Heterosymmetry Stacking: Principles and Applications

• Heterosymmetry stacking is a concept describing stacking arrangements that induce symmetry changes in multilayer structures, impacting polarization, Berry curvature, and topology.
• It unifies diverse systems including graphene polytypes, moiré stacks, and stacking-disordered crystals through modifications in inversion, mirror, and rotational symmetries.
• Experimental and theoretical studies reveal that stacking order can control electronic dipoles, excitonic behavior, and domain-wall kinetics, enabling novel device functionalities.

Searching arXiv for papers on heterosymmetry stacking and closely related stacking-order phenomena. Heterosymmetry stacking denotes stacking arrangements in which the relative registry, rotation, lateral translation, or local layer type changes the effective symmetry of a multilayer or close-packed structure. In current usage, the term covers several related situations: mixed polytypes such as ABCB tetralayer graphene, registry-contrasted moiré stacks such as AB-AB versus AB-BA twisted double bilayer graphene, bilayers whose composite point group differs from that of the constituent monolayers, and stacking-disordered crystals whose layer types are not related by in-plane translation (Singh et al., 10 Apr 2025, Layek et al., 26 Feb 2025, Hart et al., 2021). Across these settings, the central consequence is the same: stacking ceases to be a passive structural label and becomes a symmetry variable that can control polarization, Berry curvature, topology, magnetic order, excitonic dimensionality, charge-density-wave order, and diffraction.

1. Symmetry principle and scope

At its most elementary, heterosymmetry stacking is the breaking, reshaping, or recombination of symmetries by the stacking sequence itself. In graphene-based systems, this can mean mixing Bernal and rhombohedral registries within one crystal, as in ABCB tetralayer graphene, so that inversion and mirror symmetries are both lost. In moiré systems, it can mean that two stacks with similar flat-band dispersions nevertheless encode different microscopic symmetries through distinct layer–sublattice tunneling matrices, as in AB-AB and AB-BA twisted double bilayer graphene. In stacking-disordered crystals, the term is broader: “heterosymmetric” means that successive layer types are not related by in-plane translations and therefore do not share the same structure factor (Singh et al., 10 Apr 2025, Layek et al., 26 Feb 2025, Hart et al., 2021).

This diversity of usage reflects a common structural logic. The relevant symmetry may be inversion, mirror, layer-exchange, rotational, or translational equivalence; the important point is that the stacking operation changes which of these remain valid for the composite system. A plausible implication is that heterosymmetry stacking is best understood not as a single microscopic mechanism, but as a unifying symmetry framework spanning ordered polytypes, moiré registries, domain-wall heterostructures, and aperiodic stacks.

System class	Heterosymmetry stacking means	Representative consequence
Tetralayer graphene	Mixed stacking sequence such as ABCB	Intrinsic out-of-plane polarization
TDBG	AB-AB versus AB-BA registry	Odd/even field parity of BCD
Few-layer graphene or bilayer MoS$_2$	Coexisting stacking domains	Atomically sharp symmetry boundaries
Close-packed aperiodic crystals	Non-translationally equivalent layer types	Heterosymmetric diffraction
Honeycomb–kagome heterolayers	Registry-dependent point group	$\Gamma$-point topology
Magnetic bilayers	Stacking-defined bilayer point group	Symmetry-inequivalent AFM or altermagnetism

2. Graphene as the canonical electronic realization

The most direct realization of heterosymmetry stacking as an intrinsic electronic order parameter is ABCB tetralayer graphene. Uniform-registry graphene polytypes such as AB bilayer and ABAB tetralayer are inversion symmetric, while ABC trilayer is non-centrosymmetric but still mirror symmetric in a way that forbids a net out-of-plane dipole at zero field in the noninteracting limit. By contrast, ABCB combines a rhombohedral ABC block with a Bernal continuation and therefore breaks both inversion and mirror symmetry. DFT and SWMcC analyses with $\gamma_0 = 3.16\ \text{eV}$, $\gamma_1 = 390\ \text{meV}$, $\gamma_2 = -17\ \text{meV}$, $\gamma_3 = 315\ \text{meV}$, $\gamma_4 = 70\ \text{meV}$, $\gamma_5 = 30\ \text{meV}$, and $\Delta = -5\ \text{meV}$ place the conduction-band edge mainly on layer $1A$ and the valence-band edge on layer $\Gamma$0, yielding a layer-polarized electronic dipole even at charge neutrality. The polarization is written as $\Gamma$1, and the inverted BCBA state reverses its sign (Singh et al., 10 Apr 2025).

