Site-Nematic Insulator in Correlated Materials

• The site-nematic insulator is an interaction-driven insulating state characterized by spontaneous sublattice density imbalance and broken rotational symmetry (e.g., C4 or C3 symmetry).
• It arises in diverse systems such as checkerboard lattices, Mott insulators, and moiré graphene, where diagnostics include finite order parameters and anisotropic charge distributions.
• Its realization involves strong-coupling methods, including quantum Monte Carlo and Hartree–Fock analyses, that reveal transitions from gapless semimetallic states to fully gapped nematic phases.

A site-nematic insulator is an interaction-driven insulating state in which rotational symmetry is spontaneously broken through inequivalent sites, sublattices, or closely related intra-unit-cell degrees of freedom. In the most explicit usage, developed for spinless fermions on the checkerboard lattice at half filling, it denotes a strong-coupling, time-reversal-preserving, $C_4$-breaking insulator with a charge-density imbalance between the two sublattices (Liu et al., 21 Jul 2025). In other materials classes, closely related language is used for insulating states with local site inequivalence tied to magneto-structural coupling, or for gapped moiré phases with broken $C_3$ or $C_{3z}$ symmetry and anisotropic real-space charge patterns (Hussain et al., 2020, Brillaux et al., 2020, Zhang et al., 2021). This suggests that the term has both a narrow model-specific meaning and a broader descriptive role for insulating phases with locally broken rotational symmetry.

1. Conceptual definition and scope

In the checkerboard-lattice literature, the site-nematic insulator is defined as a charge-ordered insulating state with unequal occupation of the two sublattices $A$ and $B$ inside each unit cell, but without the additional spatial modulation characteristic of a stripe phase (Sur et al., 2018). The associated order parameter is the sublattice imbalance,

$\frac{\langle n_{i,A}-n_{i,B}\rangle}{2},$

which is finite in the site-nematic insulator and vanishes in the symmetric semimetal (Sur et al., 2018). In the newer constrained-path quantum Monte Carlo formulation, the same phase is diagnosed through the staggered sublattice density correlation

$$\delta_{\mathrm{SNI}} = \frac{1}{N_s} \sum_{\mathbf r} \left\langle \left(c^\dagger_{\mathbf r,1}c_{\mathbf r,1}-c^\dagger_{\mathbf r,2}c_{\mathbf r,2}\right) \left(c^\dagger_{\mathbf r_0,1}c_{\mathbf r_0,1}-c^\dagger_{\mathbf r_0,2}c_{\mathbf r_0,2}\right) \right\rangle,$$

whose finite long-range value signals site-nematic order (Liu et al., 21 Jul 2025).

In insulating transition-metal compounds, the term is used more locally. In La$_2$O$_3$Fe$_2$Se$C_3$0, the relevant phenomenon is the emergence of two inequivalent La sites below $C_3$1 K, with different quadrupole couplings and likely domain segregation, together with low-frequency fluctuations in the paramagnetic phase that are interpreted as nematic-like (Hussain et al., 2020). In the cuprate context, the closely related order is often intra-unit-cell nematicity on oxygen sites,

$C_3$2

rather than a Cu-site charge imbalance, and it appears as part of a hierarchy including spin stripes, charge stripes, and isotropic phases (Fischer et al., 2014).

The expression should not be extended indiscriminately. In Cu$C_3$3Bi$C_3$4Se$C_3$5, the phase above $C_3$6 is a vestigial nematic phase generated by superconducting fluctuations. It has $C_3$7 but $C_3$8, so it is non-superconducting but not an insulator in the localized-electron sense (Hecker et al., 2017).

2. Checkerboard-lattice realization

A canonical realization is the half-filled spinless-fermion checkerboard lattice. One formulation studies

$C_3$9

with $C_{3z}$0, $C_{3z}$1, and $C_{3z}$2, so that the noninteracting system has a quadratic band touching at half filling (Sur et al., 2018). A related formulation fixes

$C_{3z}$3

and uses nearest-neighbor repulsion $C_{3z}$4 as the control parameter; in momentum space, the noninteracting model has a quadratic band touching at $C_{3z}$5, protected by time-reversal symmetry and a fourfold lattice rotational symmetry $C_{3z}$6 (Liu et al., 21 Jul 2025).

