Quantum Geometric Kohn–Luttinger Mechanism

• Quantum Geometric Kohn–Luttinger mechanism is a repulsion-driven superconducting process where Bloch-wave overlaps and quantum metrics redefine screening and Cooper pair formation.
• It employs band-projected form factors, the Fubini–Study metric, and Berry curvature to enhance unconventional pairing and lift angular-momentum degeneracies.
• The mechanism enables strong pairing near band extrema, illustrated in models like rhombohedral graphene multilayers and flat Chern-band systems.

The quantum geometric Kohn–Luttinger mechanism is a repulsion-driven superconducting mechanism in which Bloch-wave quantum geometry reshapes screening and the Cooper-pair vertex. Its basic ingredients are band-projected form factors, the Fubini–Study metric, and Berry curvature: the modulus of the form factor modifies the momentum dependence of the screened Coulomb interaction, while its phase can split opposite angular-momentum channels and favor chiral order. Recent work has shown that quantum metric anisotropy or inhomogeneity can strongly enhance unconventional pairing, and that in flat Chern bands Landau-level-like form factors can align the attractive lobe of an RPA-screened Coulomb interaction with the form-factor peak, producing anomalously strong pairing near local band extrema (Shavit et al., 2024, Jahin et al., 2024, Zou et al., 15 Jun 2026).

1. Conventional Kohn–Luttinger physics and its quantum-geometric extension

In the ordinary Kohn–Luttinger mechanism, a purely repulsive interaction is screened by particle–hole excitations of a Fermi liquid. In RPA one writes

$V_q^{\mathrm{RPA}}=\frac{V_q}{1+\Pi_q V_q},$

with static polarization

$$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$

For “featureless” bands, $\Pi_q$ is essentially constant for $q<2k_F$, so $V_q^{\mathrm{RPA}}$ falls monotonically with $q$, and attractive angular harmonics emerge only at quite high order in perturbation theory. It is therefore commonly believed that observability of the effect requires special circumstances such as proximity to van Hove singularities, significant lattice-induced band distortions, or non-trivial Fermi-surface topologies (Shavit et al., 2024).

The quantum-geometric extension alters that premise. When the overlap between Bloch states at nearby momenta is nontrivial, screening acquires explicit dependence on the band wavefunctions. In one formulation, the overlap

$\Lambda_{k,k+q}=\mathrm{Tr}[P_k P_{k+q}]$

suppresses screening at larger $q$, and for small $q$ gives

$$\Pi_q\approx \Pi_0\,[1-q_\mu q_\nu\langle g_{\mu\nu}\rangle_{\mathrm{FS}}]+O(q^3).$$

In another formulation, the phase of the overlap contributes Berry-phase structure to the pairing kernel. The result is that nontrivial geometry can generate attractive harmonics much more efficiently than in the conventional scenario and can lift the $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$0 degeneracy, favoring chiral order parameters (Shavit et al., 2024, Jahin et al., 2024).

2. Band projection, form factors, and geometric data

The formalism is naturally expressed after projection to an isolated band. For Bloch states $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$1 and dispersion $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$2, the projected density operator may be written as

$$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$3

The band-projected two-particle vertex in the Cooper channel then becomes

$$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$4

where the form factor carries both a modulus $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$5 and a phase $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$6 (Jahin et al., 2024).

For small $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$7, the overlap admits the expansion

$$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$8

with Berry connection

$$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$9

Berry curvature

$\Pi_q$0

and Fubini–Study metric

$\Pi_q$1

A closely related exponential form, useful in flat Chern bands, is

$\Pi_q$2

In this representation, the Gaussian decay controlled by $\Pi_q$3 determines the range of momentum transfers, while the phase proportional to $\Pi_q$4 selects chirality in the projected interaction (Zou et al., 15 Jun 2026).

Flat Chern-band analyses also emphasize two integrated invariants: $\Pi_q$5 The first is the Chern number, and the second is the quantum volume. Their simultaneous appearance reflects the fact that the mechanism depends on both topology and geometry, but with distinct roles for Berry curvature and metric (Zou et al., 15 Jun 2026).

3. Screening, overscreening, and the geometric figure of merit

The central modification to the conventional picture occurs in screening. In the metric-based treatments, the projected static RPA bubble is

$\Pi_q$6

and the screened interaction is

$\Pi_q$7

Because

$\Pi_q$8

the metric suppresses screening at larger $\Pi_q$9. Equivalently, since $q<2k_F$0, $q<2k_F$1 can acquire positive curvature at $q<2k_F$2, described as overscreening: $q<2k_F$3 with $q<2k_F$4 and $q<2k_F$5. If

$q<2k_F$6

the repulsion grows with $q<2k_F$7, which strongly favors sign-changing pairing (Shavit et al., 2024).

This result leads to an experimentally accessible figure of merit. Defining the Fermi-surface-averaged metric length

$q<2k_F$8

one obtains the geometric overscreening criterion

$q<2k_F$9

In rhombohedral graphene multilayers, the estimates

$V_q^{\mathrm{RPA}}$0

imply

$V_q^{\mathrm{RPA}}$1

so the criterion is met (Shavit et al., 2024).

