betaFeedback

WorksPhysics and Astronomy

Majorana-like Josephson branch structure in one-dimensional transport of cold atoms

Abstract

We study phase-controlled transport through a one-dimensional channel of attractively interacting fermions connecting two three-dimensional BCS reservoirs. The competition between channel phase stiffness and pair tunnelling at the two contacts results in the emergence of two 4π4π-periodic families of stationary states, characterised by symmetric or asymmetric phase drops at the boundaries. Advancing the reservoir phase difference by 2π2π exchanges these families so that the equilibrium energy remains 2π2π periodic, while the 4π4π periodicity might be attained in a diabatic regime. This structure admits a formal description in terms of two boundary Majorana operators, combining into a nonlocal fermion mode. Renormalization-group analysis shows that quantum fluctuations and phase slips are suppressed in the zero-temperature thermodynamic limit, preserving the branch crossings. The predicted branch exchange, metastable currents, and sweep-rate-dependent switching provide experimentally testable signatures of Majorana-like Josephson branch physics in a gapless interacting channel.

One-dimensional (1D) quantum systems provide a setting in which interactions qualitatively change the nature of both equilibrium states and transport. Instead of the quasiparticles characteristic of a Fermi liquid, their low-energy excitations are collective density and phase modes described by Tomonaga–Luttinger liquid theory [1]. In particular, connecting macroscopic superconducting reservoirs through such a 1D channel has long been known to produce transport properties very different from those of a conventional Josephson junction.

For an interacting electronic wire coupled to two superconductors, the Josephson current is strongly affected by the Luttinger-liquid correlations and by the nature of the contacts [2, 3]. In such a case, perfect Andreev reflection imposes boundary conditions which pin the bosonic phase field to the phases of the adjoining superconductors. The corresponding problem is naturally formulated as a boundary sine-Gordon theory, resulting in a nonanalytic dependence of the critical Josephson current [2, 3].

Ultracold atomic gases offer an especially clean realization of such physics. Mesoscopic transport geometries in which two macroscopic reservoirs of ultracold fermions are connected by a narrow channel, quantum wire, or quantum point contact have been realized experimentally [4, 5, 6, 7, 8, 9, 10, 11, 12]. In particular, experiments with strongly interacting Fermi superfluids have demonstrated transport dominated by pairing and Andreev processes [8, 9].

In such experiments, interaction strength, geometry, and reservoir populations can all be controlled independently. Also, transport experiments in one-dimensional attractive Fermi gases have provided evidence [13] for the Luther–Emery regime [14] with gapped spin excitations and a gapless particle-number mode. In this regime enhanced singlet-pairing correlations result in a state which is the one-dimensional analogue of a superconductor, although true long-range order is destroyed by phase fluctuations. These developments make it possible to explore superconducting boundary phenomena and strongly cor-

related one-dimensional transport in a regime that is difficult to access in electronic systems.

In this work we consider attractive spin-1/2 fermions with Fermi momentum kFk_F propagating through a 1D channel of length LL connecting two 3D reservoirs containing fermionic superfluids in the BCS state. We assume that the stationary flow is driven entirely by an imposed phase difference Φ\Phi between the reservoirs [15], as illustrated in Figure 1(a). We show that in contrast to the repulsive case considered in Refs. [2, 3], the physics is governed by the nontrivial saddle-point configuration, which is determined by the competition between the stiffness of the 1D liquid and the boundary tunnelling amplitudes. The boundary renormalization group (RG) flow favors phase pinning at low energies, confirming robustness of this stationary-state structure.

One-dimensional channel and Josephson energy branches

Figure 1. (a) A one-dimensional channel of attractively interacting fermions connects two three-dimensional BCS reservoirs with phases Φ/2\Phi/2 and −Φ/2-\Phi/2. The junctions have tunnelling amplitude jj, and the contact labels γl\gamma_l and γr\gamma_r refer to the formal Majorana representation of the branch label discussed in the text. The arrow indicates the stationary pair superflow, I=∂E/∂ΦI=\partial E/\partial\Phi (in units with ℏ=1\hbar=1). (b) Josephson energy branches labelled by P+=+1\mathcal{P}_{+}=+1 (blue, S) and P−=−1\mathcal{P}_{-}=-1 (red, A), shown for W0=π/2W_0=\pi/2; see Eqs. (12) and (14). The family-resolved energy structure is 4π4\pi-periodic, whereas the equilibrium lower envelope is 2π2\pi-periodic.

The manifestation of this physics is a 4π4\pi-periodic Josephson response, analogous to systems containing spatially separated Majorana modes [16, 17]. This resemblance can be ascribed to the emergence of the nonlocal fermionic mode formed by a pair of local boundary Majorana operators γl,γr\gamma_l,\gamma_r at the left or right contact [18], similarly to the case of a topological superconducting junction [19]. The situation considered here is different in a crucial respect: the one-dimensional bulk is not a fully gapped mean-field topological superconductor. Its spin excitations are gapped, but its number sector remains a gapless Luttinger liquid [20].

