Fractional Topological Stark Insulator

• Fractional Topological Stark Insulator is a free-fermionic phase where Stark localization converts topological flat-band states into localized states, enabling fractional charge pumping.
• A strong linear potential induces local hybridization of states from different bands, opening a real-space energy gap that protects the phase.
• Experimental lattice models, such as generalized Aubry–André systems in ultracold atom setups, concretely demonstrate the mechanism behind fractional pumping.

Searching arXiv for the cited paper and closely related context. A fractional topological Stark insulator (FTSI) is a non-interacting fractional topological phase in $(1+1)$-dimension that arises when a linear potential gradient is applied to topological flat bands. Its defining mechanism is not many-body correlation but Stark localization: the linear potential converts topological flat-band states into localized Stark states, and sufficiently strong localization causes states from different topological bands to hybridize in real space and open a real space energy gap (RSEG). In the formulation introduced in "Non-interacting fractional topological Stark insulator" (Chen et al., 29 Jul 2025), the RSEG rather than the original Bloch-band gap protects the phase, and the many-body state under topological pumping returns to itself only after multiple $2\pi$ periods, yielding fractional charge pumping.

1. Definition and conceptual position

Fractional topological phases are usually associated with strong interactions, many-body ground-state degeneracy, and fractionalized topological response, as in the fractional quantum Hall effect. The FTSI departs from that paradigm by realizing fractional response in a free-fermionic setting. Its fractionalization is not attributed to correlated quasiparticles, topological order, or spontaneous symmetry breaking; instead, it is encoded in the adiabatic evolution of occupied species of Stark-localized single-particle states (Chen et al., 29 Jul 2025).

This distinction is central to the term fractional in this context. In the FTSI, the many-body state does not form a degenerate interacting ground-state manifold. Rather, the occupied states belong to a protected multiplet of multiband Stark states, and adiabatic pumping cycles the occupied branch through orthogonal species before the original state is recovered. The central interpretive claim is therefore that Stark localization plays a role analogous to interactions: it creates a protected multiplet structure and enables a fractional topological response without many-body correlations.

A frequent misconception is to equate every fractional topological response with intrinsic topological order. The FTSI is explicitly formulated as a counterexample to that identification. Its fractional response is real, but its origin differs from conventional fractional topological phases both microscopically and topologically.

2. Microscopic setting and Stark-localization mechanism

The starting point is a one-dimensional system with a synthetic parameter $\lambda$ and Hamiltonian

$$H(\lambda)=H_0(\lambda)+F\sum_m \frac{x_m}{\mathcal L}\, c_m^\dagger c_m,$$

where $H_0(\lambda)$ is a family of topological flat-band Hamiltonians, $F$ is the linear potential strength, $x_m$ is the position of site $m$, and $\mathcal L$ is a normalization factor. The relevant low-energy sector consists of $n_b$ flat bands separated from higher bands by a large gap $2\pi$0, with total Chern number $2\pi$1. The simplest case emphasized is $2\pi$2 and $2\pi$3 (Chen et al., 29 Jul 2025).

The linear potential generates Stark ladders: each band is converted into a ladder of localized Stark states. For each band, the Stark localization length is approximately

$2\pi$4

with $2\pi$5 the band widths. Because the bands are flat or nearly flat, the Stark states are strongly localized. This is not a secondary effect. Localization reorganizes the Hilbert space by unit cell and band species, making it possible for a strong linear potential to hybridize states from different topological bands locally in real space.

The mechanism has four linked components. First, flat-band topology supplies a nontrivial Chern structure. Second, Stark localization converts extended Bloch states into real-space-localized states. Third, once the linear potential becomes sufficiently strong, states from distinct ladders overlap in energy at the same real-space location. Fourth, the resulting local interband hybridization opens a new protecting gap in real space. Within the paper’s framework, that sequence is the essential origin of the FTSI.

3. Real-space energy gap and phase structure

The real space energy gap is the gap opened between hybridized Stark states when the linear potential brings states from different bands into resonance or near resonance at the same position. For resonant Stark-localized states $2\pi$6 and $2\pi$7, the paper characterizes the RSEG through the matrix element of the linear-potential operator,

$2\pi$8

and relates the resonant-state separation to the field strength by

$2\pi$9

Near the transition, the RSEG scales as $\lambda$0 times an exponential factor involving $\lambda$1 and $\lambda$2, and because both scale like $\lambda$3, the exponential factor becomes effectively trivial, giving $\lambda$4 (Chen et al., 29 Jul 2025).

The phase diagram is organized by the competition between the original band separation and the Stark-induced real-space hybridization.

Regime	Condition	Defining feature
Integer TSI regime	$\lambda$5	Two Stark ladders remain separated; protected gap is essentially the original Bloch gap
Transition regime	Intermediate $\lambda$6	Integer-like boundary states coexist with bulk fractional Stark states
Fractional TSI regime	$\lambda$7	Stark ladders overlap; the dominant protecting gap is the RSEG

In the weak-gradient regime, the many-body state occupying the lower band remains adiabatically connected to an ordinary integer topological band insulator. The linear potential reduces the effective band separation by the potential drop across the system, but does not change the topological character of the occupied band. By contrast, once the linear potential is strong enough for the two Stark ladders to overlap, states from different bands hybridize within the same unit cell and the system enters the fractional TSI regime. The transition is controlled by the collapse of the RSEG as $\lambda$8.

