betaFeedback

WorksPhysics and Astronomy

Finite-Temperature Transconductance of Multiterminal Josephson Junctions with Topological Andreev Spectra

Abstract

Multiterminal Josephson junctions (MTJJs) provide a controllable platform for engineering topological Andreev spectra. At zero temperature, Andreev bound states (ABSs) with nonzero Chern numbers give rise to quantized transconductance between superconducting terminals. We derive the finite-temperature transconductance of MTJJs using linear-response theory. We first demonstrate the destructive role of temperature: changes in the thermal occupations of the ABSs destroy exact transconductance quantization. We then show that finite temperature can also play a constructive role by making the transconductance sensitive to finite-energy topological crossings hidden from the zero-temperature response. Our results establish the thermal limitations of transconductance quantization and identify finite-temperature transport as a probe of otherwise inaccessible topology in Andreev spectra.

Introduction

Topological band theory provides a framework for phases whose transport can be governed by integer topological invariants [1]. Representative examples include topological insulators [2], Weyl semimetals [3], and topological superconductors [4]. The paradigmatic quantized response is the integer quantum Hall effect, for which σxy=(e2/h)ν\sigma_{xy} = (e^2/h)\nu [5, 6]. Here, ν\nu is an integer Chern number, ee is the electron charge, and hh is Planck’s constant.

Synthetic topological systems realize analogous band structures in controllable parameter spaces rather than exclusively in real space [7, 8]. Examples include Floquet systems [9, 10], photonic platforms [11, 12], and superconducting circuits [13, 14, 15]. Their tunability also enables effective topological physics in dimensions that are difficult to access in ordinary materials [16, 17, 18, 19].

Multiterminal Josephson junctions (MTJJs) are another tunable platform for exploring unconventional superconducting transport and phase-controlled Andreev physics [20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41] and, in particular, synthetic topology [42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64]. In an NN-terminal junction, the N−1N-1 independent superconducting phase differences act as synthetic quasimomenta, while the Andreev bound states (ABSs) form an effective band structure in this phase space. For N≥4N \ge4, the phase space is at least three-dimensional and can host Weyl points (WPs), i.e., conical degeneracies carrying integer topological charge [42, 43, 47]. These WPs can result in a nonzero ground-state Chern number or redistribute Chern numbers among individual ABSs.

In the adiabatic transport regime, MTJJ topology manifests itself in the dc transconductance. Incommensurate voltages applied to two superconducting (SC) leads make the corresponding SC phases densely cover a two-dimensional phase Brillouin zone (BZ), i.e., a torus. The BZ-averaged transconductance is then quantized as Gij=(4e2/h)CijG^{ij} = (4e^2/h)C^{ij}, where CijC^{ij} is the ground-state Chern number. This is the synthetic-phase analogue of the quantized Hall response [43].

Experimental observation of this quantization is challenging. Finite bias can drive nonadiabatic ABS transitions and restrict the adiabatic voltage window [44], while level broadening can smear the response [52]. The measurement is also intrinsically nonlocal because it correlates a voltage on one terminal with the current in another. These limitations have motivated complementary probes, including tunneling [59, 62] and microwave spectroscopy [50, 51].

Quantization can nevertheless be robust in special regimes, for example with Majorana-hosting SC terminals [53] or when the scattering matrix of the normal region lies near a chiral point [63]. For time-reversal-symmetric junctions, continuum states above the SC gap Δ\Delta can also preserve a quantized contribution [49].

A further open question concerns the effect of finite temperature on transconductance quantization. At finite temperature, ABS occupations are no longer step functions, so different states acquire unequal thermal weights. How this thermal reweighting modifies the quantized response has not been systematically investigated. In addition, it remains unclear whether transconductance can probe the topology of individual ABSs rather than only that of the ground state. This question is particularly relevant when individual ABSs carry nonzero Chern numbers whose sum vanishes, so that the zero-temperature transconductance contains no signature of their individual topology.

Here we address both questions in a four-terminal MTJJ within the framework of effective quantum dot (QD) models [50, 58]. First, we determine how thermal occupation modifies the quantized transconductance and show that exact quantization is lost once the temperature becomes comparable to the relevant ABS energy scale. Second, we show that finite-temperature transconductance can reveal individual-ABS topology that is absent from the zero-temperature response. Thus, the same thermal reweighting that limits quantization can also expose otherwise hidden structure of the Andreev spectrum.

The paper is organized as follows. In § 2 we discuss the synthetic topology of MTJJs. In § 3 we introduce the three effective QD models used throughout the paper and summarize their topological properties. In § 4 we derive the finite-temperature transconductance within linear-response theory. In § 5 we demonstrate the thermal breakdown of the transconductance quantization. In § 6 we investigate the finite-temperature signatures of individual-ABS topology. In § 7 we discuss the coexistence of ground-state and individual-ABS topology. In § 8 we discuss the experimental implications. In § 9 we present our conclusions. Technical derivations are collected in the Appendices.

Topology of MTJJs

We first summarize the synthetic topology of MTJJs. This section concerns the zero-temperature limit, T=0T = 0; the finite-temperature response is introduced in § 4.

Synthetic topology

We consider NN SC terminals coupled through a normal region; see Figure 1 for an example of four-terminal geometry. SC terminals α=0,…,N−1\alpha= 0, \ldots, N - 1 are ss-wave superconductors with gaps Δα\Delta_{\alpha} and phases φα\varphi_{\alpha}. Gauge invariance leaves N−1N - 1 independent SC phase differences, and we set φ0=0\varphi_{0} = 0. We assume MM relevant ABSs, which are described by a 2M×2M2M \times2M Bogoliubov–de Gennes (BdG) Hamiltonian H^eff\hat{H}^{\mathrm{eff}}. Particle-hole symmetry gives MM negative-energy states and MM positive-energy partners. Energies are measured relative to the common chemical potential EFE_{\mathrm{F}}.

Figure 1. Schematic four-terminal MTJJ. Four ss-wave SC leads with phases φα\varphi_\alpha (α=0,…,3\alpha= 0,\ldots,3) are coupled through a normal region hosting the ABSs described by H^eff\hat{H}^{\mathrm{eff}}. We choose the gauge φ0=0\varphi_0 = 0. Voltages V1V_1 and V2V_2 drive φ1\varphi_1 and φ2\varphi_2, while φ3\varphi_3 is controlled by the flux Φ\Phi through the loop connecting leads 0 and 3. The current IαI_\alpha is measured in lead α\alpha, and the junction is at temperature T≠0T \ne0.

The matrix elements of H^eff\hat{H}^{\mathrm{eff}} depend on the SC phases φα\varphi_{\alpha}. These phases define a synthetic parameter space, in which the ABS spectrum can acquire nontrivial topology. For the kkth ABS eigenstate ∣k⟩\lvert k\rangle of H^eff\hat{H}^{\mathrm{eff}}, with energy EkE_{k}, we define the Berry curvature in the (φi,φj)(\varphi_{i},\varphi_{j}) phase plane as

Bkij=2Im⁡[(∂φi⟨k∣)∂φj∣k⟩],(1)B_{k}^{ij} = 2\operatorname{Im}\left[\left(\partial_{\varphi_{i}}\langle k\lvert\right)\partial_{\varphi_{j}}\lvert k\rangle\right], \tag*{(1)}

and the corresponding Chern number as

Ckij=12π∫02πdφi∫02πdφjBkij.(2)C_{k}^{ij} = \frac{1}{2\pi}\int_{0}^{2\pi}d\varphi_{i}\int_{0}^{2\pi}d\varphi_{j}B_{k}^{ij}. \tag*{(2)}

The derivatives and integrals are taken over the two SC phases spanning the chosen two-dimensional synthetic BZ.

At T=0T = 0, the ground-state Berry curvature and Chern number are obtained by summing over the occupied negative-energy states:

Bij=∑Ek<0Bkij,Cij=∑Ek<0Ckij,(3)B^{ij} = \sum_{E_{k}<0}B_{k}^{ij}, \qquad C^{ij} = \sum_{E_{k}<0}C_{k}^{ij}, \tag*{(3)}

A nonzero CkijC_{k}^{ij} identifies a topologically nontrivial kkth ABS, whereas Cij≠0C^{ij} \ne0 identifies a topologically nontrivial ground state. Importantly, individual ABSs can carry nonzero Chern numbers whose sum cancels, yielding a topologically trivial ground state with Cij=0C^{ij} = 0.

We now focus on a four-terminal MTJJ (N=4N = 4). Topology is evaluated on the (φ1,φ2)(\varphi_{1},\varphi_{2}) phase torus, with φ3\varphi_{3} as a control parameter, so we study Ck12(φ3)C_{k}^{12}(\varphi_{3}) and C12(φ3)C^{12}(\varphi_{3}). We call a parameter regime topological when at least one ABS has Ck12≠0C_{k}^{12} \ne0 over a finite interval of φ3\varphi_{3}. We distinguish a ground-state topological regime, C12≠0C^{12} \ne0, from an individual-ABS topological regime, C12=0C^{12} = 0 and Ck12≠0C_{k}^{12} \ne0 for at least one band.

Different types of WPs

The synthetic topology of MTJJs is governed by WPs in the ABS spectrum [43]. We distinguish two types. Zero-energy WPs occur when a positive-energy ABS meets its particle-hole partner at E=0E = 0; see Figure 2(a). Near a zero-energy WP at φZE\boldsymbol{\varphi}_{\mathrm{ZE}}, the effective Hamiltonian is

Figure 2. Schematic ABS spectra versus the control phase φ3\varphi_3 at fixed φ1\varphi_1 and φ2\varphi_2. At T=0T = 0, states with E<0E < 0 (E>0E > 0) are occupied (unoccupied). (a) At the zero-energy WP φ3=φZE\varphi_3 = \varphi_{\mathrm{ZE}}, the negative- and positive-energy branches change their Chern numbers by −QZE-Q_{\mathrm{ZE}} and +QZE+Q_{\mathrm{ZE}}, giving ΔC12=−QZE\Delta C^{12} = -Q_{\mathrm{ZE}}. (b) Particle-hole-related finite-energy WPs occur at E=±εFEE = \pm\varepsilon_{\mathrm{FE}} and φ3=φFE\varphi_3 = \varphi_{\mathrm{FE}}. At each crossing, the lower and upper branches change their Chern numbers by −QFE-Q_{\mathrm{FE}} and +QFE+Q_{\mathrm{FE}}, while the ground-state Chern number remains unchanged, ΔC12=0\Delta C^{12} = 0.

H^ZE=dZE⋅τ^,dZE=M^ZEδφ,(4)\hat{H}_{\mathrm{ZE}} = \boldsymbol{d}_{\mathrm{ZE}} \cdot\hat{\boldsymbol{\tau}}, \qquad\boldsymbol{d}_{\mathrm{ZE}} = \hat{M}_{\mathrm{ZE}}\delta\boldsymbol{\varphi}, \tag*{(4)}

where δφ=φ−φZE\delta\boldsymbol{\varphi} = \boldsymbol{\varphi} - \boldsymbol{\varphi}_{\mathrm{ZE}}, τ^=(τ^x,τ^y,τ^z)\hat{\boldsymbol{\tau}} = (\hat{\tau}_x,\hat{\tau}_y,\hat{\tau}_z) is a vector of Pauli matrices, and M^ZE\hat{M}_{\mathrm{ZE}} is a real 3×33 \times3 velocity matrix. A zero-energy WP has topological charge

QZE=sgn⁡det⁡M^ZE.(5)Q_{\mathrm{ZE}} = \operatorname{sgn}\det\hat{M}_{\mathrm{ZE}}. \tag*{(5)}

Crossing the zero-energy WP as φ3\varphi_3 changes the Chern numbers of the negative- and positive-energy branches by −QZE-Q_{\mathrm{ZE}} and +QZE+Q_{\mathrm{ZE}}, respectively. At T=0T = 0, only the former is occupied, so the ground-state Chern number changes by −QZE-Q_{\mathrm{ZE}}.

Finite-energy WPs occur when two ABSs within the same T=0T = 0 occupation sector cross; see Figure 2(b). Particle-hole symmetry generates a partner crossing at the opposite energy. In a particle-hole-related basis, the four relevant BdG states are described by

H^FE=(h^FE00−h^FE∗),h^FE=EFEτ^0+dFE⋅τ^(6)\hat{H}_{\mathrm{FE}} = \begin{pmatrix} \hat{h}_{\mathrm{FE}} & 0 \\ 0 & -\hat{h}_{\mathrm{FE}}^{*} \end{pmatrix}, \qquad \hat{h}_{\mathrm{FE}} = E_{\mathrm{FE}}\hat{\tau}_0 + \boldsymbol{d}_{\mathrm{FE}}\cdot\hat{\boldsymbol{\tau}} \tag*{(6)}

where

EFE=εFE+vFE⋅δφ,dFE=M^FEδφ,(7)E_{\mathrm{FE}} = \varepsilon_{\mathrm{FE}} + \boldsymbol{v}_{\mathrm{FE}}\cdot\delta\boldsymbol{\varphi}, \qquad \boldsymbol{d}_{\mathrm{FE}} = \hat{M}_{\mathrm{FE}}\delta\boldsymbol{\varphi}, \tag*{(7)}

Here εFE>0\varepsilon_{\mathrm{FE}} > 0 is the positive-energy crossing energy, vFE\boldsymbol{v}_{\mathrm{FE}} describes the cone tilt, M^FE\hat{M}_{\mathrm{FE}} is a 3×33 \times3 velocity matrix, and δφ=φ−φFE\delta\boldsymbol{\varphi} = \boldsymbol{\varphi} - \boldsymbol{\varphi}_{\mathrm{FE}}. The topology is encoded in dFE\boldsymbol{d}_{\mathrm{FE}}, giving the finite-energy WP charge

QFE=sgn⁡det⁡M^FE.(8)Q_{\mathrm{FE}} = \operatorname{sgn}\det\hat{M}_{\mathrm{FE}}. \tag*{(8)}

The particle-hole-related finite-energy WPs at ±εFE\pm\varepsilon_{\mathrm{FE}} carry the same charge QFEQ_{\mathrm{FE}}. At each crossing, the lower and upper branches change their Chern numbers by −QFE-Q_{\mathrm{FE}} and +QFE+Q_{\mathrm{FE}}. Because both negative-energy branches are occupied at T=0T = 0, these changes cancel in the ground-state sum. Finite-energy WPs therefore redistribute Chern numbers among individual ABSs without changing the ground-state Chern number.