At strong coupling, the interactions favor localized charge configurations that spontaneously break $C_{3z}$7. In the two-interaction classical limit, the competing insulating energies are

$C_{3z}$8

so the site-nematic insulator is one of the strong-coupling endpoints of the phase diagram (Sur et al., 2018). In the constrained-path quantum Monte Carlo study with nearest-neighbor repulsion only, the half-filled system displays three interaction-induced phases: a quantum anomalous Hall insulator at weak coupling, a bond-nematic Dirac semimetal at intermediate coupling, and a site-nematic insulator at strong coupling (Liu et al., 21 Jul 2025).

The symmetry content is sharp. The site-nematic insulator breaks $C_{3z}$9 lattice rotation symmetry, preserves time-reversal symmetry, and exhibits an imbalance of electron density between sublattices 1 and 2 (Liu et al., 21 Jul 2025). It differs from the bond-nematic Dirac semimetal because both phases break the same $A$0 symmetry, but the bond-nematic state remains gapless with two Dirac cones, whereas in the site-nematic insulator the Dirac cones annihilate and a full gap opens (Liu et al., 21 Jul 2025). It also differs from the stripe insulator, which has additional translation-symmetry breaking; in the stripe state the neighboring-site difference

$A$1

is finite, while it is zero in the site-nematic insulator (Sur et al., 2018).

The numerical evidence is correspondingly specific. On finite torus geometries, the bond-nematic Dirac semimetal to site-nematic insulator transition occurs near $A$2, depending on size and boundary conditions, and is identified as first order (Liu et al., 21 Jul 2025). Finite-size scaling of the site-nematic order parameter gives

$A$3

at $A$4, supporting long-range order in the thermodynamic limit (Liu et al., 21 Jul 2025). At the same time, weak-coupling analytical work rejects a weak-coupling site-nematic instability in this model: the nematic semimetal proposed for $A$5 at weak coupling is absent, and the quantum anomalous Hall state is the only weak-coupling instability of the spinless quadratic-band-touching semimetal (Sur et al., 2018).

3. Mott and charge-transfer settings

A different, experimentally anchored realization appears in the Mott insulator La$A$6O$A$7Fe$A$8Se$A$9. $B$0La NQR shows a single La site in the paramagnetic phase and two inequivalent La sites, La1 and La2, in the antiferromagnetically ordered phase below $B$1 K (Hussain et al., 2020). The two sites have different quadrupole couplings,

$B$2

implying different local electric-field gradients and therefore distinct lattice and/or charge configurations (Hussain et al., 2020). Their comparable intensity supports an intrinsic bulk origin rather than impurity physics, and the authors suggest distorted and undistorted domains, or possibly a charge-disproportionated phase-separated state (Hussain et al., 2020). Within the paper’s interpretation, this is the microscopic signature of a site-nematic Mott insulator: local $B$3 symmetry is broken and atomic sites become inequivalent without requiring a large global orthorhombic distortion (Hussain et al., 2020).

The temperature dependence links this site inequivalence to magnetism. The quadrupole frequency of La2 follows the magnetic order-parameter-like evolution below $B$4, which the authors interpret as evidence that the magnetic order parameter drives the structural distortion (Hussain et al., 2020). Dynamically, the transverse relaxation rate $B$5 begins increasing around $B$6 K, has a clear maximum around $B$7 K, diverges at $B$8, and reflects fluctuations with characteristic times in the tens of microseconds, corresponding to MHz-scale frequencies far below the Heisenberg exchange scale (Hussain et al., 2020). In the fast-fluctuation regime,

$B$9

which the paper identifies as strongly reminiscent of nematic fluctuations in frustrated spin systems (Hussain et al., 2020).

The cuprate literature develops the same theme more microscopically, though not always as a standalone insulating phase. In a three-band Emery model, quadrupolar fluctuations in the oxygen $\frac{\langle n_{i,A}-n_{i,B}\rangle}{2},$0-orbitals inevitably generate a biquadratic coupling between Cu spins, and this coupling favors local stripe-like magnetic fluctuations that enhance the $\frac{\langle n_{i,A}-n_{i,B}\rangle}{2},$1 nematic susceptibility (Orth et al., 2017). The resulting susceptibility,

$\frac{\langle n_{i,A}-n_{i,B}\rangle}{2},$2

is amplified by the oxygen-fluctuation-induced biquadratic exchange, but its peak remains controlled by the magnetic scale $\frac{\langle n_{i,A}-n_{i,B}\rangle}{2},$3, not by the biquadratic coupling itself (Orth et al., 2017). This mechanism explains enhanced nematic fluctuations near an antiferromagnetic Mott insulator, rather than a spontaneous long-range site-nematic phase by itself (Orth et al., 2017).

