Hyperbolic Exciton Polaritons

• Hyperbolic exciton polaritons are exciton–photon hybrid modes with hyperbolic isofrequency contours arising from anisotropic excitonic resonances.
• They manifest in diverse platforms including semiconductor Bragg structures, monolayer black phosphorus, and organic J-aggregate films.
• Tuning parameters such as radiative decay, bias fields, and magnetic fields control dispersion, refraction, and nonlinear polaritonic behavior.

Searching arXiv for recent and foundational papers on hyperbolic exciton polaritons and closely related platforms. Searching arXiv for recent and foundational papers on hyperbolic exciton polaritons and closely related platforms. Hyperbolic exciton polaritons are exciton–photon hybrid modes whose constant-frequency contours in momentum space are hyperbolic rather than elliptic because the relevant optical response is strongly anisotropic and changes sign between principal directions. In the literature, this designation encompasses several realizations: Bragg-exciton polaritons in semiconductor photonic crystals with opposite-sign effective masses (Sedov et al., 2016, Sedov et al., 2014), in-plane exciton polaritons in monolayer black phosphorus driven by anisotropic excitonic conductivity (Wang et al., 2021), surface exciton polaritons in natural or artificial hyperbolic excitonic media such as J-aggregate organic films (Thomas et al., 9 Jun 2025, Gama et al., 23 Dec 2025), hybrid exciton–hyperbolic-phonon-polariton states in biased bilayer graphene encapsulated by hBN (Eini et al., 5 Jun 2025), and more recent magnetically controlled magnetoexciton-polariton variants in graphene and van der Waals semiconductors (Domina et al., 30 Jun 2025, Jia et al., 13 Oct 2025). Across these platforms, the central feature is the same: excitonic resonances reshape the dielectric response or the polaritonic band geometry so that energy flow, refraction, confinement, and density of states acquire the characteristic properties of hyperbolic media.

1. Concept and defining criteria

The broad electromagnetic criterion for hyperbolicity is anisotropy with opposite-sign principal response components. For an anisotropic medium with relative permittivity tensor

$$\boldsymbol{\varepsilon}_r= \begin{pmatrix} \varepsilon_{xx} & 0 & 0\ 0 & \varepsilon_{yy} & 0\ 0 & 0 & \varepsilon_{zz} \end{pmatrix},$$

the extraordinary-wave dispersion in a uniaxial case is

$$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$

so the isofrequency surface is hyperbolic when the relevant permittivity components have opposite signs (Jia et al., 2022). In two-dimensional or effectively anisotropic sheet systems, the same condition is frequently expressed in terms of in-plane optical conductivities, for example

$$\mathrm{Im}(\sigma_{xx})\cdot \mathrm{Im}(\sigma_{yy})<0,$$

which yields open in-plane hyperbolic contours rather than closed elliptical ones (Wang et al., 2021, Jia et al., 2022).

In excitonic systems, the anisotropic response is supplied by exciton resonances rather than free-carrier plasmons or optical phonons. A generic excitonic contribution is of Lorentz form,

$$\varepsilon_i(\omega)=\varepsilon_{i,\infty}+\sum_j \frac{f_{ij}}{\omega_{ij}^2-\omega^2-i\gamma_{ij}\omega},$$

so directional differences in oscillator strength, resonance energy, and damping can drive one component through zero while another remains of opposite sign (Jia et al., 2022). This is the basic mechanism behind natural hyperbolic exciton polaritons in layered perovskites, Bi-based chalcogenides, and monolayer black phosphorus (Jia et al., 2022, Wang et al., 2021).

