Papers
Topics
Authors
Recent
Search
2000 character limit reached

Majorana Fermions: Concepts and Realizations

Updated 9 November 2026
  • Majorana fermions are fermionic excitations that are their own antiparticles, characterized by the Majorana condition $\psi=\psi^c$, and in condensed matter, by particle–hole-related creation and annihilation operators coinciding at zero energy.
  • Majorana fermions find applications in topological quantum computing due to their non-Abelian exchange statistics, potential for producing quasiparticles and as topological states in superconductors.
  • In two-dimensional chiral superconductors, Majorana fermions can occur at boundaries or vortex cores, while in three-dimensional systems, they can form bulk nodes and surface arcs.
  • keywords_homework_list_keywords_keywords_keyword_selector

Majorana fermions are fermionic excitations that coincide with their own antiparticles. In relativistic quantum field theory, this is expressed by the Majorana condition ψ=ψc\psi=\psi^c, where ψc\psi^c is the charge-conjugate field. In condensed matter, the term usually denotes an emergent Bogoliubov quasiparticle whose particle–hole-related creation and annihilation operators coincide at zero energy, γ†=γ\gamma^\dagger=\gamma. Such modes are electrically neutral relative to electron number, obey fermionic algebra, and, when spatially separated and topologically protected, can generate non-Abelian exchange statistics. The distinction between generic BdG particle–hole redundancy and an isolated, localized, topologically protected Majorana zero mode is fundamental (Sato et al., 2016).

1. Conceptual and historical foundations

Ettore Majorana proposed in 1937 that a relativistic fermion could be described by a real spinor field. The Dirac equation,

(iγμ∂μ−m)ψ=0,(i\gamma^\mu\partial_\mu-m)\psi=0,

normally treats particles and antiparticles as independent degrees of freedom. A Majorana field instead satisfies

ψ=ψc,\psi=\psi^c,

so the particle and antiparticle are identical. A massive Majorana fermion cannot carry an ordinary conserved Abelian charge, including electromagnetic charge. The Majorana mass term couples a Weyl field to its complex conjugate and violates fermion-number or lepton-number conservation by two units (Borsten et al., 2016).

The possible Majorana nature of neutrinos remains an unresolved particle-physics question. Neutrino oscillations establish nonzero neutrino masses but do not distinguish Dirac from Majorana masses. Neutrinoless double-beta decay,

(A,Z)⟶(A,Z+2)+2e−,(A,Z)\longrightarrow(A,Z+2)+2e^-,

would provide evidence for lepton-number violation and, under the standard light-neutrino mechanism, Majorana neutrinos. Its rate is proportional to the effective Majorana mass,

Γ0νββ∝∣mββ∣2,mββ=∣∑iUei2mi∣.\Gamma_{0\nu\beta\beta}\propto |m_{\beta\beta}|^2, \qquad m_{\beta\beta}=\left|\sum_i U_{ei}^2m_i\right|.

The absence of an established observation leaves the elementary-particle question open (Borsten et al., 2016).

In superconductors, Majorana behavior has a different status. Electrons remain charged Dirac fermions, but superconducting pairing breaks the effective electron-number U(1)U(1) symmetry. The Bogoliubov–de Gennes formalism introduces a Nambu spinor containing electron and hole components. Its particle–hole symmetry relates positive- and negative-energy solutions,

γE†=γ−E.\gamma_E^\dagger=\gamma_{-E}.

At zero energy this becomes

γ0†=γ0.\gamma_0^\dagger=\gamma_0.

A superconducting Majorana is therefore an emergent quasiparticle, not an elementary particle. All BdG systems possess particle–hole redundancy, including ordinary ss-wave superconductors; what is special in a topological superconductor is an isolated, spatially localized, zero-energy state protected by a bulk invariant (Sato et al., 2016).

