Semi-Dirac Anisotropy in Two-Dimensional Systems

• Semi-Dirac anisotropy is characterized by a band structure where quasiparticles disperse linearly in one direction and quadratically in the perpendicular direction.
• It emerges from the merging of Dirac cones, linking anisotropic semimetallic behavior with topological transitions and unconventional Landau quantization.
• Experimental platforms, such as strained graphene and Ag₃C₂₀ monolayers, demonstrate its impact on anisotropic optical, transport, and plasmonic responses.

Semi-Dirac anisotropy denotes the defining directional asymmetry of a two-dimensional band touching in which quasiparticles disperse linearly in one momentum direction and quadratically in the orthogonal one. In the contemporary literature this structure appears both as a low-energy fixed-point kinematics and as a critical state associated with the merging of two Dirac cones, so it links anisotropic semimetallic band geometry, topological reconstruction, unconventional Landau quantization, and strongly direction-dependent optical and transport response (Asafov et al., 2024, Uryszek et al., 2019).

1. Definition and canonical Hamiltonians

A standard continuum description is the universal merging Hamiltonian adopted from Montambaux and collaborators,

$$H(\mathbf{k})=(\Delta+a k_x^2)\sigma_x+v k_y \sigma_y,$$

with $a>0$, $v>0$, and tunable merging parameter $\Delta$. Its spectrum,

$$\epsilon_\pm(\mathbf{k})=\pm \sqrt{(\Delta+a k_x^2)^2+v^2 k_y^2},$$

is quadratic along $k_x$ and linear along $k_y$ at $\Delta=0$, the strict semi-Dirac point (Asafov et al., 2024). A closely related field-theoretic convention instead writes

$$H_0(k_x,k_y)=v k_x \sigma_x+\frac{k_y^2}{2m}\sigma_y,$$

so the linear and quadratic axes are exchanged purely by notation; the underlying anisotropy is the same (Uryszek et al., 2019). More general two-band models add an isotropic or momentum-dependent mass term, for example

$$H(\mathbf{k})=V_x k_x^2\sigma_x+V_y k_y\sigma_y+\left(M_0-M_{1x}k_x^2-M_{1y}k_y^2\right)\sigma_z,$$

which preserves the mixed linear–quadratic structure while enabling gap opening, band inversion, and edge topology (Olmos et al., 2024).

This anisotropy is stronger than an ordinary velocity anisotropy of a Dirac cone. One axis carries a finite Dirac velocity, while the orthogonal axis is governed by an inverse mass scale. Correspondingly, the density of states is intermediate between Dirac and parabolic limits,

$a>0$0

rather than $a>0$1 or $a>0$2 (Ly et al., 2022). The same mixed scaling underlies the energy dependence of the effective velocities: along the linear direction the characteristic velocity is finite, whereas along the quadratic direction it vanishes at the touching point and grows away from it (Banerjee et al., 2012).

2. Cone merging, band inversion, and topological reconstruction

In the universal merging picture, $a>0$3 organizes three regimes. For $a>0$4, two Dirac points exist and are separated along the quadratic axis by

$a>0$5

At $a>0$6, the cones merge into a semi-Dirac point. For $a>0$7, the system is gapped (Asafov et al., 2024). This makes semi-Dirac anisotropy inseparable from a topological Lifshitz transition: the linear dispersion retained in the transverse direction coexists with quadratic dispersion generated along the merging direction after the annihilation of opposite-chirality Dirac cones.

The same structure reappears in explicit materials and engineered lattices. In strained Kekulé-O graphene, the gap at the folded $a>0$8 point closes at a critical strain, producing a semi-Dirac spectrum that is quadratic along one axis and linear along the other; beyond the transition the touching splits again into two anisotropic Dirac cones at $a>0$9 (Mohammadi, 9 Jun 2026). In the synthesized Ag$v>0$0C$v>0$1 monolayer, compressive biaxial strain of $v>0$2 or compressive uniaxial $v>0$3-strain of $v>0$4 merges two type-I Dirac cones at the $v>0$5 point into a semi-Dirac state whose orthogonal branch is additionally nearly flat, leading the authors to describe it as a “double semi-Dirac cone” (Ly et al., 2022). In the boron monolayer hr-sB, tilted semi-Dirac cones coexist with Dirac nodal lines; there the crossing along $v>0$6-$v>0$7 is linear in one direction, quadratic in the perpendicular direction, and tilted in energy-momentum space (Zhang et al., 2016).

A recurrent misconception is to treat the semi-Dirac point as merely a distorted Dirac cone. The literature instead treats it as a critical object with different scaling exponents, different density of states, and often a different topological role. In Ag$v>0$8C$v>0$9, for example, the semi-Dirac point is discussed as a quantum Lifshitz transition rather than as a Weyl-like node carrying a simple quantized topological charge (Ly et al., 2022).