A distinct but closely related route appears in photonic-crystal and Bragg-polariton platforms. There, hyperbolicity is encoded not directly as a closed-form dielectric tensor, but in the polariton band structure through an anisotropic effective mass tensor with opposite signs along orthogonal directions. For the lowest branch near a saddle point, one may write

$$E_1(\mathbf{q}) \simeq E_0 + \frac{\hbar^2}{2m_\parallel}(k_x^2+k_y^2)+\frac{\hbar^2}{2m_\perp}k_z^2,$$

with $m_\parallel>0$ and $m_\perp<0$, which produces hyperbolic isofrequency surfaces in momentum space (Sedov et al., 2014). The corresponding low-energy expansion in a one-dimensional resonant hyperbolic metamaterial is

$$\omega(K,k_\rho)\approx \omega_0+\frac{\hbar}{2m_z^*}K^2+\frac{\hbar}{2m_\rho^*}k_\rho^2,$$

with $m_z^*<0$ and $m_\rho^*>0$ on the lowest branch near its saddle point (Sedov et al., 2016). This suggests that hyperbolic exciton polaritons are best understood as a class unified by hyperbolic polaritonic dispersion, while the microscopic origin of that dispersion may be either excitonic dielectric anisotropy or exciton-modified photonic band geometry.

2. Semiconductor Bragg and photonic-crystal implementations

A foundational semiconductor implementation is the one-dimensional GaN/AlGaN Bragg structure with embedded In$$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$0Ga$$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$1N quantum wells studied as a tunable resonant hyperbolic metamaterial (Sedov et al., 2016). The structure is a periodic stack with GaN layers of thickness $$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$2, refractive index $$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$3, and Al$$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$4Ga$$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$5N layers of thickness $$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$6, refractive index $$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$7, giving period $$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$8. Every GaN layer contains a single In$$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$9Ga$$\mathrm{Im}(\sigma_{xx})\cdot \mathrm{Im}(\sigma_{yy})<0,$$0N quantum well at its center. The second photonic bandgap is centered around $$\mathrm{Im}(\sigma_{xx})\cdot \mathrm{Im}(\sigma_{yy})<0,$$1, and the quantum-well exciton is tuned near its lower edge at $$\mathrm{Im}(\sigma_{xx})\cdot \mathrm{Im}(\sigma_{yy})<0,$$2 (Sedov et al., 2016).

The eigenmodes are obtained from a transfer-matrix equation

$$\mathrm{Im}(\sigma_{xx})\cdot \mathrm{Im}(\sigma_{yy})<0,$$3

with excitonic coupling entering through the quantum-well reflection coefficient

$$\mathrm{Im}(\sigma_{xx})\cdot \mathrm{Im}(\sigma_{yy})<0,$$4

Here $$\mathrm{Im}(\sigma_{xx})\cdot \mathrm{Im}(\sigma_{yy})<0,$$5 is the radiative decay rate and $$\mathrm{Im}(\sigma_{xx})\cdot \mathrm{Im}(\sigma_{yy})<0,$$6 the nonradiative decay rate (Sedov et al., 2016). The quantum wells transform the purely photonic structure into a four-branch polaritonic system, and the lowest branch near the Brillouin-zone center develops opposite curvatures along the growth and in-plane directions, with $$\mathrm{Im}(\sigma_{xx})\cdot \mathrm{Im}(\sigma_{yy})<0,$$7 and $$\mathrm{Im}(\sigma_{xx})\cdot \mathrm{Im}(\sigma_{yy})<0,$$8 (Sedov et al., 2016).