4. Multiband Stark states and topological characterization

In the fractional regime, the relevant localized eigenstates are multiband Stark states,

$\lambda$9

For the two-band case, adiabatic evolution over one $$H(\lambda)=H_0(\lambda)+F\sum_m \frac{x_m}{\mathcal L}\, c_m^\dagger c_m,$$0 cycle of $$H(\lambda)=H_0(\lambda)+F\sum_m \frac{x_m}{\mathcal L}\, c_m^\dagger c_m,$$1 obeys

$$H(\lambda)=H_0(\lambda)+F\sum_m \frac{x_m}{\mathcal L}\, c_m^\dagger c_m,$$2

Thus the occupied state does not return to itself after one cycle. It returns only after two cycles. In the general $$H(\lambda)=H_0(\lambda)+F\sum_m \frac{x_m}{\mathcal L}\, c_m^\dagger c_m,$$3-band case, the supplementary material gives

$$H(\lambda)=H_0(\lambda)+F\sum_m \frac{x_m}{\mathcal L}\, c_m^\dagger c_m,$$4

so the state closes only after $$H(\lambda)=H_0(\lambda)+F\sum_m \frac{x_m}{\mathcal L}\, c_m^\dagger c_m,$$5 cycles (Chen et al., 29 Jul 2025).

The corresponding many-body state is formed by fully occupying one species of multiband Stark states,

$$H(\lambda)=H_0(\lambda)+F\sum_m \frac{x_m}{\mathcal L}\, c_m^\dagger c_m,$$6

Its fractional behavior is therefore tied to branch occupancy rather than many-body degeneracy. After one adiabatic cycle, the many-body state evolves into an orthogonal state in which a different Stark-state species is occupied; after $$H(\lambda)=H_0(\lambda)+F\sum_m \frac{x_m}{\mathcal L}\, c_m^\dagger c_m,$$7 cycles, the original state is recovered.

The topological characterization is obtained by mapping the many-body state to an artificial band insulator in a synthetic Brillouin zone. The occupied Stark states are Fourier transformed into artificial Bloch states,

$$H(\lambda)=H_0(\lambda)+F\sum_m \frac{x_m}{\mathcal L}\, c_m^\dagger c_m,$$8

and for the two-band case one sets

$$H(\lambda)=H_0(\lambda)+F\sum_m \frac{x_m}{\mathcal L}\, c_m^\dagger c_m,$$9

The many-body Berry curvature in synthetic $H_0(\lambda)$0 space defines

$H_0(\lambda)$1

with the result

$H_0(\lambda)$2

Because the many-body state requires $H_0(\lambda)$3 cycles to close, the effective fractional topological number is $H_0(\lambda)$4. For $H_0(\lambda)$5 and $H_0(\lambda)$6, the phase is therefore a $H_0(\lambda)$7 fractional TSI.

5. Fractional pumping and representative lattice models

The principal physical signature of the FTSI is fractional charge pumping. In the integer regime, one pumping cycle yields the usual integer pumped charge set by the band Chern number. In the fractional regime, one cycle pumps only a fraction of that amount, while $H_0(\lambda)$8 cycles restore integer quantization. For the two-band case with $H_0(\lambda)$9, the reported values are a pumped charge of $F$0 after one cycle and $F$1 after two cycles. For the three-band case, the corresponding sequence is $F$2, $F$3, and $F$4 after one, two, and three cycles, respectively (Chen et al., 29 Jul 2025).

This behavior follows directly from the multiband adiabatic cycle. The many-body state traverses distinct occupied-species sectors rather than closing after a single $F$5 evolution. Fractional pumping is therefore quantized per cycle only as a fraction of the total topological transport, while the full integer response appears only after the complete $F$6-cycle orbit is finished.

The theory is demonstrated with explicit lattice models. The paper uses a $F$7 generalized Aubry–André model to realize the $F$8 fractional TSI and a $F$9 Aubry–André model to realize the $x_m$0 fractional TSI. These examples serve two purposes. They provide concrete realizations of topological flat bands subject to a linear potential gradient, and they show that the mechanism is not restricted to a single finely tuned toy model.

6. Experimental realization and relation to other fractional topological platforms

The proposed experimental implementation uses ultracold atoms in bichromatic optical lattices. In that setting the tight-binding Hamiltonian takes the Aubry–André form

$x_m$1

with rational $x_m$2. The linear potential can be generated by a magnetic field gradient, and the topological pumping signal can be measured through the center-of-mass shift of the atomic cloud. The paper also states that realistic parameters for $x_m$3 atoms give lifetimes long enough to observe the pumping plateau (Chen et al., 29 Jul 2025).

Within the broader landscape of fractional topological matter, the FTSI occupies a specific niche. Other one-dimensional fractional platforms, such as topological-insulator constrictions, realize fractional charge and spin states through domain walls between competing gapped phases and can support non-Abelian bound states with fractional charge and spin, $x_m$4 parafermions, and an $x_m$5-fold degenerate ground state in the fractional regime (Klinovaja et al., 2015). The FTSI is distinguished from those settings by the absence of interaction-driven topological order and by the fact that its fractional response originates from Stark-localized multiband structure rather than from parafermionic zero modes or correlated quasiparticles.

This comparison clarifies the scope of the term fractional topological. In interacting constriction-based settings, fractionalization is tied to many-body degeneracy, domain-wall zero modes, and non-Abelian algebra. In the FTSI, the corresponding fractional response is instead encoded in the period of adiabatic evolution and in the protected multiplet of hybridized Stark states. A plausible implication is that the FTSI enlarges the classification of fractional topological response beyond the usual interaction-based framework, while remaining sharply distinct from conventional topologically ordered fractional phases.