The central result is the emergence of a two-family Josephson energy structure, as illustrated in Figure 1(b). In the symmetric (S) family, the boundary phase drops are equal on both contacts, while in the asymmetric one they differ by 2π2\pi. The Majorana operators form the nonlocal parity operator

P=iγlγr,(1)\mathcal{P}=i\gamma_l\gamma_r, \tag*{(1)}

with eigenvalues P±=±1\mathcal{P}_{\pm}=\pm1 distinguishing these two families which are interchanged by a 2π2\pi phase shift. In the thermodynamic limit, the nonlocal parity conservation prevents these branches from hybridizing at their crossing, resulting in quasidegenerate ground states for Φ=±π\Phi=\pm\pi and producing the fractional Josephson effect. Model and effective boundary action. The setup outlined above (Figure 1(a)) is described by the action

S=Sl+Sr+Stun+SLL.(2)S=S_l+S_r+S_{\mathrm{tun}}+S_{\mathrm{LL}}. \tag*{(2)}

Here SLLS_{\mathrm{LL}} is the Luttinger-liquid action in the channel, StunS_{\mathrm{tun}} is the tunnelling action, and Sl,rS_{l,r} are the standard BCS actions in the reservoirs:

Sλ=−∫dξ1 dξ2 Ψ‾λ(ξ1)Gλ−1(ξ1,ξ2)Ψλ(ξ2).(3)S_{\lambda}=-\int d\xi_1\,d\xi_2\,\overline{\Psi}_{\lambda}(\xi_1)G_{\lambda}^{-1}(\xi_1,\xi_2)\Psi_{\lambda}(\xi_2). \tag*{(3)}

Here λ=l,r\lambda=l,r labels the left and right reservoirs, ξ≡(r,τ)\xi\equiv(\mathbf{r},\tau), ∫dξ≡∫0βdτ∫dr\int d\xi\equiv\int_0^\beta d\tau\int d\mathbf{r} (τ\tau is imaginary time, β=1/kBT\beta=1/k_{\mathrm{B}}T is inverse temperature) and the Nambu spinors and BCS kernel are

Ψλ=(Ψλ↑Ψ‾λ↓),Ψ‾λ=(Ψ‾λ↑,Ψλ↓),Gλ−1=(−∂+ΔλΔλ∗−∂−),\Psi_\lambda=\begin{pmatrix}\Psi_{\lambda\uparrow}\\ \overline{\Psi}_{\lambda\downarrow}\end{pmatrix},\qquad \overline{\Psi}_\lambda=(\overline{\Psi}_{\lambda\uparrow},\Psi_{\lambda\downarrow}),\qquad G_\lambda^{-1}=\begin{pmatrix}-\partial_+&\Delta_\lambda\\ \Delta_\lambda^*&-\partial_-\end{pmatrix},

where ∂±=∂τ±(∇2/2m+μ)\partial_\pm=\partial_\tau\pm(\nabla^2/2m+\mu). We assume in what follows that Δλ=∣Δ∣eiφλ\Delta_\lambda=|\Delta|e^{i\varphi_\lambda}, i.e. the magnitude of the order parameter, as well as the chemical potential μ\mu and the normal density of states ν0\nu_0, are the same in both reservoirs, implying that the current through the 1D channel is driven by a constant phase difference φl−φr≡Φ\varphi_l-\varphi_r\equiv\Phi. We choose, without loss of generality, φl=−φr=Φ/2\varphi_l=-\varphi_r=\Phi/2.

The reservoir fields are coupled to the fermionic fields in the 1D channel, ψ(x,τ)=(ψ↑(x,τ),ψ‾↓(x,τ))T\psi(x,\tau)=(\psi_\uparrow(x,\tau),\overline{\psi}_\downarrow(x,\tau))^{\mathsf{T}} by the tunnelling action

Stun=−j∫0βdτ∑λ=l,r[Ψ‾λ(τ)ψλ(τ)+h.c.].(4)S_{\mathrm{tun}}=-j\int_0^\beta d\tau\sum_{\lambda=l,r}\left[\overline{\Psi}_\lambda(\tau)\psi_\lambda(\tau)+\mathrm{h.c.}\right]. \tag*{(4)}

Here Ψ‾λ(τ)\overline{\Psi}_\lambda(\tau) and ψλ(τ)\psi_\lambda(\tau) are the reservoir and channel fields evaluated at the reservoir contact point rλr_\lambda corresponding to the channel endpoint xλ=±L/2x_\lambda=\pm L/2. The tunnelling amplitudes, jj, are assumed to be spin-independent and equal on the left and right barriers.

