betaFeedback

WorksPhysics and Astronomy

Gate-tunable supercurrent switching in a three-terminal quantum dot transistor

Abstract

We propose and analyze a three-terminal superconducting transistor switch in which a gate-defined quantum dot region interrupts a flux-controlled loop. The output supercurrent is switched off completely when the two loop arms couple to the dot with equal tunneling strengths and half a flux quantum threads the loop. The device then operates as a symmetric SQUID at half flux. The condition follows from a single geometric factor χχ, which reduces the three-terminal problem to an equivalent two-terminal Anderson impurity model and makes the switching mechanism independent of the Coulomb interaction, the dot level position, the total hybridization and temperature. These enter only through a prefactor and thus decide how tolerant the device is to imperfect barrier tuning. Numerical renormalization group calculations across the full coupling parameter space show that the resonance is embedded in a broad region of elevated switching ratio at intermediate hybridization Γ≈ΔΓ\approx Δ, while both weaker and stronger coupling reduce it to a thin isolated line. Finite temperature smooths the underlying singlet-doublet transition and lowers the supercurrent amplitude, but affects the open and closed states comparably and leaves the switching ratio largely intact. Within a zero-bandwidth two-level model, a second dot level is shown to open a leakage channel that grows as the two levels approach in energy. Raising the output barrier and the realistic presence of inter-level Coulomb repulsion, however, suppress the leakage.

Introduction

Semiconductor field-effect transistors form the foundation of present-day digital electronics, with physical gate lengths in high-volume manufacturing reaching ≈20\approx20 nm while maintaining on/off current ratios of 10510^5–10710^7 between the conducting and blocking states [1, 2]. However, both the switching and leakage currents of such devices dissipate power, and this constraint severely limits their applications at cryogenic temperatures. A dilution refrigerator provides only hundreds of microwatts of cooling power at 100 mK, while a single complementary metal–oxide–semiconductor (CMOS) control channel dissipates milliwatts. Control electronics for solid-state qubits is therefore kept mostly at room temperature and connects to the cold stages through coaxial lines that conduct heat and carry thermal photons downward. Cryogenic CMOS controllers relocate part of this electronics to the 4 K stage, where more cooling power is available, but this stage remains hot on the scale of the qubit and the signal lines must still be attenuated on their way down [3]. Superconducting (SC) circuit elements operating natively at the qubit stage would remove this thermal boundary entirely, making SC classical logic a natural candidate for integration with quantum computation nodes. Such elements operate with supercurrents that must be controllably suppressed (the OFF state) or allowed to flow (the ON state), with a switching ratio of at least several orders of magnitude required for reliable control. While ratios of this size have been reported in gate-controlled suppression of the supercurrent in SC heterostructures [4, 5, 6, 7, 8, 9, 10, 11], the microscopic origin of such gate-controlled-supercurrent effects remains under debate [12, 13, 14, 15, 16]; see Ref.[17] for a recent review. SC elements based on Josephson junctions, which switch via the phase rather than the condensate, are thus a more natural pathway to overcome both limitations.

Josephson junctions are often realized either as superconductor–insulator–superconductor (SIS) tunnel junctions, such as Al/AlOx_x/Al stacks [18], or as superconductor–normal-metal–superconductor (SNS) junctions, in which the supercurrent is carried through a weak link [19]. Electrostatic control of the weak link was first demonstrated in planar junctions defined in GaAs/AlGaAs [20] and InAs [21] two-dimensional electron gases (2DEGs) with related Josephson field-effect transistor implementations demonstrated subsequently [22]. Later developments involved carbon nanotubes [23] and semiconductor nanowires [24]. A complementary approach employs electrostatic gates to control Josephson transport through quantum dots by tuning discrete levels and/or their tunnel coupling to superconducting electrodes [25, 26, 27, 28]. Among these, planar 2DEGs combine a proximity-induced gap and high mobility with lithographic flexibility; InAs quantum wells with epitaxially grown Al are the most developed realization [29, 30], with InSb [31], HgTe [32] and planar Ge [33, 34, 35] representing emerging alternatives. The versatility and lithographic flexibility of planar 2DEG platforms extend naturally to multi-terminal geometries [36, 37, 38, 39, 40, 41], which give access to phenomena with no counterpart in two-terminal devices. These include coherent transport of multiple Cooper pairs [36, 42, 43, 44, 45] and the engineering of Andreev bound state (ABS) bands in a multi-dimensional phase space [39, 40, 46, 47, 48, 49, 50, 51], as well as supercurrent diode and transistor effects. The latter were recently identified as phase-controlled phenomena in three-terminal structures whose central region hosts a single electronic level, modeled as a quantum dot (QD) with non-negligible Coulomb interaction [52, 53, 54].

The reported effect was, however, restricted to a single configuration of the QD and its couplings to the SC leads. The resulting switching ratio reached at most one order of magnitude, well short of the requirement set out above. Whether this reflects an intrinsic bound of the multi-terminal mechanism or merely the particular parameters chosen remains open. The device nevertheless offers two physically distinct sets of parameters, both addressable in realistic setups. While the SC phases of the leads, controlled by magnetic fluxes, underlie the transistor effect reported in Ref. [54], the electrostatic properties of the QD, together with its tunnel couplings to the SC leads, remain unexplored in this context.

The paper is organized as follows. In § 2, we introduce the three-terminal device architecture in which two SC terminals form a flux-controlled loop interrupted by a QD region, and review in § 2 A the single-level theory of the multi-terminal Anderson model that allows to reduce the complexity of supercurrent calculations by introducing the concept of the geometric factor χ\chi. Section III contains the central analytic result of the paper: a supercurrent switching mechanism arising when both arms of the loop couple to the QD with equal tunneling strengths. The tolerance of this resonance condition to imperfect barrier tuning is examined in Sec. IV across the full coupling parameter space, where Sec. IV A traces the observed structures back to the underlying singlet-doublet quantum phase transitions and Sec. IV B addresses their temperature dependence. Deviations from the ideal single-level scenario are then analyzed within a zero-bandwidth two-level model in § 7, separately for leakage caused by the energetic proximity of an additional level (§ 7 A) and for the role of inter-level Coulomb repulsion (§ 7 B). In § 8, we discuss the technological feasibility of the proposed device within standard lithographic techniques for Al/InAs 2DEG heterostructures. Conclusions are given in § 9.

Three-terminal SC transistor device

A classical transistor switch comprises input, output, and control (base) terminals, with the latter designated for controlling the input and output currents. While this

concept translates readily into the SC scenario as demonstrated in Ref. [54], its straightforward application is impractical. In detail, an incoming supercurrent entering the SC transistor through the input electrode would inevitably be split into the remaining SC electrodes. Consequently, the incoming current JinJ_{\mathrm{in}} would not convert fully into the output lead as Jout=JinJ_{\mathrm{out}}=J_{\mathrm{in}}. Instead, a fraction of JinJ_{\mathrm{in}} would always leak into the base electrode. Crucially, such leakage current would generically be of the same order as the one reaching the output, and would moreover vary with the SC phases. Consequently, the conversion ratio Jout/JinJ_{\mathrm{out}}/J_{\mathrm{in}} would neither be fixed nor close to unity. Additionally, in such a conventional design, the OFF state has to be reached through the details of 00-π\pi quantum phase transitions (QPTs), which are device and material specific [54].

