Frustrated Josephson Tunneling

• Frustrated Josephson tunneling is defined by conflicting phase minimization in superconductors, leading to emergent BKT criticality and a U(1)×Z₂ symmetry structure.
• It arises from diverse mechanisms including magnetic-flux frustration, alternating-sign couplings, and topological constraints in various lattice geometries.
• Experimental probes reveal unique signatures such as linewidth anomalies, chiral edge currents, and vortex dynamics that inform future quantum device applications.

Frustrated Josephson tunneling denotes Josephson transport in which the superconducting phases, or more generally the internal order-parameter variables that enter the Josephson energy, cannot simultaneously minimize all coupling terms. In the literature this occurs in several distinct but structurally related settings: magnetic-flux frustration in Josephson junction arrays (JJAs), sign competition between $0$- and $\pi$-junction couplings, fluxoid-quantization constraints in loop geometries, positive interband Josephson couplings in multiband superconductors, and anisotropic couplings between triplet $d$-vectors. The resulting phenomenology includes Berezinskii–Kosterlitz–Thouless (BKT) criticality, emergent $\mathrm U(1)\times \mathbb Z_2$ structure, flat-band crossovers, $\varphi$-contacts, spontaneous currents, edge-current phases, pinning–depinning critical currents, irrational-flux superconducting dips, and nonintegral flux trapping (Cosmic et al., 2020, Monaco et al., 2011, Bojesen et al., 2014, Etter et al., 2014, Frazier et al., 27 Apr 2026).

1. Microscopic origin of frustration

In a JJA, frustration enters directly through the gauge-invariant phase difference. The phase-only Hamiltonian is

$$H_{\rm JJA} = \frac{(2e)^2}{2} \sum_{\langle i,j\rangle} n_i C^{-1}_{ij} n_j - E_J\sum_{\langle i,j\rangle} \cos(\phi_i-\phi_j-A_{ij}),$$

with

$$A_{ij} = \frac{2\pi}{\Phi_0}\int_i^j \vec A\cdot d\vec l,$$

so magnetic flux through a plaquette shifts the preferred bond phases. In the classical regime this maps to an XY model with $\vec S_i=(\cos\phi_i,\sin\phi_i)$, making the Josephson phase the XY spin angle and the flux the source of frustration (Cosmic et al., 2020).

A second mechanism is alternating-sign Josephson coupling. In one-dimensional diamond and sawtooth arrays, the couplings $\alpha_{ij}$ alternate in sign and are parametrized by $\alpha=1-2f$. The frustration parameter $\pi$0 controls a crossover from non-frustrated to frustrated regimes, with the lowest linear band becoming completely flat at a critical $\pi$1; in the diamond chain $\pi$2, and in the sawtooth chain $\pi$3 (Andreanov et al., 2018). Closely related sign competition appears in quasi-one-dimensional ladders with $\pi$4 inter-rung couplings and frustrating $\pi$5 intra-rung couplings, where nontrivial phase differences produce chiral edge currents (Marques et al., 2016).

A third mechanism is topological rather than local. In long Josephson tunnel junctions with doubly connected electrodes, fluxoid quantization imposes a global constraint on the phase profile, so the gauge-invariant phase difference is restricted not only by the sine-Gordon equation and boundary conditions but also by loop topology. The paper identifies this as a phase-frustration constraint and attributes the reduction of critical current to the impossibility of satisfying the local static equations, the fluxoid constraint, and the minimum-energy condition simultaneously (Monaco et al., 2011). An analogous topological incompatibility appears in Pb/Ru/Sr$\pi$6RuO$\pi$7 junctions when the bulk state of Sr$\pi$8RuO$\pi$9 is taken to be chiral $d$0-wave: the coupling term $d$1 favors a winding phase for $d$2, while the gradient term favors $d$3 constant, creating a topologically frustrated junction (Etter et al., 2014).