Experimentally, dual-gated, non-aligned ABCB tetralayer graphene encapsulated in hexagonal boron nitride shows pronounced hysteresis in resistance under both top and bottom gate modulation, with persistence up to at least $\Gamma$2. The gate controls are $\Gamma$3 and $\Gamma$4. Sweeping $\Gamma$5 at $\Gamma$6 and $\Gamma$7 at $\Gamma$8 gives stable hysteresis offsets $\Gamma$9 and $\gamma_0 = 3.16\ \text{eV}$0, while the estimated areal polarization is $\gamma_0 = 3.16\ \text{eV}$1. The microscopic interpretation is reversible layer-polarized charge reordering coupled to gate-driven interlayer sliding between ABCB and BCBA, captured by a double-well Landau free energy $\gamma_0 = 3.16\ \text{eV}$2 (Singh et al., 10 Apr 2025).

A crucial contrast appears in twisted double bilayer graphene near $\gamma_0 = 3.16\ \text{eV}$3, where heterosymmetry stacking does not create ferroelectricity directly but reprograms quantum geometry. AB-AB and AB-BA TDBG both form inversion-broken moiré superlattices, yet their Berry curvature responses to perpendicular electric field differ categorically. The Berry curvature dipole,

$\gamma_0 = 3.16\ \text{eV}$4

is antisymmetric under $\gamma_0 = 3.16\ \text{eV}$5 in AB-AB and symmetric in AB-BA. Experimentally, the nonlinear Hall signal $\gamma_0 = 3.16\ \text{eV}$6 changes sign with $\gamma_0 = 3.16\ \text{eV}$7 versus $\gamma_0 = 3.16\ \text{eV}$8 in AB-AB, but retains the same sign in AB-BA. Devices at $\gamma_0 = 3.16\ \text{eV}$9 and $\gamma_1 = 390\ \text{meV}$0 were measured with $\gamma_1 = 390\ \text{meV}$1 at $\gamma_1 = 390\ \text{meV}$2, and the intrinsic contribution was isolated through $\gamma_1 = 390\ \text{meV}$3, with $\gamma_1 = 390\ \text{meV}$4 as the BCD indicator (Layek et al., 26 Feb 2025).

Taken together, these graphene cases establish two distinct electronic roles for heterosymmetry stacking. In ABCB graphene, the stacking sequence itself is the source of a switchable polar degree of freedom. In TDBG, the registry does not primarily alter the band dispersion, but instead fixes the transformation properties of Berry curvature and its first moment under field reversal. This suggests that heterosymmetry can act either in real space, by creating a layer dipole, or in momentum space, by dictating how quantum geometry transforms.

3. Domain walls, sliding transitions, and stack-selective kinetics

Heterosymmetry stacking is often realized not as a uniform bulk phase but as an in-plane heterostructure separated by movable boundaries. In few-layer graphene, ABA and ABC regions can coexist within the same flake, forming a stacking-fault soliton or domain wall. Joule heating in hBN-encapsulated devices converts ABC to ABA controllably: in a $\gamma_1 = 390\ \text{meV}$5 device ramped to $\gamma_1 = 390\ \text{meV}$6, the first local transition after $\gamma_1 = 390\ \text{meV}$7 corresponds to a local temperature of $\gamma_1 = 390\ \text{meV}$8 and a domain-wall pressure of $\gamma_1 = 390\ \text{meV}$9, while complete conversion at $\gamma_2 = -17\ \text{meV}$0 corresponds to $\gamma_2 = -17\ \text{meV}$1 and $\gamma_2 = -17\ \text{meV}$2. Raman mapping uses the ratio $\gamma_2 = -17\ \text{meV}$3, with ABC for $\gamma_2 = -17\ \text{meV}$4 and ABA for $\gamma_2 = -17\ \text{meV}$5. In ultrafast TEM, $\gamma_2 = -17\ \text{meV}$6, $\gamma_2 = -17\ \text{meV}$7 laser-pulse trains at $\gamma_2 = -17\ \text{meV}$8 over $\gamma_2 = -17\ \text{meV}$9 drive partial conversion, while dark-field imaging resolves the domain wall with $\gamma_3 = 315\ \text{meV}$0 spatial resolution (Latychevskaia et al., 2019).