Thus zero-energy WPs change the ground-state Chern number, whereas finite-energy WPs redistribute Chern numbers among individual ABSs while leaving the ground-state Chern number unchanged. Importantly, finite-energy WPs can result in nontrivial individual-ABS topology even when the ground state is topologically trivial.

Effective QD models

We model the ABS spectra using effective QD models [50, 58], with MM single-level QDs tunnel coupled to NN SC leads. Integrating out the leads in the atomic limit, Δα→∞\Delta_{\alpha} \to\infty, yields an effective BdG Hamiltonian H^Deff\hat{H}_{D}^{\mathrm{eff}} for resonant ABSs with E≪ΔαE \ll\Delta_{\alpha}; Appendix § 11.§ 1 gives the derivation. We compute the Chern numbers of the ABSs from the effective Hamiltonians using the discretized Fukui algorithm [65].

We focus on three four-terminal QD models with distinct topological properties. The symmetric single-QD model (M=1M = 1) is the minimal realization of nontrivial ground-state topology [50]; see Figure 3(a). The symmetric normal-coupled double-QD model (M=2M = 2) provides a minimal multiband realization of ground-state topology [58]; see Figure 3(b). Finally, the antisymmetric SC-coupled double-QD model supports finite-energy WPs and hence nontrivial individual-ABS topology; see Figure 3(c). Appendix § 11.§ 2 summarizes the corresponding effective Hamiltonians.

Figure 3. Four-terminal QD models: (a) symmetric single-QD, (b) symmetric normal-coupled double-QD, and (c) antisymmetric SC-coupled double-QD. Each QD represents one fermionic level. Panels (a) and (b) use equal on-site energies ε\varepsilon and dot–lead couplings ww; panel (c) uses opposite dot energies and strong/weak couplings wSw_{S} and wW<wSw_{W}<w_{S}. Both double-QD models contain normal hopping rNr_{N}; the shared leads 1 and 2 in panel (c) additionally generate the nonlocal SC coupling rSCr_{\mathrm{SC}}. Neighboring SC leads in panel (a) are directly coupled by tt.

Symmetric single-QD model

The symmetric single-QD model shown in Figure 3(a) is the four-terminal model with one topological resonant ABS [50]. A dot with on-site energy ε\varepsilon couples equally to the four SC terminals with amplitude ww, while neighboring terminals are directly coupled by tt. The effective dot Hamiltonian is

H^Deff=d⋅τ^,(9)\hat{H}^{\mathrm{eff}}_{D}=\mathbf{d}\cdot\hat{\boldsymbol{\tau}}, \tag*{(9)}

with

d=(∑a=03Γcos⁡φa−∑a=03Γsin⁡φaε−2τΓ∑a=03cos⁡(φa−φa+1)).(10)\mathbf{d}= \left( \begin{array}{c} \sum_{a=0}^{3}\Gamma\cos\varphi_{a}\\ -\sum_{a=0}^{3}\Gamma\sin\varphi_{a}\\ \varepsilon-2\tau\Gamma\sum_{a=0}^{3}\cos\left(\varphi_{a}-\varphi_{a+1}\right) \end{array} \right). \tag*{(10)}

Here Γ=πN0w2\Gamma=\pi N_{0}w^{2}, τ=πN0t\tau=\pi N_{0}t, and N0N_{0} is the normal-state density of states per spin at the Fermi level. The sum in Eq. (10) assumes that the index aa is cyclic, so that φ4≡φ0\varphi_{4}\equiv\varphi_{0}.

The ground-state Chern number C12(φ3)C^{12}(\varphi_{3}) and a representative zero-energy WP are shown in Figure 4(a) and Figure 4(d), respectively. Because the single-QD model has only one positive-energy ABS [red curve in Figure 4(d)], it cannot host finite-energy WPs, whereas zero-energy WPs occur over a broad parameter range; see Appendix § 12.§ 1.

Figure 4. Chern numbers (a)–(c) and representative ABS spectra (d)–(f) versus φ3\varphi_{3}. Ground-state and individual-ABS topological regions are shaded light blue and light red, respectively. Red and blue denote the ABS branches whose Chern numbers are shown; additional branches are black. Dashed lines in (d)–(f) mark the first Chern-number jumps in (a)–(c). (a,d) Symmetric single-QD: ε/Γ=−0.2\varepsilon/\Gamma=-0.2, τ=0.1\tau=0.1; panel (d), at φ1=π\varphi_{1}=\pi, φ2=4π/3\varphi_{2}=4\pi/3, shows a zero-energy WP. (b,e) Symmetric normal-coupled double-QD: ε/Γ=1\varepsilon/\Gamma=1, rN/Γ=1.5r_{N}/\Gamma=1.5; panel (e), at φ1=4.33\varphi_{1}=4.33, φ2=π\varphi_{2}=\pi, shows a zero-energy WP. (c,f) Antisymmetric SC-coupled double-QD: ε/ΓS=1.2\varepsilon/\Gamma_{S}=1.2, ΓW/ΓS=0.3\Gamma_{W}/\Gamma_{S}=0.3, rN/ΓS=0.1r_{N}/\Gamma_{S}=0.1. Panel (c) shows C12C^{12}, C112C_{1}^{12}, and C212C_{2}^{12}, with C112=−C212C_{1}^{12}=-C_{2}^{12} and C12=0C^{12}=0 in the individual-ABS topological region. Panel (f), at φ1=2.35\varphi_{1}=2.35, φ2=5.5\varphi_{2}=5.5, shows the occupied-sector finite-energy WP and its particle-hole-related crossing in the unoccupied sector (black).

Double-QD model

To study multiband topology (M>1M>1), in particular individual-ABS topology, we next consider double-QD models with two resonant ABS pairs. A general double-QD model consisting of two QDs coupled to SC leads has an effective BdG Hamiltonian of the block form

H^Deff=(H^1V^12V^12†H^2),(11)\hat{H}^{\mathrm{eff}}_{D}= \begin{pmatrix} \hat{H}_{1} & \hat{V}_{12}\\ \hat{V}_{12}^{\dagger} & \hat{H}_{2} \end{pmatrix}, \tag*{(11)}

where the diagonal and off-diagonal blocks respectively are given by

H^i=(εiγiγi∗−εi),V^12=(rNrSCrSC∗−rN).(12)\hat{H}_{i}= \begin{pmatrix} \varepsilon_{i} & \gamma_{i}\\ \gamma_{i}^{*} & -\varepsilon_{i} \end{pmatrix}, \qquad \hat{V}_{12}= \begin{pmatrix} r_{N} & r_{\mathrm{SC}}\\ r_{\mathrm{SC}}^{*} & -r_{N} \end{pmatrix}. \tag*{(12)}

Here εi\varepsilon_{i} and γi\gamma_{i} are the on-site energy and induced pairing amplitude on dot i=1,2i=1,2, respectively; rNr_{N} is the normal interdot hopping and rSCr_{\mathrm{SC}} is a nonlocal SC interdot coupling generated when the dots share common SC leads.

We consider two double-QD geometries. In the normal-coupled model [Figure 3(b)], the two QDs couple to disjoint pairs of SC leads, so rSC=0r_{\mathrm{SC}}=0 while rN≠0r_{N}\ne0. In the SC-coupled model [Figure 3(c)], the QDs share SC leads, which generates a nonlocal SC coupling rSC≠0r_{\mathrm{SC}}\ne0 in addition to rN≠0r_{N}\ne0. As we will see below, the difference in interdot coupling leads to qualitatively different ABS topology in these two models.

Symmetric normal-coupled double-QD model

We first consider the symmetric normal-coupled double-QD model shown in Figure 3(b). It contains two dots with equal on-site energies [58]. Dot 1 couples to terminals 0 and 1, and dot 2 to terminals 2 and 3, all with amplitude ww. Thus, in Eq. (12), ε1=ε2=ε\varepsilon_{1}=\varepsilon_{2}=\varepsilon, γ1=Γ(1+eiφ1)\gamma_{1}=\Gamma(1+e^{i\varphi_{1}}), and γ2=Γ(eiφ2+eiφ3)\gamma_{2}=\Gamma(e^{i\varphi_{2}}+e^{i\varphi_{3}}), with Γ=πN0w2\Gamma=\pi N_{0}w^{2}. The dots are coupled only by the normal hopping rNr_{N}, so rSC=0r_{\mathrm{SC}}=0.

This model describes two ABSs primarily formed between leads (0, 1) and (2, 3), respectively, while the normal interdot hopping rNr_{N} introduces a weak hybridization between these two states. Over a finite range of rNr_{N}, this hybridization produces zero-energy WPs and a nontrivial ground-state Chern number [58]. Figure 4(b) and Figure 4(e) show C12(φ3)C^{12}(\varphi_{3}) and a representative zero-energy WP. The lowest positive- and negative-energy ABSs [red and blue curves in Figure 4(e), respectively] acquire nonzero Chern numbers through zero-energy WPs, whereas finite-energy WPs are absent and the higher-energy pair [black curves in Figure 4(e)] remains topologically trivial; see also Appendix § 12.§ 2.1.

Antisymmetric SC-coupled double-QD model

The normal-coupled double-QD model does not support robust finite-energy WPs and hence does not realize individual-ABS topology. We therefore introduce an SC-coupled double-QD model with a nonlocal interdot pairing rSC≠0r_{\mathrm{SC}} \ne0 in addition to normal hopping rN≠0r_{\mathrm{N}} \ne0. The purely SC-coupled limit (rN=0r_{\mathrm{N}} = 0) is also insufficient to produce robust finite-energy WPs; see Appendix § 12.§ 2.2. Thus both couplings are required in our working model. Starting from the geometry of Ref. [58], we allow the central leads 1 and 2 to couple to both dots, which generates the phase-dependent nonlocal pairing rSCr_{\mathrm{SC}}.

For concreteness, we consider below the antisymmetric geometry with opposite dot energies; see Figure 3(c). Dot 1 couples strongly to terminals 0 and 1 with amplitude wSw_{\mathrm{S}} and weakly to terminal 2 with wW<wSw_{\mathrm{W}} < w_{\mathrm{S}}; dot 2 couples strongly to terminals 2 and 3 and weakly to terminal 1. The two resulting ABSs are therefore centered on the (0, 1) and (2, 3) terminal pairs but hybridize both through the shared SC terminals and through direct normal hopping. In Eqs. (11) and (12), the parameters are ϵ1=−ϵ2=−ϵ\epsilon_1 = -\epsilon_2 = -\epsilon, γ1=ΓS(1+eiφ1)+ΓWeiφ2\gamma_1 = \Gamma_{\mathrm{S}}(1 + e^{i\varphi_1}) + \Gamma_{\mathrm{W}}e^{i\varphi_2}, γ2=ΓS(eiφ2+eiφ3)+ΓWeiφ1\gamma_2 = \Gamma_{\mathrm{S}}(e^{i\varphi_2} + e^{i\varphi_3}) + \Gamma_{\mathrm{W}}e^{i\varphi_1} and rSC=ΓNL(eiφ1+eiφ2)r_{\mathrm{SC}} = \Gamma_{\mathrm{NL}}(e^{i\varphi_1} + e^{i\varphi_2}), with ΓS=πN0wS2\Gamma_{\mathrm{S}} = \pi N_0 w_{\mathrm{S}}^2, ΓW=πN0wW2\Gamma_{\mathrm{W}} = \pi N_0 w_{\mathrm{W}}^2, and ΓNL=ΓSΓW\Gamma_{\mathrm{NL}} = \sqrt{\Gamma_{\mathrm{S}}\Gamma_{\mathrm{W}}}.

Figure 4(c) and Figure 4(f) show the individual-ABS Chern numbers and representative finite-energy WPs in the ABS spectrum, respectively. In Figure 4(f) the two occupied topological ABSs are shown in red and blue, while their particle-hole partners are shown in black. For these parameters, finite-energy WPs generate nontrivial individual-ABS topology while the absence of zero-energy WPs keeps the ground-state Chern number trivial. As we will discuss in § 7, other parameter choices yield ground-state topology and regimes in which ground-state and individual-ABS topology coexist.

Transconductance: linear response theory

In this section, we first recall the zero-temperature quantized transconductance of Ref. [43] and then derive its finite-temperature generalization from the Kubo formula. The resulting expression also separates two ways in which thermal occupations modify the topological response.

In our consideration, we assume quasistatic thermal equilibrium: the ABS relaxation rate is much larger than the Josephson drive frequencies, so the instantaneous occupations are given by the Fermi function. Additionally, we retain only the ABS contribution to the transconductance and neglect continuum states. In the atomic limit used here, the continuum is pushed to infinite energy and its thermal response is therefore negligible.