Internal order-parameter structure can itself be the source of frustration. In multiband London models with positive interband Josephson couplings, each Josephson term prefers a phase difference of $d$4, but with three components the pairwise preferences cannot all be satisfied. The ground state therefore acquires two opposite chiralities and an effective $d$5 symmetry structure (Bojesen et al., 2014). In triplet systems, the Josephson energy depends not only on the gauge phase difference but also on the relative orientation of the $d$6-vectors. The effective bond energy can be written as

$d$7

where $d$8 is an emergent geometric phase generated by the internal spin texture of the Cooper pair; nonzero loop sums of $d$9 create frustrated circulating-current states (Frazier et al., 27 Apr 2026). Related work derives Heisenberg-like, Dzyaloshinskii–Moriya-like, and $\mathrm U(1)\times \mathbb Z_2$0-type anisotropic couplings between $\mathrm U(1)\times \mathbb Z_2$1-vectors from frustrated spin textures in the barrier, again turning Josephson tunneling into a competition among incompatible preferred configurations (Frazier et al., 18 Jun 2025).

2. Lattice realizations and effective models

The canonical two-dimensional realization is the square-lattice JJA implementing the XY model. At zero flux the couplings are unfrustrated, $\mathrm U(1)\times \mathbb Z_2$2, while at half flux per plaquette the fully frustrated XY model has $\mathrm U(1)\times \mathbb Z_2$3 with one effectively antiferromagnetic bond per plaquette and a doubly degenerate ground state apart from global $\mathrm U(1)\times \mathbb Z_2$4 rotation (Cosmic et al., 2020). This realization directly connects frustrated Josephson tunneling to the longstanding problem of coupled $\mathrm U(1)\times \mathbb Z_2$5 and $\mathrm U(1)\times \mathbb Z_2$6 ordering.

One-dimensional geometries display a different but related structure. In diamond and sawtooth chains the linearized Josephson dynamics has multiband dispersions $\mathrm U(1)\times \mathbb Z_2$7, and the lowest band becomes flat precisely at the crossover into the frustrated regime. In the same parameter range the ordinary phase correlator $\mathrm U(1)\times \mathbb Z_2$8 crosses from long-range to short-range behavior, while higher harmonics such as $\mathrm U(1)\times \mathbb Z_2$9 in the diamond chain and $\varphi$0 in the sawtooth chain recover long-range order deep in the frustrated regime (Andreanov et al., 2018). This establishes a direct correspondence between spectral flattening and the loss of ordinary phase coherence.

Open-boundary ladders introduce boundary-specific frustrated states. In a quasi-one-dimensional Josephson ladder with diagonal couplings, the Josephson energy maps to the expectation value of a two-band tight-binding Hamiltonian subject to a uniform-density constraint. Above $\varphi$1, where $\varphi$2 is the ratio of intra-rung to inter-rung coupling strength in the mapped model, circulating currents nucleate at the edges and then propagate inward as $\varphi$3 increases; for an infinite ladder the edge-current phase disappears at $\varphi$4, coincident with the opening of a gap between the flat and dispersive bands (Marques et al., 2016).

More constrained frustrated networks generate effective Ising descriptions. In the vertex-sharing Kagome lattice with periodically arranged $\varphi$5- and $\varphi$6-junctions, each frustrated triangle supports either a vortex or an antivortex, so the low-energy sector is represented by an Ising variable $\varphi$7. Flux quantization around the hexagons then produces highly anisotropic long-range interactions between these effective spins (Neyenhuys et al., 2023). In frustrated sawtooth arrays of small quantum junctions, the frustrated regime $\varphi$8 produces $\varphi$9 degenerate classical minima and low-energy vortex/antivortex doublets per triangle, again motivating effective spin models for the coherent regime (Fistul et al., 2024). A related transmission-line embedding yields an effective long-range $$H_{\rm JJA} = \frac{(2e)^2}{2} \sum_{\langle i,j\rangle} n_i C^{-1}_{ij} n_j - E_J\sum_{\langle i,j\rangle} \cos(\phi_i-\phi_j-A_{ij}),$$0 chain,

