Three-Channel Kondo Model

• The Three-Channel Kondo Model is a quantum impurity system where a localized spin couples antiferromagnetically to three independent conduction channels.
• It displays diverse low-energy behaviors including overscreened non-Fermi-liquid fixed points and two-stage or fully screened Fermi-liquid regimes based on symmetry and impurity spin.
• Realizations span FePc/Au(111), Floquet-engineered systems, and Y-junction spin chains, offering tunable quantum phase transitions and distinctive transport signatures.

The three-channel Kondo model denotes a class of quantum-impurity problems in which a localized spin or pseudospin couples antiferromagnetically to three independent electronic channels. Its infrared behavior is not unique: in the channel-symmetric spin-$\tfrac12$ case, the impurity is overscreened and the flow is to a non-Fermi-liquid fixed point, while in other three-channel settings the same channel multiplicity can produce two-stage screening of an $S=1$ moment or a fully screened Fermi liquid. The topic therefore encompasses a family of Hamiltonians, fixed points, and realizations rather than a single universal phenomenology (Zheng et al., 2022, Fernández et al., 2018, Hanl et al., 2014).

1. Canonical definition and screening classes

A standard formulation of the three-channel Kondo model takes a single spin-$\tfrac12$ impurity $\mathbf S_0$ coupled to three independent conduction channels, represented in one numerical implementation by three tight-binding chains labeled top, middle, and bottom: $$H_{3{\rm CK}} \;=\; \sum_{a\in\{t,m,b\}} H_a \;+\; H_{\rm int},$$

$$H_a \;=\; -\,t \sum_{\langle i,j\rangle,\sigma} \bigl(c_{a\,i\sigma}^\dagger\,c_{a\,j\sigma} + \mathrm{h.c.}\bigr), \qquad H_{\rm int} \;=\; J_t\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{t1} + J_m\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{m1} + J_b\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{b1}.$$

At perfect symmetry, $J_t=J_m=J_b$, the impurity is overscreened and the system flows to a non-Fermi-liquid fixed point; breaking the channel symmetry drives a zero-temperature impurity quantum phase transition to a Fermi-liquid phase in which only a single channel screens the impurity (Zheng et al., 2022).

The same three-channel label also appears in models with larger local moments. In the FePc/Au(111) problem, the impurity supports a net $S=1$ moment formed by one hole in $z^2$ and one in a $\pi$ orbital, and the low-energy description contains one $S=1$0 channel and two degenerate $S=1$1 channels. There the Kondo stage takes place in two stages, with different energy scales $S=1$2, rather than at a single overscreened fixed point (Fernández et al., 2018).

By contrast, the fully screened three-channel Kondo model has a Fermi-liquid low-energy expansion. For $S=1$3, the equilibrium Fermi-liquid coefficients are

$S=1$4

characterizing the $S=1$5, $S=1$6, and $S=1$7 corrections to the resonance and resistivity (Hanl et al., 2014). This juxtaposition shows that the infrared regime depends on impurity spin, channel equivalence, and the microscopic route by which the three-channel structure is generated.

2. Microscopic Hamiltonians and effective Kondo descriptions

One important route to three-channel physics starts from multiorbital Anderson impurities. For FePc on Au(111), the impurity Hamiltonian contains the $S=1$8 orbital and two degenerate $S=1$9 orbitals,

$\tfrac12$0

with conduction continua $\tfrac12$1, local interactions $\tfrac12$2, $\tfrac12$3, and $\tfrac12$4, and orbital-resolved hybridizations $\tfrac12$5. In the deep-Kondo regime, a Schrieffer–Wolff transformation yields

$\tfrac12$6

with two independent exchange constants $\tfrac12$7 and $\tfrac12$8 (Fernández et al., 2018).

A different microscopic origin arises in Floquet engineering. Starting from an infinite-$\tfrac12$9 Anderson model with a single impurity level, periodic driving produces symmetry-distinct Floquet sidebands. Under unpolarized light, the relevant irreducible representations are $\mathbf S_0$0, $\mathbf S_0$1, and the twofold-degenerate $\mathbf S_0$2, leading to an effective three-channel Kondo Hamiltonian

$\mathbf S_0$3

with the three-channel degeneracy condition

$\mathbf S_0$4

In this construction, the channel multiplicity is generated dynamically by the drive rather than by three pre-existing metallic bands (Quito et al., 2021).