Performing Gaussian integration over the reservoir fields reduces the first three terms in Eq. (2) to the effective boundary action:

Sb=−12j2∑λ=l,r∫dτ dτ′ ψ‾λ(τ)Gλ(τ−τ′)ψλ(τ′).(5)S_b=-\frac{1}{2}j^2\sum_{\lambda=l,r}\int d\tau\,d\tau'\,\overline{\psi}_\lambda(\tau)G_\lambda(\tau-\tau')\psi_\lambda(\tau'). \tag*{(5)}

Here Gλ=(GλFλFλ†−Gλ)G_\lambda=\begin{pmatrix}G_\lambda&\mathcal{F}_\lambda\\ \mathcal{F}_\lambda^\dagger&-G_\lambda\end{pmatrix} is the Nambu Green’s function of lead λ\lambda. Only anomalous propagators, Fl,r\mathcal{F}_{l,r}, carry the superconducting phases. For a bulk BCS lead, the anomalous propagators have the Matsubara-frequency representation, reduced in the low-energy limit ∣ωn∣,T≪Δ|\omega_n|,T\ll\Delta, to

Fl,r(ωn)=−πν0Δe±iΦ/2ωn2+Δ2≃−πν0e±iΦ/2,(6)\mathcal{F}_{l,r}(\omega_n)=-\pi\nu_0\frac{\Delta e^{\pm i\Phi/2}}{\sqrt{\omega_n^2+\Delta^2}}\simeq-\pi\nu_0e^{\pm i\Phi/2}, \tag*{(6)}

where ν0\nu_0 is the normal-state density of states in the reservoirs. Then Fl,r(τ−τ′)≃−πν0e±iΦ/2δ(τ−τ′)\mathcal{F}_{l,r}(\tau-\tau')\simeq-\pi\nu_0e^{\pm i\Phi/2}\delta(\tau-\tau') and the phase-dependent part of the boundary action (5) becomes local in imaginary time:

Sb≃πν0j22∑λ=l,r∫0βdτ[e−iφλψλ↓(τ)ψλ↑(τ)+h.c.].(7)S_b\simeq\frac{\pi\nu_0j^2}{2}\sum_{\lambda=l,r}\int_0^\beta d\tau\left[e^{-i\varphi_\lambda}\psi_{\lambda\downarrow}(\tau)\psi_{\lambda\uparrow}(\tau)+\mathrm{h.c.}\right]. \tag*{(7)}

We assume from the outset that attractively interacting fermions in the 1D channel are in the Luther–Emery regime where the backscattering interaction opens a spin gap Δσ\Delta_\sigma. In this case, as we show in the End Matter, Eq. (7) reduces after bosonizing the channel fields ψ\psi to

Sb=−J0∫0βdτ[cos⁡χl(τ)+cos⁡χr(τ)],(8)S_b=-J_0\int_0^\beta d\tau\left[\cos\chi_l(\tau)+\cos\chi_r(\tau)\right], \tag*{(8)}

where J0∼ν0κFj2J_0\sim\nu_0\kappa_Fj^2 is the effective tunnelling energy. The boundary phase drops χλ(τ)\chi_\lambda(\tau) are related to the boundary values φρ\varphi_\rho of the bosonic phase field φρ(x,τ)\varphi_\rho(x,\tau) in the number (ρ\rho) sector by

χl(τ)=12Φ−2φρl(τ),χr(τ)=12Φ+2φρr(τ).(9)\chi_l(\tau)=\frac{1}{2}\Phi-2\varphi_\rho^l(\tau),\qquad\chi_r(\tau)=\frac{1}{2}\Phi+2\varphi_\rho^r(\tau). \tag*{(9)}

In the low-energy Andreev strong-coupling limit, the boundary dynamics can be expressed solely in terms of the number-phase field φρ\varphi_\rho, as shown in the End Matter. The bosonized Luttinger-liquid action entering Eq. (2) then reduces to

SLL[φρ]=Kπv∫0βdτ∫−L/2L/2dx[(∂τφρ)2+v2(∂xφρ)2].(10)S_{\mathrm{LL}}[\varphi_\rho]=\frac{K}{\pi v}\int_0^\beta d\tau\int_{-L/2}^{L/2}dx\left[(\partial_\tau\varphi_\rho)^2+v^2(\partial_x\varphi_\rho)^2\right]. \tag*{(10)}

Here KK and vv are the Luttinger parameter and (renormalized) Fermi velocity in the channel. Equations (8)–(10), which govern transport of cold fermions with attraction, correspond to those in Ref. [21] for transport of cold bosons with repulsion.

Saddle-point configuration. For the attractive case considered here, K>1K>1, perturbative approach in weak tunnelling, used for repulsive interactions in Refs. [2, 3], is not applicable in the limit L→∞L\to\infty, thus motivating a nonperturbative saddle-point treatment.

A static saddle point is found by varying the bulk action, Eq. (10), which gives ∂x2φρ(x)=0\partial_x^2\varphi_\rho(x)=0. Hence the phase profile is linear along the channel. With the boundary conditions (9), we find the saddle-point solution as

φρ0(x)=(χ+−12Φ)xL+χ−2,χ±≡χr±χl2,(11)\varphi_{\rho0}(x)=\frac{\left(\chi_+-\frac{1}{2}\Phi\right)x}{L}+\frac{\chi_-}{2},\qquad\chi_\pm\equiv\frac{\chi_r\pm\chi_l}{2}, \tag*{(11)}

describing uniform superflow through the one-dimensional channel.