$$H_{\rm JJA} = \frac{(2e)^2}{2} \sum_{\langle i,j\rangle} n_i C^{-1}_{ij} n_j - E_J\sum_{\langle i,j\rangle} \cos(\phi_i-\phi_j-A_{ij}),$$1

with $$H_{\rm JJA} = \frac{(2e)^2}{2} \sum_{\langle i,j\rangle} n_i C^{-1}_{ij} n_j - E_J\sum_{\langle i,j\rangle} \cos(\phi_i-\phi_j-A_{ij}),$$2 at long distance (Pernack et al., 2024).

Aperiodicity offers a further variant. In a one-dimensional ladder with two incommensurate plaquette areas, the ground-state energy $$H_{\rm JJA} = \frac{(2e)^2}{2} \sum_{\langle i,j\rangle} n_i C^{-1}_{ij} n_j - E_J\sum_{\langle i,j\rangle} \cos(\phi_i-\phi_j-A_{ij}),$$3 becomes quasiperiodic in frustration and its deep minima are well captured by an independent-plaquette model. The dominant frequencies in the power spectrum occur at $$H_{\rm JJA} = \frac{(2e)^2}{2} \sum_{\langle i,j\rangle} n_i C^{-1}_{ij} n_j - E_J\sum_{\langle i,j\rangle} \cos(\phi_i-\phi_j-A_{ij}),$$4, $$H_{\rm JJA} = \frac{(2e)^2}{2} \sum_{\langle i,j\rangle} n_i C^{-1}_{ij} n_j - E_J\sum_{\langle i,j\rangle} \cos(\phi_i-\phi_j-A_{ij}),$$5, and $$H_{\rm JJA} = \frac{(2e)^2}{2} \sum_{\langle i,j\rangle} n_i C^{-1}_{ij} n_j - E_J\sum_{\langle i,j\rangle} \cos(\phi_i-\phi_j-A_{ij}),$$6, where $$H_{\rm JJA} = \frac{(2e)^2}{2} \sum_{\langle i,j\rangle} n_i C^{-1}_{ij} n_j - E_J\sum_{\langle i,j\rangle} \cos(\phi_i-\phi_j-A_{ij}),$$7 is the area ratio (Azizi et al., 2011).

3. Ordered states, degeneracy, and phase transitions

In the unfrustrated square JJA, the finite-temperature transition is the standard BKT transition driven by vortex–antivortex unbinding. In the circuit-QED realization, the transition occurs near $$H_{\rm JJA} = \frac{(2e)^2}{2} \sum_{\langle i,j\rangle} n_i C^{-1}_{ij} n_j - E_J\sum_{\langle i,j\rangle} \cos(\phi_i-\phi_j-A_{ij}),$$8, and the correlation-function exponent reaches the BKT value $$H_{\rm JJA} = \frac{(2e)^2}{2} \sum_{\langle i,j\rangle} n_i C^{-1}_{ij} n_j - E_J\sum_{\langle i,j\rangle} \cos(\phi_i-\phi_j-A_{ij}),$$9 at the transition (Cosmic et al., 2020). At full frustration $$A_{ij} = \frac{2\pi}{\Phi_0}\int_i^j \vec A\cdot d\vec l,$$0, the same platform identifies a transition near $$A_{ij} = \frac{2\pi}{\Phi_0}\int_i^j \vec A\cdot d\vec l,$$1, consistent with prior numerical studies for the fully frustrated XY model. The same work emphasizes that the exact relationship between the $$A_{ij} = \frac{2\pi}{\Phi_0}\int_i^j \vec A\cdot d\vec l,$$2 and $$A_{ij} = \frac{2\pi}{\Phi_0}\int_i^j \vec A\cdot d\vec l,$$3 transitions remains debated.