2. Algebra, BdG structure, and topological classification

A Majorana operator is Hermitian and satisfies a Clifford algebra. Depending on normalization,

γi†=γi,{γi,γj}=2δij,\gamma_i^\dagger=\gamma_i, \qquad \{\gamma_i,\gamma_j\}=2\delta_{ij},

or {γi,γj}=δij\{\gamma_i,\gamma_j\}=\delta_{ij}. Two Majorana operators form an ordinary complex fermion,

d=γa+iγb2,nd=d†d=1+iγaγb2.d=\frac{\gamma_a+i\gamma_b}{2}, \qquad n_d=d^\dagger d=\frac{1+i\gamma_a\gamma_b}{2}.

The fermionic information is stored in the joint parity of the pair rather than in either Majorana separately. For a set of spatially separated modes, local perturbations cannot generally shift a single nondegenerate zero mode away from zero unless another Majorana is coupled to it or the protecting bulk gap closes (Stanescu et al., 2013).

The BdG Hamiltonian satisfies particle–hole symmetry,

CH(k)C−1=−H(−k),{\cal C}H({\bm k}){\cal C}^{-1}=-H(-{\bm k}),

with an antiunitary operator such as C=τxK{\cal C}=\tau_xK. Consequently, if ∣un(k)⟩|u_n({\bm k})\rangle has energy En(k)E_n({\bm k}), then C∣un(k)⟩{\cal C}|u_n({\bm k})\rangle has energy −En(k)-E_n({\bm k}). A zero-energy eigenstate can be chosen to obey the self-conjugacy condition

C∣u0⟩=∣u0⟩.{\cal C}|u_0\rangle=|u_0\rangle.

The topological protection of Majorana modes is described by the symmetry class and dimensionality of the gapped BdG Hamiltonian. Class D, characterized by particle–hole symmetry with C2=+1{\cal C}^2=+1 and absent time-reversal symmetry, supports a Z2\mathbb Z_2 invariant in one dimension and an integer Chern invariant in two dimensions. Class BDI can occur when an additional effective chiral or time-reversal-like symmetry is present, giving an integer invariant in one dimension. Interactions can reduce free-fermion classifications, including Z→Z8\mathbb Z\to\mathbb Z_8 for one-dimensional BDI systems and Z→Z16\mathbb Z\to\mathbb Z_{16} for three-dimensional DIII superconductors (Sato et al., 2016).

For a one-dimensional class-D system, the nontrivial Z2\mathbb Z_2 phase supports one Majorana mode at each end of a finite wire. In two dimensions, a nonzero Chern number produces chiral Majorana edge modes and can protect vortex-core Majorana zero modes. In three-dimensional systems, class-DIII winding numbers protect helical Majorana surface modes. Gapless systems can also host Majorana nodes carrying Berry-monopole or Chern charge, with associated surface arcs (Kozii et al., 2016).

The canonical one-dimensional example is the Kitaev chain,

H=∑i[−t(ai†ai+1+h.c.)−μ(ai†ai−12)+Δaiai+1+h.c.].H=\sum_i\left[-t(a_i^\dagger a_{i+1}+\mathrm{h.c.}) -\mu\left(a_i^\dagger a_i-\frac12\right) +\Delta a_i a_{i+1}+\mathrm{h.c.}\right].

Writing

ai=12(γ2i−1+iγ2i),a_i=\frac12(\gamma_{2i-1}+i\gamma_{2i}),

the topological phase occurs for

2∣t∣>∣μ∣,Δ≠0.2|t|>|\mu|, \qquad \Delta\neq0.

At t=Δ>0t=\Delta>0 and μ=0\mu=0, neighboring bulk Majoranas pair while one Majorana remains unpaired at each end. The bulk gap closes at 2t=±μ2t=\pm\mu, separating trivial and nontrivial phases. In a finite chain, the end-mode splitting is exponentially small but nonzero, with an oscillatory dependence on microscopic parameters (Borsten et al., 2016).