This anisotropy is large. The reported effective-mass ratios are $$\mathrm{Im}(\sigma_{xx})\cdot \mathrm{Im}(\sigma_{yy})<0,$$9 without quantum wells, $$\varepsilon_i(\omega)=\varepsilon_{i,\infty}+\sum_j \frac{f_{ij}}{\omega_{ij}^2-\omega^2-i\gamma_{ij}\omega},$$0 for $$\varepsilon_i(\omega)=\varepsilon_{i,\infty}+\sum_j \frac{f_{ij}}{\omega_{ij}^2-\omega^2-i\gamma_{ij}\omega},$$1, and $$\varepsilon_i(\omega)=\varepsilon_{i,\infty}+\sum_j \frac{f_{ij}}{\omega_{ij}^2-\omega^2-i\gamma_{ij}\omega},$$2 for $$\varepsilon_i(\omega)=\varepsilon_{i,\infty}+\sum_j \frac{f_{ij}}{\omega_{ij}^2-\omega^2-i\gamma_{ij}\omega},$$3 (Sedov et al., 2016). The corresponding equifrequency contours in $$\varepsilon_i(\omega)=\varepsilon_{i,\infty}+\sum_j \frac{f_{ij}}{\omega_{ij}^2-\omega^2-i\gamma_{ij}\omega},$$4 are hyperbolic near the saddle point, and the structure exhibits strong negative refraction and bias-controlled group-velocity reduction (Sedov et al., 2016).

An earlier closely related formulation introduced semiconductor Bragg mirrors with periodically arranged quantum wells as “quantum hyperbolic metamaterials” supporting Bragg exciton polaritons (Sedov et al., 2014). In that work, the lower branch near $$\varepsilon_i(\omega)=\varepsilon_{i,\infty}+\sum_j \frac{f_{ij}}{\omega_{ij}^2-\omega^2-i\gamma_{ij}\omega},$$5 is characterized by

$$\varepsilon_i(\omega)=\varepsilon_{i,\infty}+\sum_j \frac{f_{ij}}{\omega_{ij}^2-\omega^2-i\gamma_{ij}\omega},$$6

with

$$\varepsilon_i(\omega)=\varepsilon_{i,\infty}+\sum_j \frac{f_{ij}}{\omega_{ij}^2-\omega^2-i\gamma_{ij}\omega},$$7

for the representative GaN/AlGaN–InGaN system (Sedov et al., 2014). In dimensionless form, the linear lower-branch equation becomes

$$\varepsilon_i(\omega)=\varepsilon_{i,\infty}+\sum_j \frac{f_{ij}}{\omega_{ij}^2-\omega^2-i\gamma_{ij}\omega},$$8

so the constant-$$\varepsilon_i(\omega)=\varepsilon_{i,\infty}+\sum_j \frac{f_{ij}}{\omega_{ij}^2-\omega^2-i\gamma_{ij}\omega},$$9 surfaces are hyperboloids (Sedov et al., 2014). The same work further showed that this hyperbolic polariton fluid supports X-wave solutions, a Ginzburg–Landau–Higgs mapping, kink solutions, oscillons, and cat-state constructions in the nonlinear regime (Sedov et al., 2014).

A more recent photonic-crystal realization moves from single-particle propagation to condensate hydrodynamics. In a GaAs/AlGaAs photonic-crystal waveguide with a one-dimensional grating, the lower exciton-polariton band near $$E_1(\mathbf{q}) \simeq E_0 + \frac{\hbar^2}{2m_\parallel}(k_x^2+k_y^2)+\frac{\hbar^2}{2m_\perp}k_z^2,$$0 is saddle-shaped,

$$E_1(\mathbf{q}) \simeq E_0 + \frac{\hbar^2}{2m_\parallel}(k_x^2+k_y^2)+\frac{\hbar^2}{2m_\perp}k_z^2,$$1

with positive effective mass along one in-plane direction and negative along the orthogonal direction (Georgakilas et al., 2024). That work demonstrated an optically tunable dimer of hyperbolic exciton-polariton condensates whose coupling continuously crosses over from evanescent to ballistic as the dimer angle is varied relative to the grating (Georgakilas et al., 2024). This suggests that hyperbolic exciton polaritons are not only a linear-wave phenomenon but also a platform for driven-dissipative quantum-fluid physics.

3. Natural and artificial excitonic hyperbolic media

Natural hyperbolic exciton polaritons in the visible and near-infrared have been reviewed in the context of anisotropic two-dimensional materials (Jia et al., 2022). The review identifies several excitonic platforms.