Zero-temperature transconductance

Within adiabatic perturbation theory [43], a slowly varying SC phase results in

Iα=2eℏ∑Ek<0∂Ek∂φα+4e2ℏBαβVβ,(13)I_{\alpha} = \frac{2e}{\hbar}\sum_{E_k<0}\frac{\partial E_k}{\partial\varphi_{\alpha}} + \frac{4e^2}{\hbar}B^{\alpha\beta}V_{\beta}, \tag*{(13)}

where IαI_{\alpha} is the SC current in lead α\alpha and VβV_{\beta} is the voltage applied to lead β\beta. At T=0T = 0, the sum runs over the occupied, negative-energy BdG states.

For two incommensurate voltages, the corresponding phases densely cover the synthetic BZ. The first term in Eq. (13) averages to zero, while the Berry-curvature term yields the quantized dc transconductance

I‾α=GαβVβ,Gαβ=4e2hCαβ,(14)\overline{I}_{\alpha} = G^{\alpha\beta}V_{\beta}, \qquad G^{\alpha\beta} = \frac{4e^2}{h}\mathcal{C}^{\alpha\beta}, \tag*{(14)}

where Iˉα\bar{I}_{\alpha} denotes the BZ-averaged current. Thus the transconductance between distinct terminals is quantized in units of 4e2/h4e^{2}/h and is determined by the ground-state Chern number CαβC^{\alpha\beta} [43].

Finite-temperature transconductance

To obtain finite-temperature transconductance, we use linear-response theory. We first define the ABS susceptibility as [39, 66]

Iα(ω)=∑βχαβABS(ω)Vβ(ω)−iω.(15)I^{\alpha}(\omega)=\sum_{\beta}\chi_{\alpha\beta}^{\mathrm{ABS}}(\omega)\frac{V_{\beta}(\omega)}{-i\omega}. \tag*{(15)}

This susceptibility is the ABS-mediated current response in lead α\alpha to an ac voltage Vβ(ω)V_{\beta}(\omega) applied to lead β\beta.

In the Lehmann representation, the retarded Kubo response is

χαβABS(ω)=∑k≠k′Ikk′αIk′kβnF(Ek)−nF(Ek′)ℏω+Ek−Ek′+i0.(16)\chi_{\alpha\beta}^{\mathrm{ABS}}(\omega)=\sum_{k\ne k'}I_{kk'}^{\alpha}I_{k'k}^{\beta}\frac{n_{\mathrm{F}}(E_k)-n_{\mathrm{F}}(E_{k'})}{\hbar\omega+E_k-E_{k'}+i0}. \tag*{(16)}

see Appendix § 13.§ 1. Here Ikk′α=(2e/ℏ)⟨k∣∂αH^Deff∣k′⟩I_{kk'}^{\alpha}=(2e/\hbar)\langle k|\partial_{\alpha}\hat{H}_{D}^{\mathrm{eff}}|k'\rangle is the current-operator matrix element between BdG eigenstates ∣k⟩|k\rangle and ∣k′⟩|k'\rangle, and nF(Ek)=1/(eEk/T+1)n_{\mathrm{F}}(E_k)=1/(e^{E_k/T}+1) is the Fermi function (we set the Boltzmann constant kB=1k_{\mathrm{B}}=1). The sum includes both positive- and negative-energy states, and +i0+i0 implies the retarded prescription.

In the adiabatic limit, ω→0\omega\to0, we expand the ABS susceptibility to first order in frequency:

χαβABS(ω)=χαβ(0)−iωχαβ(1)+O(ω2).(17)\chi_{\alpha\beta}^{\mathrm{ABS}}(\omega)=\chi_{\alpha\beta}^{(0)}-i\omega\chi_{\alpha\beta}^{(1)}+O(\omega^{2}). \tag*{(17)}

Here χαβ(0)\chi_{\alpha\beta}^{(0)} is the static susceptibility and χαβ(1)\chi_{\alpha\beta}^{(1)} is the leading dynamical term. Together with the diamagnetic and population contributions, the static response determines the inverse Josephson inductance Lαβ−1L_{\alpha\beta}^{-1}. Here we focus on the topological term linear in ω\omega:

χαβ(1)=4e2hBTαβ,(18)\chi_{\alpha\beta}^{(1)}=\frac{4e^{2}}{h}B_{T}^{\alpha\beta}, \tag*{(18)}

which for T≠0T\ne0 defines the finite-temperature Berry curvature

BTij=Im⁡∑k≠k′[nF(Ek)−nF(Ek′)]⟨k∣∂iH^Deff∣k′⟩⟨k′∣∂jH^Deff∣k⟩(Ek−Ek′)2.(19)B_{T}^{ij}=\operatorname{Im}\sum_{k\ne k'}\frac{\left[n_{\mathrm{F}}(E_k)-n_{\mathrm{F}}(E_{k'})\right]\langle k|\partial_i\hat{H}_{D}^{\mathrm{eff}}|k'\rangle\langle k'|\partial_j\hat{H}_{D}^{\mathrm{eff}}|k\rangle}{(E_k-E_{k'})^{2}}. \tag*{(19)}

Averaging over the BZ gives the finite-temperature transconductance

GTαβ=4e2hCTαβ,CTαβ=12π∫02πdφα∫02πdφβBTαβ.(20)G_{T}^{\alpha\beta}=\frac{4e^{2}}{h}C_{T}^{\alpha\beta},\qquad C_{T}^{\alpha\beta}=\frac{1}{2\pi}\int_{0}^{2\pi}d\varphi_{\alpha}\int_{0}^{2\pi}d\varphi_{\beta}B_{T}^{\alpha\beta}. \tag*{(20)}

For compactness, we refer to the BZ integral CTαβC_{T}^{\alpha\beta} as the finite-temperature Chern number. Equation (20) is then formally identical to the T=0T=0 result under Cαβ↦CTαβC^{\alpha\beta}\mapsto C_{T}^{\alpha\beta}. At T>0T>0, however, CTαβC_{T}^{\alpha\beta} is generally noninteger and is not a topological invariant.

At T=0T=0, the finite-temperature Berry curvature and finite-temperature Chern number reduce to the Berry curvature and Chern number in Eq. (3); see Appendix § 13.§ 2. At T>0T>0, Fermi factors reweight the ABS contributions, so the finite-temperature and zero-temperature responses no longer coincide. Exact transconductance quantization is therefore lost, and a finite transconductance can also appear where the zero-temperature ground-state Chern number vanishes.

Mechanisms of temperature-induced changes in finite-temperature Berry curvature

Equation (19) reveals two distinct thermal contributions to the finite-temperature Berry curvature and, after BZ averaging, to the finite-temperature transconductance. The first (occupied–unoccupied) contribution comes from matrix elements between the states on opposite sides of the Fermi level at T=0T = 0. Thermal occupation reduces their Fermi-factor difference, suppressing this contribution. The effect is especially pronounced near zero-energy WPs, where the excitation energy vanishes.

The second (same-sector) contribution comes from matrix elements between two states in the same T=0T = 0 occupation sector. Such terms vanish at zero temperature because the two Fermi factors are equal, but become finite when the levels are thermally populated unequally. This mechanism requires at least two positive-energy ABSs (and their particle-hole partners) and is enhanced near finite-energy WPs. In a multiband (M>1M > 1) spectrum both mechanisms generally coexist. To isolate them analytically, we use the single-QD model for the occupied–unoccupied contribution and the four-level Hamiltonian in Eq. (6) for the same-sector contribution.

Single-QD model: occupied–unoccupied contribution

For a general single-QD, the 2×22 \times2 BdG Hamiltonian is determined by a single vector d\mathbf{d} [see Eq. (9)]. Equation (19) then reduces to

BT12=B12tanh⁡∣d∣2T,(21)B_{T}^{12} = B^{12} \tanh\frac{|\mathbf{d}|}{2T}, \tag*{(21)}

where ∣d∣|\mathbf{d}| is the phase-dependent ABS energy.

Equation (21) shows directly that temperature suppresses the Berry-curvature response. The negative-energy state loses occupation while its positive-energy particle-hole partner acquires occupation and contributes with the opposite Berry curvature. Their combined finite-temperature Berry curvature is therefore reduced relative to the T=0T = 0 Berry curvature.

The suppression is strongly nonuniform in phase space. Regions with ∣d∣≫T|\mathbf{d}| \gg T are essentially unchanged, whereas low-energy regions with ∣d∣≲T|\mathbf{d}| \lesssim T are strongly reweighted. In particular, near a zero-energy WP, ∣dZE∣→0|\mathbf{d}_{\mathrm{ZE}}| \to0, the finite-temperature Berry curvature becomes

BT12=−δφ3det⁡M^ZE4T∣dZE∣2.(22)B_{T}^{12} = -\frac{\delta\varphi_{3}\det\hat{M}_{\mathrm{ZE}}}{4T|\mathbf{d}_{\mathrm{ZE}}|^{2}}. \tag*{(22)}

Relative to the zero-temperature Berry curvature, this result is suppressed by ∣dZE∣/(2T)→0|\mathbf{d}_{\mathrm{ZE}}|/(2T) \to0. The finite-temperature Berry curvature nevertheless remains singular at the zero-energy WP. A more useful measure is the finite-temperature Berry-curvature flux through a constant-energy sphere centered on the zero-energy WP:

QT∣dZE∣=−QZEtanh⁡∣dZE∣2T,(23)Q_{T}^{|\mathbf{d}_{\mathrm{ZE}}|} = -Q_{\mathrm{ZE}} \tanh\frac{|\mathbf{d}_{\mathrm{ZE}}|}{2T}, \tag*{(23)}

This flux tends to zero as ∣dZE∣/T→0|\mathbf{d}_{\mathrm{ZE}}|/T \to0. Thus any finite temperature smooths the zero-energy WP contribution after phase averaging: the finite-temperature Chern number becomes continuous and no longer exhibits a quantized jump at the zero-energy WP.

The single-QD therefore gives a simple phase-resolved picture of thermal suppression. Because the suppression varies across phase space, however, the BZ-integrated finite-temperature Chern number can either decrease or increase depending on how regions of positive and negative Berry curvature are reweighted; see § 5 below. We also note that this monotonic suppression of the finite-temperature Berry curvature is specific to the single-QD model. In multiband systems, different terms in Eq. (19) can have opposite signs, so thermal reweighting can either suppress or enhance the finite-temperature Berry curvature.

Double-QD model: thermally induced same-sector contribution

To isolate the same-sector contribution, we use the four-level Hamiltonian from Eq. (6). Beyond the immediate vicinity of a finite-energy WP, it can be viewed as a generic two-ABS model (plus particle-hole partners) parameterized by EFEE_{\mathrm{FE}} and dFE\mathbf{d}_{\mathrm{FE}}. It describes either an individual-ABS topological case, when ∣dFE∣=0|\mathbf{d}_{\mathrm{FE}}|=0 somewhere in phase space, or a topologically trivial avoided-crossing case, when ∣dFE∣≠0|\mathbf{d}_{\mathrm{FE}}|\ne0 everywhere. We assume EFE−∣dFE∣>0E_{\mathrm{FE}}-|\mathbf{d}_{\mathrm{FE}}|>0, excluding zero-energy WPs.

Because Eq. (6) is block diagonal in the two T=0T=0 occupation sectors, only same-sector matrix elements contribute, and Eq. (19) reduces to

BT12=BFE12(tanh⁡EFE+∣dFE∣2T−tanh⁡EFE−∣dFE∣2T),(24)B_{T}^{12}=B_{\mathrm{FE}}^{12}\left(\tanh\frac{E_{\mathrm{FE}}+|\mathbf{d}_{\mathrm{FE}}|}{2T}-\tanh\frac{E_{\mathrm{FE}}-|\mathbf{d}_{\mathrm{FE}}|}{2T}\right), \tag*{(24)}

where BFE12=−dFE⋅[∂φ1dFE×∂φ2dFE]/(2∣dFE∣3)B_{\mathrm{FE}}^{12}=-\mathbf{d}_{\mathrm{FE}}\cdot[\partial_{\varphi_1}\mathbf{d}_{\mathrm{FE}}\times\partial_{\varphi_2}\mathbf{d}_{\mathrm{FE}}]/(2|\mathbf{d}_{\mathrm{FE}}|^3) is the Berry curvature of the lowest negative-energy ABS. Unlike Eq. (21), this contribution vanishes at T=0T=0 and is generated by unequal thermal populations at T>0T>0. At fixed phase, it is nonmonotonic in TT, with asymptotic limits

BT12=BFE12×{2e(−EFE+∣dFE∣)/T,EFE,∣dFE∣≫T,4(∣dFE∣/T)e−EFE/T,EFE≫T≫∣dFE∣,∣dFE∣/T,EFE,∣dFE∣≪T.(25)B_{T}^{12}=B_{\mathrm{FE}}^{12}\times \begin{cases} 2e^{(-E_{\mathrm{FE}}+|\mathbf{d}_{\mathrm{FE}}|)/T}, & E_{\mathrm{FE}},|\mathbf{d}_{\mathrm{FE}}|\gg T,\\ 4(|\mathbf{d}_{\mathrm{FE}}|/T)e^{-E_{\mathrm{FE}}/T}, & E_{\mathrm{FE}}\gg T\gg|\mathbf{d}_{\mathrm{FE}}|,\\ |\mathbf{d}_{\mathrm{FE}}|/T, & E_{\mathrm{FE}},|\mathbf{d}_{\mathrm{FE}}|\ll T. \end{cases} \tag*{(25)}

For the first asymptotic expression, we additionally require the physically relevant condition EFE−∣dFE∣≫TE_{\mathrm{FE}}-|\mathbf{d}_{\mathrm{FE}}|\gg T. The finite-temperature Berry curvature vanishes both in the low-temperature limit, EFE,∣dFE∣≫TE_{\mathrm{FE}},|\mathbf{d}_{\mathrm{FE}}|\gg T, and in the high-temperature limit, EFE,∣dFE∣≪TE_{\mathrm{FE}},|\mathbf{d}_{\mathrm{FE}}|\ll T, where all four occupations approach one half.