Spin-chain constructions supply yet another mapping. In the image-impurity boundary condition geometry, $\mathbf S_0$5 identical spin-$\mathbf S_0$6 chains couple at one end to a physical impurity and at the other to an image impurity. After continuum reduction to $\mathbf S_0$7 WZW currents and folding, one obtains the standard overscreened $\mathbf S_0$8-channel Kondo Hamiltonian. For $\mathbf S_0$9, this realizes the three-channel non-Fermi-liquid fixed point in a purely spin-chain setting (Gaines et al., 3 Jun 2025).

A topological-Kondo realization is obtained at a Y junction of three inhomogeneous spin-$$H_{3{\rm CK}} \;=\; \sum_{a\in\{t,m,b\}} H_a \;+\; H_{\rm int},$$0 chains. There the low-energy boundary Hamiltonian,

$$H_{3{\rm CK}} \;=\; \sum_{a\in\{t,m,b\}} H_a \;+\; H_{\rm int},$$1

maps exactly onto an anisotropic three-channel Kondo Hamiltonian with bare couplings

$$H_{3{\rm CK}} \;=\; \sum_{a\in\{t,m,b\}} H_a \;+\; H_{\rm int},$$2

In this realization, the tilting angles $$H_{3{\rm CK}} \;=\; \sum_{a\in\{t,m,b\}} H_a \;+\; H_{\rm int},$$3 act as direct control parameters for channel anisotropy (Giuliano et al., 2020).

3. Fixed points, entropy, and low-energy scaling

For overscreened three-channel Kondo criticality, the Affleck–Ludwig boundary entropy takes the form

$$H_{3{\rm CK}} \;=\; \sum_{a\in\{t,m,b\}} H_a \;+\; H_{\rm int},$$4

which for $$H_{3{\rm CK}} \;=\; \sum_{a\in\{t,m,b\}} H_a \;+\; H_{\rm int},$$5 becomes

$$H_{3{\rm CK}} \;=\; \sum_{a\in\{t,m,b\}} H_a \;+\; H_{\rm int},$$6

This result is obtained explicitly in the spin-chain image-impurity construction and is also the residual entropy reported for the Floquet-engineered three-channel point (Gaines et al., 3 Jun 2025, Quito et al., 2021).

The same golden-ratio entropy appears in the Ho-ion realization, but with a different microscopic origin. In the seven-orbital Anderson model for Ho$$H_{3{\rm CK}} \;=\; \sum_{a\in\{t,m,b\}} H_a \;+\; H_{\rm int},$$7, the local $$H_{3{\rm CK}} \;=\; \sum_{a\in\{t,m,b\}} H_a \;+\; H_{\rm int},$$8 triplet behaves as an effective $$H_{3{\rm CK}} \;=\; \sum_{a\in\{t,m,b\}} H_a \;+\; H_{\rm int},$$9, and for $$H_a \;=\; -\,t \sum_{\langle i,j\rangle,\sigma} \bigl(c_{a\,i\sigma}^\dagger\,c_{a\,j\sigma} + \mathrm{h.c.}\bigr), \qquad H_{\rm int} \;=\; J_t\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{t1} + J_m\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{m1} + J_b\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{b1}.$$0 channels the Affleck–Ludwig formula gives

$$H_a \;=\; -\,t \sum_{\langle i,j\rangle,\sigma} \bigl(c_{a\,i\sigma}^\dagger\,c_{a\,j\sigma} + \mathrm{h.c.}\bigr), \qquad H_{\rm int} \;=\; J_t\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{t1} + J_m\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{m1} + J_b\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{b1}.$$1

Numerical renormalization-group flows identify a three-channel Kondo phase with $$H_a \;=\; -\,t \sum_{\langle i,j\rangle,\sigma} \bigl(c_{a\,i\sigma}^\dagger\,c_{a\,j\sigma} + \mathrm{h.c.}\bigr), \qquad H_{\rm int} \;=\; J_t\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{t1} + J_m\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{m1} + J_b\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{b1}.$$2 over a relatively wide region of the $$H_a \;=\; -\,t \sum_{\langle i,j\rangle,\sigma} \bigl(c_{a\,i\sigma}^\dagger\,c_{a\,j\sigma} + \mathrm{h.c.}\bigr), \qquad H_{\rm int} \;=\; J_t\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{t1} + J_m\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{m1} + J_b\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{b1}.$$3 plane (Hotta, 3 Sep 2025).