Layered Ruddlesden–Popper perovskites of composition (BA)$$E_1(\mathbf{q}) \simeq E_0 + \frac{\hbar^2}{2m_\parallel}(k_x^2+k_y^2)+\frac{\hbar^2}{2m_\perp}k_z^2,$$2(MA)$$E_1(\mathbf{q}) \simeq E_0 + \frac{\hbar^2}{2m_\parallel}(k_x^2+k_y^2)+\frac{\hbar^2}{2m_\perp}k_z^2,$$3Pb$$E_1(\mathbf{q}) \simeq E_0 + \frac{\hbar^2}{2m_\parallel}(k_x^2+k_y^2)+\frac{\hbar^2}{2m_\perp}k_z^2,$$4I$$E_1(\mathbf{q}) \simeq E_0 + \frac{\hbar^2}{2m_\parallel}(k_x^2+k_y^2)+\frac{\hbar^2}{2m_\perp}k_z^2,$$5 were reported to exhibit hyperbolic regimes centered at about 513 nm for $$E_1(\mathbf{q}) \simeq E_0 + \frac{\hbar^2}{2m_\parallel}(k_x^2+k_y^2)+\frac{\hbar^2}{2m_\perp}k_z^2,$$6 and 571 nm for $$E_1(\mathbf{q}) \simeq E_0 + \frac{\hbar^2}{2m_\parallel}(k_x^2+k_y^2)+\frac{\hbar^2}{2m_\perp}k_z^2,$$7, with ellipsoidal isofrequency surfaces at 400 nm and hyperboloidal ones near the exciton resonance (Jia et al., 2022). The paper further notes that the photonic density of states is greatly enhanced near the excitonic hyperbolic resonance (Jia et al., 2022).

Bi$$E_1(\mathbf{q}) \simeq E_0 + \frac{\hbar^2}{2m_\parallel}(k_x^2+k_y^2)+\frac{\hbar^2}{2m_\perp}k_z^2,$$8Se$$E_1(\mathbf{q}) \simeq E_0 + \frac{\hbar^2}{2m_\parallel}(k_x^2+k_y^2)+\frac{\hbar^2}{2m_\perp}k_z^2,$$9 has been identified as supporting hyperbolic edge exciton polaritons, experimentally probed through energy-dispersive cathodoluminescence, with edge-bound modes around 4 eV propagating along cube edges and reflecting from corners (Jia et al., 2022). Monolayer black phosphorus was predicted to sustain in-plane hyperbolic exciton polaritons in the range 1.703–1.844 eV because of sign-changing imaginary optical conductivity along the armchair direction (Jia et al., 2022, Wang et al., 2021).

Monolayer black phosphorus is the most explicit natural two-dimensional HEP platform in the supplied literature. Its in-plane conductivity tensor is

$m_\parallel>0$0

with AC and ZZ denoting armchair and zigzag axes (Wang et al., 2021). Polarization-resolved reflection spectroscopy on monolayer samples revealed a strong 1s exciton near 1.69 eV and weaker 2s and 3s states near 1.92 eV and 2.04 eV. The extracted exciton binding energy is about 452 meV (Wang et al., 2021). Most importantly, the paper reports that $m_\parallel>0$1 from 1.703 eV to 1.844 eV, while $m_\parallel>0$2 throughout the studied range, yielding

$m_\parallel>0$3

in that interval (Wang et al., 2021). The resulting loss-function analysis shows open hyperbolic isofrequency contours at 1.71, 1.75, and 1.79 eV, with asymptote angles of 58°, 51°, and 47° (Wang et al., 2021). The predicted polariton quality factor reaches about 35.5 at small wavevector (Wang et al., 2021).