3. Principal condensed-matter realization mechanisms

The most developed realization combines a semiconductor with strong Rashba spin–orbit coupling, proximity-induced ss-wave pairing, and Zeeman splitting. A representative continuum BdG Hamiltonian is

HBdG(kx)=(ℏ2kx22m∗−μ)τz+αkxσyτz+Γσx+Δτx,H_{\mathrm{BdG}}(k_x)= \left(\frac{\hbar^2k_x^2}{2m^*}-\mu\right)\tau_z +\alpha k_x\sigma_y\tau_z +\Gamma\sigma_x +\Delta\tau_x,

where Γ=12gμBB\Gamma=\tfrac12g\mu_BB, α\alpha is the Rashba coefficient, and Δ\Delta is the induced pairing amplitude. The magnetic field must have a component perpendicular to the effective spin–orbit field. In the ideal single-band limit, the topological condition is

Γ2>Δ2+μ2,\Gamma^2>\Delta^2+\mu^2,

with a bulk-gap closing at

Γc=Δ2+μ2.\Gamma_c=\sqrt{\Delta^2+\mu^2}.

Above the transition, a finite topological segment supports one Majorana mode at each end (Stanescu et al., 2013).

A related two-dimensional architecture uses a Rashba semiconductor quantum well coupled to an ss-wave superconductor and exchange-coupled to a ferromagnetic insulator or subject to an external Zeeman field. For an InAs-like quantum well, the normal-state bands are

Ekβ=k22m′−μ+βα2k2+Vz2,β=±1.{\cal E}_k^\beta=\frac{k^2}{2m'}-\mu+ \beta\sqrt{\alpha^2k^2+V_z^2}, \qquad \beta=\pm1.

The regime

Vz>μV_z>\mu

places only the lower band at the Fermi level. Projecting the singlet proximity pairing onto this single band yields

Δk∝ky−ikx,\Delta_{\mathbf k}\propto k_y-i k_x,

which is equivalent, up to phase conventions, to chiral px+ipyp_x+ip_y pairing. The physical edge then supports chiral interface states without requiring vortices. Their dispersion is

∣ε(θ)∣=αΔkF∣sin⁡θ∣2Vz2+α2kF2,|\varepsilon(\theta)| = \frac{\alpha\Delta k_F|\sin\theta|} {2\sqrt{V_z^2+\alpha^2k_F^2}},

so the energy vanishes at normal incidence but is generally finite at nonzero momentum parallel to the edge. These states form an edge-localized subgap continuum rather than an isolated, immobile vortex zero mode (Linder et al., 2010).

Helical magnetism provides another route. In singlet superconductors coexisting with a magnetic spiral, a spin-dependent gauge transformation removes the spatially varying magnetism and generates a momentum-dependent exchange field. Projection onto spin-split bands converts singlet pairing into an effective odd-parity amplitude,

Δeff(p)∝Δ hz(p)2∣h(p)∣.\Delta_{\rm eff}(\mathbf p) \propto \Delta\,\frac{h_z(\mathbf p)}{2|\mathbf h(\mathbf p)|}.

Because hz(p)h_z(\mathbf p) changes sign under pz→−pzp_z\to-p_z, the projected pairing is nonchiral pp-wave. Surfaces can then host a finite density of zero-energy Majorana modes over a region of conserved surface momentum. Confinement into a wire discretizes the momenta and can isolate a discrete number of modes. The representative scale Kξ≫1K\xi\gg1 allows rapid magnetic modulation to coexist with singlet superconductivity (Martin et al., 2011).

Magnetic-atom chains on superconducting substrates realize related physics through hybridized Yu–Shiba–Rusinov states. A magnetic impurity produces subgap Shiba states with energies

E±Δ=±1−α21+α2,α=πνFJS.\frac{E_\pm}{\Delta}=\pm\frac{1-\alpha^2}{1+\alpha^2}, \qquad \alpha=\pi\nu_FJS.

A chain of impurities hybridizes these states into a one-dimensional band. Exchange coupling, spin–orbit interaction or helical order, and superconducting pairing can produce an effective spinless odd-parity superconducting sector. A nonzero winding number and a bulk gap then imply end Majorana modes. Experiments on Fe chains on Pb observed end-localized zero-bias conductance enhancements, but the observations remain consistent with, rather than conclusive of, topological Majorana modes (Pawlak et al., 2019).