The leading boundary scaling also has a characteristic three-channel value. In the high-transparency three-channel charge-Kondo problem, the leading irrelevant operator has scaling dimension

$$H_a \;=\; -\,t \sum_{\langle i,j\rangle,\sigma} \bigl(c_{a\,i\sigma}^\dagger\,c_{a\,j\sigma} + \mathrm{h.c.}\bigr), \qquad H_{\rm int} \;=\; J_t\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{t1} + J_m\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{m1} + J_b\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{b1}.$$4

and the low-energy conductance approaches the fixed-point value with exponent $$H_a \;=\; -\,t \sum_{\langle i,j\rangle,\sigma} \bigl(c_{a\,i\sigma}^\dagger\,c_{a\,j\sigma} + \mathrm{h.c.}\bigr), \qquad H_{\rm int} \;=\; J_t\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{t1} + J_m\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{m1} + J_b\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{b1}.$$5. The same work reports the exact zero-frequency conductance

$$H_a \;=\; -\,t \sum_{\langle i,j\rangle,\sigma} \bigl(c_{a\,i\sigma}^\dagger\,c_{a\,j\sigma} + \mathrm{h.c.}\bigr), \qquad H_{\rm int} \;=\; J_t\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{t1} + J_m\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{m1} + J_b\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{b1}.$$6

together with a residual boundary entropy difference

$$H_a \;=\; -\,t \sum_{\langle i,j\rangle,\sigma} \bigl(c_{a\,i\sigma}^\dagger\,c_{a\,j\sigma} + \mathrm{h.c.}\bigr), \qquad H_{\rm int} \;=\; J_t\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{t1} + J_m\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{m1} + J_b\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{b1}.$$7

The FRG values $$H_a \;=\; -\,t \sum_{\langle i,j\rangle,\sigma} \bigl(c_{a\,i\sigma}^\dagger\,c_{a\,j\sigma} + \mathrm{h.c.}\bigr), \qquad H_{\rm int} \;=\; J_t\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{t1} + J_m\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{m1} + J_b\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{b1}.$$8, $$H_a \;=\; -\,t \sum_{\langle i,j\rangle,\sigma} \bigl(c_{a\,i\sigma}^\dagger\,c_{a\,j\sigma} + \mathrm{h.c.}\bigr), \qquad H_{\rm int} \;=\; J_t\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{t1} + J_m\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{m1} + J_b\,\mathbf{S}_0\!\cdot\!\mathbf{s}_{b1}.$$9, and $J_t=J_m=J_b$0 reproduce these benchmarks within a few percent (Paris et al., 3 Sep 2025).

Not all three-channel models flow to this non-Fermi-liquid point. In FePc/Au(111), poor-man’s scaling and slave-boson mean-field theory yield two distinct Kondo scales,

$J_t=J_m=J_b$1

and the low-temperature state is described within SBMFA as a Fermi liquid with two-stage screening rather than overscreening (Fernández et al., 2018).

4. Channel asymmetry, impurity quantum phase transitions, and order parameters

A central issue in the three-channel Kondo problem is the fate of the channel-symmetric non-Fermi-liquid state under asymmetry. In the chain representation with

$J_t=J_m=J_b$2

the parameter $J_t=J_m=J_b$3 gives exact three-channel symmetry, $J_t=J_m=J_b$4 favors a one-channel Fermi liquid, and $J_t=J_m=J_b$5 favors a two-channel overscreened phase. The transition at $J_t=J_m=J_b$6 was characterized by the spin-correlation ratio

$J_t=J_m=J_b$7

where $J_t=J_m=J_b$8 is the total impurity–electron spin correlation in the bottom channel and $J_t=J_m=J_b$9 is the sum over all channels. At symmetry, $S=1$0; in the Fermi-liquid phase, $S=1$1; and for $S=1$2, $S=1$3 (Zheng et al., 2022).

Finite-size scaling near the critical point uses

$S=1$4

and yields

$S=1$5

for the three-channel model. The reported $S=1$6 matches the conformal-field-theory prediction $S=1$7 with $S=1$8 for $S=1$9, linking the finite-size numerics directly to the boundary critical theory (Zheng et al., 2022).

A distinct 3CK–Fermi-liquid boundary appears in Ho-based multiorbital Anderson models. For the simplified $z^2$0 setting, tuning either the crystalline-electric-field mixing parameter $z^2$1 or the hybridization $z^2$2 produces critical values

$z^2$3

where the system leaves the $z^2$4-triplet three-channel Kondo regime and enters a local-singlet Fermi liquid. The same study interpreted the intermediate plateau at the boundary as a four-channel Kondo critical point of a spin-1 impurity (Hotta, 2021).

The high-transparency charge-Kondo formulation extends this theme further by finding, for interacting leads, a continuous line of nontrivial intermediate fixed points for $z^2$5, separating ballistic and insulating regimes. This broadens the notion of “three-channel criticality” from an isolated fixed point to a family of interaction-controlled boundary phases (Paris et al., 3 Sep 2025).