A related strong-coupling three-band model for hole-doped charge-transfer insulators finds that doped oxygen holes frustrate the Cu antiferromagnetic background through Kondo-type Cu–O exchange, stabilizing a hierarchy of spin stripes, charge stripes, intra-unit-cell nematic order, and isotropic phases (Fischer et al., 2014). The order parameter

$\frac{\langle n_{i,A}-n_{i,B}\rangle}{2},$4

measures oxygen-site inequivalence within a CuO$\frac{\langle n_{i,A}-n_{i,B}\rangle}{2},$5 unit cell (Fischer et al., 2014). The work supports a robust intra-unit-cell nematic sector tied to charge-transfer insulating physics, while also emphasizing that the clearest explicitly insulating states are stripe insulators rather than an isolated pure nematic insulator across generic parameters (Fischer et al., 2014).

4. Moiré graphene nematic insulators

In moiré graphene systems, “nematic insulator” refers to gapped correlated states that spontaneously break the rotational symmetry of the moiré lattice. Near charge neutrality in twisted bilayer graphene, a generalized Bistritzer–MacDonald continuum model with corrugation effects and symmetry-classified contact interactions yields two classes of quartic interactions: one favoring gap opening and one favoring nematic ordering (Brillaux et al., 2020). The combined group-theory and renormalization-group analysis concludes that proximity to the first magic angle favors a layer-polarized, gapped state with a spatial modulation of interlayer correlations, termed a nematic insulator (Brillaux et al., 2020). In the paper’s language, the state is insulating because a charge gap opens, and nematic because $\frac{\langle n_{i,A}-n_{i,B}\rangle}{2},$6 is broken by stripe-like modulation of interlayer density or interlayer correlations (Brillaux et al., 2020).

At half filling of the flat bands in twisted bilayer-monolayer graphene and twisted double-bilayer graphene under finite displacement fields, unrestricted Hartree–Fock calculations find spin-polarized, $\frac{\langle n_{i,A}-n_{i,B}\rangle}{2},$7-broken insulator states with zero total Chern number (Zhang et al., 2021). The dominant order parameters are of the form

$\frac{\langle n_{i,A}-n_{i,B}\rangle}{2},$8

and the resulting real-space charge density becomes stripe-like and nematic (Zhang et al., 2021). If only the dominant intra-valley Coulomb interaction is included, the spin-polarized insulator is quasi-degenerate with a valley-polarized state, but adding atomic on-site Hubbard interactions lowers the energy of the spin-polarized state by about $\frac{\langle n_{i,A}-n_{i,B}\rangle}{2},$9 meV per electron, making it the unique ground state (Zhang et al., 2021). In some displacement-field regimes these nematic insulators are also quantum valley Hall states, with zero total Chern number but nontrivial valley Chern numbers (Zhang et al., 2021).