We therefore abandon the conventional input-output-base concept and instead connect two SC terminals (originally designated in Ref. [54] as input and base) to a flux-controlled loop. The motivation for this choice is the well-known interference effect in a dc SQUID with identical junctions [55, 56]. When its loop is threaded by half a flux quantum and the loop inductance is negligible, the supercurrents in the two arms interfere destructively: the transport critical current vanishes, while a current keeps circulating within the loop. A flux-controlled loop therefore provides a natural OFF state that does not rely on material-specific QPTs. In conventional SQUIDs, however, the junctions are fixed by fabrication, and any mismatch between them leaves a residual critical current at half flux [57, 58].

In the present device, the loop is interrupted by a QD region, which acts as the common node of the two arms, as shown in Figure 1(a). The role of the two SQUID junctions is then taken by the gate-tunable barriers between the QD and terminals 1 and 3, so that their symmetry can be restored in situ rather than being fixed by fabrication. A third gate-tunable barrier couples the QD to the output terminal 2. Since the two loop terminals are joined externally, they present a single current path to the rest of the circuit, and charge conservation alone then guarantees complete channeling of the incoming supercurrent through the device:

Scheme of a three-terminal superconducting quantum-dot transistor and current convention

Figure 1. (a) Scheme of a three-terminal superconducting (SC) quantum-dot (QD) transistor in a planar 2DEG heterostructure. SC terminals 1 and 3, with phases φ1\varphi_1 and φ3\varphi_3, form a flux-controlled loop, while terminal 2 with phase φ2\varphi_2, serves as the output electrode. A magnetic flux ΦB\Phi_{\mathrm{B}} threading the loop fixes the phase difference φ3−φ1=2πΦB/Φ0\varphi_3-\varphi_1=2\pi\Phi_{\mathrm{B}}/\Phi_0, where Φ0=h/2e\Phi_0=h/2e is the superconducting flux quantum. The central QD region is defined electrostatically: VQDV_{\mathrm{QD}} controls the QD potential, whereas V1V_1, V2V_2, and V3V_3 tune the tunnel barriers between the QD and the corresponding superconducting terminals. The inset illustrates a cross section of the gate-defined junction, and the color code identifies the 2DEG, in-situ superconductor, insulating barrier and buffer layers, electrostatic gates, and the additional SC layer. Further details on the fabrication requirements are given in § 8. (b) Current convention used throughout the manuscript. The terminal currents JiJ_i are defined as positive when flowing toward the QD region. The incoming current therefore splits between the two arms of the loop as Jin=J1+J3J_{\mathrm{in}}=J_1+J_3, recombines at the QD, and exits through terminal 2 as Jout=−J2J_{\mathrm{out}}=-J_2.

Jin=Jout=−J2=J1+J3.(1)J_{\mathrm{in}} = J_{\mathrm{out}} = -J_2 = J_1 + J_3. \tag*{(1)}

Crucially, thanks to the loop architecture the charge conservation ensures that input and output currents are the same irrespective of the dot occupation, the Coulomb interaction, the hybridization strengths, the SC phases, or temperature. The supercurrent passing through the device is then controlled solely via the magnetic flux ΦB\Phi_{\mathrm{B}} penetrating the loop. Assuming a loop inductance small enough that the phase drop generated by the circulating supercurrent is negligible, the flux fixes the phase difference between the two loop terminals as

φ3−φ1=2πΦBΦ0,(2)\varphi_3 - \varphi_1 = 2\pi\frac{\Phi_{\mathrm{B}}}{\Phi_0}, \tag*{(2)}

with Φ0=h/2e\Phi_0 = h/2e the SC flux quantum. The flux-controlled loop thus serves a dual purpose: it sets the phase difference entering (2) and thereby the operating point of the switch, while simultaneously controlling how the input supercurrent splits between the two arms [54, 59]. In operation, the device is switched between its ON and OFF states by threading integer or half-integer flux quanta through the loop, while the phase φ2\varphi_2 across the output branch is set by the external circuit.

In what follows, we label the three SC terminals as j∈{1,2,3}j \in\{1, 2, 3\}. Through each of the terminals, a supercurrent JjJ_j passes in a direction selected so that its positive-values indicate flow towards the QD-lead interface [Fig. 1(b)]. The terminals 1 and 3 are designated to form the flux-controlled loop, while the SC terminal 2 serves as the output electrode. Current conservation requires the input supercurrent JinJ_{\mathrm{in}} to split into supercurrents J1J_1 and J3J_3 upon entering the device. Each of these supercurrents is then passing through the tunnel barrier between the QD region and the corresponding SC terminal. In the QD region, the two supercurrents J1J_1 and J3J_3 recombine into the output supercurrent JoutJ_{\mathrm{out}}, which due to the selected convention about the flow direction fulfills Jout=−J2J_{\mathrm{out}} = -J_2. Crucially, its magnitude as well as the split of the incoming supercurrent are governed by quantum phenomena dictated by the multi-terminal Josephson effect.

Single-level in the QD region

The transport across the tunneling barriers connecting the SC leads and the QD region is dependent on the properties of the QD region. We first consider a scenario where one tunes the QD region by gate voltage so that a single electronic level resides in the transport window and remaining levels are well separated [60]. We denote the energy of such a dominant level by εd\varepsilon_d and consider a repulsive Coulomb interaction UU for two electrons occupying it simultaneously. The full Hamiltonian of the three-terminal device then reads

H=Hd+∑j=13(Hj,SC+Hj,T),(3)H = H_d + \sum_{j=1}^{3} \left(H_{j,\mathrm{SC}} + H_{j,\mathrm{T}}\right), \tag*{(3)}

where HdH_d describes the dominant electronic level in the QD region, Hj,SCH_{j,\mathrm{SC}} is the conventional Bardeen-Cooper-Schrieffer (BCS) Hamiltonian for the SC lead jj and Hj,TH_{j,\mathrm{T}} defines the tunnel-coupling between the lead jj and the electronic level. The electrons residing in the QD level therefore obey

Hd=εd∑σdσ†dσ+Un↑n↓,(4)H_d = \varepsilon_d \sum_{\sigma} d^\dagger_{\sigma} d_{\sigma} + U n_{\uparrow} n_{\downarrow}, \tag*{(4)}

with dσ†d^\dagger_{\sigma} (dσd_{\sigma}) creating (annihilating) a dot electron of spin σ∈{↑,↓}\sigma\in\{\uparrow,\downarrow\} and nσ=dσ†dσn_{\sigma} = d^\dagger_{\sigma}d_{\sigma} representing the occupation number operator. Each SC follows a BCS Hamiltonian