5. Physical realizations and engineered platforms

The three-channel Kondo model appears in several experimentally or numerically accessible settings. The table summarizes representative realizations documented in the recent literature.

Realization	Microscopic degrees of freedom	Reported hallmark
FePc on Au(111)	$z^2$6 and two degenerate $z^2$7 orbitals	Two-stage screening with $z^2$8
Ho$z^2$9 impurity Anderson model	$\pi$0 and $\pi$1 conduction channels, local $\pi$2 triplet	$\pi$3
Spin-chain IIBC	$\pi$4 critical Heisenberg chains plus impurity/image spins	$\pi$5
Floquet-engineered Kondo model	Unpolarized-light-generated $\pi$6, $\pi$7, and $\pi$8 channels	Three-channel degeneracy tuned by $\pi$9
Y junction of spin chains	Majorana zero modes and chiral fields at a junction	Tunable $S=1$00 via $S=1$01
Three-channel charge-Kondo circuit	Quantum island coupled to three QPCs or edge channels	Universal $S=1$02

In FePc/Au(111), the “on-top” adsorption geometry provides the orbital structure required for one $S=1$03 channel and two degenerate $S=1$04 channels, and the STM spectrum between $S=1$05 mV and $S=1$06 mV is reproduced by a slave-boson mean-field solution of the corresponding Anderson model (Fernández et al., 2018).

In Ho-based realizations, the seven-orbital Anderson description with ten local $S=1$07 electrons supports a $S=1$08 triplet ground multiplet under cubic crystalline electric field, and the three-channel Kondo phase occupies a relatively wide region in the $S=1$09 plane. The same phase diagram also contains adjacent Fermi-liquid regimes and an unexpected two-channel Kondo pocket (Hotta, 3 Sep 2025).

Engineered platforms broaden the range of accessible control parameters. In Floquet engineering, light polarization, frequency, and amplitude tune the channel content, and unpolarized light is particularly useful for inducing three-channel degeneracies. In the Y-junction topological-Kondo realization, the bare couplings are explicit functions of $S=1$10, so the Kondo screening length can be tuned by phase control alone. In spin chains with image-impurity boundary conditions, the three-channel fixed point is realized without itinerant electrons at all, using a continuum map to the multichannel Kondo boundary problem (Quito et al., 2021, Giuliano et al., 2020, Gaines et al., 3 Jun 2025).

6. Spectroscopy, transport, and recurrent points of confusion

In FePc/Au(111), the differential conductance is modeled as a superposition of two Fano resonances,

$S=1$11

with widths

$S=1$12

The broad asymmetric Fano–Kondo peak at $S=1$13 meV is attributed to the $S=1$14 channel, while the narrow antiresonance centered near zero bias arises from the two $S=1$15 channels. The same analysis emphasizes strong interference between channels, with each Kondo scale suppressed by the presence of the others (Fernández et al., 2018).

Charge-Kondo transport yields a complementary diagnostic set. In the high-transparency regime, the linear conductance and impurity entropy both exhibit single-parameter scaling in $S=1$16 or $S=1$17, with

$S=1$18

The zero-temperature conductance approaches

$S=1$19

and the low-frequency correction is controlled by the exponent $S=1$20. For the three-QPC thermoelectric circuit, the Seebeck coefficient was found to scale as

$S=1$21

providing a transport probe of the non-Fermi-liquid regime (Paris et al., 3 Sep 2025, Nguyen et al., 2019).

In the Y-junction realization, the screening length $S=1$22 is extracted from the crossover of equilibrium spin currents. In the fully isotropic case,

$S=1$23

while for anisotropic bare couplings an explicit logarithmic formula replaces the isotropic form. As the chain length $S=1$24 crosses $S=1$25, the current shows an upturn from the $S=1$26 decay of the uncoupled regime into a plateau or crossover characteristic of strong coupling (Giuliano et al., 2020).

A recurrent source of confusion is the entropy quoted for “the” three-channel Kondo state. Spin-impurity, spin-chain, Floquet, and Ho-ion formulations report the Affleck–Ludwig value $S=1$27. The high-transparency charge-Kondo analysis instead reports a boundary entropy difference $S=1$28, while the bosonized three-QPC thermoelectric treatment associates the strong-coupling state with $S=1$29 in its boundary-CFT description. These results refer to different objects and conventions, so identical numerical values are not expected across all realizations (Gaines et al., 3 Jun 2025, Paris et al., 3 Sep 2025, Nguyen et al., 2019).