The distinction between ordinary phase order and hidden higher-harmonic order is central in one-dimensional frustrated arrays. For the diamond and sawtooth chains, low-temperature ordinary correlations decay slowly for $$A_{ij} = \frac{2\pi}{\Phi_0}\int_i^j \vec A\cdot d\vec l,$$4, but for $$A_{ij} = \frac{2\pi}{\Phi_0}\int_i^j \vec A\cdot d\vec l,$$5 the factor $$A_{ij} = \frac{2\pi}{\Phi_0}\int_i^j \vec A\cdot d\vec l,$$6 suppresses or oscillates the correlators and drives $$A_{ij} = \frac{2\pi}{\Phi_0}\int_i^j \vec A\cdot d\vec l,$$7 short-ranged. Near $$A_{ij} = \frac{2\pi}{\Phi_0}\int_i^j \vec A\cdot d\vec l,$$8, however, $$A_{ij} = \frac{2\pi}{\Phi_0}\int_i^j \vec A\cdot d\vec l,$$9 in the diamond chain and $\vec S_i=(\cos\phi_i,\sin\phi_i)$0 in the sawtooth chain regain long-range behavior. The paper interprets the diamond-chain result as a $\vec S_i=(\cos\phi_i,\sin\phi_i)$1-to-$\vec S_i=(\cos\phi_i,\sin\phi_i)$2 type effect: ordinary phase coherence is short-ranged while pair-like coherence persists (Andreanov et al., 2018). A common misconception is therefore that frustration simply destroys order; in these arrays it can instead transfer coherence into higher harmonics.

Checkerboard frustration from four-terminal junctions modifies the usual commensuration logic. In the Nb-Pt-Nb four-terminal array, alternating plaquettes carry frustrations $\vec S_i=(\cos\phi_i,\sin\phi_i)$3 and $\vec S_i=(\cos\phi_i,\sin\phi_i)$4, and superconducting resistance dips occur when $\vec S_i=(\cos\phi_i,\sin\phi_i)$5, equivalently $\vec S_i=(\cos\phi_i,\sin\phi_i)$6. Because $\vec S_i=(\cos\phi_i,\sin\phi_i)$7 with $\vec S_i=(\cos\phi_i,\sin\phi_i)$8, the dip positions

$\vec S_i=(\cos\phi_i,\sin\phi_i)$9

are generally irrational in the single-plaquette frustration variable. The paper interprets this as stabilization of the BKT superconducting phase even when the flux per individual plaquette is irrational (Teller et al., 18 Mar 2025).

Quantum frustrated arrays exhibit further ordered phases. In the sawtooth arrays of small junctions, the non-frustrated regime supports disordered (insulating) and ordered (superconducting) phases controlled by the competition between Josephson coupling and charging energies, while the frustrated regime supports a spatially disordered vortex–antivortex superposition phase and, for long arrays with strong effective inter-cell coupling, a stripe phase of vortices and antivortices (Fistul et al., 2024). In the multiband $\alpha_{ij}$0 problem, the field couples mainly to the $\alpha_{ij}$1 sector, so vortex-lattice melting and chirality restoration need not coincide; the resulting chiral metallic phase is a resistive vortex liquid that still breaks time-reversal symmetry (Bojesen et al., 2014).

4. Dynamical response and nonequilibrium phase evolution

A major development has been the use of weak microwave spectroscopy to detect frustrated Josephson criticality without driving the array far from equilibrium. In the $\alpha_{ij}$2 plaquette JJA inside a three-dimensional cavity, the measured observable is the complex reflection coefficient $\alpha_{ij}$3, whose linewidth is governed by the dissipative susceptibility $\alpha_{ij}$4. The effective total linewidth is

$\alpha_{ij}$5

As temperature increases, the linewidth broadens, peaks near the transition, and narrows again. The paper interprets this nonmonotonic behavior as a signature of vortex dynamics: bound pairs at low temperature, strong dissipation from unbinding near the transition, and screened free-vortex response above it (Cosmic et al., 2020). The same work states explicitly that this probe is primarily sensitive to the dynamics of the $\alpha_{ij}$6 part, not directly to the $\alpha_{ij}$7 chirality sector.