Hj,SC=∑kσϵkjckjσ†ckjσ−∑k(Δjckj↑†c−kj↓†+H.c.),(5)H_{j,\mathrm{SC}} = \sum_{k\sigma} \epsilon_{kj} c^\dagger_{kj\sigma} c_{kj\sigma} - \sum_k \left(\Delta_j c^\dagger_{kj\uparrow} c^\dagger_{-kj\downarrow} + \mathrm{H.c.}\right), \tag*{(5)}

while the coupling of the level to the SC lead jj obeys

Hj,T=∑kσ(Vkj∗ckjσ†dσ+Vkjdσ†ckjσ),(6)H_{j,\mathrm{T}} = \sum_{k\sigma} \left(V^*_{kj} c^\dagger_{kj\sigma} d_{\sigma} + V_{kj} d^\dagger_{\sigma} c_{kj\sigma}\right), \tag*{(6)}

where ckjσ†c^\dagger_{kj\sigma} (ckjσc_{kj\sigma}) creates (annihilates) an electron of spin σ\sigma, quasi-momentum kk and energy ϵkj\epsilon_{kj} in lead jj. The leads couple to the QD only and not directly to each other. All terminals are assumed to be made of the same SC material, reflecting the typical experimental scenario. The SC leads thus share the gap magnitude Δ\Delta and the bandwidth 2B2B. Their BCS phases φj\varphi_j on the other hand are individually controlled by flux or current bias. The coupling of the level to lead jj is characterized by the tunneling strength Γj≡π∑k∣Vkj∣2δ(ω−ϵkj)\Gamma_j \equiv\pi\sum_k |V_{kj}|^2\delta(\omega-\epsilon_{kj}), presumed energy independent for simplicity, with Γ≡∑jΓj\Gamma\equiv\sum_j \Gamma_j denoting the total hybridization of the level with the three leads. In what follows, the device geometry is conveniently expressed through the relative hybridization strengths γj≡Γj/Γ\gamma_j \equiv\Gamma_j/\Gamma, which fulfill γ1+γ2+γ3=1\gamma_1+\gamma_2+\gamma_3=1. Their configurations are thus fully defined by specifying only two of them, for which we choose γ2\gamma_2 and γ3\gamma_3 throughout the paper.

Following Refs. [54, 59, 61], we resolve the problem by solving for its Green’s function. This is facilitated by the Hamiltonians (5) and (6) being quadratic. Consequently, the SC environment can be integrated out exactly, and its entire effect on the QD level reduces to a corresponding self-energy contribution to the retarded QD Green’s function. The latter is then used to resolve the full many-body problem, giving access to the spectrum, the ground state character, and the free energy. Applying the Nambu formalism with spinor D†=(d↑†,d↓)D^\dagger=(d_\uparrow^\dagger,d_\downarrow) allows to obtain the Green’s function of the dd-electrons as

Gd(ω)=1ω1−Hd,0−Σ0(ω)−ΣU(ω)+iη(7)\mathbf{G}_d(\omega)=\frac{1}{\omega\mathbf{1}-\mathbf{H}_{d,0}-\mathbf{\Sigma}_0(\omega)-\mathbf{\Sigma}_U(\omega)+i\eta} \tag*{(7)}

where boldface denotes 2×22\times2 Nambu matrices, η\eta is infinitesimally small, and

Hd,0=(εd00−εd)(8)\mathbf{H}_{d,0}=\begin{pmatrix}\varepsilon_d&0\\0&-\varepsilon_d\end{pmatrix} \tag*{(8)}

is the Nambu representation of the non-interacting part of (4). The Coulomb interaction UU enters the Green’s function via ΣU(ω)\mathbf{\Sigma}_U(\omega), which needs to be obtained numerically. On the other hand, as already explained, the SC environment is quadratic and contributes Σ0(ω)\mathbf{\Sigma}_0(\omega) hybridization self-energy which can be directly obtained by setting U=0U=0 in (3). A standard equation-of-motion technique then yields

Σ0(ω+)=Γ(ωΔχΔχ∗ω)F(ω+),(9)\mathbf{\Sigma}_0(\omega^+)=\Gamma\begin{pmatrix}\omega&\Delta\chi\\\Delta\chi^*&\omega\end{pmatrix}F(\omega^+), \tag*{(9)}

where F(ω+)F(\omega^+) has a parametric dependence only on Δ\Delta and BB. In the wide-band limit B→∞B\to\infty, it reads

F(ω+)={1Δ2−ω2,for ∣ω∣<Δ,i sgn⁡(ω)ω2−Δ2,for ∣ω∣>Δ.(10)F(\omega^+)=\begin{cases} \frac{1}{\sqrt{\Delta^2-\omega^2}}, & \text{for }|\omega|<\Delta,\\ \frac{i\,\operatorname{sgn}(\omega)}{\sqrt{\omega^2-\Delta^2}}, & \text{for }|\omega|>\Delta. \end{cases} \tag*{(10)}

The off-diagonal elements of (9) clearly depend on a single complex-valued and manifestly gauge-dependent quantity

χ≡∑j=13γjeiφj,(11)\chi\equiv\sum_{j=1}^{3}\gamma_j e^{i\varphi_j}, \tag*{(11)}

which is a sum of three phasors of lengths γj\gamma_j and phases φj\varphi_j. The underlying gauge freedom allows rotating χ\chi onto the real axis via φj→φj−Arg⁡(χ)\varphi_j\to\varphi_j-\operatorname{Arg}(\chi). Such a global shift of all SC phases leaves the physics unchanged, as only phase differences are physically observable. This results in a fully gauge-invariant formulation of (9) which allows to replace in Eq. (9) the gauge dependent χ\chi with its gauge-independent amplitude

χ≡∣χ∣=1−4∑j>l=13γjγlsin⁡2φj−φl2.(12)\chi\equiv|\chi|=\sqrt{1-4\sum_{j>l=1}^{3}\gamma_j\gamma_l\sin^2\frac{\varphi_j-\varphi_l}{2}}. \tag*{(12)}

It only involves phase differences and hybridization strengths and is bounded by 0≤χ≤10\le\chi\le1.

Most importantly, the description via Eq. (12) extends beyond the non-interacting case. To this end, we notice that the SC environment enters the QD problem (7) exclusively through Σ0(ω)\Sigma_0(\omega), so the interaction self-energy ΣU(ω)\Sigma_U(\omega) knows nothing about the individual γj\gamma_j and φj\varphi_j. Consequently, it is only χ\chi that determines the full Gd(ω)G_d(\omega), the free energy F(T)F(T) and all QD observables expressible via Gd(ω)G_d(\omega) once UU, εd\varepsilon_d, Γ\Gamma and TT are fixed. Consequently, two different configurations {γj,φj}\{\gamma_j,\varphi_j\} with equal χ\chi possess identical many-body spectra, which allows to project any multi-terminal configuration onto a two-terminal one with equal couplings, provided the leads are not directly coupled, as first derived in Ref. [54].

The geometric factor of such a two-terminal reference problem is χ=∣cos⁡(φ/2)∣\chi=|\cos(\varphi/2)| with φ\varphi its phase difference, so it covers the full range 0≤χ≤10\le\chi\le1 and thus all configurations of the present three-terminal device. It is additionally a numerically well tractable model that can be solved by standard Numerical Renormalization Group (NRG) [62, 63, 64] or Quantum Monte Carlo (QMC) [65, 66, 67] techniques in an unbiased way. For instance, for given UU and Γ\Gamma, the two-terminal phase difference φ∗\varphi^* at which a singlet-doublet transition occurs gives the corresponding critical χ∗=cos⁡(φ∗/2)\chi^*=\cos(\varphi^*/2). Plotting isolines of χ=χ∗\chi=\chi^* from the analytic expression (12) then yields the phase diagram of our device.