Substituting the saddle-point configuration (11) into Eqs. (8) and (10), we obtain the energy

EJc=(12Φ−χ+)2−2W0cos⁡χ+cos⁡χ−,(12)\frac{E}{J_c}=\left(\frac{1}{2}\Phi-\chi_+\right)^2-2W_0\cos\chi_+\cos\chi_-, \tag*{(12)}

where Jc≡Jc(L)=Kv/(πL)J_c\equiv J_c(L)=Kv/(\pi L) is the phase stiffness of a channel of length LL, and W0≡J0/JcW_0\equiv J_0/J_c is the dimensionless tunnelling strength. The first term in Eq. (12) is the energy of superfluid flow through the channel and the second is an effective two-parameter washboard potential. Minimizing EE with respect to χ+\chi_+ and χ−\chi_- gives

W0sin⁡χ+cos⁡χ−=12Φ−χ+,cos⁡χ+sin⁡χ−=0.(13)\begin{aligned} W_0\sin\chi_+\cos\chi_-&=\frac{1}{2}\Phi-\chi_+,\\ \cos\chi_+\sin\chi_-&=0. \tag*{(13)} \end{aligned}

The locally stable minima have χ−=0\chi_-=0 or π\pi modulo 2π2\pi, with cos⁡χ+cos⁡χ−>0\cos\chi_+\cos\chi_->0 [22]. This gives two stationary-state families:

12Φ−χ+S,A={+W0sin⁡χ+S,χ−=0,−W0sin⁡χ+A,χ−=π,(14)\frac{1}{2}\Phi-\chi_+^{S,A}= \begin{cases} +W_0\sin\chi_+^S,\quad\chi_-=0,\\ -W_0\sin\chi_+^A,\quad\chi_-=\pi, \end{cases} \tag*{(14)}

corresponding, respectively, to symmetric (SS) and asymmetric (AA) phase drops at the two boundaries. In the

symmetric family χr=χl\chi_r=\chi_l, while in the asymmetric family χr=χl+2π\chi_r=\chi_l+2\pi.

An important consequence of Eq. (14) is that the two stationary-state families have energies which are intrinsically 4π4\pi-periodic in the reservoir phase difference Φ\Phi: a shift Φ→Φ+2π\Phi\to\Phi+2\pi does not return a given family to itself; instead, it interchanges the symmetric and asymmetric families, ES,A(Φ+2π)=EA,S(Φ)E_{S,A}(\Phi+2\pi)=E_{A,S}(\Phi). A further shift by 2π2\pi returns to the original family, ES,A(Φ+4π)=ES,A(Φ)E_{S,A}(\Phi+4\pi)=E_{S,A}(\Phi), see Figure 1(b) and Figure 2(b), where we draw the branches only in the regions where solutions are (meta)stable.

Figure 2. (a) Colour density plot of the washboard energy in Eq. (12) at Φ=0\Phi=0 and W0=12W_0=12. Dots mark locally stable minima, labelled (P,m)(\mathcal{P},m): n=2mn=2m for P+=+1\mathcal{P}_+=+1 (blue dots, S) and n=2m+1n=2m+1 for P−=−1\mathcal{P}_-=-1 (red dots, A). Solid arrows connect opposite winding-parity families (Δn=±1\Delta n=\pm1, one-contact phase slips); dashed arrows connect different winding states within the same family (Δn=2\Delta n=2, Δm=1\Delta m=1, phase slips at both contacts). (b) Saddle-point energies for W0=4W_0=4. In this strong-coupling regime, each family contains several metastable branches distinguished by the integer winding number mm.

Winding families and Majorana representation. The branch label admits a formal representation using two self-adjoint Majorana operators γl\gamma_l and γr\gamma_r, with {γλ,γλ′}=2δλλ′\{\gamma_\lambda,\gamma_{\lambda'}\}=2\delta_{\lambda\lambda'}. Introducing nonlocal fermion operators

f=γl+iγr2,f†=γl−iγr2,(15)f = \frac{\gamma_l + i\gamma_r}{2}, \qquad f^\dagger= \frac{\gamma_l - i\gamma_r}{2}, \tag*{(15)}

we express the parity operator of Eq. (1) as P=2f†f−1\mathcal{P}=2f^\dagger f-1.

For sufficiently large dimensionless tunnelling strength WW, each of the two families, labelled by eigenvalues ±1\pm1 of operator P\mathcal{P}, contains several metastable minima as in Figure 2, corresponding to the phase winding, χ+→χ++2πm\chi_+ \to\chi_+ + 2\pi m, where mm is an integer winding index.

Avoided crossings between different branches could be generated by phase-slip processes (instantons) connecting different minima. Equivalently, we can consider fluctuations beyond the saddle-point approximation via the RG approach.

