uniform effective force

$A$7

with simulation values

$A$8

on a lattice with $A$9. An initial Gaussian-like wavepacket is prepared predominantly in the lowest Floquet band,

$B$0

with $B$1 and $B$2 (Zuo et al., 1 Aug 2025).

The force is chosen so that the Bloch periods along $B$3 and $B$4,

$B$5

satisfy

$B$6

hence

$B$7

This ensures adiabatic motion, suppresses Landau–Zener transitions, and makes the wavepacket effectively sweep the whole Brillouin zone quasi-uniformly.

Over one Bloch period, the dispersive contribution to the displacement vanishes because the $B$8-space trajectory is closed. The net drift comes solely from anomalous velocity, and quasi-quantized numbers are defined as

$B$9

When the wavepacket samples the Brillouin zone quasi-uniformly, $j=1,2,3$0, the Chern number of the occupied band.

The diagnostic contrast is direct. In the FCI regime, the wavepacket shows oscillatory motion plus a unidirectional drift orthogonal to the effective force, with

$j=1,2,3$1

close to the lowest-band Chern number $j=1,2,3$2. In the FATI regime, the same protocol yields

$j=1,2,3$3

so the wavepacket returns on average to its initial position after each Bloch period (Zuo et al., 1 Aug 2025). The significance is that the distinction is purely bulk-based: finite quasi-quantized transverse drift identifies an FCI, while zero net drift identifies a FATI, even though both phases possess chiral edge states.

This bulk use of Bloch oscillations is conceptually related to earlier waveguide-array work in which a constant gradient produced Bloch oscillations in a straight exciton-polariton microcavity array, although that system was not a straight–curved Floquet lattice (Beierlein et al., 2020).

5. Edge spectra, transport, and sublattice asymmetry

In ribbon geometry, the Floquet quasienergy spectra display gap-crossing chiral edge states in both topological phases. In the FCI, the number and chirality of these states agree with the sums of Chern numbers below each gap. In the FATI, all five bands satisfy $j=1,2,3$4, yet each gap still hosts pairs of chiral edge states because the gap winding numbers satisfy $j=1,2,3$5 for all gaps (Zuo et al., 1 Aug 2025).

With fully open boundaries, edge and bulk states separate clearly in real space. Chiral edge states localize predominantly along boundaries, with sublattice-resolved asymmetry: some edges are populated mainly on Kagome sites and others on honeycomb sites. Edge states with opposite group velocity localize on different sublattice types and different boundaries.

Real-space transport simulations further show that edge propagation alone is not a phase discriminator. In the FCI, excitation of three neighboring honeycomb boundary sites with phases $j=1,2,3$6 launches clockwise perimeter transport with minimal backscattering and only a small fraction leaking into the bulk. In the FATI, excitation of three Kagome sites with $j=1,2,3$7 also yields robust unidirectional boundary propagation, with even less bulk leakage. Since both phases exhibit qualitatively similar corner-avoiding chiral transport, the decisive distinction must come from the bulk Bloch-oscillation response rather than from edge phenomenology.

6. Role of the hybrid design and research context

The hybrid straight–curved architecture is the core mechanism of the platform rather than an incidental geometrical detail. First, it produces a two-dimensional Floquet effect using only one-dimensional periodic curving in the transverse plane, avoiding full three-dimensional helices and thereby simplifying fabrication and modeling (Zuo et al., 1 Aug 2025). Second, the asynchronous $j=1,2,3$8-shifted Kagome motion makes the strongest couplings circulate around each hexagon, generating a loop-like sequence that breaks time-reversal symmetry and yields an effective synthetic flux. Third, varying $j=1,2,3$9 and $a=\sqrt{3}\,d,$0 continuously deforms the micromotion from a regime with nonzero band Chern numbers to one with vanishing band Chern numbers but nontrivial gap winding numbers, so both FCI and FATI phases occur within the same device class.

This position within the broader curved-waveguide literature is precise. General moment-based theory for curved arrays represents bending as an effective transverse gradient and describes Bloch oscillations, dynamic localization, and shape-invariant discrete Bessel beams through a $a=\sqrt{3}\,d,$1-parameter formalism (Longhi, 2010). Periodically curved zigzag arrays can rectify light refraction through resonant coupling sequences (Longhi, 2010). Artificial-gauge-field silicon arrays use sinusoidal trajectories to suppress coupling in dense straight and bent waveguide arrays (Zhou et al., 2022). The hybrid honeycomb–Kagome structure differs from these settings by combining a static backbone with asynchronously curved auxiliary sites, so that topology is encoded in loop-resolved Floquet micromotion rather than only in a curvature-induced effective force.

Several extensions are stated explicitly. The geometry suggests inclusion of optical nonlinearity (Kerr) to explore nonlinear Floquet topological insulators, nonlinear Thouless pumps, and Floquet solitons. The same strategy is proposed for other platforms, including ultracold atoms in optical lattices, superconducting circuits, and digital quantum simulators, as well as for other lattice geometries such as Lieb, dice, higher-order topological lattices, multi-frequency drives, and three-dimensional photonic lattices (Zuo et al., 1 Aug 2025). A plausible implication is that the hybrid straight–curved waveguide array functions as a compact Floquet-engineering template in which bulk transport, rather than edge transport alone, becomes the decisive probe of whether topology is carried by Chern numbers or by winding of the full evolution operator.