A different class of excitonic hyperbolic media is fully organic. Artificial organic hyperbolic metamaterials based on alternating J-aggregate carbocyanine dyes and polyelectrolytes have been shown to exhibit a uniaxial tensor

$m_\parallel>0$4

with $m_\parallel>0$5 in excitonic Reststrahlen-like bands (Gama et al., 23 Dec 2025). For j560, j590, and j620 J-aggregate systems, the reported hyperbolic windows are 526–558 nm, 529–583 nm, and 578–613 nm, respectively (Gama et al., 23 Dec 2025). These films support hyperbolic surface exciton polaritons and, for j560, an additional near-zero-permittivity surface mode (Gama et al., 23 Dec 2025). Structural characterization links the optical anisotropy to preferential in-plane molecular orientation and lamellar stacking (Gama et al., 23 Dec 2025).

A related experimental study on neat TDBC J-aggregates reported what it described as the first experimental study of hyperbolic surface exciton polaritons (Thomas et al., 9 Jun 2025). In that system, the in-plane permittivity is fitted by a Lorentz model with main oscillator at $m_\parallel>0$6 eV, $m_\parallel>0$7 eV, $m_\parallel>0$8, while the out-of-plane component is approximately constant at $m_\parallel>0$9 (Thomas et al., 9 Jun 2025). The material is identified as type-II hyperbolic, with $m_\perp<0$0 and $m_\perp<0$1 from about 2.11–2.43 eV, and $m_\perp<0$2 from about 2.11–2.26 eV (Thomas et al., 9 Jun 2025). Prism-coupled spectroscopic ellipsometry showed a single phase singularity for the hyperbolic surface exciton polariton branch, in contrast to the two singularities found for non-hyperbolic surface plasmon or phonon polaritons (Thomas et al., 9 Jun 2025). This suggests that phase topology in the $m_\perp<0$3 plane can discriminate hyperbolic from non-hyperbolic surface-polariton responses.

4. Magnetoexciton and hybrid hyperbolic regimes

A more recent direction extends HEP concepts into magnetic and hybrid mid-infrared regimes. One realization uses charge-neutral graphene nanoribbon metasurfaces under perpendicular magnetic field to form quantum hyperbolic magnetoexciton polaritons (Domina et al., 30 Jun 2025). In that system, interband Landau-level transitions in charge-neutral graphene act as magnetoexcitons, and the metasurface anisotropy drives a topological transition of isofrequency curves from closed to open as the field is increased from 6 T to 9 T at 25 THz (Domina et al., 30 Jun 2025). The real-space wavefronts evolve from nearly isotropic to hyperbolic-like rays, and at $m_\perp<0$4 T the IFCs can flatten into nearly parallel lines, producing canalization along the ribbon direction (Domina et al., 30 Jun 2025). Although the underlying matter excitation is a Landau-quantized interband magnetoexciton rather than a conventional semiconductor exciton, the work explicitly frames these modes as a new platform for hyperbolic exciton-like polaritons (Domina et al., 30 Jun 2025).

A related theoretical proposal studies hyperbolic magnetoexciton polaritons in monolayer WTe$m_\perp<0$5, MoS$m_\perp<0$6, and phosphorene under Shubnikov–de Haas conditions (Jia et al., 13 Oct 2025). There the conductivity tensor is obtained from a Landau-level Kubo formula, and the resulting surface-polariton IFCs include two-fold hyperbolas, one-sheet hyperbolas, witch-of-Agnesi curves, and twisted pincerlike contours (Jia et al., 13 Oct 2025). Reported group velocities are as low as $m_\perp<0$7 for the $m_\perp<0$8 transition in WTe$m_\perp<0$9, with lifetime $$\omega(K,k_\rho)\approx \omega_0+\frac{\hbar}{2m_z^*}K^2+\frac{\hbar}{2m_\rho^*}k_\rho^2,$$0, and for phosphorene the paper reports lifetimes reaching about $$\omega(K,k_\rho)\approx \omega_0+\frac{\hbar}{2m_z^*}K^2+\frac{\hbar}{2m_\rho^*}k_\rho^2,$$1 ms (Jia et al., 13 Oct 2025). Since these are theoretical predictions tied to specific low-temperature high-field conditions, they should be understood as a proposed HEP regime rather than an established experimental standard.