We show in the End Matter that such fluctuations are strongly suppressed in the thermodynamic limit thus restricting the scope for avoided crossings. The boundary RG flow for K>1K>1 increases the dimensionless cosine term toward strong coupling. In its weak-coupling scaling regime, W∼W0(L/a)1−1/KW\sim W_0(L/a)^{1-1/K}, with aa being an ultraviolet cutoff length [23]. This flow favors phase pinning. We show that together with the strong-coupling phase-slip picture [24], it supports the persistence of metastability and suppressed switching at low energies.

At finite size, phase-slip-induced mixing can produce avoided crossings between winding states. Slow evolution through such a crossing follows the hybridized state, whereas sufficiently rapid evolution can retain the metastable branch, as in Landau–Zener dynamics [25, 26]. This might provide a direct probe of the emergent Majorana physics.

The critical Josephson current behaves differently in such diabatic and adiabatic regimes, being of the order of J0(a/L)1/KJ_0(a/L)^{1/K} in the former, and Kv/LKv/L in the latter, as demonstrated in the End Matter. In both regimes it is parametrically larger than the critical perturbative current proportional to J02/LJ_0^2/L.

There is a useful comparison with the number-conserving spinless-fermion model of Ref. [27], where Majorana-like edge physics is diagnosed through parity-dependent edge-to-edge single-particle correlations. Identifying whether the winding-family label has a direct physical fermionic interpretation remains an open question.

Our results motivate phase-controlled transport experiments that probe metastable winding states, current reversals, and switching as a function of sweep rate, channel length, and temperature. These measurements would test the Majorana-like Josephson branch structure.

Acknowledgments

We are grateful to Boris Altshuler and Ivan Khaymovich for useful discussion. This work was supported by the Leverhulme Trust via grant No. RPG-2024-401. son energy structure can also arise without Majorana zero modes [28].

End Matter

Effective action. The fermionic fields in the 1D channel, ψˉ(x,τ)=(ψˉ↑,ψˉ↓)\bar{\psi}(x,\tau) = (\bar{\psi}_{\uparrow},\bar{\psi}_{\downarrow}) and their Hermitian conjugate, are bosonised in the usual way and expressed via phase fields, ϕρ,σ\phi_{\rho,\sigma}, and canonically conjugate displacement fields, ϑρ,σ\vartheta_{\rho,\sigma}, in the number (ρ\rho) and spin (σ\sigma) sectors as

ψs(x,τ)=kFπ∑η=±1ei(ηkF−π2)xe−i(ηϑρ−ϕρ)e−is(ηϑσ−ϕσ)\psi_s(x,\tau)=\sqrt{\frac{k_F}{\pi}}\sum_{\eta=\pm1}e^{i(\eta k_F-\frac{\pi}{2})x}e^{-i(\eta\vartheta_\rho-\phi_\rho)}e^{-is(\eta\vartheta_\sigma-\phi_\sigma)}

with s=±1s=\pm1 for spin ↑,↓\uparrow,\downarrow, η=±1\eta=\pm1 for right- and left-movers. Klein factors and particle-number zero modes are suppressed as they have no impact on this phase-only treatment of the pair-tunnelling action.

The attractive spin-1/2 channel is assumed to be in the Luther–Emery regime, with a spin gap Δσ\Delta_\sigma and a gapless number sector [14]. The slowly varying singlet-pair operator is proportional to e2iϕρcos⁡(2ϑσ)e^{2i\phi_\rho}\cos(2\vartheta_\sigma). Pinning of the spin displacement field ϑσ\vartheta_\sigma replaces its contribution by a nonzero ground-state expectation value. Its magnitude, together with short-distance normalization factors, is absorbed into the contact pair-tunnelling energy. The remaining number-sector action is

SLL=1π∫0βdτ∫−L/2L/2dx[−2i∂xϕρ∂τϑρ+vK(∂xϕρ)2+vK(∂xϑρ)2],(16)S_{\mathrm{LL}}=\frac{1}{\pi}\int_0^\beta d\tau\int_{-L/2}^{L/2}dx\left[-2i\partial_x\phi_\rho\partial_\tau\vartheta_\rho+vK(\partial_x\phi_\rho)^2+\frac{v}{K}(\partial_x\vartheta_\rho)^2\right], \tag*{(16)}

where KK and vv are the Luttinger parameter and (renormalized) Fermi velocity in the channel.

Gaussian elimination of the displacement-field fluctuations in Eq. (16) gives the phase-only action of Eqs. (8) and (10) with the effective boundary conditions of Eq. (9).

Suppression of fluctuations. Consider fluctuations of the boundary and channel fields around the saddle-point solution (11). Integrating out the fluctuating channel fields in the Gaussian phase action (10) yields in the limit ∣ω∣≫ω0=v/L|\omega|\gg\omega_0=v/L the Ohmic (Caldeira–Leggett) action [29]

Sfl=K∫∣ω∣dω(2π)2[∣δχ+(ω)∣2+∣δχ−(ω)∣2],(17)S_{\mathrm{fl}}=K\int\frac{|\omega|d\omega}{(2\pi)^2}\left[|\delta\chi_+(\omega)|^2+|\delta\chi_-(\omega)|^2\right], \tag*{(17)}

in terms of the fluctuating boundary fields, δχ±(τ)=χ±(τ)−χ±S,A\delta\chi_\pm(\tau)=\chi_\pm(\tau)-\chi_\pm^{\mathrm{S,A}}. Their fluctuations are governed by the sum of the Ohmic action and of the following effective boundary action,