3. Anisotropic scaling and critical field theory

Semi-Dirac anisotropy can define the fixed point itself rather than appear as a perturbation. In the Yukawa theories for charge-density-wave, spin-density-wave, and superconducting instabilities, the RG shell is taken in

$\Delta$0

with anisotropic rescaling

$\Delta$1

subject to

$\Delta$2

At the fixed-point convention $\Delta$3, one has $\Delta$4 already at mean-field level (Uryszek et al., 2019). This implies different scaling dimensions for the two spatial directions and transfers the microscopic anisotropy directly into critical behavior.

The bosonic sector is correspondingly non-isotropic. After incorporating the infrared fermionic polarization, the critical propagator becomes

$\Delta$5

which is non-analytic in $\Delta$6 and $\Delta$7 but analytic and quadratic in $\Delta$8 (Uryszek et al., 2019). The order-parameter correlations inherit this structure. The correlation lengths satisfy

$\Delta$9

with

$$\epsilon_\pm(\mathbf{k})=\pm \sqrt{(\Delta+a k_x^2)^2+v^2 k_y^2},$$0

At $$\epsilon_\pm(\mathbf{k})=\pm \sqrt{(\Delta+a k_x^2)^2+v^2 k_y^2},$$1, $$\epsilon_\pm(\mathbf{k})=\pm \sqrt{(\Delta+a k_x^2)^2+v^2 k_y^2},$$2, $$\epsilon_\pm(\mathbf{k})=\pm \sqrt{(\Delta+a k_x^2)^2+v^2 k_y^2},$$3, and therefore

$$\epsilon_\pm(\mathbf{k})=\pm \sqrt{(\Delta+a k_x^2)^2+v^2 k_y^2},$$4

The mean-field Ginzburg–Landau expansion is also nonstandard,

$$\epsilon_\pm(\mathbf{k})=\pm \sqrt{(\Delta+a k_x^2)^2+v^2 k_y^2},$$5

with direction-dependent gradient terms scaling as $$\epsilon_\pm(\mathbf{k})=\pm \sqrt{(\Delta+a k_x^2)^2+v^2 k_y^2},$$6 and $$\epsilon_\pm(\mathbf{k})=\pm \sqrt{(\Delta+a k_x^2)^2+v^2 k_y^2},$$7 (Uryszek et al., 2019).

These results show that semi-Dirac anisotropy is not washed out by interactions in the controlled large-$$\epsilon_\pm(\mathbf{k})=\pm \sqrt{(\Delta+a k_x^2)^2+v^2 k_y^2},$$8 treatment. The reported $$\epsilon_\pm(\mathbf{k})=\pm \sqrt{(\Delta+a k_x^2)^2+v^2 k_y^2},$$9 corrections to $k_x$0, $k_x$1, $k_x$2, and $k_x$3 are numerically small, so the anisotropic universality class remains intact rather than flowing to an isotropic Dirac or conventional bosonic fixed point (Uryszek et al., 2019).

4. Magnetic quantization, quantum Hall response, and Landau-level collapse

In zero electric field, semiclassical Landau quantization follows the Peierls–Onsager rule

$k_x$4

where $k_x$5 is the area enclosed by a constant-energy orbit in momentum space (Asafov et al., 2024). For semi-Dirac dispersion this area scales as

$k_x$6

so the Landau levels obey the characteristic law

$k_x$7

intermediate between graphene’s $k_x$8 and the linear-in-$k_x$9 spectrum of a parabolic 2D electron gas (Asafov et al., 2024). The same $k_y$0 behavior is confirmed in magneto-optical treatments and tight-binding-based ribbon studies (Zhou et al., 2021, Sinha et al., 2020).

The most distinctive magnetic manifestation of semi-Dirac anisotropy in crossed fields is the directional collapse of Landau levels. In the Lifshitz–Kaganov form of semiclassical quantization, the drift velocity is

$k_y$1

and closed orbits are quantized through the effective energy $k_y$2 (Asafov et al., 2024). Because the band grows quadratically along one axis but only linearly along the other, the collapse criterion is axis selective. Along the quadratic direction, $k_y$3 always dominates a linear drift term, so orbits remain closed. Along the Dirac-like direction, collapse occurs once the drift component exceeds the intrinsic Dirac velocity: $k_y$4 With $k_y$5, this implies a critical in-plane field

$k_y$6

while an electric field parallel to the quadratic axis does not induce collapse (Asafov et al., 2024). This contrasts sharply with isotropic Dirac systems, where the collapse criterion is direction independent.