Another important hybrid regime is the strong coupling between biased-bilayer-graphene excitons and hBN hyperbolic phonon polaritons in the mid-infrared (Eini et al., 5 Jun 2025). The excitonic response is modeled by a conductivity

$$\omega(K,k_\rho)\approx \omega_0+\frac{\hbar}{2m_z^*}K^2+\frac{\hbar}{2m_\rho^*}k_\rho^2,$$2

while hBN supplies type-I or type-II hyperbolicity depending on the Reststrahlen band (Eini et al., 5 Jun 2025). In a symmetric hBN/BBLG/hBN stack, even modes with finite $$\omega(K,k_\rho)\approx \omega_0+\frac{\hbar}{2m_z^*}K^2+\frac{\hbar}{2m_\rho^*}k_\rho^2,$$3 at the graphene plane hybridize strongly with the excitons, while odd modes with $$\omega(K,k_\rho)\approx \omega_0+\frac{\hbar}{2m_z^*}K^2+\frac{\hbar}{2m_\rho^*}k_\rho^2,$$4 do not (Eini et al., 5 Jun 2025). The resulting hybridized exciton–HPhP branches show clear anticrossing and can be tuned between lower and upper Reststrahlen bands by changing the bilayer bias from 58 meV to 115 meV (Eini et al., 5 Jun 2025). This system is not a purely excitonic hyperbolic medium in the same sense as black phosphorus or TDBC, but it is a direct example of excitons inheriting and strongly modifying hyperbolic polaritonic dispersion.

5. Propagation phenomena: refraction, canalization, X-waves, and singular optics

Negative refraction is among the most direct manifestations of hyperbolic dispersion. In the resonant one-dimensional semiconductor hyperbolic metamaterial, full-wave transfer-matrix simulations of a 30-period slab show negative refraction of a Gaussian beam for both the QW-free and excitonic structures, with stronger exciton–photon coupling modifying the refraction angle and reducing beam blurring (Sedov et al., 2016). The same work demonstrated group-velocity control through the exciton radiative rate $$\omega(K,k_\rho)\approx \omega_0+\frac{\hbar}{2m_z^*}K^2+\frac{\hbar}{2m_\rho^*}k_\rho^2,$$5, with $$\omega(K,k_\rho)\approx \omega_0+\frac{\hbar}{2m_z^*}K^2+\frac{\hbar}{2m_\rho^*}k_\rho^2,$$6 decreasing as $$\omega(K,k_\rho)\approx \omega_0+\frac{\hbar}{2m_z^*}K^2+\frac{\hbar}{2m_\rho^*}k_\rho^2,$$7 increases at fixed $$\omega(K,k_\rho)\approx \omega_0+\frac{\hbar}{2m_z^*}K^2+\frac{\hbar}{2m_\rho^*}k_\rho^2,$$8 eV (Sedov et al., 2016).

Canalization arises when the IFC becomes nearly flat over a broad $$\omega(K,k_\rho)\approx \omega_0+\frac{\hbar}{2m_z^*}K^2+\frac{\hbar}{2m_\rho^*}k_\rho^2,$$9-space segment, making the group velocity nearly collinear for many plane-wave components. In the graphene nanoribbon quantum-magnetoexciton metasurface, increasing the period from 150 nm to 200–240 nm at $m_z^*<0$0 T and 25 THz flattens the IFCs into nearly parallel lines and produces strongly collimated propagation along the ribbons (Domina et al., 30 Jun 2025). In a more conventional exciton-polariton condensate system, birefringent CsPbBr$m_z^*<0$1 in a planar microcavity exhibits a hyperbolic–flat–parabolic evolution of lower-polariton IFCs due to TE–TM splitting and birefringence (Ren et al., 29 Jan 2026). The transverse curvature of the y-polarized branch is