Seff=Jc∫0βdτ{(12Φ−χ+)2−2Wcos⁡χ+cos⁡χ−},(18)S_{\mathrm{eff}}=J_c\int_0^\beta d\tau\left\{\left(\frac{1}{2}\Phi-\chi_+\right)^2-2W\cos\chi_+\cos\chi_-\right\}, \tag*{(18)}

obtained by integrating energy of Eq. (12) over the imaginary time. The nonlinear term in Eq. (18) can be treated within the standard RG analysis, with the dimensionless tunnelling WW being the running parameter.

As only the boundary modes enter the action (17)–(18), the bulk Luttinger parameter KK is not renormalized. The leading weak-coupling RG equation for the running dimensionless boundary coupling WW is obtained similarly to the Kane–Fisher healing of a weak link for K>1K>1 [30, 31]:

dln⁡Wdln⁡b=1−1K,(19)\frac{d\ln W}{d\ln b}=1-\frac{1}{K}, \tag*{(19)}

where bb is the length rescaling factor. In the weak-coupling scaling regime, this leads to

W(L)∼W0(a)(La)1−1/K,(20)W(L)\sim W_0(a)\left(\frac{L}{a}\right)^{1-1/K}, \tag*{(20)}

where W0(a)=J0/Jc(a)≡πaJ0/KvW_{0}(a)=J_{0}/J_{c}(a)\equiv\pi aJ_{0}/Kv with aa being the ultraviolet cutoff. This RG flow is analogous to dimensionless boundary tunnelling growth toward strong coupling, indicating a flow toward the Andreev strong-coupling regime discussed in [32]. The crossover length, LcL_{c}, to the strong coupling is estimated from W(Lc)∼1W(L_{c})\sim1 yielding

Lc∼aW0(a)−K/(K−1)∼a[KvπaJ0]K/(K−1).(21)L_{c}\sim aW_{0}(a)^{-K/(K-1)}\sim a\left[\frac{Kv}{\pi aJ_{0}}\right]^{K/(K-1)} . \tag*{(21)}

In the strong-coupling regime, we start with the pinned phase drops parameterized by the winding index nn as χ+=πn\chi_{+}=\pi n and χ−=0, π\chi_{-}=0,\ \pi. The leading winding energies are

En(0)(Φ)=−2J0+Jc(Φ2−πn)2,(22)E_{n}^{(0)}(\Phi)=-2J_{0}+J_{c}\left(\frac{\Phi}{2}-\pi n\right)^{2}, \tag*{(22)}

which holds for fixed winding and phase bias when L≫LcL\gg L_{c}, with the phase gradients still in the low-energy regime.

An elementary 2π2\pi slip at the left or right contact changes (χ+,χ−)(\chi_{+},\chi_{-}) by (π,−π)(\pi,-\pi) or (π,π)(\pi,\pi), respectively, with the opposite signs for an anti-slip. Both processes change nn by one. Substituting into Eq. (17) a slip–anti-slip pair at one contact λ=l,r\lambda=l,r, separated by imaginary time τ\tau gives for τc≪∣τ∣≪L/v\tau_{c}\ll|\tau|\ll L/v

Spair(τ)=2Score+2Kln⁡(∣τ∣τc).(23)S_{\mathrm{pair}}(\tau)=2S_{\mathrm{core}}+2K\ln\left(\frac{|\tau|}{\tau_{c}}\right). \tag*{(23)}

Here τc\tau_{c} is the short-time cutoff of the strong-coupling description, and ScoreS_{\mathrm{core}} is the core action for one slip. Equation (17) identifies the phase-slip operator dimension as KK. Its dimensionless fugacity is

dyλdln⁡b=(1−K)yλ ⇒ yλ(L)∼cλ(LcL)K−1,(24)\frac{d y_{\lambda}}{d\ln b}=(1-K)y_{\lambda}\ \Rightarrow\ y_{\lambda}(L)\sim c_{\lambda}\left(\frac{L_{c}}{L}\right)^{K-1}, \tag*{(24)}

where cλc_{\lambda} are matching constants. This dilute-slip expansion for yλ(E)≲1y_{\lambda}(E)\lesssim1 shows that the pinned fixed point is stable for K>1K>1.

The two limiting descriptions are related in the same manner as weak tunnelling and weak backscattering in the Kane–Fisher problem [30, 31], with the strong-coupling endpoint corresponding to phase pinning and Andreev reflection [32].