Multi-Terminal SC Transistor Effect

To completely resolve the supercurrents in the three-terminal device, the formalism of the geometric factor χ\chi represents the crucial tool. Following Ref. [54], we calculate the supercurrent flowing from terminal jj to the QD by differentiating the total free energy F(T)F(T) with respect to φj\varphi_j and using χ\chi in the chain rule:

Jj(T)=2eℏ∂F(T)∂φj=2eℏ∂F(T)∂χ∂χ∂φj=JT(χ)χj′.(13)J_j(T)=\frac{2e}{\hbar}\frac{\partial F(T)}{\partial\varphi_j}=\frac{2e}{\hbar}\frac{\partial F(T)}{\partial\chi}\frac{\partial\chi}{\partial\varphi_j}=\mathcal{J}_T(\chi)\chi'_j. \tag*{(13)}

This defines two functions

JT(χ)≡J0Δ∂F(T)∂χ,(14)\mathcal{J}_T(\chi)\equiv\frac{J_0}{\Delta}\frac{\partial F(T)}{\partial\chi}, \tag*{(14)}
χj′≡∂χ∂φj,(15)\chi'_j\equiv\frac{\partial\chi}{\partial\varphi_j}, \tag*{(15)}

where J0≡2eΔ/ℏJ_0\equiv2e\Delta/\hbar serves as the unit of supercurrent throughout the paper. Crucially, the prefactor JT(χ)\mathcal{J}_T(\chi) depends only on χ\chi while χj′\chi'_j is fully analytic and encodes the details about any NN-terminal SC environment defined by its relative hybridizations γj\gamma_j and phases φj\varphi_j. Consequently, at fixed UU, ϵd\epsilon_d, Γ\Gamma and TT we can calculate JT(χ)\mathcal{J}_T(\chi) from the simplest model that encompasses all values of χ\chi in its allowed range of [0,1][0,1]. To this end the two-terminal coupling symmetric model is best suited. Using NRG or QMC, one can then determine the corresponding Josephson current J(T)J(T) at phase difference φ\varphi, which appears on the left-hand side of Eq. (13). In this system the geometric factor becomes χ=cos⁡(φ/2)\chi=\cos(\varphi/2) yielding χ′=−sin⁡(φ/2)/2\chi'=-\sin(\varphi/2)/2 where we suppressed the lead indices in both the Josephson current and the derivative function. The universal prefactor is then extracted as

JT(χ)=J(T)χ′.(16)\mathcal{J}_T(\chi)=\frac{J(T)}{\chi'}. \tag*{(16)}

The resulting functions JT(χ)\mathcal{J}_T(\chi) are shown in Figs. 2(a1)-(a10). They have been obtained by employing NRG Ljubljana [68] to the coupling-symmetric two-terminal model at zero temperature with U/Δ=3U/\Delta=3 and Γ\Gamma varying accordingly. For such a parameter choice we can observe characteristic discontinuities at critical values χ∗\chi^* [Figs. 2(a4)-(a10)]. When 0≤χ≤χ∗0\leq\chi\leq\chi^* the universal function is negative yielding a characteristic π\pi supercurrent which signals an underlying doublet ground state. For χ∗≤χ≤1\chi^*\leq\chi\leq1 the universal function yields a characteristic 00 supercurrent indicating a singlet ground state. Notably, there is a characteristic disparity between the considerably smaller magnitude of JT=0(χ)\mathcal{J}_{T=0}(\chi) in the doublet regions and its maximum in the singlet ground state. This effect was exploited in Ref. [54] to design a supercurrent transistor effect that allowed for suitably chosen γj\gamma_j to drive a three-terminal device into a region of exclusively doublet ground state, where the output supercurrent was strongly suppressed.

The SC transistor effect of Ref. [54] relied thus on delicate tuning of UU, Γ\Gamma and the relative hybridizations γj\gamma_j. It was exclusively facilitated by the presence of QPTs and would thus rapidly diminish at finite temperature. To overcome such severe limitations, we instead turn to the analytic factor that describes geometric details of the three-terminal device. In the gauge φ1=0\varphi_1=0 we can use φ2\varphi_2 and φ3\varphi_3 directly as the gauge-invariant phase differences. Together with γ1+γ2+γ3=1\gamma_1+\gamma_2+\gamma_3=1, this yields

χj′=−γjχ∑l≠jγlsin⁡(φj−φl),(17)\chi'_j=-\frac{\gamma_j}{\chi}\sum_{l\ne j}\gamma_l\sin(\varphi_j-\varphi_l), \tag*{(17)}

where the geometric factor χ\chi itself appears in the denominator. Specifically for the output terminal we obtain

χ2′=−γ2χ[γ1sin⁡(φ2)+γ3sin⁡(φ2−φ3)],(18)\chi'_2=-\frac{\gamma_2}{\chi}\left[\gamma_1\sin(\varphi_2)+\gamma_3\sin(\varphi_2-\varphi_3)\right], \tag*{(18)}

which exhibits fundamentally different behavior as φ3=0\varphi_3=0 and φ3=±π\varphi_3=\pm\pi. For φ3=0\varphi_3=0 we obtain

χ2′∣φ3=0=−γ2(1−γ2)sin⁡(φ2)χ,(19)\left.\chi'_2\right|_{\varphi_3=0}=-\frac{\gamma_2(1-\gamma_2)\sin(\varphi_2)}{\chi}, \tag*{(19)}

whereas for φ3=±π\varphi_3=\pm\pi

χ2′∣φ3=±π=−γ2(γ1−γ3)sin⁡(φ2)χ.(20)\left.\chi'_2\right|_{\varphi_3=\pm\pi}=-\frac{\gamma_2(\gamma_1-\gamma_3)\sin(\varphi_2)}{\chi}. \tag*{(20)}

Along the trajectory with φ3=0\varphi_3=0, the geometric current factor vanishes identically only in the degenerate case γ2=1\gamma_2=1, which corresponds to terminals 1 and 3 being completely disconnected from the system and is thus a trivial regime. In contrast, along the trajectory φ3=±π\varphi_3=\pm\pi the numerator of (20) vanishes identically when

γ3=1−γ22=γ1,(21)\gamma_3=\frac{1-\gamma_2}{2}=\gamma_1, \tag*{(21)}

leading to a perfectly closed supercurrent switch (J2=0J_2=0) via a gate-control of the hybridization strengths.