Honeycomb lattices offer a lattice realization of the same ingredients. A zigzag ribbon with Zeeman field, Rashba coupling, and induced ss-wave pairing maps at k=0k=0 and t3=2tt_3=2t onto a one-dimensional monomer chain. The gap-closing condition is

(μ−∑i=13ti)2+Δ2=VZ2.\left(\mu-\sum_{i=1}^3t_i\right)^2+\Delta^2=V_Z^2.

In the topological regime, zero-energy edge branches appear on opposite ribbon boundaries and have unit integrated Majorana polarization. Numerical calculations also find related edge modes in isotropic lattices and in some armchair geometries (Dutreix et al., 2013).

4. Spatial modes, vortex lattices, and higher-dimensional quasiparticles

Majorana modes occur at boundaries, vortices, interfaces, domain walls, and defects of topological superconductors. In a two-dimensional chiral pp-wave phase, a boundary supports a chiral branch,

E=Δ0ky,E=\Delta_0 k_y,

with

γky=γ−ky†.\gamma_{k_y}=\gamma^\dagger_{-k_y}.

Only the zero-momentum state is exactly at zero energy; the complete edge branch is a dispersive chiral Majorana mode (Sato et al., 2016).

Vortex-core Majoranas are localized in both spatial directions and are tied to superconducting phase winding. By contrast, edge states are localized transverse to the boundary but extended along it. The distinction is important experimentally: a broad edge-localized subgap conductance enhancement should not be interpreted as an isolated quantized zero-bias peak (Linder et al., 2010).

In a vortex lattice, isolated zero modes hybridize through tunneling. For a px+ipyp_x+ip_y superconductor, the amplitude has an oscillatory and exponentially decaying dependence on vortex separation,

t(r)∼cos⁡(kFr+π4)e−r/ξ.t(r)\sim \cos\left(k_{\mathrm F}r+\frac{\pi}{4}\right)e^{-r/\xi}.

Positional disorder can therefore change both the magnitude and sign of tunneling. On a triangular lattice, random sign disorder creates domains of opposite Chern number, ν=+1\nu=+1 and ν=−1\nu=-1. Their interfaces support chiral Majorana modes. Weak disorder produces localized subgap states, while stronger disorder produces percolating domain walls and extended low-energy states. A percolation transition occurs near

pc≃0.15,p_c\simeq0.15,

where pp is the sign-flip probability. The strongest zero-energy enhancement can occur below this threshold, near pe≃0.11p_e\simeq0.11, where long narrow domain walls generate anomalously soft one-dimensional modes (Kraus et al., 2011).

A different phase-sensitive tunneling mechanism in vortex arrays produces macroscopically degenerate flat bands. The inter-vortex amplitude satisfies

tAB∝isin⁡(ϕA−ϕB2),t_{AB}\propto i\sin\left(\frac{\phi_A-\phi_B}{2}\right),

where ϕA−ϕB\phi_A-\phi_B is determined by the global superconducting phase profile. Certain nearest-neighbor links therefore vanish exactly. In triangular and square vortex lattices, the resulting superlattices contain localizable zero-energy Majorana combinations. For the triangular lattice, one band is flat at zero energy while another is dispersive; the square lattice contains site-localized zero modes together with additional zero-energy states and a gapless band (Biswas, 2013).

Three-dimensional chiral superconductors can host bulk Majorana nodes. Strong spin–orbit coupling and odd-parity chiral pairing may be non-unitary, causing spin-selective gapping. One pseudospin sector can remain gapless at point nodes while the other is gapped. Linear nodes have C=±1C=\pm1 and density of states

N(E)∼E2,N(E)\sim E^2,

whereas quadratic tangential nodes have C=±2C=\pm2 and

N(E)∼E.N(E)\sim E.