Critical Josephson current. The phase-controlled pair-number current at T=0T=0 is

I=∂E∂Φ=(Φ2−χ+)Jc(L)=±J(L)sin⁡χ+,(25)I=\frac{\partial E}{\partial\Phi}=\left(\frac{\Phi}{2}-\chi_{+}\right)J_{c}(L)=\pm J(L)\sin\chi_{+}, \tag*{(25)}

where Jc(L)=Kv/(πL)J_{c}(L)=Kv/(\pi L) and J(L)=Jc(L)W(L)∼J0(a/L)1/KJ(L)=J_{c}(L)W(L)\sim J_{0}(a/L)^{1/K}, as follows from Eqs. (14) with the renormalized value of dimensionless tunnelling W(L)W(L), Eq. (20).

There are two limiting current protocols to be distinguished. In a diabatic sweep through a small avoided crossing, the system can retain its winding branch rather than follow the hybridized equilibrium state. Equation (25) bounds the magnitude of the stationary current by J(L)J(L). If the chosen branch can be followed up to ∣sin⁡χ+∣=1|\sin\chi_{+}|=1, its limiting current is

Id=J(L)∼J0(a/L)1/K.(26)I_{\mathrm{d}}=J(L)\sim J_{0}(a/L)^{1/K}. \tag*{(26)}

For unequal tunnelling energies JlJ_{l} and JrJ_{r}, the stationary current must satisfy I=Jlsin⁡χl=Jrsin⁡χrI=J_{l}\sin\chi_{l}=J_{r}\sin\chi_{r}, so ∣I∣≤min⁡(Jl,Jr)|I|\leq\min(J_{l},J_{r}), i.e. the diabatic current is limited by the weaker contact.

In an adiabatic limit, finite mixing or relaxation permits the system to follow the equilibrium lower-energy envelope, the maximum current occurs at the first branch crossing, Φ=±π\Phi=\pm\pi. For large renormalized WW, Eq. (14) gives χ+≃π/(2W)\chi_{+}\simeq\pi/(2W) on the symmetric branch approaching Φ=π\Phi=\pi, yielding

Ia≃πJc2=Kv2L.(27)I_{\mathrm{a}}\simeq\frac{\pi J_{c}}{2}=\frac{Kv}{2L}. \tag*{(27)}

For both protocols, the limiting currents are parametrically larger than the critical perturbative current for repulsive interactions, Icrit∼(J02/ϵF2)(v/L)(a/L)2(1/K−1)I_{\mathrm{crit}}\sim(J_{0}^{2}/\epsilon_{F}^{2})(v/L)(a/L)^{2(1/K-1)} found in [2, 3]. In contrast to the perturbative result, neither the diabatic nor the adiabatic critical current goes over to the noninteracting limit by simply taking K→1K\to1 within the strong-coupling approximation. The reason for this is that the limits K→1K\to1 and L→∞L\to\infty do not commute.

Indeed, for K>1K>1 the coupling is relevant and reaches the Andreev strong-coupling regime at the crossover length LcL_{c}, Eq. (21). Hence, at any fixed K>1K>1, taking L→∞L\to\infty drives the junction to the Andreev fixed point, where the adiabatic envelope current is controlled by the phase stiffness, JcJ_{c}, Eq. (27). For weak bare tunnelling, W0(a)<1W_{0}(a)<1, at fixed LL the crossover scale LcL_{c} diverges when K→1K\to1 from above. Thus the system remains in the weak-coupling regime and the usual perturbative result proportional to J02J_{0}^{2} can be recovered.