The origin of the effect is transparent at the level of the geometric factor χ\chi itself. Setting φ3=±π\varphi_3=\pm\pi in (11) gives χ=(γ1−γ3)+γ2eiφ2\chi=(\gamma_1-\gamma_3)+\gamma_2e^{i\varphi_2}. At γ1=γ3\gamma_1=\gamma_3 the two phasors coming from the flux-loop cancel exactly leaving χ=γ2\chi=\gamma_2, which thus becomes independent of φ2\varphi_2. The free energy of the QD in Eq. (13), being a function of χ\chi alone, is then insensitive to the output phase. The output current J2J_2 must therefore vanish identically. At the same time, (17) returns χ1′=−χ3′=γ1γ2sin⁡(φ2)/χ≠0\chi'_1=-\chi'_3=\gamma_1\gamma_2\sin(\varphi_2)/\chi\ne0 meaning that the corresponding supercurrents J1J_1 and J3J_3 are non-zero but fulfill J1=−J3J_1=-J_3. Consequently, the flux-loop sustains a purely circulating supercurrent while no charge is transmitted into the output lead. Such a scenario can be interpreted on a more general footing. By understanding the tunneling barriers between the electronic level residing in the QD region and leads 1 and 3 as individual Josephson junctions, we notice that the condition γ1=γ3\gamma_1=\gamma_3 essentially requires them to be identical. In the closed state, the flux loop thus operates in a regime analogous to a dc SQUID with identical junctions at half a flux quantum. For negligible loop inductance, the two arm supercurrents of such a SQUID are exactly opposite at half flux, so that interference suppresses the transport critical current completely while a circulating current persists [Jaklevic-1964, Zimmerman-1966, Fulton-1972]. Junction asymmetry or finite loop inductance leaves a residual critical current [57, 56, 58], in close analogy to the case γ1≠γ3\gamma_1\ne\gamma_3 case of the present three-terminal device.

Resonance width of the switching ratio

The multi-terminal SC transistor effect can thus be understood in purely geometric terms with phases being operated in-situ by flux- or current drive and tunnel barriers by gate-voltages. While the former switch the device between open and closed states, the latter is required to tune to the ideal closed regime given by γ1=γ3\gamma_1=\gamma_3 condition of Eq. (21). Since γ2=1−γ1−γ3\gamma_2=1-\gamma_1-\gamma_3, it is natural to select relative hybridizations γ2\gamma_2 and γ3\gamma_3 for characterization of the device. The parameter γ2\gamma_2 is expected to directly influence the output supercurrent J2J_2 while γ3\gamma_3 governs directly supercurrents in the loop. In the γ2\gamma_2-γ3\gamma_3 parametric plane, condition (21) defines then a line γ3=(1−γ2)/2\gamma_3=(1-\gamma_2)/2 along which the three-terminal device switches between perfectly closed (φ3=±π\varphi_3=\pm\pi) and open (φ3=0\varphi_3=0) states.

Since the output supercurrent in the open state (denoted as JonJ_{\mathrm{on}}) is finite but its counterpart in the closed state (JoffJ_{\mathrm{off}}) vanishes, their ratio defining the switching ratio diverges, i. e. R=Jon/Joff→∞R=J_{\mathrm{on}}/J_{\mathrm{off}}\to\infty. Eq. (21) acts thus as a resonance condition for tuning the couplings to obtain the ideal SC transistor characterized by R→∞R\to\infty. However, exact control over tunneling barriers in experiments is challenging to achieve. The central question is

thus how broad is the resonance R→∞R\to\infty with regard to tuning of the tunneling barriers.

To this end, we turn our attention to the detailed interaction effects encoded in JT(χ)\mathcal{J}_{T}(\chi), which was obtained numerically as explained in Sec. III with results plotted in Figure 2(aFigure 1-(a10). The model parameters are selected as U/Δ=3U/\Delta=3, while the central QD is set to half-filling, i. e. εd=−U/2\varepsilon_d=-U/2. Then we vary the ratio U/ΓU/\Gamma. The output supercurrent in the three-terminal scenario is then obtained by multiplying the numerically obtained JT=0(χ)\mathcal{J}_{T=0}(\chi) function with the proper geometric current factor χ2\chi_2. This way, we can generate J2J_2 currents at any γ2\gamma_2-γ3\gamma_3 configuration of the three-terminal device at given UU, εd\varepsilon_d and Γ\Gamma. From the φ2\varphi_2 dependence of J2J_2 at φ3=0\varphi_3=0 and φ3=±π\varphi_3=\pm\pi one then reads off the critical values of supercurrent corresponding to the maximum allowed supercurrent JonJ_{\mathrm{on}} in the open state and the corresponding maximum allowed supercurrent JoffJ_{\mathrm{off}} in the closed state. The resulting switching ratio RR is then shown in Figure 2(bFigure 1-(b10) for the triangular regions of allowed γ2\gamma_2 and γ3\gamma_3 strengths.

Figure 2. (a1)-(a10) Universal functions JT=0(χ)\mathcal{J}_{T=0}(\chi) calculated from the coupling-symmetric two-terminal counterpart by employing NRG Ljubljana. (b1)-(b10) The switching ratio R=Jon/JoffR=J_{\mathrm{on}}/J_{\mathrm{off}} calculated from the corresponding values of critical currents ∣J2c∣\lvert J_2^c\rvert in the corresponding state with fixed φ3\varphi_3. Current calculations employ the mapping technique of [54]. Along the analytically obtained sweet line γ3=(1−γ2)/2\gamma_3=(1-\gamma_2)/2, a region of significantly increased transistor yield is apparent. All maps show a triangular region of increased ratio RR, which is, however, suppressed as Γ\Gamma increases.

In all selected cases of Γ/Δ\Gamma/\Delta, dark-red lines emerge at the position predicted by (21). Although sweet lines are present regardless of exact model parameters, the behavior of RR around their position shows complex behavior due to Γ\Gamma-UU interplay. For Γ/Δ<1\Gamma/\Delta< 1 [Figs. 2(b1)-(b3)], the resonances are sharp and their width decreases monotonically as γ2→1\gamma_2 \to1. At Γ=Δ\Gamma= \Delta [Fig. 2(b4)], bimodal behavior emerges with a broad resonance for γ2<0.65\gamma_2 < 0.65 that abruptly switches to a significantly narrower width at γ2∼0.65\gamma_2 \sim0.65 until γ2→1\gamma_2 \to1. The broad nature is associated with a green-shaded inner triangle region where the switching ratio is significant. Sharp resonance occurs in the outer dark blue-shaded region with R≈1R \approx1 that shows only irrelevant transistor effect outside of the resonance condition. Increasing Γ\Gamma expands this region irrelevant for supercurrent switching at the expense of the inner triangle. At Γ=4Δ\Gamma= 4\Delta [Fig. 2(b10)] the remaining inner triangle region has almost entirely vanished.

Consequently, Figure 2(bFigure 1-(b10) clearly define the operating regime of the device. The sweet line itself is present throughout, but only at intermediate hybridization, Γ≈Δ\Gamma\approx\Delta, is it surrounded by a broad region of elevated RR. Both weaker and stronger coupling shrink this region: at small Γ\Gamma [Figs. 2(b1)-(b3)] the resonance appears as a thin line, while at large Γ\Gamma the broad resonant feature in the green-shaded triangle is progressively diminished in favor of the outer region where R≈1R \approx1. Operating the device at intermediate Γ\Gamma therefore allows one to exploit the broad resonances for an effective tuning of the switching ratio. Tuning of the γ2\gamma_2 values, on the other hand, matters only with respect to the bimodal behavior, and a proper value shall only ensure that the device is operated within the green-shaded area.