Bulk nodes imply Majorana surface arcs connecting projections of nodes with opposite charge. Anisotropic NMR relaxation is proposed as a diagnostic: for quadratic nodes, the relaxation rate behaves as T3T^3 for nuclear polarization perpendicular to the nodal axis and is suppressed to T4/Δ~0T^4/\widetilde\Delta_0 for polarization along the axis (Kozii et al., 2016).

Other proposed generalizations include spinon Majorana fermions in frustrated Z2Z_2 spin liquids coupled to topological-insulator surfaces. In that setting, the excitations are combinations of neutral spinons and antispinons rather than electrons and holes. A topological proximity effect transfers Dirac dispersion and spin–momentum locking to the spinons; a time-reversal-breaking mass or defect can then support chiral spinon Majorana modes (Wang et al., 2016). Birefringent Majorana fermions extend the effective-field concept further by proposing self-conjugate modes with two distinct velocities,

E=±v1∣k∣,E=±v2∣k∣,v1≠v2.E=\pm v_1|\mathbf k|, \qquad E=\pm v_2|\mathbf k|, \qquad v_1\neq v_2.

This remains a phenomenological construction without a specified microscopic material realization (Zhang, 2022).

5. Interactions, statistics, and quantum-information applications

For well-separated Majorana zero modes, exchanging two modes implements

γi⟶γj,γj⟶−γi.\gamma_i\longrightarrow\gamma_j, \qquad \gamma_j\longrightarrow-\gamma_i.

The corresponding braid operator can be written

Uij=exp⁡(−π4γiγj)=12(1−γiγj).U_{ij}= \exp\left(-\frac{\pi}{4}\gamma_i\gamma_j\right) = \frac{1}{\sqrt2}(1-\gamma_i\gamma_j).

Different exchanges do not generally commute, producing non-Abelian statistics. The resulting degeneracy is protected when the modes remain separated, the bulk gap remains open, and fermion parity is conserved (Sato et al., 2016).

Majorana braiding alone is not computationally universal. It supplies a restricted Clifford-like operation set that can be efficiently classically simulated. Additional non-topological resources, magic states, collective measurements, or more powerful anyons are required for universal quantum computation. Nevertheless, Majorana operations can generate correlations beyond local hidden-variable theories. With five shared Majorana pairs—ten modes in total—a magic-square construction violates a Bell inequality with ideal value G(P∗)=9G(P^*)=9 against a classical bound

G(P)≤7.G(P)\leq7.

Four shared pairs admit a local hidden-variable model under the restricted measurement set, making ten modes necessary and sufficient for the proposed bipartite nonlocality demonstration (Campbell et al., 2013).

Interactions among Majoranas generate physics beyond quadratic BdG theory. A general interacting Hamiltonian contains bilinear and four-Majorana terms,

H=∑ijitijγiγj+∑ijklgijklγiγjγkγl+⋯ .H=\sum_{ij}it_{ij}\gamma_i\gamma_j+ \sum_{ijkl}g_{ijkl}\gamma_i\gamma_j\gamma_k\gamma_l+\cdots.

One-dimensional interacting Majorana chains exhibit Ising criticality, tricritical-Ising transitions with central charge c=7/10c=7/10, critical phases with c=3/2c=3/2, and symmetry-broken gapped phases. Two-dimensional models can support chiral Majorana edge modes, Z2\mathbb Z_2 topological order, Majorana surface codes, and proposed Fibonacci phases. The Sachdev–Ye–Kitaev model provides an exactly solvable all-to-all interacting Majorana system with non-Fermi-liquid behavior, finite zero-temperature entropy, and maximal Lyapunov growth λL=2πT\lambda_L=2\pi T (Rahmani et al., 2018).

Interactions can also generate odd-frequency correlations. For two end Majoranas μ1\mu_1 and μ2\mu_2, the cross-end anomalous correlator

M12(τ)=⟨T μ1(τ)μ2(0)⟩M^{12}(\tau)=\langle T\,\mu_1(\tau)\mu_2(0)\rangle

can acquire a stable odd-frequency component through boson-mediated interactions. In the model studied, the odd-frequency paired phase appears above a finite critical coupling gcg_c, while the even-frequency solution is energetically unfavorable (Huang et al., 2015).