Driven frustrated junctions can generate oscillatory phase dynamics even without harmonic forcing. For a multichannel junction between an $\alpha_{ij}$8-wave superconductor and a three-band superconductor, the energy

$\alpha_{ij}$9

has fourteen candidate ground states, including trivial $\alpha=1-2f$0 states and nontrivial $\alpha=1-2f$1 states. When the couplings are driven across boundaries in parameter space, the phase variables begin to oscillate and the voltage oscillates accordingly. The paper distinguishes coherent few-harmonic spectra from broader multi-harmonic spectra, depending on whether the trajectory settles into one minimum or moves among several nearby frustrated minima (Guarcello et al., 2022).

In transmission-line-embedded frustrated sawtooth chains, each triangular cell is a two-level system consisting of clockwise and anticlockwise persistent-current states. Coherent tunneling between them is described by the local $\alpha=1-2f$2 term, while the transmission line induces nonlocal exchange couplings that can dominate when the array length exceeds the screening scale $\alpha=1-2f$3. Exact diagonalization identifies a local-tunneling-dominated paramagnet, a compressible superfluid, and a weakly compressible superfluid (Pernack et al., 2024).

Constraint-induced quantum dynamics also appears in the Kagome array. There the quantum regime reduces to a transverse-field Ising model in which $\alpha=1-2f$4 distinguishes vortex from antivortex and $\alpha=1-2f$5 generates macroscopic tunneling between them. In the strong-interaction limit the classical degeneracy is lifted and the ground state becomes an entangled superposition of classical vortex/antivortex patterns (Neyenhuys et al., 2023).

5. Multiband, topological, and triplet frustrated junctions

Multiband point contacts realize frustrated tunneling through interference among channel currents. In the dirty point contact between a single-band superconductor and a three-band superconductor, the total Josephson current is the sum of three band-resolved contributions with phase shifts $\alpha=1-2f$6, $\alpha=1-2f$7, and $\alpha=1-2f$8. When the three-band bank has broken time-reversal symmetry, the bulk equilibrium phases $\alpha=1-2f$9 are nontrivial and occur in conjugate pairs, and the point contact can become a $\pi$00-contact with energy minima at finite phase shift rather than at $\pi$01 or $\pi$02 (Yerin et al., 2014). Near $\pi$03 the current-phase relation becomes close to sinusoidal, but the phase offsets remain controlled by the internal multiband state.

A related but more explicitly frustrated formulation treats positive interband Josephson couplings in a three-component London model. Because the field couples only to the overall $\pi$04 phase while the chiral $\pi$05 sector remains indirect, one may observe a vortex lattice or vortex liquid together with a separate Ising-like chirality transition (Bojesen et al., 2014). This separates superconducting phase rigidity from time-reversal breaking.

Topological frustration in extended junctions produces a different critical-current mechanism. In the cylindrical Pb/Ru/Sr$\pi$06RuO$\pi$07 geometry, the chiral phase of Sr$\pi$08RuO$\pi$09 generates a spontaneous flux pattern on the interface. In a perfectly rotationally symmetric frustrated junction, any finite applied current shifts this flux pattern, so no static dc supercurrent can be sustained. When angular inhomogeneity pins the spontaneous flux line, a finite critical current reappears, but the limiting mechanism is then pinning–depinning of spontaneous flux rather than conventional vortex entry (Etter et al., 2014).