$m_z^*<0$2

so the IFC can be hyperbolic, flat, or parabolic depending on energy (Ren et al., 29 Jan 2026). In that experiment, flat IFC condensation at 2.326 eV gave a collimation factor $m_z^*<0$3, hyperbolic IFCs at 2.323 eV gave $m_z^*<0$4, and parabolic IFCs at 2.334 eV gave $m_z^*<0$5 relative to arc-shaped reference contours (Ren et al., 29 Jan 2026). Because the perovskite is explicitly described as non-hyperbolic in the bulk-permittivity sense, this platform is best viewed as a hyperbolic exciton-polariton band geometry rather than a natural hyperbolic excitonic medium.

Linear hyperbolic dispersion also supports non-diffracting X-waves. In the Bragg-polariton quantum hyperbolic metamaterial, the stationary linear equation

$m_z^*<0$6

admits an X-wave solution

$m_z^*<0$7

with

$m_z^*<0$8

(Sedov et al., 2014). This is a specific demonstration that hyperbolic polaritonic dispersion can by itself stabilize localized wave packets without relying on nonlinearity.

An additional optical consequence of hyperbolicity is the emergence of unusual phase singularities and shear-like asymmetries. Hyperbolic surface exciton polaritons in TDBC exhibit a single phase singularity in ellipsometric phase response, unlike non-hyperbolic surface polaritons, which show two (Thomas et al., 9 Jun 2025). More generally, work on vortex-induced shear polaritons shows that vortex excitation of hyperbolic media can generate asymmetric hyperbolic shear-polariton patterns even without intrinsic off-diagonal tensor elements (Xue et al., 2022). That paper concerns phonon polaritons rather than excitons, but it suggests that structured excitation could supply an additional control knob for HEP wavefront engineering.

6. Tunability, nonlinearity, and relation to adjacent polariton classes

Tunability in excitonic hyperbolic systems depends strongly on platform. In the GaN/AlGaN Bragg structure, the key control parameter is the exciton radiative decay rate $m_z^*<0$9, related to the radiative lifetime by

$m_\rho^*>0$0

and controlled through the quantum-confined Stark effect under normal electric field $m_\rho^*>0$1 via

$m_\rho^*>0$2

(Sedov et al., 2016). Increasing $m_\rho^*>0$3 increases the real part of the effective quantum-well permittivity near resonance, for example from $m_\rho^*>0$4 at $m_\rho^*>0$5 eV and $m_\rho^*>0$6 meV to $m_\rho^*>0$7 at the same energy and $m_\rho^*>0$8 meV (Sedov et al., 2016). This tunes the hyperbolic dispersion window, negative-refraction angle, and group velocity (Sedov et al., 2016).

Electrical tunability is also central in biased-bilayer graphene, where the interband exciton energies $m_\rho^*>0$9 shift with displacement field and can be brought into resonance with hBN hyperbolic phonon-polariton branches (Eini et al., 5 Jun 2025). Magnetic tunability plays the analogous role in graphene nanoribbon magnetoexciton metasurfaces and vdW-semiconductor HMEP proposals (Domina et al., 30 Jun 2025, Jia et al., 13 Oct 2025).

Nonlinearity is especially prominent in Bragg exciton polaritons. The mean-field lower-branch dynamics obey

$$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$00

with $$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$01, $$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$02, and a projected exciton–exciton interaction coefficient

$$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$03

(Sedov et al., 2014). After rescaling, this maps to a Ginzburg–Landau–Higgs equation with a Mexican-hat potential,

$$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$04

supporting kinks and oscillons (Sedov et al., 2014). This is a nonlinear hyperbolic polariton regime not usually associated with natural hyperbolic phonon or plasmon polaritons.