A nearby semimetallic counterpart also exists. Self-consistent Hartree–Fock calculations for magic-angle twisted bilayer graphene at charge neutrality, restricted to valley-$$\delta_{\mathrm{SNI}} = \frac{1}{N_s} \sum_{\mathbf r} \left\langle \left(c^\dagger_{\mathbf r,1}c_{\mathbf r,1}-c^\dagger_{\mathbf r,2}c_{\mathbf r,2}\right) \left(c^\dagger_{\mathbf r_0,1}c_{\mathbf r_0,1}-c^\dagger_{\mathbf r_0,2}c_{\mathbf r_0,2}\right) \right\rangle,$$0-preserving orders, find three classes of competitive states: $$\delta_{\mathrm{SNI}} = \frac{1}{N_s} \sum_{\mathbf r} \left\langle \left(c^\dagger_{\mathbf r,1}c_{\mathbf r,1}-c^\dagger_{\mathbf r,2}c_{\mathbf r,2}\right) \left(c^\dagger_{\mathbf r_0,1}c_{\mathbf r_0,1}-c^\dagger_{\mathbf r_0,2}c_{\mathbf r_0,2}\right) \right\rangle,$$1-breaking insulators, spin- or valley-polarized insulators, and $$\delta_{\mathrm{SNI}} = \frac{1}{N_s} \sum_{\mathbf r} \left\langle \left(c^\dagger_{\mathbf r,1}c_{\mathbf r,1}-c^\dagger_{\mathbf r,2}c_{\mathbf r,2}\right) \left(c^\dagger_{\mathbf r_0,1}c_{\mathbf r_0,1}-c^\dagger_{\mathbf r_0,2}c_{\mathbf r_0,2}\right) \right\rangle,$$2-breaking semimetals whose gaplessness is protected by the topology of the moiré flat bands (Liu et al., 2019). Weak strains that break $$\delta_{\mathrm{SNI}} = \frac{1}{N_s} \sum_{\mathbf r} \left\langle \left(c^\dagger_{\mathbf r,1}c_{\mathbf r,1}-c^\dagger_{\mathbf r,2}c_{\mathbf r,2}\right) \left(c^\dagger_{\mathbf r_0,1}c_{\mathbf r_0,1}-c^\dagger_{\mathbf r_0,2}c_{\mathbf r_0,2}\right) \right\rangle,$$3 stabilize both the nematic semimetal and, somewhat unexpectedly, the $$\delta_{\mathrm{SNI}} = \frac{1}{N_s} \sum_{\mathbf r} \left\langle \left(c^\dagger_{\mathbf r,1}c_{\mathbf r,1}-c^\dagger_{\mathbf r,2}c_{\mathbf r,2}\right) \left(c^\dagger_{\mathbf r_0,1}c_{\mathbf r_0,1}-c^\dagger_{\mathbf r_0,2}c_{\mathbf r_0,2}\right) \right\rangle,$$4-breaking insulators (Liu et al., 2019). This places the moiré nematic insulator in a landscape where gapped and gapless rotational-symmetry-broken states can be very close in energy.

5. Diagnostics and empirical signatures

The identification of site-nematic insulating behavior depends strongly on the microscopic platform, but the core diagnostic is always a static or fluctuating anisotropy tied to site, sublattice, or local orbital inequivalence.

Platform	Primary diagnostic	Interpretation
Checkerboard lattice	Finite $$\delta_{\mathrm{SNI}} = \frac{1}{N_s} \sum_{\mathbf r} \left\langle \left(c^\dagger_{\mathbf r,1}c_{\mathbf r,1}-c^\dagger_{\mathbf r,2}c_{\mathbf r,2}\right) \left(c^\dagger_{\mathbf r_0,1}c_{\mathbf r_0,1}-c^\dagger_{\mathbf r_0,2}c_{\mathbf r_0,2}\right) \right\rangle,$$5; full gap	Sublattice density imbalance with broken $$\delta_{\mathrm{SNI}} = \frac{1}{N_s} \sum_{\mathbf r} \left\langle \left(c^\dagger_{\mathbf r,1}c_{\mathbf r,1}-c^\dagger_{\mathbf r,2}c_{\mathbf r,2}\right) \left(c^\dagger_{\mathbf r_0,1}c_{\mathbf r_0,1}-c^\dagger_{\mathbf r_0,2}c_{\mathbf r_0,2}\right) \right\rangle,$$6
La$$\delta_{\mathrm{SNI}} = \frac{1}{N_s} \sum_{\mathbf r} \left\langle \left(c^\dagger_{\mathbf r,1}c_{\mathbf r,1}-c^\dagger_{\mathbf r,2}c_{\mathbf r,2}\right) \left(c^\dagger_{\mathbf r_0,1}c_{\mathbf r_0,1}-c^\dagger_{\mathbf r_0,2}c_{\mathbf r_0,2}\right) \right\rangle,$$7O$$\delta_{\mathrm{SNI}} = \frac{1}{N_s} \sum_{\mathbf r} \left\langle \left(c^\dagger_{\mathbf r,1}c_{\mathbf r,1}-c^\dagger_{\mathbf r,2}c_{\mathbf r,2}\right) \left(c^\dagger_{\mathbf r_0,1}c_{\mathbf r_0,1}-c^\dagger_{\mathbf r_0,2}c_{\mathbf r_0,2}\right) \right\rangle,$$8Fe$$\delta_{\mathrm{SNI}} = \frac{1}{N_s} \sum_{\mathbf r} \left\langle \left(c^\dagger_{\mathbf r,1}c_{\mathbf r,1}-c^\dagger_{\mathbf r,2}c_{\mathbf r,2}\right) \left(c^\dagger_{\mathbf r_0,1}c_{\mathbf r_0,1}-c^\dagger_{\mathbf r_0,2}c_{\mathbf r_0,2}\right) \right\rangle,$$9Se$_2$0	One La site above $_2$1, two below; slow $_2$2 fluctuations	Local site inequivalence and magnetically induced distortion
Twisted graphene	Broken $_2$3 or $_2$4; stripe-like/local filling anisotropy	Gapped nematic moiré state