Equation (24) applies to both topological crossings and trivial avoided crossings. At the same time, near a finite-energy WP in the individual-ABS topological regime, ∣dFE∣→0|\mathbf{d}_{\mathrm{FE}}|\to0, substituting the linearized parameters from Eq. (7) gives

BT12=−δφ3det⁡M^FE2T∣dFE∣2cosh⁡2(EFE/2T),(26)B_{T}^{12}=-\frac{\delta\varphi_3\det\widehat{M}_{\mathrm{FE}}}{2T|\mathbf{d}_{\mathrm{FE}}|^2\cosh^2(E_{\mathrm{FE}}/2T)}, \tag*{(26)}

The expression scales as (1/T)exp⁡(−EFE/T)(1/T)\exp(-E_{\mathrm{FE}}/T) for T/EFE≪1T/E_{\mathrm{FE}}\ll1 and as 1/T1/T for T/EFE≫1T/E_{\mathrm{FE}}\gg1. As for a zero-energy WP, the finite-temperature Berry curvature is singular at the crossing, although the thermally active region shrinks at low temperature. The corresponding finite-temperature Berry-curvature flux through an energy sphere centered on the finite-energy WP is finite:

QT∣dFE∣=−QFE(tanh⁡EFE+∣dFE∣2T−tanh⁡EFE−∣dFE∣2T),(27)Q_{T}^{|\mathbf{d}_{\mathrm{FE}}|}=-Q_{\mathrm{FE}}\left(\tanh\frac{E_{\mathrm{FE}}+|\mathbf{d}_{\mathrm{FE}}|}{2T}-\tanh\frac{E_{\mathrm{FE}}-|\mathbf{d}_{\mathrm{FE}}|}{2T}\right), \tag*{(27)}

where, for simplicity, we consider an untilted finite-energy WP. The finite-temperature Berry-curvature flux vanishes in both the low- and high-temperature limits. Consequently, the finite-temperature Chern number has no quantized jump associated with a finite-energy WP at any temperature.

These results show that same-sector thermal reweighting can generate a finite-temperature Berry curvature that is strictly absent at T=0T=0. Because this response is highly nonuniform in phase space, however, the corresponding finite-temperature Chern number must be evaluated numerically.

Violation of the transconductance quantization

In this section, we investigate the thermal breakdown of transconductance quantization in four-terminal MTJJs. In terms of the finite-temperature Chern number, the breakdown manifests itself as

CT12≠C12,T≠0.(28)C_T^{12} \ne C^{12}, \qquad T \ne0. \tag*{(28)}

For the QD models considered here, the breakdown is dominated by thermal suppression of the occupied–unoccupied contribution.

We evaluate Eqs. (19) and (20) numerically for the symmetric single-QD and normal-coupled double-QD models. Figure 5(a) and Figure 5(d) show CT12(φ3)C_T^{12}(\varphi_3) in these cases, respectively. Both models behave similarly: at low TT, deviations from quantization are small except near zero-energy WPs, where the excitation gap closes and even weak thermal occupation smooths the response. As TT increases, the deviation extends throughout the whole ground-state topological region.

Figure 5. Finite-temperature Chern number CT12C_T^{12} for the symmetric single-QD (top row) and symmetric normal-coupled double-QD (bottom row). Panels (a) and (d) show CT12(φ3)C_T^{12}(\varphi_3) at the indicated temperatures; light-blue shading marks the ground-state topological regions. Parameters are ϵ/Γ=−0.2\epsilon/\Gamma=-0.2, τ=0.1\tau=0.1 (top) and ϵ/Γ=1\epsilon/\Gamma=1, rN/Γ=1.5r_N/\Gamma=1.5 (bottom). At T=0T=0, CT12=C12C_T^{12}=C^{12}; finite temperature smooths the zero-energy-WP steps and suppresses the plateau. Dashed lines at φ3=4\varphi_3=4 in (a,d) correspond to the topological cuts in (b,e), while dash-dotted lines at φ3=5.5\varphi_3=5.5 correspond to the trivial cuts in (c,f). Dotted lines in (a,d) delimit the nearly quantized regions. The topological cuts decrease monotonically with TT; the trivial cuts are nonmonotonic, with maxima marked by black dots.

Figure 5(b),(Figure 5c) and (Figure 5e),(Figure 5f) show CT12(T)C_T^{12}(T) at fixed φ3\varphi_3 for the symmetric single-QD and normal-coupled double-QD models, respectively. In a ground-state topological region [Figure 5(b) and (Figure 5e) which correspond to the dashed vertical lines in Figure 5(a) and (Figure 5d), respectively], CT12C_T^{12} decreases monotonically because thermal occupation suppresses the dominant Berry curvature contribution; Appendix § 14.§ 1 shows the corresponding finite-temperature Berry-curvature maps. In a ground-state trivial region [Figure 5(c) and (Figure 5f) which correspond to the dash-dotted vertical lines in Figure 5(a) and (Figure 5d), respectively], positive- and negative-Berry-curvature contributions cancel at T=0T=0 but are thermally reweighted at different rates. The resulting finite-temperature transconductance is finite and nonmonotonic and vanishes again at high temperature as the ABS occupations become equal.

Finite-temperature detection of individual-ABS topology

We next ask whether finite-temperature transconductance can reveal individual-ABS topology that is invisible in the ground-state Chern number. We use the antisymmetric SC-coupled double-QD model in a regime with finite-energy WPs but no zero-energy WPs, so C12=0C^{12}=0 although individual ABSs carry nonzero Chern numbers; the response is therefore governed by the thermally induced same-sector contribution. We then compare this regime with a topologically trivial state close to the topological transition to assess how specifically the finite-temperature transconductance identifies individual-ABS topology.

Finite-temperature Chern number of the individual-ABS topological state

Using Eqs. (19) and (20), we compute CT12(φ3)C_{\mathrm{T}}^{12}(\varphi_3) for the antisymmetric SC-coupled double-QD model. Figure 6(a) shows finite-temperature peaks precisely in the interval where the individual ABSs carry nonzero Chern numbers. The underlying finite-temperature Berry curvature is thermally generated and its phase-space maps are shown in Appendix § 14.§ 2.

Figure 6. Finite-temperature Chern number CT12C_T^{12} in an individual-ABS topological regime (top row) and a trivial regime (bottom row). Panels (a,b) correspond to the antisymmetric SC-coupled double-QD with ϵ/ΓS=1.2\epsilon/\Gamma_S = 1.2, ΓW/ΓS=0.3\Gamma_W/\Gamma_S = 0.3, and rN/ΓS=0.1r_N/\Gamma_S = 0.1. In (a), the light-red interval marks nonzero individual-ABS Chern numbers; vertical lines indicate its boundaries and the dashed line at φ3=4\varphi_3 = 4 defines the cut used in (b). Thermally induced peaks appear in CT12(φ3)C_T^{12}(\varphi_3) in (a), with the corresponding nonmonotonic CT12(T)C_T^{12}(T) shown in (b). Panels (c,d) correspond to the symmetric normal-coupled double-QD in the trivial regime with ϵ/Γ=1\epsilon/\Gamma= 1 and rN/Γ=0.8r_N/\Gamma= 0.8 in (c). The dashed line at φ3=5\varphi_3 = 5 defines the cut used in (d), where CT12(T)C_T^{12}(T) is shown for several rNr_N. Similar thermal peaks appear in (c), with a nonmonotonic CT12(T)C_T^{12}(T) dependence in (d); the peak magnitude decreases as the system is tuned away from the transition at rN/Γ=1r_N/\Gamma= 1.

At fixed φ3\varphi_3 inside this interval, Figure 6(b) shows a nonmonotonic CT12(T)C_{\mathrm{T}}^{12}(T) with a maximum at finite temperature. Unlike the trivial-region maxima in Figure 5(c) and (Figure 5f), this maximum is not caused by unequal suppression of pre-existing regions of positive and negative Berry curvature. Instead, same-sector thermal reweighting generates finite-temperature Berry curvature with nonuniform phase-space weight; see Appendix § 14.§ 2. Thus individual-ABS topology produces two characteristic signatures in this model: peaks in CT12(φ3)C_{\mathrm{T}}^{12}(\varphi_3) and a nonmonotonic CT12(T)C_{\mathrm{T}}^{12}(T).

Individual-ABS topology vs. trivial topology

Thermally induced finite-temperature transconductance features do not by themselves imply nontrivial topology. To test their specificity, we compare them with a topologically trivial system that nevertheless has a nonzero phase-resolved Berry curvature distribution.

For comparison, we choose the symmetric normal-coupled double-QD model on the trivial side of its topological transition, rN/Γ<1r_{\mathrm{N}}/\Gamma< 1 [58]. Figure 6(c) and (Figure 6d) again show phase-dependent peaks of CT12(φ3)C_{\mathrm{T}}^{12}(\varphi_3) and a nonmonotonic CT12(T)C_{\mathrm{T}}^{12}(T), although all individual-ABS Chern numbers vanish. Appendix § 14.§ 3 shows the corresponding finite-temperature Berry-curvature maps. These peaks are generally weaker and decrease rapidly as rNr_{\mathrm{N}} is tuned farther into the trivial side of the transition, becoming negligible sufficiently far from the phase boundary.

Pronounced finite-temperature transconductance peaks are therefore not a unique diagnostic of nonzero individual-ABS Chern numbers. Phase-resolved measurements of the finite-temperature Berry curvature can, however, distinguish the two cases in these simple QD models despite their similar averaged finite-temperature transconductance; see Appendices § 14.§ 2 and § 14.§ 3.

Coexistence of ground-state and individual-ABS topology

Section 5 showed the thermal suppression of finite-temperature transconductance associated with ground-state topology (zero-energy WPs), whereas Sec. 6 showed thermally generated signatures of individual-ABS topology (finite-energy WPs). We now consider the antisymmetric SC-coupled double-QD in regimes where both types of WPs coexist.

Figures 7(a)–(c) show the ground-state and individual-ABS Chern numbers for three parameter sets. The ground-state and individual-ABS topological intervals are either separated [Figure 7(a)] or overlap [Figure 7(b) and (Figure 7c)], with different individual-ABS Chern number patterns in the two overlapping cases.

Figure 7. Coexisting ground-state and individual-ABS topology in the antisymmetric SC-coupled double-QD model with ΓW/ΓS=0.3\Gamma_W/\Gamma_S = 0.3. Ground-state, individual-ABS, and overlapping topological regions are shaded light blue, light red, and light purple, respectively; dashed lines mark the boundaries of the individual-ABS topological regions. Panels (a)–(c) show C12C^{12} (black), C112C_1^{12} (blue), and C212C_2^{12} (red), while panels (d)–(f) show CT12(φ3)C_T^{12}(\varphi_3). Panels (a,d) use ε/ΓS=0.1\varepsilon/\Gamma_S = 0.1 and rN/ΓS=1.2r_N/\Gamma_S = 1.2; panels (b,e), ε/ΓS=0.1\varepsilon/\Gamma_S = 0.1 and rN/ΓS=0.6r_N/\Gamma_S = 0.6; and panels (c,f), ε/ΓS=0.2\varepsilon/\Gamma_S = 0.2 and rN/ΓS=0.1r_N/\Gamma_S = 0.1. In (a,d), the two topological intervals are separated; in (b,e) and (c,f), they overlap. Thermal individual-ABS contributions appear separately in (d), as a peak on the suppressed ground-state response in (e), and as a dip in (f).

The corresponding finite-temperature Chern numbers are shown in Figure 7(d)–Figure 7(f). When the intervals are separated [Figure 7(d)], the two thermal signatures are also separated: ∣CT12∣\lvert C_T^{12}\rvert decreases in the ground-state topological interval, while finite-temperature peaks appear in the individual-ABS topological interval. When the intervals overlap [Figure 7(e) and (Figure 7f)], these trends superpose. Depending on the ordering of the individual-ABS Chern numbers, the finite-energy-WP-related contribution appears as either a peak [Figure 7(e)] or a dip [Figure 7(f)] on top of the overall thermal suppression of the ground-state response.

These results show that the finite-temperature response can be understood as a superposition of two physically distinct mechanisms: suppression of the contribution associated with ground-state topology and generation of a contribution associated with individual-ABS topology. Their relative sign and phase-space overlap determine whether the latter appears as a peak or a dip in the total response.

Discussion

Strength of the temperature-induced effects

The preceding sections identified both phase-resolved (§ 4.3) and phase-averaged (§ 5–§ 7) thermal effects. We now discuss their magnitude.

Temperature can strongly modify the phase-resolved response. In the single-QD model, ∣d∣≪T|d| \ll T strongly suppresses the finite-temperature Berry curvature, whereas near a finite-energy WP even a small temperature can generate a large finite-temperature Berry curvature. The phase-resolved response can therefore show either pronounced suppression or enhancement, see Appendix § 14.