A Majorana path integral requires care because the field and its conjugate are constrained. A Faddeev–Jackiw treatment introduces the constraint f†−f=0f^\dagger-f=0 and yields the effective kinetic term

Leff=∑ai2f˙afa−H(f).L_{\rm eff}=\sum_a\frac{i}{2}\dot f_a f_a-H(f).

The corresponding quadratic functional integral is naturally associated with a real Grassmann field and, in general, Pfaffian rather than determinant structures (Greco, 2016).

6. Experimental signatures, limitations, and unresolved questions

The most widely used probe is tunneling spectroscopy. An isolated Majorana coupled to a normal lead produces resonant Andreev reflection and, in the ideal zero-temperature and zero-broadening limit,

G(0)=2e2h.G(0)=\frac{2e^2}{h}.

Real measurements are reduced by temperature, tunnel-barrier transmission, intrinsic broadening, finite-size splitting, quasiparticle poisoning, background conductance, and multichannel transport. A zero-bias conductance peak is therefore necessary in some proposed measurements but not sufficient evidence for a Majorana zero mode (Stanescu et al., 2013).

In Al–InAs nanowires, field-induced zero-bias peaks appeared near the expected Zeeman threshold, remained pinned over gate-voltage intervals, split at larger fields, disappeared with increasing temperature, and showed strong magnetic-field orientation dependence. The reported peaks were far below 2e2/h2e^2/h and the experiments did not establish nonlocality or non-Abelian statistics; the results were characterized as strong supporting evidence rather than proof (Das et al., 2012).

Magnetic-chain experiments use STM and STS to resolve spatially dependent local density of states. End-localized zero-energy peaks, chain-length dependence, disappearance when superconductivity is suppressed, spin-polarized spectra, and short- and long-range localization scales are consistent with end Majoranas. However, conventional Andreev states, partially separated Andreev states, Kondo resonances, Tamm–Shockley states, accidental Shiba crossings, disorder, and finite-size broadening can produce similar signatures (Pawlak et al., 2019).

Other proposed diagnostics include bulk-gap closing and reopening, correlated splitting oscillations at both ends, nonlocal conductance, crossed Andreev reflection, parity-sensitive interference, fractional Josephson periodicity, thermal Hall transport, spin-resolved spectroscopy, angle-resolved surface spectroscopy, and braiding or fusion experiments. The fractional Josephson effect is associated with

E(ϕ)∝cos⁡ϕ2,I(ϕ)∝sin⁡ϕ2,E(\phi)\propto\cos\frac{\phi}{2}, \qquad I(\phi)\propto\sin\frac{\phi}{2},

and a 4π4\pi rather than 2π2\pi phase periodicity, but quasiparticle poisoning and conventional mechanisms can mimic such responses (Sato et al., 2016).

The principal theoretical and experimental limitations are multiband occupancy, smooth confinement, interface inhomogeneity, disorder, orbital magnetic-field effects, uncertain electrostatics, finite-size hybridization, soft induced gaps, quasiparticle poisoning, and imperfect parity conservation. In vortex lattices, positional disorder can generate localized zero modes, extended domain-wall states, and a divergent zero-energy density of states. In interacting systems, quadratic approximations may fail and free-fermion topological classifications may be reduced.

The central criterion remains the combination of a BdG particle–hole structure, a nontrivial bulk invariant, a bulk gap or controlled nodal topology, and a boundary or defect supporting the corresponding self-conjugate excitation. A local zero-bias peak alone does not establish that criterion. Definitive identification requires correlated, nonlocal, topological, or non-Abelian evidence that excludes ordinary subgap mechanisms.

Topic to Video (Beta)

No one has generated a video about this topic yet.

Whiteboard

No one has generated a whiteboard explanation for this topic yet.

Follow Topic

Get notified by email when new papers are published related to Majorana Fermions.