Triplet Josephson networks extend frustration to the internal spin geometry of Cooper pairs. For unitary triplet pairing, anisotropic Josephson coupling can be written in terms of Heisenberg-like, antisymmetric Dzyaloshinskii–Moriya-like, and $\pi$10-type terms coupling neighboring $\pi$11-vectors. In a three-grain ring this yields a critical antisymmetric coupling $\pi$12, above which the minimum-energy texture becomes a chiral $\pi$13 configuration and can trap $\pi$14 for the appropriate sign of $\pi$15 (Frazier et al., 27 Apr 2026). More generally, the accumulated geometric phase need not be quantized, so the trapped flux can be a nonintegral multiple of $\pi$16. Parallel work derives the anisotropic $\pi$17-vector couplings microscopically from frustrated noncollinear spin textures at the barrier and connects them to spatially modulated triplet pairing, anomalous vortices, and a Josephson diode effect (Frazier et al., 18 Jun 2025).

Atomic-scale frustrated coupling has also been reported in a conventional-to-multiband setting. In scanned Josephson tunneling microscopy on Nb/FeSe, the net Josephson current is a quantum-coherent sum of two tunneling channels into two FeSe condensates with opposite signs, so the junction approaches a $\pi$18-to-$\pi$19 transition as the relative transparency of the channels changes. The measured inequality

$\pi$20

is presented as a direct signature of destructive interference from opposite-sign channels, and the extracted condensate-resolved superfluid modulations are spatially anti-correlated with cross-correlation coefficient about $\pi$21 (Sharma et al., 22 Jul 2025).

6. Experimental signatures and conceptual significance

The experimental signatures of frustrated Josephson tunneling are unusually diverse because the frustration can reside in geometry, topology, band structure, or internal pairing degrees of freedom. In JJAs and XY simulators, the central observables are linewidth anomalies in microwave reflection, vortex-sensitive damping, and transition temperatures inferred from $\pi$22 (Cosmic et al., 2020). In four-terminal arrays, dc resistance oscillations show a beating envelope and superconducting dips at checkerboard cancellation points $\pi$23, permitting an estimate of the weak-link area through $\pi$24 and $\pi$25 (Teller et al., 18 Mar 2025). In long junctions on superconducting loops, discrete critical-current shifts track trapped fluxoids, while the offset current $\pi$26 improves sensitivity (Monaco et al., 2011).

In boundary-sensitive and topological settings, the critical current itself becomes diagnostic. The Pb/Ru/Sr$\pi$27RuO$\pi$28 analysis attributes the anomalous temperature dependence of $\pi$29 near the bulk $\pi$30 to a crossover from an unfrustrated topologically trivial regime to a topologically frustrated chiral regime (Etter et al., 2014). In edge-current ladders, the onset, propagation, and eventual disappearance of chiral edge currents track the change in the underlying two-band spectrum (Marques et al., 2016).

Phase-sensitive spectroscopy is especially powerful in multiband systems. In dirty point contacts to three-band superconductors, persistent $\pi$31-contact behavior from low temperature to near $\pi$32 is proposed as a diagnostic of broken time-reversal symmetry (Yerin et al., 2014). In atomic-scale Nb/FeSe junctions, sublinear $\pi$33 versus $\pi$34, violation of the ordinary multiband lower bound, and local reconstruction of band-resolved transparencies provide direct evidence for frustrated Josephson interference in an $\pi$35 superconductor (Sharma et al., 22 Jul 2025).

Across these realizations, one recurring theme is that frustration does not define a single phase but a constraint structure. It can generate BKT transitions, chirality order, flat bands, quasiperiodic energy landscapes, spontaneous supercurrents, entangled vortex–antivortex states, or nonintegral flux trapping, depending on which variables are overconstrained and which degrees of freedom remain soft. A second recurring theme is sector selectivity: some probes couple mainly to the global $\pi$36 dynamics, while others expose chirality, fluxoid number, channel interference, or $\pi$37-vector texture. This suggests that the principal open problems are not only to identify new frustrated Josephson platforms, but also to disentangle which broken symmetry or constrained variable is actually being measured in each experiment (Cosmic et al., 2020, Bojesen et al., 2014, Frazier et al., 27 Apr 2026).