Perfect Klein Tunneling in Dirac and Analog Systems

• Perfect Klein tunneling is the phenomenon where massless Dirac quasiparticles achieve unity transmission at normal incidence, regardless of barrier height or width.
• Mechanisms such as pseudospin conservation, spin-momentum locking, and topological mass inversion underpin reflectionless transport in various Dirac and analog media.
• Experimental realizations in graphene, photonic crystals, and metamaterials substantiate theoretical predictions and broaden the concept to multifaceted tunneling scenarios.

Perfect Klein tunneling is the unit transmission of a relativistic or Dirac quasiparticle through a potential barrier under conditions that suppress backscattering, most characteristically at normal incidence in a massless Dirac system. In the canonical barrier problem, the phenomenon appears as $T(0)=1$ independently of barrier height and width, in sharp contrast with nonrelativistic Schrödinger tunneling. In condensed-matter and wave analogues, the effect is realized by Dirac cones, pseudospin or spin-momentum locking, mass-sign changes, or specially tailored constitutive parameters, and it has been studied in graphene, topological insulators, bilayer systems, photonic crystals, phononic and elastic metamaterials, and related platforms (Zhang et al., 2021, Nakatsugawa et al., 2023).

1. Dirac-barrier formulation

The standard description of perfect Klein tunneling is the scattering of a two-component Dirac spinor from a one-dimensional barrier. In photonic graphene, paraxial beam propagation in a refractive-index-modulated medium obeys

$$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$

and, near the $K$ points of a honeycomb lattice, reduces to the massless Dirac Hamiltonian

$$\hat H\psi=[\,c\,\sigma\!\cdot\!\hat p+V(x)\,]\psi=E\psi,$$

with $\sigma=(\sigma_x,\sigma_y)$ acting on the $A/B$ sublattice pseudospin (Zhang et al., 2021).

For a barrier of height $V_0$ and width $d$, transverse-momentum conservation yields the Dirac-Snell relation

$E\sin\phi=-(E-V_0)\sin\phi_t.$

The closed-form transmission probability depends on $\phi$, $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$0, $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$1, and the longitudinal momentum inside the barrier; in the limit $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$2, one recovers perfect Klein tunneling,

$$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$3

which is the defining normal-incidence result in the massless case (Zhang et al., 2021).

The same structure appears in other Dirac media. In honeycomb photonic crystals with a massive Dirac Hamiltonian

$$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$4

the normal-incidence transmission reduces, in the massless limit $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$5, to

$$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$6

This form makes explicit that, for massless Dirac waves, unity transmission is barrier-independent rather than a fine-tuned resonance (Nakatsugawa et al., 2023).

2. Symmetry protection and matching mechanisms

In graphene-type systems, perfect Klein tunneling is conventionally attributed to pseudospin conservation. At normal incidence, the incident and transmitted pseudospins are aligned, so the reflected state is excluded by chirality mismatch. The same logic appears in the deformed $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$7-$$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$8 lattice: in the Dirac phase, at $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$9, one has $K$0 for arbitrary $K$1 and arbitrary barrier height $K$2, because the incident and transmitted pseudospins $K$3 are perfectly aligned (Mandhour et al., 2020).

A distinct but closely related mechanism appears in topological-insulator surface states, where the relevant conserved quantity is real spin rather than sublattice pseudospin. For a three-dimensional topological-insulator PN junction, the surface Hamiltonian

$K$4

implies spin-momentum locking, and at normal incidence $K$5 for all values of barrier height or width. The suppression of normal reflection follows because the spin does not flip for $K$6, so backscattering is forbidden (Xie et al., 2017).

An algebraic formulation of the same reflectionlessness is given by the supersymmetric treatment of graphene and metallic nanotubes. For the one-dimensional massless Dirac Hamiltonian with scalar potential,

$K$7

the unitary chiral rotation

$K$8

satisfies

$K$9

Because the interacting Hamiltonian is unitarily equivalent to the free massless Dirac operator, the right-moving and left-moving channels decouple, and the transmitted amplitude differs only by a pure phase; hence $$\hat H\psi=[\,c\,\sigma\!\cdot\!\hat p+V(x)\,]\psi=E\psi,$$0 for all smooth $$\hat H\psi=[\,c\,\sigma\!\cdot\!\hat p+V(x)\,]\psi=E\psi,$$1 at normal incidence (Jakubsky et al., 2010).