Origin of the resonance-width structures

We now analyze the origin of the structures observed in the switching-ratio maps. The geometric current factor χ22\chi_2^2 clearly defines the resonance line throughout the entire γ2\gamma_2-γ3\gamma_3 plane while the universal function JT(χ)\mathcal{J}_{\mathrm{T}}(\chi) serves as a modulating prefactor. Its role will thus be crucial especially away from the resonance line given by Eq. (21). Consequently, both the inner triangle of elevated RR and the outer region with R≈1R \approx1, as observed for example in Figure 2(bFigure 4, must stem from the singlet-doublet QPTs encoded in JT(χ)\mathcal{J}_{\mathrm{T}}(\chi).

Figure 4. (a1)–(a5) Universal functions JT(χ)J_T(\chi) obtained from NRG Ljubljana at temperatures T/Δ∈{0,0.02,0.04,0.06,0.08}T/\Delta\in\{0, 0.02, 0.04, 0.06, 0.08\} (panels left to right). Other model parameters are selected according to Figure 2(aFigure 4: U/Δ=3U/\Delta= 3, εd=−U/2\varepsilon_d = -U/2 and Γ/Δ=1.0\Gamma/\Delta= 1.0. The discrete jump present at T=0T = 0 progressively smoothens into a continuous crossover and the overall amplitude decreases with increasing T/ΔT/\Delta. (b1)–(b5) Switching ratio R=Jon/JoffR = J_{\mathrm{on}}/J_{\mathrm{off}} across the γ2\gamma_2–γ3\gamma_3 parameter space. The sweet line and the surrounding region of significant transistor yield remain visible up to the highest temperature shown. The resonance width remains stable with only a minor decrease in the width throughout the temperature range shown. (c1)–(c5) Current-phase maps of J2J_2 at γ2=0.2\gamma_2 = 0.2 and γ3=0.3\gamma_3 = 0.3, showing the open (ON) and closed (OFF) regimes. With increasing T/ΔT/\Delta, an overall reduction in the magnitude of supercurrents is observed. Since this affects all regimes, the switching ratio RR remains stable.

We therefore select the Γ/Δ=1\Gamma/\Delta= 1 case where these distinct regions of switching ratio are all co-existing and select four representative points W-Z according to Figure 2(bFigure 4. W lies in the green-shaded inner triangle and close to the sweet line, X lies within the green-shaded inner triangle but does not fulfill (21), Y lies in the outer dark-blue boundary away from the resonance (21) while Z lies close to the resonance condition but still in the same region. For this discussion, we again choose the gauge φ1=0\varphi_1 = 0, so values of φ2\varphi_2 and φ3\varphi_3 directly represent the gauge-independent phase differences of our device.

Their variation shows singlet (green) and doublet (violet) ground states emerging in phase diagrams 3(a1)-(a4). These are obtained from the geometric factor χ\chi with numeric input of χ∗=0.721\chi^{*}=0.721 from the corresponding NRG calculation. Using additionally JT=0J_{T}=0 obtained from the same NRG calculation, the corresponding J2J_{2} supercurrents are derived in Figure 3(bFigure 1-(bFigure 4. The side panels display J2cJ_{2}^{c} as a function of the φ3\varphi_{3}-configuration of the device.

Figure 3. (a1)-(a4) Phase diagrams in the φ2\varphi_2-φ3\varphi_3 plane for cases W, X, Y and Z. Only singlet (green color) and doublet (violet) ground states occur. Stripes and islands occur in the phase diagrams in accordance with [54]. (b1)-(b4) The corresponding φ2\varphi_2-φ3\varphi_3 maps of the J2J_2 currents corresponding to panels (a1)-(a4). The insets to the right of each panel show the magnitude of critical currents ∣J2c∣\lvert J_2^c\rvert for a fixed value of φ2\varphi_2.

Comparing the first cases W and X, we notice that the overall structure of singlet islands immersed in a continuum of doublet ground states is preserved. Nevertheless, the geometrically dependent switching mechanism crucially alters the J2J_{2} profile in the doublet continuum. Being close to the resonance condition, case W develops a horizontally aligned area of strongly suppressed J2J_{2} along φ3=±π\varphi_{3}=\pm\pi. These are absent in X, as a geometrically-dependent suppression mechanism is unavailable because of the distance from the resonance condition (21). As the inset of Figure 3(bFigure 2 shows, J2cJ_{2}^{c} still reaches its minima at φ3=±π\varphi_{3}=\pm\pi, but unlike in W, the output supercurrent remains small but still non-zero. As a consequence, the transistor switching ratio does not increase without bound and instead reaches ≈13\approx13.

While the involvement of the geometric mechanism is superior to the one relying solely on the QPT mechanism of Ref. [54], for supercurrent transistor switching, the latter is sufficient to generate the distinct triangular-shaped regions of ratio R>1R>1 (green-shaded inner triangles) as observed in Figure 2. This pattern is tightly connected to the underlying phase diagram in the φ2\varphi_{2}-φ3\varphi_{3} with singlet islands (green) being well separated from each other by broad regions of doublet phase when φ3\varphi_{3} is tuned.

However, once such a structure is broken, as in the case Y, φ2\varphi_{2} tunes only between alternating singlet and doublet stripes regardless of φ3\varphi_{3}. Consequently, only an insignificant variation in J2cJ_{2}^{c} is observed as shown in the inset of Figure 3(bFigure 3. The resulting transistor switching ratio drops to ≈1.3\approx1.3. In general, striped phase diagrams are associated with the outer regions of the suppressed transistor ratio as previously observed in Figure 2. Unlike in the inner triangular regions, strong interactions and the underlying quantum phases form no suitable structures to generate the transistor effect.

As the case Z demonstrates, the geometric mechanism generating supercurrent switching is still present and leads to the observed thin resonances around the condition (21). As shown in the inset of Figure 3(bFigure 4, the critical J2cJ_{2}^{c} current is still relatively large in the open regime of the three-terminal device when compared to cases W and X. However, Figure 3(bFigure 4 also clearly demonstrates that zero-current regions are extremely narrow when tuned via φ3\varphi_{3}, which is detrimental for gate-tuning in real devices.

Temperature dependence of the resonance-width structures

The structures identified in Sec. IV A rest on the discontinuities in JT=0(χ)J_{T=0}(\chi) that stem from singlet-doublet transitions. These become smoothed out at any finite temperature, which inevitably affects the coupling dependencies of the switching ratio. Nevertheless, the geometric current factor χ2c\chi_{2}^{c} is temperature independent, so the resonance condition (21) itself cannot be affected. What remains to be examined are thus the regions of elevated RR surrounding it.

Let us therefore demonstrate such effects, focusing on the case of a sizable resonance width provided by Γ/Δ=1.0\Gamma/\Delta=1.0, as evaluated in Figure 2(bFigure 4 and select a range of temperatures up to T/Δ=0.08T/\Delta=0.08. First the corresponding JTJ_{T} functions are obtained as shown in Figure 4(aFigure 1-(aFigure 5. The calculations have been performed as previously by employing the spectrally equivalent coupling-symmetric SC-AIM using the full density-matrix numerical renormalization group approach implemented within NRG Ljubljana.