In the checkerboard model, the order parameter itself is a density-density correlation of sublattice imbalance, and finite-size scaling supplies a direct thermodynamic test. The constrained-path quantum Monte Carlo results show long-range site-nematic order in the strong-coupling regime, while the simultaneous disappearance of bond-nematic order tracks the transition from the bond-nematic Dirac semimetal into the gapped site-nematic state (Liu et al., 21 Jul 2025).

In La$_2$5O$_2$6Fe$_2$7Se$_2$8, the evidence is spectroscopic and dynamical. Two inequivalent quadrupole frequencies below $_2$9 imply distinct local electric-field gradients, while the activated slowing down seen in $_3$0 reveals low-frequency fluctuations already in the paramagnetic phase (Hussain et al., 2020). The coexistence of local inequivalence and slow fluctuations is central to the paper’s site-nematic interpretation (Hussain et al., 2020).

In moiré graphene, the observables are spatial and topological. The charge-neutral twisted-bilayer analysis predicts a gapped, layer-polarized state with broken $_3$1 and spatial modulation of interlayer correlations (Brillaux et al., 2020). In twisted multilayer graphene, the broken-$_3$2 insulator produces stripe-like charge density and can carry nontrivial valley topology even when the total Chern number is zero (Zhang et al., 2021). In the magic-angle bilayer Hartree–Fock study, the local filling fraction

$_3$3

displays clear $_3$4-breaking patterns compatible with STM observations near charge neutrality (Liu et al., 2019).

6. Relation to neighboring phases and common confusions

The site-nematic insulator is often confused with several related but distinct phases. The first distinction is with the stripe insulator. In the checkerboard model, the site-nematic insulator has finite sublattice imbalance but zero stripe order, whereas the stripe insulator has both sublattice imbalance and translation-symmetry breaking along one direction (Sur et al., 2018). The second distinction is with the bond-nematic Dirac semimetal: both break $_3$5, but the bond-nematic state remains gapless, while the site-nematic state is fully gapped after Dirac-cone annihilation (Liu et al., 21 Jul 2025).

A different confusion concerns spin-nematic insulators. In the spin-1 bilinear-biquadratic model on the triangular lattice, the spin-nematic insulator is a ferro-quadrupolar state with no conventional magnetic dipole order,

$_3$6

in the zero-field limit, but with quadrupolar order and hidden spin symmetry breaking (Ishikawa et al., 2024). This is not a charge-ordered site-nematic insulator; its order parameter is quadrupolar in spin space, and its experimental probe in the cited work is spin pumping from a ferromagnetic insulator into a spin-nematic insulator (Ishikawa et al., 2024).

The vestigial nematic phase proposed for Cu$_3$7Bi$_3$8Se$_3$9 is likewise distinct. There the nematic order is composite and fluctuation-induced, derived from a two-component superconducting order parameter,

$_2$0

with $_2$1 and no superconducting long-range order above $_2$2 (Hecker et al., 2017). The phase is “insulating” only in the sense that it is not superconducting; it is not an electronically localized insulator (Hecker et al., 2017).

Finally, nematic order in Dirac systems need not imply an insulating phase. On the surface of a three-dimensional topological insulator, time-reversal-invariant nematic order spontaneously breaks rotational symmetry but distorts rather than generically gaps the Dirac cone (Lundgren et al., 2017). At the Dirac point, the paper finds a first-order transition between isotropic and nematic Dirac semimetals, while at finite doping it finds a second-order transition between isotropic and nematic helical Fermi liquids (Lundgren et al., 2017). This suggests that “nematic” and “insulating” should be treated as separate descriptors unless the underlying work explicitly establishes both.

Across the literature, the most restrictive meaning of site-nematic insulator remains the checkerboard-lattice phase with sublattice density imbalance and broken $_2$3 symmetry. Broader usages extend the idea to Mott, charge-transfer, and moiré systems whenever an insulating state acquires local site inequivalence or symmetry-breaking anisotropy in real space.