The BZ-averaged effect can also be large, as the violation of quantization in § 5 demonstrates. The thermally generated signal associated with individual-ABS topology is smaller in our examples: its typical maximum is CT12∼0.1C_T^{12} \sim0.1. This is not a fundamental bound; it reflects the detailed ABS dispersion of the chosen model. A larger response would require a broad phase-space region in which one relevant ABS is thermally active while the other remains effectively unoccupied. In this case, the finite-temperature Chern number could reach CT12→1C_T^{12} \to1.

Reconstruction of the individual-ABS Chern numbers from the finite-temperature transconductance

The experimentally relevant inverse problem is whether finite-temperature transconductance measurements can reconstruct ABS topology. At T=0T = 0, the averaged transconductance directly measures the ground-state Chern number [43]. At low temperature and sufficiently far from a zero-energy WP, this quantization remains a good approximation. At higher temperature, the sharp Chern number steps are smoothed, but their locations can still be inferred from the crossover of C12(φ3)C^{12}(\varphi_3) and from its characteristic temperature dependence.

Reconstructing individual-ABS Chern numbers is less direct. As shown in § 6.2, thermally induced peaks also occur on the trivial side near a topological transition. Their amplitude decreases as the system is tuned away from the transition, so a pronounced finite-temperature transconductance feature signals either proximity to a transition or topological ABS bands but does not uniquely determine the individual-ABS Chern numbers.

The ambiguity increases when several resonant ABSs contribute. In the antisymmetric SC-coupled double-QD model, zero-energy WPs and finite-energy WPs remain distinguishable because their phase intervals and temperature dependences can be tracked separately (§ 7), whereas generic multiband junctions, in principle, may produce similar finite-temperature transconductance curves from different microscopic spectra. Finite-temperature transconductance should therefore be viewed as a sensitive probe: it robustly identifies ground-state topology and highlights phase regions consistent with nontrivial individual-ABS structure, but additional information is required for an unambiguous individual-ABS topological assignment.

Alternative measurement protocols

Because the BZ-averaged finite-temperature transconductance does not uniquely determine individual-ABS Chern numbers, it is most useful in combination with phase-resolved and spectroscopic probes. The finite-temperature Berry-curvature maps in Appendices D.2 and D.3 distinguish, for the present models, thermally generated same-sector response which is a characteristic thermal feature of the individual-ABS topology from unequal suppression of pre-existing Berry curvature which is a characteristic thermal feature of the trivial or ground-state topology. However, this distinction need not remain unique in a more complicated multiband case. Finite-temperature transconductance can therefore first identify promising control-phase intervals. Tunneling spectroscopy [59, 62] can then test for finite-energy ABS crossings, while microwave spectroscopy provides a complementary check because the transitions at ω=2E1\omega= 2E_1, 2E22E_2, and E1+E2E_1 + E_2 merge at E1=E2E_1 = E_2. The protocols of Refs. [50, 51] can additionally reconstruct the Berry curvature of a selected ABS from transition matrix elements, providing an independent test of the individual-ABS topology.

Conclusions

We studied the finite-temperature transconductance of four-terminal MTJJs with topological Andreev spectra. Their synthetic topology is controlled by WPs in SC-phase space: zero-energy WPs change the ground-state Chern number, whereas finite-energy WPs redistribute Chern numbers among individual ABSs without changing the ground-state sum. Using effective single-QD and double-QD models, we identified regimes with zero-energy WPs, finite-energy WPs, and their coexistence; the antisymmetric SC-coupled double-QD with both normal hopping and nonlocal SC coupling is the minimal model used here to realize robust finite-energy crossings and nontrivial individual-ABS Chern numbers.

We derived the finite-temperature transconductance using linear-response theory. Thermal occupation has two consequences: it suppresses contributions connecting states on opposite sides of the Fermi level and activates contributions between states in the same T=0T = 0 occupation sector. The first mechanism manifests itself in the breakdown of the exact transconductance quantization, while the second generates a finite response near finite-energy WPs even when the T=0T = 0 ground-state Chern number is zero.

For the single-QD and normal-coupled double-QD models, quantization remains accurate when TT is small compared with the relevant ABS energies, except near zero-energy WPs where the gap closes. In ground-state topological regions, ∣CT12∣|C_T^{12}| decreases with temperature; in trivial regions, unequal thermal reweighting can produce a nonmonotonic finite response.

For individual-ABS topology, the antisymmetric SC-coupled double-QD develops finite-temperature peaks in CT12C_T^{12} within the phase interval bounded by finite-energy WPs, and a nonmonotonic temperature dependence. Similar features can arise in a topologically trivial model close to a transition, so the BZ-averaged finite-temperature transconductance is not by itself a unique individual-ABS topology diagnostic. At the same time, phase-resolved finite-temperature response measurements can help distinguish between these two scenarios.

When zero-energy WPs and finite-energy WPs coexist, their thermal signatures superpose as suppression of the ground-state response together with peaks or dips associated with individual-ABS topology. These results establish finite temperature both as a limitation on quantized transconductance and as a practical source of otherwise hidden information about the topology of Andreev spectra.

Acknowledgements

We thank Daniel Bobok, Gustavo Diniz, Joren Harms, Michael Hein, Tomáš Novotný, Aleksandr Svetogorov, Peter Zalom, and Martin Žonda for useful discussions.

Funding information We acknowledge support by the DAAD PPP (Projekt-ID: 57753337) and the Deutsche Forschungsgemeinschaft (DFG; German Research Foundation) by SFB 1432 (Project No. 425217212).

Effective Hamiltonians of QD models

Derivation scheme

We consider MM single-level QDs, indexed by i,j=1,…,Mi,j = 1,\ldots,M, coupled to NN SC leads, indexed by α,β=0,…,N−1\alpha,\beta= 0,\ldots,N-1; see Figure 8. We use the Green’s function technique to integrate out SC leads and derive the effective BdG Hamiltonian for the resonant ABSs [50, 51, 58].

Figure 8. Generic effective QD model for an MTJJ. The NN SC leads have phases φα\varphi_\alpha (α=0,…,N−1\alpha= 0,\ldots,N-1), while the MM resonant ABSs are represented by single-level QDs with on-site energies ϵi\epsilon_i (i=1,…,Mi = 1,\ldots,M). Dot ii couples to lead α\alpha with amplitude wiαw_{i\alpha}; dots ii and jj may be coupled by normal hopping rijr_{ij}, and SC leads α\alpha and β\beta may be directly coupled by tαβt_{\alpha\beta}.

In the Nambu representation, the full BdG Hamiltonian has the block form

HBdG=(H^DH^DSH^DS†H^S)(29)\mathcal{H}_{\mathrm{BdG}} = \begin{pmatrix} \hat{H}_{D} & \hat{H}_{DS} \\ \hat{H}_{DS}^{\dagger} & \hat{H}_{S} \end{pmatrix} \tag*{(29)}

where H^D\hat{H}_{D} is 2M×2M2M \times2M dot block, H^S\hat{H}_{S} is 2N×2N2N \times2N SC lead block, and H^DS\hat{H}_{DS} is 2M×2N2M \times2N dot-leads block. We use τ^i\hat{\tau}_i for Pauli matrices in Nambu space. The blocks are given by

(H^D)ij=(ϵiδij+rij)τ^3,(30)(\hat{H}_D)_{ij}=(\epsilon_i\delta_{ij}+r_{ij})\hat{\tau}_3, \tag*{(30)}
(H^S0)αβ=[ξkτ^3+Δαeiφατ^3τ^1]δαβ,(31)(\hat{H}_{S0})_{\alpha\beta}=\left[\xi_k\hat{\tau}_3+\Delta_\alpha e^{i\varphi_\alpha\hat{\tau}_3}\hat{\tau}_1\right]\delta_{\alpha\beta}, \tag*{(31)}
(H^S−S)αβ=tαβτ^3,(32)(\hat{H}_{S-S})_{\alpha\beta}=t_{\alpha\beta}\hat{\tau}_3, \tag*{(32)}
(H^DS)iα=wiατ^3.(33)(\hat{H}_{DS})_{i\alpha}=w_{i\alpha}\hat{\tau}_3. \tag*{(33)}

Here ξk=ℏ2k2/(2m)−EF\xi_k=\hbar^2k^2/(2m)-E_F, ϵi\epsilon_i is on-site dot energy, rijr_{ij} is the normal interdot hopping, tαβt_{\alpha\beta} is direct lead–lead hopping, and wiαw_{i\alpha} is the dot–lead tunneling amplitude; see Figure 8. We assume real, spin-preserving tunneling amplitudes.

Integrating out the SC leads gives the dot self-energy

Σ^D(E)=H^DSg^SS(E)H^DS†,(34)\hat{\Sigma}_D(E)=\hat{H}_{DS}\hat{g}_{SS}(E)\hat{H}_{DS}^{\dagger}, \tag*{(34)}

where, to first order in weak direct lead–lead tunneling,

g^SS(E)≃g^S(E)+g^S(E)H^S−Sg^S(E).(35)\hat{g}_{SS}(E)\simeq\hat{g}_S(E)+\hat{g}_S(E)\hat{H}_{S-S}\hat{g}_S(E). \tag*{(35)}

In the atomic limit, E≪ΔαE\ll\Delta_\alpha, the Green function of an isolated SC lead reduces to

(g^S)αβ=−πN0eiφατ^3τ^1δαβ.(36)(\hat{g}_S)_{\alpha\beta}=-\pi N_0e^{i\varphi_\alpha\hat{\tau}_3}\hat{\tau}_1\delta_{\alpha\beta}. \tag*{(36)}

The self-energy is then energy independent, and the effective dot Hamiltonian is

H^Deff=H^D+Σ^D.(37)\hat{H}^{\mathrm{eff}}_D=\hat{H}_D+\hat{\Sigma}_D. \tag*{(37)}

It contains the bare dot levels and hoppings together with proximity-induced pairing generated by virtual tunneling through the SC leads.

Effective Hamiltonians of the single-QD and double-QD models

We now apply Appendix § 11.§ 1 to the single-QD, normal-coupled double-QD, and SC-coupled double-QD models. The resulting effective Hamiltonians are summarized below.

Single-QD model

For the single-QD, one dot couples to four SC terminals, and neighboring terminals are directly coupled. Hence tαβ≠0t_{\alpha\beta} \ne0 only for β=α±1\beta= \alpha\pm1, with terminal indices understood modulo four. Keeping both lead–lead and dot–lead tunneling in the self-energy yields

H^Deff=d⋅τ^,d=(∑α=03Γαcos⁡φα−∑α=03Γαsin⁡φαϵ1−2∑α=03τα(α+1)Γα(α+1)cos⁡(φα−φα+1)).(38)\hat{H}^{\mathrm{eff}}_{D} = \mathbf{d} \cdot\hat{\boldsymbol{\tau}}, \qquad \mathbf{d} = \begin{pmatrix} \sum_{\alpha=0}^{3} \Gamma_{\alpha} \cos\varphi_{\alpha} \\ -\sum_{\alpha=0}^{3} \Gamma_{\alpha} \sin\varphi_{\alpha} \\ \epsilon_{1} - 2\sum_{\alpha=0}^{3} \tau_{\alpha(\alpha+1)}\Gamma_{\alpha(\alpha+1)}\cos(\varphi_{\alpha}-\varphi_{\alpha+1}) \end{pmatrix}. \tag*{(38)}

Here Γα=πN0w1α2\Gamma_{\alpha} = \pi N_{0}w_{1\alpha}^{2}, τα,α+1=πN0tα,α+1\tau_{\alpha,\alpha+1} = \pi N_{0}t_{\alpha,\alpha+1}, and Γα(α+1)=πN0w1αw1(α+1)\Gamma_{\alpha(\alpha+1)} = \pi N_{0}w_{1\alpha}w_{1(\alpha+1)}. Equation (A.10) reduces to Eq. (10) when tα(α+1)=tα(α−1)=tt_{\alpha(\alpha+1)} = t_{\alpha(\alpha-1)} = t and w1α=ww_{1\alpha} = w for all α=0,…,3\alpha= 0,\ldots,3.

Normal-coupled double-QD model

The normal-coupled double-QD contains two dots: dot 1 couples to terminals 0 and 1, and dot 2 to terminals 2 and 3. The dots are directly coupled by r12=rNr_{12} = r_{\mathrm{N}}. As in [58], the dot–lead self-energy gives the block Hamiltonian of Eq. (11). In the general case, ϵ1≠ϵ2\epsilon_{1} \ne\epsilon_{2}, γ1=∣γ1∣eiψ1=Γ10+Γ11eiφ1\gamma_{1} = \lvert\gamma_{1}\rvert e^{i\psi_{1}} = \Gamma_{10} + \Gamma_{11}e^{i\varphi_{1}}, and γ2=∣γ2∣eiψ2=Γ22eiφ2+Γ23eiφ3\gamma_{2} = \lvert\gamma_{2}\rvert e^{i\psi_{2}} = \Gamma_{22}e^{i\varphi_{2}} + \Gamma_{23}e^{i\varphi_{3}}, with Γiα=πN0wiα2\Gamma_{i\alpha} = \pi N_{0}w_{i\alpha}^{2}. The symmetric normal-coupled double-QD follows for ϵ1=ϵ2=ϵ\epsilon_{1} = \epsilon_{2} = \epsilon and Γ10=Γ11=Γ22=Γ23=Γ\Gamma_{10} = \Gamma_{11} = \Gamma_{22} = \Gamma_{23} = \Gamma.