Figure 5. Current-phase relation of the output supercurrent J2J_2 in the three-terminal structure with two energy levels εA\varepsilon_A and εB\varepsilon_B. The level A is kept at half-filling by setting εA/Δ=−UA/(2Δ)=−1.5\varepsilon_A/\Delta=-U_A/(2\Delta)=-1.5 (UA/Δ=3U_A/\Delta=3), while UB/Δ=2U_B/\Delta=2 and εB\varepsilon_B is decreased from εB/Δ=3\varepsilon_B/\Delta=3 (leftmost column) towards εB/Δ=0\varepsilon_B/\Delta=0 (rightmost column). Each panel shows J2(φ2)J_2(\varphi_2) in the open regime (black) as well as in the closed regime at g1Bg_{1B}-g3Bg_{3B} asymmetry values α∈{0,0.2,0.4}\alpha\in\{0,0.2,0.4\} (blue, orange, red) and t/Δ=0.6t/\Delta=0.6. Closing of the gap between the two levels reshapes J2(φ2)J_2(\varphi_2) by removing the 00-π\pi transition observed when the two levels are most separated, which keeps level A dominant. At εB=0\varepsilon_B=0 both levels contribute comparably. This is then detrimental to the supercurrent transistor switching ratio.

At T=0T=0, the JT=0J_{T=0} function possesses a discrete jump coinciding with the boundaries of the 0 and π\pi phases. As shown in Figure 4(aFigure 2-(aFigure 5, this discontinuity disappears once T>0T>0. The supercurrent jump is instead smoothed out while simultaneously the amplitude of JTJ_{T} decreases. This significantly reduces the critical supercurrents both in the 0 and π\pi phases.

Using (13), the resulting dependencies of the switching ratio on the relative couplings are then obtained as shown in Figure 4(bFigure 1-(bFigure 5. The overall reduction of critical supercurrents is seen to have only a relatively minor impact on the resonance width in the selected temperature range, which is due to simultaneous reduction of critical currents in both the open and closed regimes of the device, as can be inferred from Figure 4(cFigure 1-(cFigure 5. From the practical perspective, temperature is therefore not detrimental to the switching ratio. Nevertheless, the overall suppression of critical supercurrents diminishes the magnitude of supercurrents transmitted by the device, which sets a practical lower bound on the operating temperature for applications requiring a minimum current throughput.

Beyond the single-level scenario

The switching mechanism established in Secs. III and IV rests on the presence of a single dominant level in the QD region. Realistic devices, however, host additional levels, and although these in principle can be pushed away in energy by electrostatic gating, they typically cannot be removed from the transport window entirely. To assess their influence on the SC transistor effect, we introduce a second level to the QD region as a minimal extension of (3).

We designate label A for the dominant level, which in 2DEG-based devices is typically the last one to hold electrons before the central region is fully depleted electrostatically. The nearest level in the hierarchy is then denoted by label B. We assume that both levels couple to the three SC leads, which keep their labels j∈{1,2,3}j \in\{1, 2, 3\} according to Figure 1. Crucially, while barriers between the leads and the QD region are gate-tunable they cannot be individually adjusted with respect to the levels A and B. Moreover, the orbital structure of level B typically differs from that of A. Therefore, a barrier configuration that equalizes tunnelings between the leads forming the flux-loop and level A, generically leaves the corresponding tunneling rates for level B asymmetric. We can thus realistically not expect that the geometric mechanism would suppress transport through both levels simultaneously in the closed regime, which results in inevitable leakage currents through the output terminal 2.

To qualitatively understand which effects contribute the most to the leakage, we employ the zero-bandwidth (ZBW) approximation for the SC leads. Technically, we thus replace all of the kk modes appearing in Eq. (5) by a single effective mode whose off-diagonal pairing is Δeiφ\Delta e^{i\varphi} with φ\varphi being the original BCS phase and Δ\Delta the bare gap parameter. The ZBW description involving two levels coupled to three SC leads then reads

HZBW=HA+HB+∑j=1,2,3(Hj,SC+Hj,At+Hj,Bt),(22)H_{\mathrm{ZBW}}=H_A+H_B+\sum_{j=1,2,3}\left(H_{j,\mathrm{SC}}+H^t_{j,A}+H^t_{j,B}\right), \tag*{(22)}

where the levels A and B of energies εA\varepsilon_A and εB\varepsilon_B follow

HA=∑σεAdσ,A†dσ,A+UAd↑,A†d↑,Ad↓,A†d↓,A,(23)H_A=\sum_{\sigma}\varepsilon_A d^\dagger_{\sigma,A}d_{\sigma,A}+U_A d^\dagger_{\uparrow,A}d_{\uparrow,A}d^\dagger_{\downarrow,A}d_{\downarrow,A}, \tag*{(23)}
HB=∑σεBdσ,B†dσ,B+UBd↑,B†d↑,Bd↓,B†d↓,B.(24)H_B=\sum_{\sigma}\varepsilon_B d^\dagger_{\sigma,B}d_{\sigma,B}+U_B d^\dagger_{\uparrow,B}d_{\uparrow,B}d^\dagger_{\downarrow,B}d_{\downarrow,B}. \tag*{(24)}

with dσ,m†d^\dagger_{\sigma,m} (dσ,md_{\sigma,m}) creating (annihilating) a dot electron of spin σ\sigma on level m∈{A,B}m\in\{\mathrm{A},\mathrm{B}\}. Coulomb repulsion terms UAU_A and UBU_B are generally not equal and represent material properties of a given device. The ZBW Hamiltonian for lead j∈{1,2,3}j\in\{1,2,3\} contains the Hamiltonian for the three spinful modes created via cjσ†c^\dagger_{j\sigma}, which obey

Hj,SC=Δ(eiφjcj↑†cj↓†+e−iφjcj↓cj↑)(25)H_{j,\mathrm{SC}}=\Delta\left(e^{i\varphi_j}c^\dagger_{j\uparrow}c^\dagger_{j\downarrow}+e^{-i\varphi_j}c_{j\downarrow}c_{j\uparrow}\right) \tag*{(25)}

with the gauge φ1=0\varphi_1=0 selected throughout this section. Six couplings gjAg_{jA} and gjBg_{jB} with j∈{1,2,3}j\in\{1,2,3\} describe how leads (25) couple the two levels via

Hj,At=gjA∑σdσ,A†cjσ+h.c.,(26)H^t_{j,A}=g_{jA}\sum_{\sigma}d^\dagger_{\sigma,A}c_{j\sigma}+\mathrm{h.c.}, \tag*{(26)}
Hj,Bt=gjB∑σdσ,B†cjσ+h.c.,(27)H^t_{j,B}=g_{jB}\sum_{\sigma}d^\dagger_{\sigma,B}c_{j\sigma}+\mathrm{h.c.}, \tag*{(27)}

where the coupling constants gjAg_{jA} and gjBg_{jB} should be understood as phenomenological quantities which are only indirectly related to the microscopic tunneling rates. Since we aim to characterize leakage processes implicated by the coupling-asymmetries of the additional level B, we subject level A to the resonant condition (21) by demanding g1A=g3Ag_{1A}=g_{3A}. Moreover, we select the center of a triangle-shaped region of the γ2\gamma_2-γ3\gamma_3 plane by setting g1A=g2A=g3A=gg_{1A}=g_{2A}=g_{3A}=g. The total tunneling rate of level A in the ZBW approximation therefore reads