In flat Chern bands, the screened interaction is often written using a dual-gate-screened Coulomb form,

$V_q^{\mathrm{RPA}}$2

together with an intraband Lindhard function projected by the Landau-level form factor,

$V_q^{\mathrm{RPA}}$3

so that

$V_q^{\mathrm{RPA}}$4

Near small Fermi pockets, this screened interaction develops its deepest negative dip at $V_q^{\mathrm{RPA}}$5, which is the ingredient later aligned with the form-factor peak (Zou et al., 15 Jun 2026).

4. Effective pairing kernel, angular harmonics, and channel selection

Once screening is determined, the superconducting problem reduces to an eigenvalue problem for the projected pairing kernel. For Kramers-partner pairing, one convenient Fermi-surface representation is

$V_q^{\mathrm{RPA}}$6

with partial-wave expansion

$V_q^{\mathrm{RPA}}$7

Overscreening guarantees that at least one $V_q^{\mathrm{RPA}}$8, producing attraction in the $V_q^{\mathrm{RPA}}$9 channel. Near $q$0, the linearized gap equation takes the form

$q$1

and with eigenvalue $q$2 one has

$q$3

Quantum-metric anisotropy in the TM model or metric inhomogeneity in the BHZ model can greatly increase $q$4 at low $q$5, typically $q$6, boosting $q$7 by orders of magnitude compared to “pure” KL (Shavit et al., 2024).

Berry-phase structure modifies not just the magnitude but also the symmetry of the instability. In the lowest-Landau-level toy model with

$q$8

the static bubble becomes

$q$9

and for odd $\Lambda_{k,k+q}=\mathrm{Tr}[P_k P_{k+q}]$0 one finds

$\Lambda_{k,k+q}=\mathrm{Tr}[P_k P_{k+q}]$1

The critical temperature then obeys

$\Lambda_{k,k+q}=\mathrm{Tr}[P_k P_{k+q}]$2

Because $\Lambda_{k,k+q}=\mathrm{Tr}[P_k P_{k+q}]$3 oscillates and has maxima at $\Lambda_{k,k+q}=\mathrm{Tr}[P_k P_{k+q}]$4, $\Lambda_{k,k+q}=\mathrm{Tr}[P_k P_{k+q}]$5 exhibits resonant peaks as $\Lambda_{k,k+q}=\mathrm{Tr}[P_k P_{k+q}]$6 is varied. “Ideal geometry,” corresponding to $\Lambda_{k,k+q}=\mathrm{Tr}[P_k P_{k+q}]$7, maximizes $\Lambda_{k,k+q}=\mathrm{Tr}[P_k P_{k+q}]$8 and hence $\Lambda_{k,k+q}=\mathrm{Tr}[P_k P_{k+q}]$9; the same work states that the ideal band geometry that favors a fractional Chern insulator in the flat-band limit has an optimal $q$0 (Jahin et al., 2024).

5. Flat Chern bands, Skyrmion lattices, and the geometric alignment principle

A particularly explicit realization is provided by the Skyrmion lattice model. For each pseudospin $q$1, the real-space Hamiltonian is

$q$2

where $q$3 acts on the two-component spin degree of freedom and $q$4 is the periodic skyrmion texture. In the adiabatic strong-coupling limit, the noncoplanar texture produces uniform Berry curvature and maps onto Landau-level physics. A convenient gauge for the lowest-band Bloch functions is

$q$5

with guiding-center operators

$q$6

For an ideal lowest Landau level, the form factor is exactly

$q$7

Near a local band extremum, two simplifications occur: the magnitude $q$8 becomes essentially $q$9-independent, producing an emergent momentum-space translational symmetry, and the $q$0 phase is governed by the local Berry curvature $q$1 (Zou et al., 15 Jun 2026).

The pairing enhancement near band extrema is organized by what the work terms the “geometric alignment principle.” As $q$2, the form-factor magnitude peaks at $q$3 with width $q$4, while the RPA attraction has its deepest dip at $q$5 due to overscreening at small $q$6. These peaks align, maximizing

$q$7

which enters the projected pairing interaction and the coupling constants $q$8. This alignment is identified as the root of anomalously strong pairing at band extrema, driven purely by quantum-geometric overlaps (Zou et al., 15 Jun 2026).