SC-coupled double-QD model

The SC-coupled double-QD model extends the normal-coupled double-QD by allowing the central terminals 1 and 2 to couple to both dots. The shared leads generate a phase-dependent nonlocal SC coupling rSCr_{\mathrm{SC}} in addition to the normal hopping rNr_{\mathrm{N}}. The effective Hamiltonian retains the block structure of Eq. (11), with rSC=∣rSC∣eiψSC=ΓNL(1)eiφ1+ΓNL(2)eiφ2r_{\mathrm{SC}} = \lvert r_{\mathrm{SC}}\rvert e^{i\psi_{\mathrm{SC}}} = \Gamma_{\mathrm{NL}}^{(1)}e^{i\varphi_{1}} + \Gamma_{\mathrm{NL}}^{(2)}e^{i\varphi_{2}}, where ΓNL(1)=πN0w11w21\Gamma_{\mathrm{NL}}^{(1)} = \pi N_{0}w_{11}w_{21} and ΓNL(2)=πN0w12w22\Gamma_{\mathrm{NL}}^{(2)} = \pi N_{0}w_{12}w_{22}. The local pairings are γ1=Γ10+Γ11eiφ1+Γ12eiφ2\gamma_{1} = \Gamma_{10} + \Gamma_{11}e^{i\varphi_{1}} + \Gamma_{12}e^{i\varphi_{2}} and γ2=Γ21eiφ1+Γ22eiφ2+Γ23eiφ3\gamma_{2} = \Gamma_{21}e^{i\varphi_{1}} + \Gamma_{22}e^{i\varphi_{2}} + \Gamma_{23}e^{i\varphi_{3}} in Eq. (12). The antisymmetric model of the main text follows for ϵ1=−ϵ2=−ϵ\epsilon_{1} = -\epsilon_{2} = -\epsilon, w10=w11=w22=w23=wSw_{10} = w_{11} = w_{22} = w_{23} = w_{\mathrm{S}}, and w12=w21=wWw_{12} = w_{21} = w_{\mathrm{W}}, with ΓS=πN0wS2\Gamma_{\mathrm{S}} = \pi N_{0}w_{\mathrm{S}}^{2} and ΓW=πN0wW2\Gamma_{\mathrm{W}} = \pi N_{0}w_{\mathrm{W}}^{2}.

WPs in different QD models

This appendix summarizes the conditions for zero- and finite-energy WPs in the single-QD and double-QD models.

Zero-energy WPs in the single-QD model

With only one positive-energy ABS, the single-QD can host zero-energy WPs but not finite-energy WPs. Zero-energy WPs require d=0\mathbf{d} = 0, i.e., three simultaneous trigonometric conditions dx=dy=dz=0d_{x} = d_{y} = d_{z} = 0. For a generic single-QD these conditions are solved numerically; in the symmetric model, symmetry allows an analytic criterion [50]. The result is −8<ϵ/(τΓ)<0-8 < \epsilon/(\tau\Gamma) < 0.

WPs in double-QD models

Before specializing the double-QD model, we summarize the degeneracy conditions for a generic 4×44 \times4 double-QD BdG Hamiltonian.

A double-QD has a 4×44 \times4 BdG Hamiltonian with two particle-hole pairs, ±E1ABS\pm E_1^{\mathrm{ABS}} and ±E2ABS\pm E_2^{\mathrm{ABS}}, so its characteristic polynomial can be written as

det⁡(H^Deff−E)=E4−AE2+B,(39)\det(\hat{H}^{\mathrm{eff}}_D-E)=E^4-AE^2+B, \tag*{(39)}

with

A=(E1ABS)2+(E2ABS)2,B=(E1ABSE2ABS)2.(40)A=(E_1^{\mathrm{ABS}})^2+(E_2^{\mathrm{ABS}})^2,\qquad B=(E_1^{\mathrm{ABS}}E_2^{\mathrm{ABS}})^2. \tag*{(40)}

A zero-energy WP requires

B=0,(41)B=0, \tag*{(41)}

whereas a finite-energy WP requires

A2−4B=0.(42)A^2-4B=0. \tag*{(42)}

The coefficients AA and BB can be expressed in terms of the microscopic parameters by matching powers of EE in Eq. (B.1). The resulting formulas are lengthy, so below we give the normal-coupled double-QD limit and the purely SC-coupled limit which clarify why robust finite-energy WPs in our working model require both normal and SC interdot coupling.

Normal-coupled double-QD model

For the normal-coupled double-QD model, Eq. (B.3) becomes

(rN2−ϵ1ϵ2−∣γ1∣∣γ2∣)2+(∣γ1∣ϵ2−∣γ2∣ϵ1)2+4rN2∣γ1∣∣γ2∣cos⁡2[(ψ1−ψ2)/2]=0,(43)\left(r_N^2-\epsilon_1\epsilon_2-\lvert\gamma_1\rvert\lvert\gamma_2\rvert\right)^2+\left(\lvert\gamma_1\rvert\epsilon_2-\lvert\gamma_2\rvert\epsilon_1\right)^2+4r_N^2\lvert\gamma_1\rvert\lvert\gamma_2\rvert\cos^2\left[(\psi_1-\psi_2)/2\right]=0, \tag*{(43)}

The finite-energy crossing condition, Eq. (B.4), becomes

(∣γ1∣2−∣γ2∣2+ϵ12−ϵ22)2+4rN2[(ϵ1+ϵ2)2+(∣γ1∣cos⁡ψ1−∣γ2∣cos⁡ψ2)2+(∣γ1∣sin⁡ψ1−∣γ2∣sin⁡ψ2)2]=0.(44)\left(\lvert\gamma_1\rvert^2-\lvert\gamma_2\rvert^2+\epsilon_1^2-\epsilon_2^2\right)^2+4r_N^2\left[(\epsilon_1+\epsilon_2)^2+\left(\lvert\gamma_1\rvert\cos\psi_1-\lvert\gamma_2\rvert\cos\psi_2\right)^2+\left(\lvert\gamma_1\rvert\sin\psi_1-\lvert\gamma_2\rvert\sin\psi_2\right)^2\right]=0. \tag*{(44)}

As shown in Ref. [58], Eq. (B.5) can be satisfied over a broad parameter range. The finite-energy crossing condition is much more restrictive and, in particular, requires ϵ1=−ϵ2\epsilon_1=-\epsilon_2. Thus, robust finite-energy WPs are excluded in the generic normal-coupled double-QD case considered in the main text, apart from such fine-tuned limits.

Purely SC-coupled double-QD model

For the purely SC-coupled limit, the zero-energy condition is

(∣rSC∣2+ϵ1ϵ2−∣γ1∣∣γ2∣)2+(∣γ1∣ϵ2+∣γ2∣ϵ1)2+4∣γ1∣∣γ2∣∣rSC∣2sin⁡2(ψ1+ψ2−2ψSC2)=0.(45)\left(\lvert r_{\mathrm{SC}}\rvert^2+\epsilon_1\epsilon_2-\lvert\gamma_1\rvert\lvert\gamma_2\rvert\right)^2+\left(\lvert\gamma_1\rvert\epsilon_2+\lvert\gamma_2\rvert\epsilon_1\right)^2+4\lvert\gamma_1\rvert\lvert\gamma_2\rvert\lvert r_{\mathrm{SC}}\rvert^2\sin^2\left(\frac{\psi_1+\psi_2-2\psi_{\mathrm{SC}}}{2}\right)=0. \tag*{(45)}

The finite-energy condition is

(∣γ1∣2+ϵ12−∣γ2∣2−ϵ22)2+4∣rSC∣2[(ϵ1−ϵ2)2+(∣γ1∣cos⁡ψ1+∣γ2∣cos⁡(2ψSC−ψ2))2+(∣γ1∣sin⁡ψ1+∣γ2∣sin⁡(2ψSC−ψ2))2]=0.(46)\begin{aligned} &\left(\lvert\gamma_1\rvert^2+\epsilon_1^2-\lvert\gamma_2\rvert^2-\epsilon_2^2\right)^2\\ &\quad+4\lvert r_{\mathrm{SC}}\rvert^2\left[(\epsilon_1-\epsilon_2)^2+\left(\lvert\gamma_1\rvert\cos\psi_1+\lvert\gamma_2\rvert\cos(2\psi_{\mathrm{SC}}-\psi_2)\right)^2\right.\\ &\qquad\left.+\left(\lvert\gamma_1\rvert\sin\psi_1+\lvert\gamma_2\rvert\sin(2\psi_{\mathrm{SC}}-\psi_2)\right)^2\right]=0. \tag*{(46)} \end{aligned}

The finite-energy condition is again restrictive. For instance, it requires ϵ1=ϵ2\epsilon_1=\epsilon_2. As a result, satisfying Eq. (B.8) implies fine-tuning of the system’s parameters, whereas allowing both normal hopping and nonlocal SC coupling removes these fine-tuning constraints and produces finite-energy WPs over a broader parameter range, as used in the main text.

Linear response theory

This appendix gives the finite-temperature linear-response derivation used in § 4.

ABS susceptibility at finite temperature

First, we derive Eq. (16) from the Kubo formula. A phase variation enters the effective Hamiltonian as

δH^Deff(t)=∑α∂H^Deff∂φαδφα(t)=∑αℏ2eI^αδφα(t),(47)\delta\hat{H}_{D}^{\mathrm{eff}}(t)=\sum_{\alpha}\frac{\partial\hat{H}_{D}^{\mathrm{eff}}}{\partial\varphi_{\alpha}}\delta\varphi_{\alpha}(t)=\sum_{\alpha}\frac{\hbar}{2e}\hat{I}_{\alpha}\delta\varphi_{\alpha}(t), \tag*{(47)}

where

I^α=2eℏ∂H^Deff∂φα(48)\hat{I}_{\alpha}=\frac{2e}{\hbar}\frac{\partial\hat{H}_{D}^{\mathrm{eff}}}{\partial\varphi_{\alpha}} \tag*{(48)}

is the current operator in lead α\alpha. In frequency space, the Josephson relation gives

δφα(ω)=2eℏVα(ω)−iω,(49)\delta\varphi_{\alpha}(\omega)=\frac{2e}{\hbar}\frac{V_{\alpha}(\omega)}{-i\omega}, \tag*{(49)}

and therefore

δH^Deff(ω)=∑αI^αVα(ω)−iω.(50)\delta\hat{H}_{D}^{\mathrm{eff}}(\omega)=\sum_{\alpha}\hat{I}_{\alpha}\frac{V_{\alpha}(\omega)}{-i\omega}. \tag*{(50)}

Thus the ABS susceptibility in Eq. (15) is the current response to Vα(ω)/(−iω)V_{\alpha}(\omega)/(-i\omega), or equivalently the phase-response kernel multiplied by 2e/ℏ2e/\hbar.

To derive Eq. (16), we use the Matsubara technique followed by analytical continuation to real frequencies. We first recall the general Kubo response for an arbitrary operator A^\hat{A}. If the perturbation is

δH^(t)=B^f(t),(51)\delta\hat{H}(t)=\hat{B}f(t), \tag*{(51)}

then the linear response of A^\hat{A} is determined by the imaginary-time correlator

χAB(τ)=−⟨TτδA^(τ)δB^(0)⟩,(52)\chi_{AB}(\tau)=-\langle T_{\tau}\delta\hat{A}(\tau)\delta\hat{B}(0)\rangle, \tag*{(52)}

where

δA^=A^−⟨A^⟩,δB^=B^−⟨B^⟩.(53)\delta\hat{A}=\hat{A}-\langle\hat{A}\rangle,\qquad\delta\hat{B}=\hat{B}-\langle\hat{B}\rangle. \tag*{(53)}

Here A^\hat{A} is an arbitrary operator whose response is measured, while B^\hat{B} is the operator coupled to the external perturbation.

For the ABS susceptibility we take

A^=I^α,B^=I^β.(54)\hat{A}=\hat{I}_{\alpha},\qquad\hat{B}=\hat{I}_{\beta}. \tag*{(54)}

Therefore the Kubo part of the ABS response is

χαβABS(τ)=−⟨TτδI^α(τ)δI^β(0)⟩.(55)\chi_{\alpha\beta}^{\mathrm{ABS}}(\tau)=-\langle T_{\tau}\delta\hat{I}_{\alpha}(\tau)\delta\hat{I}_{\beta}(0)\rangle. \tag*{(55)}

In Matsubara-frequency representation this gives, for iΩm≠0i\Omega_m\ne0,

χαβABS(iΩm)=∑k≠k′Ikk′αIk′kβnF(Ek)−nF(Ek′)iℏΩm+Ek−Ek′.(56)\chi_{\alpha\beta}^{\mathrm{ABS}}(i\Omega_m)=\sum_{k\ne k'}I_{kk'}^{\alpha}I_{k'k}^{\beta}\frac{n_{\mathrm{F}}(E_k)-n_{\mathrm{F}}(E_{k'})}{i\hbar\Omega_m+E_k-E_{k'}}. \tag*{(56)}

The usual analytical continuation from nonzero Matsubara frequencies,

iΩm⟶ω+i0,(57)i\Omega_m\longrightarrow\omega+i0, \tag*{(57)}

then gives Eq. (16) from the main text at ω≠0\omega\ne0.

Reproduction of the transconductance quantization at T=0T = 0

Finally, we verify that the finite-temperature transconductance reduces to the quantized zero-temperature transconductance. The first-order term in frequency is obtained by expanding Eq. (16) with respect to ω\omega:

χαβ(1)=−iℏ∑k≠k′nF(Ek)−nF(Ek′)(Ek−Ek′+i0)2Ikk′αIk′kβ.(58)\chi_{\alpha\beta}^{(1)}=-i\hbar\sum_{k\ne k'}\frac{n_{\mathrm{F}}(E_k)-n_{\mathrm{F}}(E_{k'})}{(E_k-E_{k'}+i0)^2}I_{kk'}^{\alpha}I_{k'k}^{\beta}. \tag*{(58)}

Generally, it can be written in the form of Eq. (18) with the finite-temperature Berry curvature introduced by Eq. (19).

At zero temperature, nF(Ek)=θ(−Ek)n_{\mathrm{F}}(E_k)=\theta(-E_k), and therefore

B0αβ=2Im⁡∑Ek<0Ek′>0⟨k∣∂αHDeff∣k′⟩⟨k′∣∂βHDeff∣k⟩(Ek−Ek′)2=∑Ek<0Bkαβ.(59)B_0^{\alpha\beta}=2\operatorname{Im}\sum_{\substack{E_k<0\\E_{k'}>0}}\frac{\langle k|\partial_{\alpha}H_{\mathrm{D}}^{\mathrm{eff}}|k'\rangle\langle k'|\partial_{\beta}H_{\mathrm{D}}^{\mathrm{eff}}|k\rangle}{(E_k-E_{k'})^2}=\sum_{E_k<0}B_k^{\alpha\beta}. \tag*{(59)}

Thus B0αβ=BαβB_0^{\alpha\beta}=B^{\alpha\beta}, and the transconductance becomes quantized, as in Eq. (14).

Finite-temperature Berry-curvature maps

This appendix shows representative finite-temperature Berry-curvature maps whose BZ integrals give the finite-temperature Chern number: the symmetric single-QD and normal-coupled double-QD models for ground-state topology, the antisymmetric SC-coupled double-QD for individual-ABS topology, and a topologically trivial normal-coupled double-QD close to its transition. These maps distinguish the two thermal mechanisms discussed in § 4.3.

Finite-temperature Berry curvature of symmetric single-QD and normal-coupled double-QD models

Figures 9 and 10 show the finite-temperature Berry curvature of the symmetric single-QD and normal-coupled double-QD models, respectively. As discussed in § 5, the thermal evolution differs between ground-state topological and trivial values of φ3\varphi_3.

In the ground-state topological region [Figure 9(a)–(Figure 9c) and 10(a)–(c)], the Berry curvature integrates to CT12=1C_{\mathrm{T}}^{12}=1. Increasing temperature predominantly suppresses the regions of positive Berry curvature, while negative contributions change more weakly. The resulting loss of positive weight produces the monotonic decrease of CT12C_{\mathrm{T}}^{12} shown in Figure 5(b) and (Figure 5e).

Zero- and finite-temperature Berry-curvature maps for the symmetric single-QD model

Figure 9. Zero- and finite-temperature Berry-curvature maps for the symmetric single-QD model with ε/Γ=−0.2\varepsilon/\Gamma=-0.2 and τ=0.1\tau=0.1. Panels (a) and (d) show the zero-temperature Berry curvature B12B^{12}; panels (b), (c), (e), and (f) show BT12B_{T}^{12} at the temperatures indicated. The top row uses φ3=4\varphi_{3}=4, inside the ground-state topological region [corresponding to Figure 5(b)], and shows predominantly thermal suppression of the regions of positive Berry curvature. The bottom row uses φ3=5.5\varphi_{3}=5.5, in the trivial region [Figure 5(c)], where regions of positive and negative Berry curvature are suppressed at different rates.

In the ground-state trivial region [Figure 9(d)–(Figure 9f) and 10(d)–(f)], positive- and negative-Berry-curvature contributions cancel at T=0T=0, giving C12=0C^{12}=0. At low temperature, the negative contribution is suppressed more strongly on average, so CT12C_{\mathrm{T}}^{12} initially increases. At higher temperature, the relative suppression changes and CT12C_{\mathrm{T}}^{12} passes through a maximum before decreasing, as in Figure 5(c) and (Figure 5f).

Although the single-QD and normal-coupled double-QD models show similar CT12(φ3)C_{\mathrm{T}}^{12}(\varphi_3) trends, their finite-temperature Berry-curvature maps differ. As discussed in § 4.3.1, Eq. (21) makes the single-QD response purely suppressive. By contrast, the normal-coupled double-QD model can show phase-resolved enhancement [the appearance of the light-blue region in Figure 10(f), which is absent in Figure 10(d)] because several ABS contributions with different signs are thermally reweighted.

Zero- and finite-temperature Berry-curvature maps

Figure 10. Zero- and finite-temperature Berry-curvature maps for the symmetric normal-coupled double-QD model with ϵ/Γ=1\epsilon/\Gamma= 1 and rN/Γ=1.5r_N/\Gamma= 1.5. Panels (a) and (d) show B12B^{12} at T=0T = 0; panels (b), (c), (e), and (f) show BT12B_T^{12} at the indicated temperatures. The top row uses φ3=4\varphi_3 = 4, inside the ground-state topological region [Figure 5(e)], where the dominant positive-Berry-curvature weight is thermally suppressed. The bottom row uses φ3=5.5\varphi_3 = 5.5, in the trivial region [Figure 5(f)], where negative and positive regions are suppressed at different rates.

Finite-temperature Berry curvature of an antisymmetric SC-coupled double-QD model

Figure 11 shows the individual-ABS topological regime of the antisymmetric SC-coupled double-QD, corresponding to Figure 6(b). At T=0T=0, the Berry curvature is weak and integrates to C12=0C^{12}=0. Finite temperature generates finite-temperature Berry curvature through the same-sector contribution [Figure 11(b)], which strengthens [Figure 11(c)] and then weakens at higher TT [Figure 11(d)]. This evolution produces the maximum and subsequent decay of CT12(T)C_{T}^{12}(T) in Figure 6(b).

Figure 11. Zero- and finite-temperature Berry-curvature maps for the antisymmetric SC-coupled double-QD model with ϵ/ΓS=1.2\epsilon/\Gamma_{\mathrm{S}} = 1.2, ΓW/ΓS=0.3\Gamma_{\mathrm{W}}/\Gamma_{\mathrm{S}} = 0.3, rN/ΓS=0.1r_{\mathrm{N}}/\Gamma_{\mathrm{S}} = 0.1, and φ3=4\varphi_3 = 4, corresponding to Figure 6(b). Panel (a) shows B12B^{12} at T=0T = 0; panels (b)–(d) show BT12B_{\mathrm{T}}^{12} at increasing temperatures. The same-sector contribution produces the dominant temperature-induced change in the Berry curvature, which reaches a maximum and then decreases at higher temperature.

Finite-temperature Berry curvature of a topologically trivial normal-coupled double-QD model

Finally, Figure 12 shows the trivial normal-coupled double-QD corresponding to the black curve in Figure 6(d). At T=0T=0, regions of positive and negative Berry curvature cancel exactly, giving C12=0C^{12}=0. At low TT, the negative contribution is suppressed more strongly, so CT12C_{T}^{12} initially increases; at higher TT, the relative suppression changes and CT12C_{T}^{12} passes through a maximum before decreasing. Thus the dominant mechanism is unequal reweighting of pre-existing Berry curvature, rather than the thermally generated same-sector finite-temperature Berry curvature characteristic of the individual-ABS topological case.

Zero- and finite-temperature Berry-curvature maps

Figure 12. Zero- and finite-temperature Berry-curvature maps for the symmetric normal-coupled double-QD model in the trivial regime. Panel (a) shows B12B^{12} at T=0T = 0; panels (b) and (c) show BT12B_{\mathrm{T}}^{12} at the indicated temperatures. The model parameters are ϵ/Γ=1\epsilon/\Gamma= 1, rN/Γ=0.8r_{\mathrm{N}}/\Gamma= 0.8, and φ3=5\varphi_3 = 5, which corresponds to the black curve in Figure 6(d). The dominant effect is unequal thermal reweighting of the regions of negative and positive Berry curvature.