The relation between HEPs and other hyperbolic polaritons is therefore platform-dependent but conceptually clear. Hyperbolic phonon polaritons rely on ionic Lorentz oscillators; hyperbolic plasmon polaritons rely on collective free-carrier response; hyperbolic exciton polaritons rely on bound electron–hole resonances (Jia et al., 2022, Wang et al., 2021). A plausible implication is that HEPs naturally occupy the visible and near-infrared more often than phonon-polariton systems, while also providing stronger access to nonlinear and quantum-optical effects than purely plasmonic hyperbolic platforms. The same implication is stated explicitly for several excitonic systems in the supplied material (Jia et al., 2022, Gama et al., 23 Dec 2025).

7. Open issues and scope of the term

The term “hyperbolic exciton polariton” is used across a broader class of systems than a strict materials-based definition might imply. In some papers it refers to modes in a genuinely excitonic hyperbolic medium with opposite-sign dielectric components, as in monolayer black phosphorus, TDBC, layered perovskites, and organic hyperbolic metamaterials (Wang et al., 2021, Jia et al., 2022, Thomas et al., 9 Jun 2025, Gama et al., 23 Dec 2025). In others it refers to exciton polaritons whose band curvature is hyperbolic because of photonic-crystal engineering or birefringent cavity effects, even when the bulk permittivity itself is not hyperbolic (Sedov et al., 2016, Sedov et al., 2014, Georgakilas et al., 2024, Ren et al., 29 Jan 2026). The latter usage is explicit, for example, in the CsPbBr$$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$05 microcavity work, which states that the perovskite is non-hyperbolic while the polariton IFCs become hyperbolic-flat-parabolic because of cavity TE–TM splitting and birefringence (Ren et al., 29 Jan 2026).

Another source of ambiguity is the boundary between excitonic, magnetoexcitonic, and hybrid exciton–hyperbolic-phonon-polariton states. Charge-neutral graphene under magnetic field produces inter-Landau-level magnetoexciton polaritons whose matter component is exciton-like but Landau-quantized (Domina et al., 30 Jun 2025). Biased-bilayer graphene encapsulated by hBN yields modes that are simultaneously excitonic and hyperbolic-phononic (Eini et al., 5 Jun 2025). These are routinely grouped with HEP-related systems in recent literature because the excitonic part controls the hybridization while hyperbolic dispersion governs propagation.

The most consistent interpretation across the cited work is therefore functional rather than taxonomic: a hyperbolic exciton polariton is a polariton whose matter fraction is excitonic or exciton-like and whose isofrequency topology is hyperbolic over the spectral and momentum range of interest. Under that interpretation, the field now spans natural anisotropic crystals, organic excitonic metamaterials, semiconductor photonic crystals, microcavities with engineered band geometry, and magnetic Landau-level systems (Sedov et al., 2016, Sedov et al., 2014, Wang et al., 2021, Jia et al., 2022, Domina et al., 30 Jun 2025, Gama et al., 23 Dec 2025).

The outstanding practical issues are likewise platform-specific but recurring. The 2022 review on natural two-dimensional hyperbolic materials emphasizes limited experimental platforms for excitonic hyperbolicity, strong sensitivity to thickness, and the challenge of losses and material quality (Jia et al., 2022). Organic HSEP work highlights spectral narrowness and environmental stability of J-aggregates (Gama et al., 23 Dec 2025, Thomas et al., 9 Jun 2025). Monolayer black phosphorus offers an appealing natural HEP window but remains chemically fragile and strongly environment-dependent (Wang et al., 2021). Semiconductor Bragg systems add active electrical control but at the cost of greater architectural complexity (Sedov et al., 2016). These considerations suggest that the future of HEP research will likely combine natural anisotropic excitonic media with artificial photonic structuring, so that excitonic resonances supply tunability and nonlinearity while engineered hyperbolic dispersion supplies directional transport, large $$\frac{k_x^2+k_y^2}{\varepsilon_\parallel}+\frac{k_z^2}{\varepsilon_\perp}=k_0^2,$$06, and controllable density of states (Sedov et al., 2014, Gama et al., 23 Dec 2025, Ren et al., 29 Jan 2026).