References

  1. [1]T. Giamarchi, *Quantum Physics in One Dimension* (Oxford University Press, Oxford, 2003).
  2. [2]R. Fazio, F. W. J. Hekking, and A. A. Odintsov, Josephson current through a Luttinger liquid, Phys. Rev. Lett. **74**, 1843 (1995).DOI
  3. [3]D. L. Maslov, M. Stone, P. M. Goldbart, and D. Loss, Josephson current and proximity effect in Luttinger liquids, Phys. Rev. B **53**, 1548 (1996).pubmed.ncbi.nlm.nih.gov/9983618
  4. [4]J.-P. Brantut, J. Meineke, D. Stadler, S. Krinner, and T. Esslinger, Conduction of ultracold fermions through a mesoscopic channel, Science **337**, 1069 (2012).
  5. [5]J.-P. Brantut, C. Grenier, J. Meineke, D. Stadler, S. Krinner, C. Kollath, T. Esslinger, and A. Georges, A thermoelectric heat engine with ultracold atoms, Science **342**, 713 (2013).DOI
  6. [6]S. Krinner, D. Stadler, D. Husmann, J.-P. Brantut, and T. Esslinger, Observation of quantized conductance in neutral matter, Nature **517**, 64 (2015).
  7. [7]G. Valtolina, A. Burchianti, A. Amico, E. Neri, K. Xhani, J. A. Seman, A. Tremblay, A. Smerzi, M. Zaccanti, M. Inguscio, and G. Roati, Josephson effect in fermionic superfluids across the BEC-BCS crossover, Science **350**, 1505 (2015).DOI
  8. [8]D. Husmann, S. Uchino, S. Krinner, M. Lebrat, T. Giamarchi, T. Esslinger, and J.-P. Brantut, Connecting strongly correlated superfluids by a quantum point contact, Science **350**, 1498 (2015).
  9. [9]S. Krinner, M. Lebrat, D. Husmann, C. Grenier, J.-P. Brantut, and T. Esslinger, Mapping out spin and particle conductances in a quantum point contact, Proc. Natl. Acad. Sci. U.S.A. **113**, 8144 (2016).
  10. [10]S. Häusler, S. Nakajima, M. Lebrat, D. Husmann, S. Krinner, T. Esslinger, and J.-P. Brantut, Scanning gate microscope for cold atomic gases, Phys. Rev. Lett. **119**, 030403 (2017).DOI
  11. [11]A. Burchianti, F. Scazza, A. Amico, G. Valtolina, J. A. Seman, C. Fort, M. Zaccanti, M. Inguscio, and G. Roati, Connecting dissipation and phase slips in a Josephson junction between fermionic superfluids, Phys. Rev. Lett. **120**, 025302 (2018).
  12. [12]N. Luick, L. Sobirey, M. Bohlen, V. P. Singh, L. Mathey, T. Lompe, and H. Moritz, An ideal Josephson junction in an ultracold two-dimensional Fermi gas, Science **369**, 89 (2020).
  13. [13]M. Lebrat, P. Grisins, D. Husmann, S. Häusler, L. Corman, T. Giamarchi, J.-P. Brantut, and T. Esslinger, Band and correlated insulators of cold fermions in a mesoscopic lattice, Phys. Rev. X **8**, 011053 (2018).
  14. [14]A. Luther and V. J. Emery, Backward scattering in the one-dimensional electron gas, Phys. Rev. Lett. **33**, 589 (1974).
  15. [15]Phase imprinting provides a route to preparing a phase bias [12]. Population imbalance can also generate phase evolution [7]; a stationary phase bias requires the reservoir chemical potentials to be equal.
  16. [16]A. Y. Kitaev, Unpaired Majorana fermions in quantum wires, Phys. Usp. **44**, 131 (2001).DOI
  17. [17]L. Fu and C. L. Kane, Josephson current and noise at a superconducto[r/ quantum-spin-Hall-insulator/ superconductor junction, Phys. Rev. B **79**, 161408 (2009).
  18. [18]Although such an analogy is useful, a 4π-periodic Joseph- R. M. Lutchyn, J. D. Sau, and S. Das Sarma, Majorana fermions and a topological phase transition in semiconductor-superconductor heterostructures, Phys. Rev. Lett. 105, 077001 (2010).
  19. [20]This is similar to the 1D model of spinless fermions with attraction considered in Ref. [27].
  20. [21]D. P. Simpson, D. M. Gangardt, I. V. Lerner, and P. Krüger, One-dimensional transport of bosons between weakly linked reservoirs, Phys. Rev. Lett. 112, 100601 (2014).pubmed.ncbi.nlm.nih.gov/24679274
  21. [22]The solutions with cos χ₊ = 0 are saddle points rather than minima.
  22. [23]For repulsive interactions, K < 1, the boundary coupling instead flows toward zero, consistent with a perturbative treatment [2, 3].
  23. [24]M. Schecter, A. Kamenev, D. M. Gangardt, and A. Lamacraft, Critical velocity of a mobile impurity in one-dimensional quantum liquids, Phys. Rev. Lett. 108, 207001 (2012).pubmed.ncbi.nlm.nih.gov/23003175
  24. [25]F. Domínguez, F. Hassler, and G. Platero, Dynamical detection of Majorana fermions in current-biased nanowires, Phys. Rev. B 86, 140503(R) (2012).DOI
  25. [26]D. I. Pikulin and Y. V. Nazarov, Phenomenology and dynamics of a Majorana Josephson junction, Phys. Rev. B 86, 140504(R) (2012).DOI
  26. [27]F. Debortoli, N. Cuzzocul, L. Barbiero, and F. Grusdt, Majorana physics in a Luttinger liquid with attractive interactions, Phys. Rev. B 114, L231108 (2026).arxiv.org/abs/2607.06751
  27. [28]C.-K. Chiu and S. Das Sarma, Fractional Josephson effect with and without Majorana zero modes, Phys. Rev. B 99, 035312 (2019).DOI
  28. [29]A. O. Caldeira and A. J. Leggett, Influence of dissipation on quantum tunneling in macroscopic systems, Phys. Rev. Lett. 46, 211 (1981).DOI
  29. [30]C. L. Kane and M. P. A. Fisher, Transport in a one-channel Luttinger liquid, Phys. Rev. Lett. 68, 1220 (1992).pubmed.ncbi.nlm.nih.gov/10046110
  30. [31]C. L. Kane and M. P. A. Fisher, Transmission through barriers and resonant tunneling in an interacting one-dimensional electron gas, Phys. Rev. B 46, 15233 (1992).pubmed.ncbi.nlm.nih.gov/10003640
  31. [32]I. Affleck, J.-S. Caux, and A. M. Zagoskin, Andreev scattering and josephson current in a one-dimensional electron liquid, Phys. Rev. B 62, 1433 (2000).DOI

Paper details

vlnc:W_97gw7xcb
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents