Skyrmion Fractional Chern Insulator: An Intrinsically Multiband Route to Fractionalization in Rhombohedral Graphene

Abstract: We propose an unconventional microscopic origin for the fractional quantum anomalous Hall (FQAH) effect in rhombohedral graphene moiré superlattices: skyrmion fractionalization. We view the state at filling $ν<1$ as a metal of skyrmion vacancies, charge $+e$ objects formed by removing layer-pseudospin skyrmions from the interaction-generated skyrmion lattice Chern insulator at $ν=1$. These vacancies are intrinsically multiband degrees of freedom, absent in single Chern band-projected studies. Building on a recently proposed ideal limit, we first develop an effective field theory showing that skyrmion vacancies can themselves fractionalize, thereby inducing charge fractionalization. Focusing on $ν=\frac{2}{3}$, we then construct explicit variational trial wavefunctions for the resulting skyrmion fractional Chern insulator and provide numerical evidence, together with general arguments, showing that this process is energetically favored. Our results establish a realistic route to the FQAH that does not rely on a partially filled Chern band, but instead arises from fractionalization of collective pseudospin textures.

• skyrmion fractionalization provides an intrinsically multiband mechanism for the fractional quantum Hall effect in rhombohedral graphene
• experimental studies on twisted MoTe2 and Rhombohedral N-layer Graphene on hBN did not fully align with the single-band fractional Chern Insulator theory
• Using parton decomposition to construct effective field theories provides fresh insights into the topological order of fractional Chern Insulators in Rhombohedral Graphene

Overview and motivation

The fractional quantum anomalous Hall (FQAH) effect observed in twisted MoTe$_2$ and rhombohedral $N$-layer graphene on hBN (R$N$G/hBN) is conventionally understood as a fractional Chern insulator (FCI) arising from partial filling of a single $C=1$ band. In R5G/hBN, however, no such band exists at the single-particle level; Hartree-Fock (HF) calculations instead generate an interaction-produced $C=1$ band at filling $\nu=1$, which is then partially filled in band-projected exact diagonalization (ED). This procedure is uncontrolled, because the parent band is itself interaction-generated, and multiband ED studies report strong band-mixing effects whose microscopic meaning has remained unclear (2608.14535).

The paper proposes that the strong multiband character is not a perturbation of a single-band FCI but is intrinsic to its origin. Building on the "ideal limit" framework of Ref. tan2025ideal, the $\nu=1$ Chern insulator of R$N$G/hBN is reinterpreted as an interaction-generated layer-pseudospin skyrmion lattice (SkL) pinned by the moiré potential. Doping below $\nu=1$ then corresponds not to holes in the emergent Chern band but to skyrmion vacancies (SkVs) — charge-$+e$ defects obtained by removing zeros from the bosonic parton wavefunction. Because each SkV modifies the collective pseudospin texture, it necessarily mixes many HF bands: numerically, as the density approaches $N$0, the majority of the electron weight lies outside the SkL band. The central claim is that these SkVs can themselves form a fractional quantum Hall state, producing a skyrmion FCI (SkFCI) — a route to the FQAH that does not rely on a partially filled Chern band.

Ideal limit framework

The analysis uses the ideal-limit spinor structure of R$N$1G conduction-band Bloch states, with layer components related by derivatives acting on the all-$N$2 component. Many-body wavefunctions take a parton-like product form $N$3, where $N$4 is a fermionic LLL wavefunction in effective field $N$5 and $N$6 is a bosonic wavefunction in the opposite field; their opposite magnetic translations combine to ordinary translations. These states are exact zero modes of contact interactions.

The $N$7 Chern insulator corresponds to $N$8 being an Abrikosov antivortex lattice for the $N$9 bosons — one skyrmion per moiré unit cell — combined with a filled LLL for $N$0. Conventional FCIs are obtained by replacing the filled LLL with a Laughlin state, yielding a state at partial filling of the single SkL-defined $N$1 band. This conventional wavefunction serves as the benchmark representing what HF-ED would find.

Skyrmion vacancies as intrinsically multiband objects

SkVs are implemented by deleting zeros from $N$2, i.e., multiplying by factors $N$3 that cancel existing zeros without introducing singularities. Each removed zero reduces the total skyrmion number by one and smooths the wavefunction, which motivates kinetic favorability over holes. Crucially, this cannot be represented by any LLL factor, so SkVs are intrinsically multiband. Quantitatively, the band weight $N$4 of the SkV trial state becomes small at realistic vacancy densities, confirming that single-band projection fails to capture the doped objects.

Effective field theory and topological order

A parton decomposition $N$5 with emergent gauge field $N$6 yields, after particle-vortex duality on $N$7, a Lagrangian for the SkV field $N$8. The equations of motion show that SkVs experience a magnetic field equal to the fermion ($N$9) density, while $C=1$0 experiences the skyrmion density. Assuming $C=1$1 forms an IQH state at $C=1$2, SkVs acquire electric charge via the Hall response.

At electron filling $C=1$3 (SkV density $C=1$4), the SkVs sit at $C=1$5 and can condense into a bosonic Laughlin state. Integrating out the gapped fields gives a topological field theory identical to the conventional $C=1$6 FQH order when the skyrmion probe field $C=1$7 is ignored — consistent with the experimentally observed $C=1$8. With $C=1$9 retained, anyons carry both fractional electric and skyrmion charge $C=1$0, producing quantized skyrmion-electric and skyrmion-skyrmion Hall responses $C=1$1.

An important conceptual point follows directly: since skyrmion number is not microscopically conserved, the FCI and SkFCI are not distinct topological phases but two distinct microscopic realizations of the same topological order. They differ sharply only when skyrmion number is conserved.

Microscopic energetics and variational evidence

The paper constructs an explicit SkFCI trial wavefunction by summing over SkV lattice configurations weighted by a bosonic Laughlin factor $C=1$2, combined with the fermionic IQH parton:

$C=1$3

Each term in the sum is an electron Slater determinant, enabling brute-force energy evaluation on small tori (restricted to $C=1$4 due to computational cost). Flux threading confirms the correct many-body Chern number for $C=1$5, and the one-body density matrix confirms genuine multiband character.

Using realistic R5G dispersion and gate-screened Coulomb interactions ($C=1$6, $C=1$7 nm), the unoptimized SkFCI beats the conventional FCI across most of the $C=1$8 phase diagram. The mechanism is transparent: because the SkFCI contains fewer skyrmions it has a larger effective $C=1$9, hence a more compact momentum distribution $\nu=1$0, avoiding the costly high-energy tail of the dispersion while maintaining comparable short-range correlations.

Introducing a variational Gaussian width parameter $\nu=1$1 improves both states dramatically, yet the optimized SkFCI remains lower in energy than the optimized FCI. Both $\nu=1$2-optimized analytic states are competitive with full single-band HF-ED to within roughly 1 meV per particle, and the SkFCI actually dips below HF-ED despite having only one variational parameter — strong evidence that it captures the energetically favored multiband physics.

The role of the moiré potential, treated phenomenologically as $\nu=1$3, is decisive but bounded: a crossover occurs at $\nu=1$4 meV above which the conventional FCI wins, while recent estimates place realistic R5G/hBN at $\nu=1$5 meV, within the SkFCI-favored regime. This dependence on a phenomenological moiré strength is an assumption of the analysis rather than a first-principles result.

A diagnostic signature: interlayer correlation hole

Layer-resolved pair correlations reveal a qualitative distinction between the two routes. Acting with $\nu=1$6 projects an SkV onto site $\nu=1$7 (spinor "up"); subsequently acting with $\nu=1$8 attempts to flip the spinor "down," annihilating the state. The SkFCI therefore develops a pronounced interlayer correlation hole absent in single-band states. Quantified by

$\nu=1$9

the diagnostic gives $\nu=1$0, versus $\nu=1$1 and $\nu=1$2. The sizable HF-ED value relative to the ideal single-band FCI hints that even standard numerical approaches may already contain some SkV-like correlations, though this is suggestive rather than established. The quantity $\nu=1$3 provides an easily computable signature of doped SkVs for future studies.

Limitations and open questions

Several caveats bound the conclusions. The variational energetics are restricted to a $\nu=1$4 torus, sufficient only for trends rather than thermodynamic-limit precision; no efficient algorithm is known for evaluating the configuration-summed wavefunction at scale. Only the competition between the SkFCI and conventional FCI is considered — other competing orders (charge density waves, other FCIs, composite Fermi liquids) are excluded. The moiré potential enters phenomenologically, and the crossover at $\nu=1$5 meV sits close enough to the estimated realistic value that quantitative predictions require better control of $\nu=1$6. Finally, whether skyrmion number conservation is approximately maintained dynamically — which would make the SkFCI's fractional skyrmion Hall responses observable — is left open, as is experimental detection of the predicted $\nu=1$7 response.

Conclusion

This work establishes skyrmion fractionalization as a viable, intrinsically multiband microscopic route to the $\nu=1$8 FQAH in rhombohedral graphene. Through a parton-based effective field theory, explicit variational wavefunctions, and energetic comparison against both optimized single-band trial states and HF-ED, the authors show that doping charged skyrmion vacancies into an interaction-generated skyrmion lattice is energetically favored over hole doping under realistic parameters. The broader implication is that the FQAH need not originate from a partially filled Chern band; collective pseudospin textures can themselves fractionalize, providing a concrete explanation for the strong band mixing observed in multiband studies of R5G/hBN.