A further mechanism arises when the Dirac mass changes sign. In topological photonic crystals, a domain wall between opposite masses, $$\hat H\psi=[\,c\,\sigma\!\cdot\!\hat p+V(x)\,]\psi=E\psi,$$2 and $$\hat H\psi=[\,c\,\sigma\!\cdot\!\hat p+V(x)\,]\psi=E\psi,$$3, gives

$$\hat H\psi=[\,c\,\sigma\!\cdot\!\hat p+V(x)\,]\psi=E\psi,$$4

even for $$\hat H\psi=[\,c\,\sigma\!\cdot\!\hat p+V(x)\,]\psi=E\psi,$$5; for a finite barrier with $$\hat H\psi=[\,c\,\sigma\!\cdot\!\hat p+V(x)\,]\psi=E\psi,$$6, one finds $$\hat H\psi=[\,c\,\sigma\!\cdot\!\hat p+V(x)\,]\psi=E\psi,$$7 for all $$\hat H\psi=[\,c\,\sigma\!\cdot\!\hat p+V(x)\,]\psi=E\psi,$$8. The paper identifies this as the photonic-crystal analogue of perfect Klein tunneling without any large potential, topologically associated with the Jackiw-Rebbi index-change interface and its chiral zero mode at $$\hat H\psi=[\,c\,\sigma\!\cdot\!\hat p+V(x)\,]\psi=E\psi,$$9 (Nakatsugawa et al., 2023).

These results suggest that perfect Klein tunneling is not tied to a single microscopic symmetry. Rather, it recurs whenever the barrier problem preserves the forward channel while eliminating the backward one, whether by pseudospin matching, spin-momentum locking, unitary gauge equivalence, or topological mass inversion.

3. Angular dependence and incidence geometry

The idealized statement $\sigma=(\sigma_x,\sigma_y)$0 applies only at a special incidence condition in most massless Dirac systems. Away from normal incidence, transmission becomes angle-dependent and may fall rapidly. In photonic graphene, the first experimental observation of perfect Klein transmission in a two-dimensional photonic system was accompanied by a direct measurement of the angular dependence, including the barrier-height dependence that had not previously been measured experimentally (Zhang et al., 2021).

The key result in that system is counterintuitive from a nonrelativistic viewpoint: as the incidence angle $\sigma=(\sigma_x,\sigma_y)$1 increases, $\sigma=(\sigma_x,\sigma_y)$2 drops more rapidly for a lower barrier $\sigma=(\sigma_x,\sigma_y)$3 than for a higher barrier $\sigma=(\sigma_x,\sigma_y)$4. The physical interpretation is given directly by the Dirac-Snell law

$\sigma=(\sigma_x,\sigma_y)$5

Increasing $\sigma=(\sigma_x,\sigma_y)$6 makes $\sigma=(\sigma_x,\sigma_y)$7 more negative, increases $\sigma=(\sigma_x,\sigma_y)$8, and maintains real $\sigma=(\sigma_x,\sigma_y)$9 over a broader angular range, thereby delaying total external reflection. In the graphene interpretation offered there, high defects are therefore less deleterious to conductivity than small ones, because oblique-incidence Klein tunneling is more robust for taller barriers (Zhang et al., 2021).

Strong angular selectivity has also been measured in nanoelectromechanical metamaterials. In the gigahertz elastic-wave N-P-N device, the exact transmission coefficient

$A/B$0

gives $A/B$1 for all $A/B$2 in the N-P-N alignment, but experimentally $A/B$3 drops from $A/B$4 at $A/B$5 to $A/B$6 at $A/B$7, in excellent agreement with the analytic expression (Lee et al., 2024).

Not all Dirac materials place perfect transmission at normal incidence. In 8-Pmmn borophene, the tilted Dirac cone produces oblique Klein tunneling: the special transmitted channel has $A/B$8, but the group velocity is tilted, so the angle of perfect transmission is

$A/B$9

In the corresponding NPN junctions, the resonance condition $V_0$0 generates asymmetric perfect-transmission branches because $V_0$1 when the cone is tilted or anisotropic (Kong et al., 2021).

4. Experimental realizations

Photonic graphene provides a direct optical realization of the two-dimensional Dirac barrier problem. In a $V_0$2 cm long rubidium vapor cell at $V_0$3C, three coupling beams form a honeycomb refractive-index lattice under electromagnetically-induced transparency, and a fourth stripe-shaped beam creates a tunable barrier of width $V_0$4 mm. The probe beam excites a selected Dirac point and pseudospin, and the output intensity is imaged on a CCD. In the Dirac regime, the probe passes through barriers up to $V_0$5 with no visible attenuation at normal incidence, while at the $V_0$6 point the transmission falls exponentially with $V_0$7; quantitatively, the normalized $V_0$8 at $V_0$9 remains $d$0 for all $d$1 at the $d$2 point (Zhang et al., 2021).

Gigahertz elastic waves realize the same phenomenon in a nanoelectromechanical metamaterial. The platform is an $d$3 nm-thick c-axis AlN membrane patterned into a honeycomb of snowflake motifs with lattice constant $d$4m. In the N-P-N heterostructure, the two Dirac frequencies are $d$5 GHz and $d$6 GHz, with effective barrier height $d$7 MHz and width $d$8m. Transmission-mode microwave impedance microscopy maps the local GHz piezoelectric potential and yields reciprocal-space information. In the $d$9–$E\sin\phi=-(E-V_0)\sin\phi_t.$0 GHz NPN regime, the transmission metric is $E\sin\phi=-(E-V_0)\sin\phi_t.$1, with a peak at $E\sin\phi=-(E-V_0)\sin\phi_t.$2 GHz, while outside that regime it falls to $E\sin\phi=-(E-V_0)\sin\phi_t.$3 for $E\sin\phi=-(E-V_0)\sin\phi_t.$4 GHz and $E\sin\phi=-(E-V_0)\sin\phi_t.$5 for $E\sin\phi=-(E-V_0)\sin\phi_t.$6 GHz; independent network-analyzer $E\sin\phi=-(E-V_0)\sin\phi_t.$7 data show a matching peak at $E\sin\phi=-(E-V_0)\sin\phi_t.$8 GHz (Lee et al., 2024).

On topological-insulator surfaces, the effect has been framed as an experimentally accessible spintronic signature rather than solely as a conductance anomaly. In a TI PN junction, the asymmetry in the potentiometric signal between PP and PN junctions and its overall angular dependence were proposed as direct signatures of Klein tunneling, measurable with a ferromagnetic probe because the transmitted spin texture is tied to the momentum distribution of the surviving modes (Xie et al., 2017).

A superconducting manifestation was reported in point-contact spectroscopy on an epitaxial SmB$E\sin\phi=-(E-V_0)\sin\phi_t.$9/YB$\phi$0 heterostructure. There, perfect Andreev reflection is interpreted as a clear signature of Klein tunneling in a proximity-induced superconducting state on a topological Kondo insulator surface. For SmB$\phi$1 thicknesses $\phi$2 nm and $\phi$3 nm, the normalized differential conductance shows exact doubling, $\phi$4, for $\phi$5 mV. Fits to the Dirac-BTK theory yield $\phi$6 meV and barrier parameter $\phi$7, yet the conductance doubling remains perfect, contrary to standard BTK expectations (Lee et al., 2018).

5. Massive, topological, and multicomponent extensions

Perfect Klein tunneling is not restricted to strictly massless, gapless media. In honeycomb photonic crystals, tuning the effective Dirac mass from positive to negative distinguishes three regimes: massless perfect Klein transmission at $\phi$8, massive Klein transmission via particle-antiparticle conversion for $\phi$9 at large $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$00, and topological Klein transmission for $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$01 even at $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$02. The last case establishes that a large electrostatic barrier is not a necessary condition when the mass profile itself produces the relevant helicity matching (Nakatsugawa et al., 2023).

Bilayer graphene introduces a more subtle relation between perfect transmission, perfect reflection, and Berry phase. In gapped bilayer graphene, electrical gating can tune the Berry phase continuously from $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$03 down to $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$04, and a value $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$05 was experimentally observed. In the low-energy two-band description used there, the normal-incidence transmission satisfies

$$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$06

so $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$07 gives $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$08, yielding a transition from anti-Klein tunneling to nearly perfect Klein tunneling as the Berry phase decreases (Du et al., 2017).