References

  1. [1]A. Bansil, H. Lin and T. Das, Colloquium: Topological band theory, Rev. Mod. Phys. 88, 021004 (2016), doi:10.1103/RevModPhys.88.021004.DOI
  2. [2]M. Z. Hasan and C. L. Kane, Colloquium: Topological insulators, Rev. Mod. Phys. 82, 3045 (2010), doi:10.1103/RevModPhys.82.3045.DOI
  3. [3]B. Yan and C. Felser, Topological materials: Weyl semimetals, Annu. Rev. Condens. Matter Phys. 8, 337 (2017), doi:10.1146/annurev-conmatphys-031016-025458.DOI
  4. [4]M. Sato and Y. Ando, Topological superconductors: A review, Rep. Prog. Phys. 80(7), 076501 (2017), doi:10.1088/1361-6633/aa6ac7.DOI
  5. [5]K. v. Klitzing, G. Dorda and M. Pepper, New method for high-accuracy determination of the fine-structure constant based on quantized Hall resistance, Phys. Rev. Lett. 45, 494 (1980), doi:10.1103/PhysRevLett.45.494.DOI
  6. [6]D. J. Thouless, M. Kohmoto, M. P. Nightingale and M. den Nijs, Quantized Hall conductance in a two-dimensional periodic potential, Phys. Rev. Lett. 49, 405 (1982), doi:10.1103/PhysRevLett.49.405.DOI
  7. [7]T. Ozawa and H. M. Price, Topological quantum matter in synthetic dimensions, Nat. Rev. Phys. 1(5), 349 (2019), doi:10.1038/s42254-019-0045-3.DOI
  8. [8]J. Argüello-Luengo, U. Bhattacharya, A. Celi, R. W. Chhajlany, T. Grass, M. Płodzień, D. Rakshit, T. Salamon, P. Stornati, L. Tarruell et al., Synthetic dimensions for topological and quantum phases, Commun. Phys. 7(1), 143 (2024), doi:10.1038/s42005-024-01636-3.
  9. [9]T. Kitagawa, E. Berg, M. Rudner and E. Demler, Topological characterization of periodically driven quantum systems, Phys. Rev. B 82, 235114 (2010), doi:10.1103/PhysRevB.82.235114.DOI
  10. [10]M. S. Rudner and N. H. Lindner, Band structure engineering and non-equilibrium dynamics in Floquet topological insulators, Nat. Rev. Phys. 2(5), 229 (2020), doi:10.1038/s42254-020-0170-z.DOI
  11. [11]A. B. Khanikaev, S. Hossein Mousavi, W.-K. Tse, M. Kargarian, A. H. MacDonald and G. Shvets, Photonic topological insulators, Nat. Mater. 12(3), 233 (2013), doi:10.1038/nmat3520.DOI
  12. [12]L. Lu, J. D. Joannopoulos and M. Soljačić, Topological photonics, Nat. Photonics 8(11), 821 (2014), doi:10.1038/nphoton.2014.248.DOI
  13. [13]V. Fatemi, A. R. Akhmerov and L. Bretheau, Weyl Josephson circuits, Phys. Rev. Res. 3, 013288 (2021), doi:10.1103/PhysRevResearch.3.013288.DOI
  14. [14]L. Peyruchat, J. Griesmar, J.-D. Pillet and Ç. Ö. Girit, Transconductance quantization in a topological Josephson tunnel junction circuit, Phys. Rev. Res. 3, 013289 (2021), doi:10.1103/PhysRevResearch.3.013289.
  15. [15]T. Herrig and R.-P. Riwar, Cooper-pair transistor as a minimal topological quantum circuit, Phys. Rev. Res. 4, 013038 (2022), doi:10.1103/PhysRevResearch.4.013038.DOI
  16. [16]Y. E. Kraus, Z. Ringel and O. Zilberberg, Four-dimensional quantum Hall effect in a two-dimensional quasicrystal, Phys. Rev. Lett. 111, 226401 (2013), doi:10.1103/PhysRevLett.111.226401.DOI
  17. [17]H. M. Price, O. Zilberberg, T. Ozawa, I. Carusotto and N. Goldman, Four-dimensional quantum Hall effect with ultracold atoms, Phys. Rev. Lett. 115, 195303 (2015), doi:10.1103/PhysRevLett.115.195303.DOI
  18. [18]M. Lohse, C. Schweizer, H. M. Price, O. Zilberberg and I. Bloch, Exploring 4D quantum Hall physics with a 2D topological charge pump, Nature 553(7686), 55 (2018), doi:10.1038/nature25000.DOI
  19. [19]O. Zilberberg, S. Huang, J. Guglielmon, M. Wang, K. P. Chen, Y. E. Kraus and M. C. Rechtsman, Photonic topological boundary pumping as a probe of 4D quantum Hall physics, Nature 553(7686), 59 (2018), doi:10.1038/nature25011.DOI
  20. [20]J. C. Cuevas and H. Pothier, Voltage-induced Shapiro steps in a superconducting multiterminal structure, Phys. Rev. B 75, 174513 (2007), doi:10.1103/PhysRevB.75.174513.DOI
  21. [21]M. Houzet and P. Samuelsson, Multiple Andreev reflections in hybrid multiterminal junctions, Phys. Rev. B 82, 060517(R) (2010), doi:10.1103/PhysRevB.82.060517.DOI
  22. [22]A. Freyn, B. Douçot, D. Feinberg and R. Mélin, Production of nonlocal quartets and phase-sensitive entanglement in a superconducting beam splitter, Phys. Rev. Lett. 106, 257005 (2011), doi:10.1103/PhysRevLett.106.257005.DOI
  23. [23]A. H. Pfeffer, J. E. Duvauchelle, H. Courtois, R. Mélin, D. Feinberg and F. Lefloch, Subgap structure in the conductance of a three-terminal Josephson junction, Phys. Rev. B 90, 075401 (2014), doi:10.1103/PhysRevB.90.075401.DOI
  24. [24]Y. Cohen, Y. Ronen, J.-H. Kang, M. Heiblum, D. Feinberg, R. Mélin and H. Shtrikman, Nonlocal supercurrent of quartets in a three-terminal Josephson junction, Proc. Natl. Acad. Sci. U.S.A. 115(27), 6991 (2018), doi:10.1073/pnas.1800044115.DOI
  25. [25]A. W. Draelos, M.-T. Wei, A. Seredinski, H. Li, Y. Mehta, K. Watanabe, T. Taniguchi, I. V. Borzenets, F. Amet and G. Finkelstein, Supercurrent flow in multiterminal graphene Josephson junctions, Nano Lett. 19(2), 1039 (2019), doi:10.1021/acs.nanolett.8b04330.DOI
  26. [26]J.-D. Pillet, V. Benzoni, J. Griesmar, J.-L. Smirr and Ç. Ö. Girit, Nonlocal Josephson effect in Andreev molecules, Nano Lett. 19(10), 7138 (2019), doi:10.1021/acs.nanolett.9b02686.DOI
  27. [27]G. V. Graziano, J. S. Lee, M. Pendharkar, C. J. Palmstrøm and V. S. Pribiag, Transport studies in a gate-tunable three-terminal Josephson junction, Phys. Rev. B 101, 054510 (2020), doi:10.1103/PhysRevB.101.054510.DOI
  28. [28]N. Pankratova, H. Lee, R. Kuzmin, K. Wickramasinghe, W. Mayer, J. Yuan, M. G. Vavilov, J. Shabani and V. E. Manucharyan, Multiterminal Josephson effect, Phys. Rev. X 10, 031051 (2020), doi:10.1103/PhysRevX.10.031051.DOI
  29. [29]E. G. Arnault, T. F. Larson, A. Seredinski, L. Zhao, S. Idris, A. McConnell, K. Watanabe, T. Taniguchi, I. Borzenets, F. Amet et al., Multiterminal inverse ac Josephson effect, Nano Lett. 21(22), 9668 (2021), doi:10.1021/acs.nanolett.1c03474.DOI
  30. [30]K.-F. Huang, Y. Ronen, R. Mélin, D. Feinberg, K. Watanabe, T. Taniguchi and P. Kim, Evidence for 4e charge of Cooper quartets in a biased multi-terminal graphene-based Josephson junction, Nat. Commun. 13(1), 3032 (2022), doi:10.1038/s41467-022-30732-7.DOI
  31. [31]G. V. Graziano, M. Gupta, M. Pendharkar, J. T. Dong, C. P. Dempsey, C. Palmstrøm and V. S. Pribiag, Selective control of conductance modes in multi-terminal Josephson junctions, Nat. Commun. 13(1), 5933 (2022), doi:10.1038/s41467-022-33682-2.DOI
  32. [32]S. Matsuo, J. S. Lee, C.-Y. Chang, Y. Sato, K. Ueda, C. J. Palmstrøm and S. Tarucha, Observation of nonlocal Josephson effect on double InAs nanowires, Commun. Phys. 5(1), 221 (2022), doi:10.1038/s42005-022-00994-0.DOI
  33. [33]M. Gupta, G. V. Graziano, M. Pendharkar, J. T. Dong, C. P. Dempsey, C. Palmstrøm and V. S. Pribiag, Gate-tunable superconducting diode effect in a three-terminal Josephson device, Nat. Commun. 14(1), 3078 (2023), doi:10.1038/s41467-023-38856-0.DOI
  34. [34]D. Z. Haxell, M. Coraiola, M. Hinderling, S. C. ten Kate, D. Sabonis, A. E. Svetogorov, W. Belzig, E. Cheah, F. Krizek, R. Schott et al., Demonstration of the nonlocal Josephson effect in Andreev molecules, Nano Lett. 23(16), 7532 (2023), doi:10.1021/acs.nanolett.3c02066.DOI
  35. [35]M. Coraiola, D. Z. Haxell, D. Sabonis, H. Weisbrich, A. E. Svetogorov, M. Hinderling, S. C. ten Kate, E. Cheah, F. Krizek, R. Schott et al., Phase-engineering the Andreev band structure of a three-terminal Josephson junction, Nat. Commun. 14(1), 6784 (2023), doi:10.1038/s41467-023-42356-6.DOI
  36. [36]S. Matsuo, T. Imoto, T. Yokoyama, Y. Sato, T. Lindemann, S. Gronin, G. C. Gardner, S. Nakosai, Y. Tanaka, M. J. Manfra et al., Phase-dependent Andreev molecules and superconducting gap closing in coherently-coupled Josephson junctions, Nat. Commun. 14(1), 8271 (2023), doi:10.1038/s41467-023-44111-3.DOI
  37. [37]M. Coraiola, A. E. Svetogorov, D. Z. Haxell, D. Sabonis, M. Hinderling, S. C. ten Kate, E. Cheah, F. Krizek, R. Schott, W. Wegscheider et al., Flux-tunable Josephson diode effect in a hybrid four-terminal Josephson junction, ACS Nano 18(12), 9221 (2024), doi:10.1021/acsnano.4c01642.DOI
  38. [38]C. G. Prosko, W. D. Huisman, I. Kulesh, D. Xiao, C. Thomas, M. J. Manfra and S. Goswami, Flux-tunable Josephson effect in a four-terminal junction, Phys. Rev. B 110, 064518 (2024), doi:10.1103/PhysRevB.110.064518.DOI
  39. [39]P. Virtanen and T. T. Heikkilä, Nonreciprocal Josephson linear response, Phys. Rev. Lett. 132, 046002 (2024), doi:10.1103/PhysRevLett.132.046002.DOI
  40. [40]S. A. Kahn, L. Stampfer, S. Marti-Sanchez, D. Olsteins, D. J. Carrad, T. Jansen, J. Johansson, J. Arbiol, P. Krogstrup and T. S. Jespersen, Phase controlled multi-terminal Josephson junction in ternary hybrid nanowire, arXiv:2511.04719 (2025), doi:10.48550/arXiv.2511.04719.DOI
  41. [41]S. Annabi, H. Riechert, K. Watanabe, T. Taniguchi, J. Griesmar, E. Arrighi, L. Bretheau and J.-D. Pillet, Fermion parity of an Andreev molecule probed by nonlocal Josephson effect, arXiv:2607.15786 (2026), doi:10.48550/arXiv.2607.15786.DOI
  42. [42]T. Yokoyama and Y. V. Nazarov, Singularities in the Andreev spectrum of a multiterminal Josephson junction, Phys. Rev. B 92, 155437 (2015), doi:10.1103/PhysRevB.92.155437.DOI
  43. [43]R.-P. Riwar, M. Houzet, J. S. Meyer and Y. V. Nazarov, Multi-terminal Josephson junctions as topological matter, Nat. Commun. 7(1), 11167 (2016), doi:10.1038/ncomms11167.DOI
  44. [44]E. Eriksson, R.-P. Riwar, M. Houzet, J. S. Meyer and Y. V. Nazarov, Topological transconductance quantization in a four-terminal Josephson junction, Phys. Rev. B 95, 075417 (2017), doi:10.1103/PhysRevB.95.075417.DOI
  45. [45]J. S. Meyer and M. Houzet, Nontrivial Chern numbers in three-terminal Josephson junctions, Phys. Rev. Lett. 119, 136807 (2017), doi:10.1103/PhysRevLett.119.136807.DOI
  46. [46]H.-Y. Xie, M. G. Vavilov and A. Levchenko, Topological Andreev bands in three-terminal Josephson junctions, Phys. Rev. B 96, 161406(R) (2017), doi:10.1103/PhysRevB.96.161406.DOI
  47. [47]H.-Y. Xie, M. G. Vavilov and A. Levchenko, Weyl nodes in Andreev spectra of multiterminal Josephson junctions: Chern numbers, conductances, and supercurrents, Phys. Rev. B 97, 035443 (2018), doi:10.1103/PhysRevB.97.035443.DOI
  48. [48]H.-Y. Xie and A. Levchenko, Topological supercurrents interaction and fluctuations in the multiterminal Josephson effect, Phys. Rev. B 99, 094519 (2019), doi:10.1103/PhysRevB.99.094519.DOI
  49. [49]E. V. Repin, Y. Chen and Y. V. Nazarov, Topological properties of multiterminal superconducting nanostructures: Effect of a continuous spectrum, Phys. Rev. B 99, 165414 (2019), doi:10.1103/PhysRevB.99.165414.DOI
  50. [50]R. L. Klees, G. Rastelli, J. C. Cuevas and W. Belzig, Microwave spectroscopy reveals the quantum geometric tensor of topological Josephson matter, Phys. Rev. Lett. 124, 197002 (2020), doi:10.1103/PhysRevLett.124.197002.DOI
  51. [51]R. L. Klees, J. C. Cuevas, W. Belzig and G. Rastelli, Ground-state quantum geometry in superconductor–quantum dot chains, Phys. Rev. B 103, 014516 (2021), doi:10.1103/PhysRevB.103.014516.DOI
  52. [52]Y. Chen and Y. V. Nazarov, Weyl point immersed in a continuous spectrum: An example from superconducting nanostructures, Phys. Rev. B 104, 104506 (2021), doi:10.1103/PhysRevB.104.104506.DOI
  53. [53]J. S. Meyer and M. Houzet, Conductance quantization in topological josephson trijunctions, Phys. Rev. B 103, 174504 (2021), doi:10.1103/PhysRevB.103.174504.DOI
  54. [54]H. Weisbrich, R. Klees, G. Rastelli and W. Belzig, Second Chern number and non-Abelian Berry phase in topological superconducting systems, PRX Quantum 2, 010310 (2021), doi:10.1103/PRXQuantum.2.010310.DOI
  55. [55]H.-Y. Xie, J. Hasan and A. Levchenko, Non-Abelian monopoles in the multiterminal Josephson effect, Phys. Rev. B 105, L241404 (2022), doi:10.1103/PhysRevB.105.L241404.DOI
  56. [56]H. Barakov and Y. V. Nazarov, Abundance of Weyl points in semiclassical multiterminal superconducting nanostructures, Phys. Rev. B 107, 014507 (2023), doi:10.1103/PhysRevB.107.014507.DOI
  57. [57]L. Peralta Gavensky, G. Usaj and C. A. Balseiro, Multi-terminal Josephson junctions: A road to topological flux networks, EPL 141(3), 36001 (2023), doi:10.1209/0295-5075/acb2f6.DOI
  58. [58]L. Teshler, H. Weisbrich, J. Sturm, R. L. Klees, G. Rastelli and W. Belzig, Ground state topology of a four-terminal superconducting double quantum dot, SciPost Phys. 15, 214 (2023), doi:10.21468/SciPostPhys.15.5.214.DOI
  59. [59]T. Antonelli, M. Coraiola, D. C. Ohnmacht, A. E. Svetogorov, D. Sabonis, S. C. ten Kate, E. Cheah, F. Krizek, R. Schott, J. C. Cuevas, W. Belzig, W. Wegscheider et al., Exploring the energy spectrum of a four-terminal Josephson junction: Toward topological Andreev band structures, Phys. Rev. X 15, 031066 (2025), doi:10.1103/qd3y-f912.DOI
  60. [60]D. C. Ohnmacht, V. Wilhelm, H. Weisbrich and W. Belzig, Non-Hermitian topology in multiterminal superconducting junctions, Phys. Rev. Lett. 134(15), 156601 (2025), doi:10.1103/PhysRevLett.134.156601.DOI
  61. [61]D. C. Ohnmacht, V. Wilhelm and W. Belzig, Reflectionless modes as a source of Weyl nodes in multiterminal Josephson junctions, arXiv:2503.10874 (2025), doi:10.48550/arXiv.2503.10874.DOI
  62. [62]W. Jung, S. Jin, S. Park, S.-H. Shin, K. Watanabe, T. Taniguchi, G. Y. Cho and G.-H. Lee, Tunneling spectroscopy of Andreev bands in multiterminal graphene-based Josephson junctions, Sci. Adv. 11(21), eads0342 (2025), doi:10.1126/sciadv.ads0342.DOI
  63. [63]S. Chakraborty, A. Mukhopadhyay, U. Khanna and S. Das, Synthetic flat bands, hierarchical topology, and phase-fluctuation-insensitive quantized transconductance in Josephson junctions, arXiv:2605.02095 (2026), doi:10.48550/arXiv.2605.02095.DOI
  64. [64]P. Zalom, G. Diniz, D. C. Ohnmacht and W. Belzig, Symmetry-induced Weyl nodes in interacting multi-terminal Josephson junctions, arXiv:2608.21200 (2026), doi:10.48550/arXiv.2608.21200.DOI
  65. [65]T. Fukui, Y. Hatsugai and H. Suzuki, Chern numbers in discretized Brillouin zone: Efficient method of computing spin Hall conductances, J. Phys. Soc. Jpn. 74(6), 1674 (2005), doi:10.1143/JPSJ.74.1674.DOI
  66. [66]P. D. Kurilovich, V. D. Kurilovich, V. Fatemi, M. H. Devoret and L. I. Glazman, Microwave response of an Andreev bound state, Phys. Rev. B 104, 174517 (2021), doi:10.1103/PhysRevB.104.174517.DOI

Paper details

vlnc:W_dbs4zzh6
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