The same framework distinguishes sharply between spin-unpolarized and spin-polarized pairing. For spin-unpolarized, time-reversal-symmetric pairing, the effective interaction depends only on $q$9, so the Berry phase drops out and the emergent translational symmetry favors an $$\Pi_q\approx \Pi_0\,[1-q_\mu q_\nu\langle g_{\mu\nu}\rangle_{\mathrm{FS}}]+O(q^3).$$0 extended-$$\Pi_q\approx \Pi_0\,[1-q_\mu q_\nu\langle g_{\mu\nu}\rangle_{\mathrm{FS}}]+O(q^3).$$1 instability concentrated at small Fermi pockets. For spin-polarized, time-reversal-broken pairing, the $$\Pi_q\approx \Pi_0\,[1-q_\mu q_\nu\langle g_{\mu\nu}\rangle_{\mathrm{FS}}]+O(q^3).$$2 phase remains in the kernel and endows it with built-in chirality, producing chiral $$\Pi_q\approx \Pi_0\,[1-q_\mu q_\nu\langle g_{\mu\nu}\rangle_{\mathrm{FS}}]+O(q^3).$$3- and $$\Pi_q\approx \Pi_0\,[1-q_\mu q_\nu\langle g_{\mu\nu}\rangle_{\mathrm{FS}}]+O(q^3).$$4-wave order without invoking spin fluctuations: the leading instability is $$\Pi_q\approx \Pi_0\,[1-q_\mu q_\nu\langle g_{\mu\nu}\rangle_{\mathrm{FS}}]+O(q^3).$$5 at large filling and $$\Pi_q\approx \Pi_0\,[1-q_\mu q_\nu\langle g_{\mu\nu}\rangle_{\mathrm{FS}}]+O(q^3).$$6 at low filling. The band-extrema enhancement also persists in higher Landau-level analogs and survives finite-temperature screening and Berezinskii–Kosterlitz–Thouless phase fluctuations (Zou et al., 15 Jun 2026).

6. Microscopic realizations and material-specific phenomenology

The principal microscopic settings discussed in the literature include the tunable-metric square-lattice model, the parabolic BHZ model, rhombohedral graphene multilayers, ABC trilayer graphene, and the flat-Chern-band Skyrmion lattice model (Shavit et al., 2024, Jahin et al., 2024, Cea et al., 2021, Zou et al., 15 Jun 2026).

System or model	Geometric ingredient	Reported consequence
TM square-lattice model	Metric anisotropy	Low-$$\Pi_q\approx \Pi_0\,[1-q_\mu q_\nu\langle g_{\mu\nu}\rangle_{\mathrm{FS}}]+O(q^3).$$7 attraction, typically $$\Pi_q\approx \Pi_0\,[1-q_\mu q_\nu\langle g_{\mu\nu}\rangle_{\mathrm{FS}}]+O(q^3).$$8-wave
BHZ model	Metric inhomogeneity peaked near $$\Pi_q\approx \Pi_0\,[1-q_\mu q_\nu\langle g_{\mu\nu}\rangle_{\mathrm{FS}}]+O(q^3).$$9	Sizable $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$00- or $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$01-wave instability
Rhombohedral graphene multilayers	Complex pseudospin form factor $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$02	Intravalley chiral $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$03; intervalley lower $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$04
ABC trilayer graphene	Bloch-overlap-controlled screening and pairing	Spin-triplet, $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$05 odd leading eigenfunction
Skyrmion lattice flat Chern band	Landau-level-like form-factor alignment	Extended-$$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$06, chiral $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$07, and chiral $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$08 channels

In the TM model,

$$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$09

and the special property $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$10 on the Fermi surface isolates the role of sea-to-Fermi-surface overlaps. In the BHZ model,

$$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$11

which is peaked near $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$12 and satisfies $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$13. These two models were introduced precisely to separate quantum-metric anisotropy from quantum-metric inhomogeneity as distinct enhancement mechanisms (Shavit et al., 2024).

For rhombohedral graphene multilayers, the effective low-energy Hamiltonian is

$$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$14

and the Fermi-surface form factor is

$$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$15

The leading intravalley instability is chiral $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$16, $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$17 grows with layer number $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$18 and with small $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$19, and as $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$20 becomes large the pseudospin polarizes, $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$21, geometry effects vanish, and $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$22 falls back toward the trivial-band KL value (Jahin et al., 2024).

ABC trilayer graphene provides a material-specific earlier realization of KL-like repulsion-driven pairing controlled by Bloch-state overlaps. In that system, the static RPA polarization contains factors $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$23, and the effective interaction inherits the same geometric structure. On a grid of $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$24 $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$25-points, the leading eigenfunction was found to have $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$26 symmetry in each valley pocket and to yield $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$27 up to $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$28 when the chemical potential sits at the van Hove energy. The maximal reported values are $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$29 at $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$30 for $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$31, and up to $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$32 for $$\Pi_q=-N_f\sum_k \frac{n(\xi_{k+q})-n(\xi_k)}{\xi_{k+q}-\xi_k}.$$33 (Cea et al., 2021).

Taken together, these studies clarify a central misconception. The conventional expectation that Kohn–Luttinger pairing is observable only near van Hove singularities or other special Fermi-surface features remains valid for featureless bands, but quantum geometry supplies an independent control parameter. Metric anisotropy or inhomogeneity can reshape screening, Berry-phase structure can select chirality, and in flat Chern bands local band extrema can replace van Hove singularities as the locus of strongest attraction through geometric alignment (Shavit et al., 2024, Zou et al., 15 Jun 2026).