For $k_y$7, two separated semiclassical orbits coexist and magnetic breakdown becomes relevant near the saddle point between cones. The quantization condition must then include tunneling between orbits, following the form used by Alexandradinata and Glazman. Numerical solutions show loops and paired levels below the saddle energy, but the collapse threshold remains $k_y$8, independent of $k_y$9, because it is fixed by the large-momentum asymptotics rather than by low-energy cone separation (Asafov et al., 2024).

Magnetotransport reflects the same anisotropy. In a semi-Dirac nanoribbon, the Hall conductivity is reported to quantize as

$\Delta=0$0

including spin degeneracy, distinct from the graphene sequence, and the longitudinal conductivities satisfy $\Delta=0$1 because the bulk Landau bands become dispersive in a direction-dependent manner (Sinha et al., 2020).

5. Optical, magneto-optical, and plasmonic signatures

The optical conductivity of a semi-Dirac point is intrinsically polarization dependent. In the merging-cone model

$\Delta=0$2

the interband conductivities differ parametrically: the $\Delta=0$3-component scales as $\Delta=0$4 while the $\Delta=0$5-component scales as $\Delta=0$6 at high frequency, and for $\Delta=0$7 the $\Delta=0$8-polarized conductivity develops a logarithmic van Hove singularity at

$\Delta=0$9

with no analogous divergence in the $$H_0(k_x,k_y)=v k_x \sigma_x+\frac{k_y^2}{2m}\sigma_y,$$0-channel (Oriekhov et al., 2022). In strained Kekulé-O graphene the same complementary critical behavior appears at the semi-Dirac transition, where

$$H_0(k_x,k_y)=v k_x \sigma_x+\frac{k_y^2}{2m}\sigma_y,$$1

and evolves under strain into a low-energy sequence of gapped absorption peaks, semi-Dirac critical scaling, and a pronounced van Hove optical resonance (Mohammadi, 9 Jun 2026).

Magneto-optics provides an even sharper directional fingerprint. In the semi-Dirac $$H_0(k_x,k_y)=v k_x \sigma_x+\frac{k_y^2}{2m}\sigma_y,$$2 model used for two-dimensional magneto-optical absorption, the Landau levels again scale as $$H_0(k_x,k_y)=v k_x \sigma_x+\frac{k_y^2}{2m}\sigma_y,$$3, but the selection rules for interband transitions depend on the polarization axis: $$H_0(k_x,k_y)=v k_x \sigma_x+\frac{k_y^2}{2m}\sigma_y,$$4 for linearly polarized light along the linear-dispersion direction, and

$$H_0(k_x,k_y)=v k_x \sigma_x+\frac{k_y^2}{2m}\sigma_y,$$5

for polarization along the parabolic direction, whereas intraband transitions obey

$$H_0(k_x,k_y)=v k_x \sigma_x+\frac{k_y^2}{2m}\sigma_y,$$6

independently of polarization (Zhou et al., 2021). The interband magneto-optical conductivity for light polarized along the linear direction is reported to be two orders of magnitude larger than that for the parabolic direction, leading to strong linear dichroism and, for the lowest transition $$H_0(k_x,k_y)=v k_x \sigma_x+\frac{k_y^2}{2m}\sigma_y,$$7, perfect linear dichroism with magnetic-field-tunable wavelength (Zhou et al., 2021).

At the level of macroscopic optics, anisotropy enters the dielectric tensor itself. In semi-Dirac heterostructures the optical conductivity tensor produces $$H_0(k_x,k_y)=v k_x \sigma_x+\frac{k_y^2}{2m}\sigma_y,$$8; this shifts Brewster angles, phase-jump loci, and resonant conditions, yielding direction-dependent Goos–Hänchen shifts and photonic spin Hall signals (Hong-Liang et al., 15 Apr 2025). The same work introduces unidirectional carrier drift along the Dirac-like direction, which adds a nonreciprocal control knob: reverse drift suppresses the photonic spin Hall peak, whereas co-drift enhances it for moderate velocities (Hong-Liang et al., 15 Apr 2025).

Collective charge dynamics is likewise anisotropic. In single-layer semi-Dirac systems, Dirac cone generation for finite negative $$H_0(k_x,k_y)=v k_x \sigma_x+\frac{k_y^2}{2m}\sigma_y,$$9 enhances the plasmon frequency range, and the plasmon spectrum is strongly anisotropic, especially for finite $$H(\mathbf{k})=V_x k_x^2\sigma_x+V_y k_y\sigma_y+\left(M_0-M_{1x}k_x^2-M_{1y}k_y^2\right)\sigma_z,$$0 with vanishing inversion terms (Giri et al., 15 Jun 2025). In bilayers, Coulomb coupling generates a second plasmon branch; rotating the upper layer controls whether the two collective oscillations are in phase or out of phase, so semi-Dirac anisotropy becomes a route to phase-tunable coupled plasmons (Giri et al., 15 Jun 2025).

6. Edge topology, disorder, and material realizations

Semi-Dirac anisotropy can localize topology to a momentum slice rather than to the entire Brillouin zone. In the minimal two-band model

$$H(\mathbf{k})=V_x k_x^2\sigma_x+V_y k_y\sigma_y+\left(M_0-M_{1x}k_x^2-M_{1y}k_y^2\right)\sigma_z,$$1

edge states appear only when the system is confined along the linear direction. A strip finite in $$H(\mathbf{k})=V_x k_x^2\sigma_x+V_y k_y\sigma_y+\left(M_0-M_{1x}k_x^2-M_{1y}k_y^2\right)\sigma_z,$$2 and periodic in $$H(\mathbf{k})=V_x k_x^2\sigma_x+V_y k_y\sigma_y+\left(M_0-M_{1x}k_x^2-M_{1y}k_y^2\right)\sigma_z,$$3 hosts in-gap edge bands with quadratic dispersion in $$H(\mathbf{k})=V_x k_x^2\sigma_x+V_y k_y\sigma_y+\left(M_0-M_{1x}k_x^2-M_{1y}k_y^2\right)\sigma_z,$$4, but the opposite strip geometry does not (Olmos et al., 2024). This is formalized by treating the 2D Hamiltonian as a family of 1D Hamiltonians at fixed $$H(\mathbf{k})=V_x k_x^2\sigma_x+V_y k_y\sigma_y+\left(M_0-M_{1x}k_x^2-M_{1y}k_y^2\right)\sigma_z,$$5 and computing the Zak phase

$$H(\mathbf{k})=V_x k_x^2\sigma_x+V_y k_y\sigma_y+\left(M_0-M_{1x}k_x^2-M_{1y}k_y^2\right)\sigma_z,$$6

Only at $$H(\mathbf{k})=V_x k_x^2\sigma_x+V_y k_y\sigma_y+\left(M_0-M_{1x}k_x^2-M_{1y}k_y^2\right)\sigma_z,$$7 is the Zak phase pinned to $$H(\mathbf{k})=V_x k_x^2\sigma_x+V_y k_y\sigma_y+\left(M_0-M_{1x}k_x^2-M_{1y}k_y^2\right)\sigma_z,$$8; away from that slice it drifts continuously because the SSH-like chiral reduction is lost (Olmos et al., 2024). For this reason the system is described there as a “high order momentum topological insulator.”

Disorder does not erase this structure uniformly. Anderson disorder that preserves the effective chiral symmetry protects the zero-mode sector more efficiently than symmetry-breaking disorder: the zero-energy density-of-states dip survives to larger disorder strength, the first $$H(\mathbf{k})=V_x k_x^2\sigma_x+V_y k_y\sigma_y+\left(M_0-M_{1x}k_x^2-M_{1y}k_y^2\right)\sigma_z,$$9 conductance plateau is more robust, and transport along the edge-state direction remains much less sensitive than transport along the orthogonal direction (Olmos et al., 2024). In the quantum diffusive regime, semi-Dirac semimetals also show anisotropic weak localization, anisotropic conductance fluctuations, and an anomalous Hall response in which intrinsic and side-jump contributions vanish while skew scattering survives (Bi et al., 2021).

Realizations and candidate platforms span both synthetic and material settings. The universal merging Hamiltonian is explicitly connected to anisotropic honeycomb-like lattices under strain or anisotropic hopping, cold-atom optical lattices, and microwave networks (Asafov et al., 2024). Semi-Dirac behavior has been modeled in black phosphorus and VO$a>0$00/TiO$a>0$01 heterostructures, while semi-Dirac criticality has also been constructed holographically as a strongly coupled phase interpolating between a Dirac semimetal and an insulator (Uryszek et al., 2019, Bahamondes et al., 2024). Ag$a>0$02C$a>0$03 provides a synthesized 2D platform where semi-Dirac states, nodal structures, and pronounced mechanical anisotropy coexist under modest strain, and hr-sB supplies a distinct example of tilted semi-Dirac cones coexisting with nodal lines (Ly et al., 2022, Zhang et al., 2016).

Semi-Dirac anisotropy is therefore best regarded not as a minor band-structure distortion but as a distinct organizing principle. It fixes the scaling of the quasiparticle spectrum, selects the direction in which Landau quantization can collapse, differentiates optical and plasmonic channels by polarization and propagation angle, and can even restrict topological protection to isolated momentum slices.