Bilayer MoS$\gamma_3 = 315\ \text{meV}$1 provides an atomically sharp variant of the same principle. There, AA′ domains are centrosymmetric and AB domains are noncentrosymmetric, so creating boundaries between them locally toggles inversion symmetry. In situ aberration-corrected STEM at $\gamma_3 = 315\ \text{meV}$2 and $\gamma_3 = 315\ \text{meV}$3 shows triangular AA′ domains nucleating inside AB regions within $\gamma_3 = 315\ \text{meV}$4, with domain area increasing from $\gamma_3 = 315\ \text{meV}$5 to $\gamma_3 = 315\ \text{meV}$6 in $\gamma_3 = 315\ \text{meV}$7, or $\gamma_3 = 315\ \text{meV}$8. The boundary width is on the order of a single unit cell, the Mo lattice remains continuous, and the transition proceeds through a two-step S migration with barriers of $\gamma_3 = 315\ \text{meV}$9 per atom. DFT-LDA further shows in-gap metallic states localized at the domain wall, forming one-dimensional conducting channels (Yan et al., 2017).

In homo-bilayer MoSe$\gamma_4 = 70\ \text{meV}$0, inversion-domain boundaries generate discrete low-symmetry stackings by imposing unusual fractional translations. The net in-plane translation is $\gamma_4 = 70\ \text{meV}$1, or $\gamma_4 = 70\ \text{meV}$2, and eight low-symmetry stackings were observed by ADF-STEM. Across sixteen DFT-relaxed stackings, stacking energy increases nearly linearly with interlayer distance with slope $\gamma_4 = 70\ \text{meV}$3, while the valence-band ordering between $\gamma_4 = 70\ \text{meV}$4 and $\gamma_4 = 70\ \text{meV}$5 changes with registry. The spectroscopic consequence is a set of discrete STS phenotypes—double-peak, band-tail, and low-conductance—rather than a continuum of moiré-like local variations (Hong et al., 2017).

These examples show that heterosymmetry stacking can be both static and kinetic. It may arise from equilibrium selection among near-degenerate registries, but it may also be written, moved, or erased by heating, illumination, electron irradiation, or defect networks. A plausible implication is that stacking boundaries should be treated as active mesoscale objects rather than merely as imperfections.

4. Optical, excitonic, topological, magnetic, and collective-order consequences

The optical signature of heterosymmetry stacking is especially clear in graphene trilayers. Electric-dipole second harmonic generation is allowed in non-centrosymmetric ABA trilayer graphene but forbidden in centrosymmetric ABC trilayer graphene. ABA belongs to the $\gamma_4 = 70\ \text{meV}$6 point group, and the in-plane response reduces effectively to $\gamma_4 = 70\ \text{meV}$7, giving

$\gamma_4 = 70\ \text{meV}$8

The extracted susceptibility is $\gamma_4 = 70\ \text{meV}$9 at $\gamma_5 = 30\ \text{meV}$0, while ABC is SHG-silent within detection limits. This makes polarization-resolved SHG a direct domain-mapping tool for stacking order (Shan et al., 2018).

In wide-gap h-BN, alternative high-symmetry stackings provide a structural counterpart. Among AA, AA′, AB, AB1′, and AB2′, the AB sequence is calculated to be slightly lower in energy than AA′ in both bulk and bilayer, with $\gamma_5 = 30\ \text{meV}$1 per unit cell in bulk and $\gamma_5 = 30\ \text{meV}$2 in bilayer relative to AA′. AA and AB2′ are high-energy and unstable, while AB and AA′ have equilibrium interlayer spacing $\gamma_5 = 30\ \text{meV}$3. The stackings also differ in bandgap character and dielectric tensor, and a modified low-pressure CVD method produces large flakes of virtually pure AB-stacked h-BN, verified by SAED and aberration-corrected HRTEM (Gilbert et al., 2018).

In WSe$\gamma_5 = 30\ \text{meV}$4/MoS$\gamma_5 = 30\ \text{meV}$5 heterobilayers, registry controls excitonic dimensionality and brightness. AA stacking is direct at $\gamma_5 = 30\ \text{meV}$6 with $\gamma_5 = 30\ \text{meV}$7, while ABW and ABSe shift the VBM to $\gamma_5 = 30\ \text{meV}$8, giving indirect $\gamma_5 = 30\ \text{meV}$9 gaps of $\Delta = -5\ \text{meV}$0 and $\Delta = -5\ \text{meV}$1, respectively. GW–BSE identifies three families of excitons: intralayer 2D excitons, dark interlayer charge-transfer excitons, and bright 3D delocalized excitons. Their binding energies depend strongly on stacking: for example, the CT exciton has $\Delta = -5\ \text{meV}$2 in AA, $\Delta = -5\ \text{meV}$3 in ABW, and $\Delta = -5\ \text{meV}$4 in ABSe, while bright 3D excitons exist in AA and ABSe but are absent in ABW. The distinction follows the symmetry-allowed interlayer hybridization pathways rather than registry alone in a geometric sense (Louafi et al., 23 Jun 2025).

Several works show that heterosymmetry stacking can also relocate the locus of topology. In honeycomb–kagome heterolayers, three rotationally symmetric stackings—HHK, THK, and BHK—produce a higher-order topological insulator, a trivial insulator, and a Dirac semimetal, respectively. The decisive inversion occurs near $\Delta = -5\ \text{meV}$5, not $\Delta = -5\ \text{meV}$6, and in the HHK phase the corner charge is $\Delta = -5\ \text{meV}$7 modulo $\Delta = -5\ \text{meV}$8, with six in-gap corner states under $\Delta = -5\ \text{meV}$9-symmetric boundaries. Twisted or lattice-mismatched HK heterostructures then reconstruct into network band structures in which THK regions act as insulating vacancies and HHK/BHK regions form conducting channels (Bark et al., 20 Feb 2025).

A closely related topological contrast appears in bilayers of time-reversed copies of the modified Haldane model. AB stacking lifts the nodal-line degeneracy and produces a class-A Chern insulator with $1A$0, whereas AA stacking remains metallic for $1A$1 or becomes trivial for finite Semenoff mass. The effective low-energy mass $1A$2 changes sign across valleys, giving two chiral edge modes per boundary in the AB case (Mannaï et al., 2022).

Magnetic and collective-order systems exhibit the same organizing principle. In Bernal-stacked MnSi$1A$3N$1A$4/FeSi$1A$5N$1A$6 heterobilayers, AB-like H3 stacking of two noncentrosymmetric monolayers makes the layers locally symmetry-inequivalent and yields “symmetry-inequivalent antiferromagnetism.” DFT-extracted exchanges include $1A$7, $1A$8, and $1A$9 for MnSi$\Gamma$00N$\Gamma$01, and a nearest interlayer exchange $\Gamma$02. Exact diagonalization finds a nondegenerate $\Gamma$03 ground state with a $\Gamma$04 gap to the first excited state (Pedroza-Rojas et al., 8 Feb 2026). In bilayer $\Gamma$05-NbSe$\Gamma$06, stacking between layer-resolved $\Gamma$07 charge-density waves lowers symmetry from $\Gamma$08 in the normal bilayer to $\Gamma$09, $\Gamma$10, $\Gamma$11, or $\Gamma$12-type CDW states depending on the blend and lateral displacement. The ground state is HC–HCS3, and the first excited states lie only $\Gamma$13 above it, with clear STM and Fourier-space fingerprints (Cossu et al., 2024).

5. General theories of stacking order, disorder, and symmetry-protected phases

A major development in the literature is the replacement of purely descriptive stacking notation by explicit symmetry and statistical formalisms. For close-packed aperiodic crystals, a first-order Markov description over block states of finite reichweite $\Gamma$14 generalizes periodic centrosymmetry to reversibility of the stacking process. Standard detailed balance is $\Gamma$15, while for close-packed blocks the relevant condition is

$\Gamma$16

When $\Gamma$17, all topologically close-packed stackings with inversion-symmetric layer types are reversible; for $\Gamma$18, reversible stackings exist but form a measure-zero set. The same framework yields an analytic scattering cross section for heterosymmetric crystals such as stacking-disordered ice I and opal CT, where layer types are not translationally equivalent (Hart et al., 2021).

A complementary equilibrium theory represents any close-packed stacking by correlation variables $\Gamma$19, with enthalpy written as

$\Gamma$20

Here $\Gamma$21 quantifies layer matching at separation $\Gamma$22, and the $\Gamma$23 vary continuously with pressure, temperature, or composition. This maps the infinite set of stackings into a finite continuous $\Gamma$24-space. Because only adjacent stability regions can be crossed under continuous changes of $\Gamma$25, direct transitions such as $\Gamma$26 are forbidden in equilibrium thermodynamics within the convergent expansion (Loach et al., 2017).

The ice literature adds a kinetic, rather than solely equilibrium, realization of heterosymmetry stacking. In situ cryo-TEM at $\Gamma$27 and $\Gamma$28 shows that heterogeneous ice formation proceeds predominantly through an Ic hemispherical core, then stacking-disordered layers, and finally Ih prismatic dendrites. The inter-branch angle is $\Gamma$29, matching the angle between fcc $\Gamma$30 planes, and the cubicity decreases from core to tip. Growth velocities are on the order of $\Gamma$31, and the fluctuating Isd region functions as a dynamic bridge between cubic and hexagonal symmetry (Huang et al., 28 Feb 2026).

General symmetry criteria have also been formulated for bilayer altermagnetism and one-dimensional symmetry-protected topology. The General Stacking Theory for altermagnetism constructs the bilayer point group from $\Gamma$32 and finds that only seven bilayer point groups—$\Gamma$33, $\Gamma$34, $\Gamma$35, $\Gamma$36, $\Gamma$37, $\Gamma$38, and $\Gamma$39—allow altermagnetic spin splitting, provided the opposite-spin coset excludes inversion $\Gamma$40 and horizontal mirror $\Gamma$41 (Pan et al., 2024). In a different setting, stacking two trivial one-dimensional BDI subsystems can generate an emergent invariant,

$\Gamma$42

with $\Gamma$43 nonzero when symmetry-preserving coupling acts between subsystems with different chiral structures and odd orbital-degree-of-freedom mismatch, thereby driving a zero-field topological superconducting phase (Choi et al., 2023).

These theories collectively show that heterosymmetry stacking is not reducible to cataloguing registries. It can be encoded as a Markov reversibility problem, a convex optimization in correlation space, a bilayer point-group construction, or an emergent topological invariant.

6. Applications, misconceptions, and open directions

One recurring misconception is that stacking-enabled functionality in van der Waals systems must be moiré-driven. ABCB tetralayer graphene directly contradicts this: the hBN was intentionally misaligned, transport showed a single charge neutrality point without moiré satellites, and a control ABCA device from the same hBN stack showed no hysteresis. The ferroelectricity there arises from stacking-order-induced symmetry breaking rather than a moiré potential (Singh et al., 10 Apr 2025). The converse misconception is that every stacking contrast yields a nontrivial phase. It does not: THK honeycomb–kagome heterolayers are trivial, AA stacking in the modified-Haldane bilayer is not a Chern insulator, and ABW WSe$\Gamma$44/MoS$\Gamma$45 suppresses the bright 3D exciton channel (Bark et al., 20 Feb 2025, Mannaï et al., 2022, Louafi et al., 23 Jun 2025).

The device implications are already concrete. ABCB graphene suggests non-volatile memory and logic based on room-temperature, dual-gate-addressable hysteresis (Singh et al., 10 Apr 2025). In TDBG, the nonlinear Hall response acts as an all-electrical stacking and symmetry sensor, separating AB-AB from AB-BA through the field parity of the Berry curvature dipole (Layek et al., 26 Feb 2025). SHG microscopy in trilayer graphene, SAED and HRTEM in h-BN, and STM/STS in NbSe$\Gamma$46, MoS$\Gamma$47, and MoSe$\Gamma$48 show that heterosymmetry stacking is not only a route to new phases but also a route to high-throughput symmetry metrology (Shan et al., 2018, Gilbert et al., 2018, Cossu et al., 2024, Yan et al., 2017, Hong et al., 2017).

Several open problems recur across the literature. In graphene ferroelectrics, endurance and retention were not directly quantified even though bistability persists to $\Gamma$49 (Singh et al., 10 Apr 2025). In TDBG, quantitative separation of intrinsic BCD from extrinsic skew and side-jump processes still benefits from scaling analyses in small $\Gamma$50-windows and improved strain control (Layek et al., 26 Feb 2025). In magnetic MA$\Gamma$51Z$\Gamma$52 bilayers, the spin model neglects SOC in the effective Hamiltonian, and larger-scale simulations are needed to assess finite-temperature behavior (Pedroza-Rojas et al., 8 Feb 2026). In ice and other close-packed materials, force-field dependence and direct extraction of kinetic exponents remain unresolved (Huang et al., 28 Feb 2026). More generally, controllable writing of stacking domains, domain walls, and registry textures remains a central engineering challenge.

The broader significance is that heterosymmetry stacking provides a symmetry-native design language. It links atomistic registry to emergent order without changing composition, and it does so across systems that would otherwise appear unrelated: graphene polytypes, moiré flat bands, CDW bilayers, magnetic nitridosilicates, excitonic TMD heterostructures, close-packed polytypes, and stacking-disordered ice. This suggests that future progress will come less from treating stacking as a secondary crystallographic detail, and more from treating it as a tunable symmetry field in its own right.