A different bilayer literature uses the term “Klein effect” for the opposite outcome. In Bernal bilayer graphene with balanced relative gaps, $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$09, the normal-incidence transmission vanishes,

$$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$10

and the effect is attributed to Berry phase $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$11 together with electron-hole and time-reversal symmetries (Park et al., 2011). This is a terminological divergence rather than a disagreement about the scattering calculation: one body of work emphasizes bilayer perfect reflection as the bilayer analogue of Klein physics, while another emphasizes the tunable recovery of monolayer-like perfect transmission when the Berry phase reaches $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$12 (Du et al., 2017, Park et al., 2011).

Multicomponent lattices further broaden the classification. In the undeformed $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$13-$$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$14 Dirac phase, perfect Klein tunneling at normal incidence persists for all $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$15, and for the dice limit $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$16 at $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$17 one obtains super-Klein tunneling, $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$18 for all $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$19. Under uniaxial deformation, however, cone merging and gap opening convert the phenomenon into anti-Klein tunneling, with $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$20 in the gapped phase and anti-super-Klein behavior in the dice limit (Mandhour et al., 2020).

6. Related phenomena, limits, and broader generalizations

A central limitation is that perfect Klein tunneling is usually not generic in angle, energy, or internal coupling. In monolayer graphene with Rashba spin-orbit coupling, and equivalently in Bernal bilayer graphene under the mapping $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$21, the normal-incidence transmission evolves from the monolayer result $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$22 at $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$23 to an anti-Klein regime when the barrier mode becomes evanescent. The crossover is non-monotonic and exhibits oscillations and resonances in $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$24 as functions of coupling and potential parameters (Dell'Anna et al., 2018).

Other extensions preserve perfect transmission only in selected internal channels. In a graphene superlattice barrier with intervalley scattering, normally incident electrons can exhibit fully valley-selective Klein tunneling: $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$25 while $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$26 for large $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$27. The angle-averaged transmission then acquires a finite valley polarization, numerically $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$28 for $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$29 and $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$30 for $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$31 (An et al., 2019).

In three-dimensional Weyl semimetals, the normal-incidence unit-transmission channel is supplemented by resonance manifolds. The condition

$$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$32

defines “magic transmission rings” in the $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$33 plane, which are the three-dimensional analogue of magic transmission angles. Magnetic-field-induced gauge shifts displace these rings and allow selective transmission of particular incident vectors (Yesilyurt et al., 2016).

Classical-wave analogues need not replicate graphene’s spinor structure. In a two-dimensional acoustic metamaterial, the constitutive-parameter conditions

$$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$34

yield exact impedance matching, $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$35, and therefore $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$36, $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$37. With anisotropic density, the construction can be extended to omnidirectional Klein-like tunneling, $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$38, $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$39 for all $$i\partial E/\partial z = -(1/2k_0)\nabla_\perp^2 E -(k_0/2)\chi(x,y)E,$$40. The mechanism is explicitly distinguished from graphene: it is scalar and rests on impedance matching and tailored negative-index dispersion rather than pseudospin conservation (Sirota, 2021).

Recent one-dimensional double-barrier work has re-opened the question of interpretation. In that model, perfect-transmission curves pass continuously from the above-barrier zone to the Klein zone, and in the Klein zone perfect transmission occurs even for subcritical barrier heights. Because the resonance conditions are direct analogues of nonrelativistic double-barrier conditions, the authors argue that coherent interference, rather than only spontaneous particle-antiparticle production, can underlie perfect Klein tunneling in that geometry (Zhang et al., 27 Feb 2026). This suggests that the traditional “paradox” language is most secure in the single-step relativistic setting, while in multibarrier Dirac systems the phenomenon can merge continuously with resonant transmission.

Perfect Klein tunneling is therefore best regarded as a family of exact-transparency results in Dirac and Dirac-analogue media. Its canonical form remains unity transmission at a special incidence condition in a massless Dirac barrier, but the modern literature shows that the same unit-transmission outcome can arise from several distinct structures: pseudospin conservation, spin-momentum locking, topological mass inversion, Berry-phase tuning, intervalley filtering, Fabry-Pérot resonance, or constitutive-parameter engineering.