ΓA,ZBW≡∑jπ(gjA)22=3πg22.(28)\Gamma_{A,\mathrm{ZBW}} \equiv\sum_j \pi\frac{(g_{jA})^2}{2} = \frac{3\pi g^2}{2}. \tag*{(28)}

Subsequently, we parametrize two types of coupling asymmetries of level B via the parameters α\alpha and β\beta via

g1B=1−α1+β t,(29)g_{1B} = \sqrt{1-\alpha}\sqrt{1+\beta}\,t, \tag*{(29)}
g2B=1−2β t,(30)g_{2B} = \sqrt{1-2\beta}\,t, \tag*{(30)}
g3B=1+α1+β t.(31)g_{3B} = \sqrt{1+\alpha}\sqrt{1+\beta}\,t. \tag*{(31)}

Clearly, α=β=0\alpha=\beta=0 corresponds to the totally symmetric coupling of the level B to the SC modes of the leads with total hybridization fulfilling

ΓB,ZBW≡∑jπ(gjB)22=3πt22(32)\Gamma_{B,\mathrm{ZBW}} \equiv\sum_j \pi\frac{(g_{jB})^2}{2} = \frac{3\pi t^2}{2} \tag*{(32)}

at arbitrary α,β≤1\alpha,\beta\le1. The effect of α\alpha is to introduce coupling asymmetry with respect to terminals 1 and 3, which is thus primarily responsible for the leakage through level B. On the other hand, increasing β\beta progressively diminishes the coupling of the output SC lead to level B. The states of the transistor are not further optimized and are selected with reference to the ideal single level scenario. In other words, ϕ3=0\phi_3=0 (ϕ3=π\phi_3=\pi) are per definition considered as the open (closed) states of the two-level system. From the resulting ground state energy EGSE_{\mathrm{GS}} we then obtain J2(ϕ2)J_2(\phi_2) via (13) as J2(ϕ2)=(2e/ℏ)∂EGS/∂ϕ2J_2(\phi_2)=(2e/\hbar)\partial E_{\mathrm{GS}}/\partial\phi_2. The obtained current-phase relations J2(ϕ2)J_2(\phi_2) are then evaluated at different asymmetry parameters α\alpha and β\beta with respect to relative energies of two levels and their possible inter-level Coulomb interactions.

Leakage current due to the proximity of additional level

We first investigate the leakage with respect to the mutual position of levels A and B. The results for J2J_2 are shown in Figure 5. The first column [Figure 5(aFigure 1-(dFigure 1] is obtained for a sizable separation of the two levels by selecting εA/Δ=−UA/2=−1.5\varepsilon_A/\Delta=-U_A/2=-1.5 and εB/Δ=3\varepsilon_B/\Delta=3. Level A resides thus at the half-filled point which is a condition we keep throughout this section. Due to the sizable distance, level B interferes with the transport via A only in a marginal way. The supercurrent J2,on(ϕ2)J_{2,\mathrm{on}}(\phi_2) in the open regime shows a clear 0-π\pi transition point at essentially the same ϕ2/π=0.6\phi_2/\pi=0.6 for all selected values of α∈{0.0,0.2,0.4}\alpha\in\{0.0,0.2,0.4\} and β∈{0.1,0.2,0.3,0.4}\beta\in\{0.1,0.2,0.3,0.4\}. Since the current-phase relations J2,off(ϕ2)J_{2,\mathrm{off}}(\phi_2) in the closed regime is almost constantly zero, the two-level system can be considered practically closed.

The energy of level εB\varepsilon_B approaches then progressively that of level εA\varepsilon_A in the subsequent columns of Figure 5. As εB→0\varepsilon_B\to0, the detrimental effects on the ideal switching mechanism become more pronounced, for two readily identifiable reasons. First, the output current J2,on(ϕ2)J_{2,\mathrm{on}}(\phi_2) in the open regime becomes increasingly reshaped resulting in a diminishing of the critical current J2,onJ_{2,\mathrm{on}} as εB→0\varepsilon_B\to0. Second, level B leaks increasingly larger supercurrents in the closed regime with Figure 5(aFigure 4-(dFigure 4 showing scenarios with comparable critical values of supercurrents in open and closed regimes. The critical current J2,offJ_{2,\mathrm{off}} then becomes comparable to its open-state counterpart J2,onJ_{2,\mathrm{on}}, reducing thus the supercurrent switching ratio towards impractical values. Crucially, we observe two main trends when changing α\alpha and β\beta. Increasing the loop coupling-asymmetry α\alpha, which breaks the geometric resonance condition g1B=g3Bg_{1B}=g_{3B} for level B, raises the leakage current J2,offJ_{2,\mathrm{off}} in the closed state. On the other hand, increasing β\beta, which reduces the coupling g2Bg_{2B} of level B to the output terminal 2, suppresses the leakage current. While the former is detrimental for supercurrent switching, the latter has benign consequences for the design and operation of the three-terminal device. Consequently, by raising the barrier between the QD region and terminal 2, both g2Ag_{2A} and g2Bg_{2B} are expected to decrease in comparison to the loop coupling strengths, which then drives β\beta effectively towards its optimal operation value. Such an approach is also consistent with the results on a single dominant level A as obtained in Figure 2, where relative tunneling rates of γ2\gamma_2 approaching 0 generally yield a wide resonance in RR, making the SC transistor effect more accessible in practical terms.

Leakage current due to the inter-level Coulomb repulsion

Inter-level Coulomb repulsion U′U' represents a ubiquitous phenomenon in systems where multiple levels are embedded in Josephson junctions. It can severely alter both the ABS spectrum and the resulting transport properties [69, 70]. To model analogous effects in the QD region of the three-terminal SC transistor, we generalize the ZBW Hamiltonian of Sec. V A by adding to (24) the term

HAB=U′(d↑,A†d↑,A+d↓,A†d↓,A)(d↑,B†d↑,B+d↓,B†d↓,B).(33)H_{\mathrm{AB}} = U'\left(d^\dagger_{\uparrow,\mathrm{A}}d_{\uparrow,\mathrm{A}}+d^\dagger_{\downarrow,\mathrm{A}}d_{\downarrow,\mathrm{A}}\right)\left(d^\dagger_{\uparrow,\mathrm{B}}d_{\uparrow,\mathrm{B}}+d^\dagger_{\downarrow,\mathrm{B}}d_{\downarrow,\mathrm{B}}\right). \tag*{(33)}

Building upon the results of Sec. V A, we fix β=0.4\beta=0.4, which corresponds to a favorable regime for the output barrier, and assess the loop-coupling asymmetries via α∈{0.0,0.2,0.4}\alpha\in\{0.0,0.2,0.4\}. The resulting current-phase relations are shown in Figure 6, whose layout is analogous to that of Figure 5: the level separation decreases from left to right as εB/Δ=3\varepsilon_B/\Delta=3 towards 0, while U′U' is fixed within each row and increases downwards from U′/Δ=0U'/\Delta=0 in Figure 6(aFigure 1-(aFigure 4 to U′/Δ=1.5U'/\Delta=1.5 in Figure 6(dFigure 1-(dFigure 4. The uppermost row thus reproduces the β=0.4\beta=0.4 results of Figure 5 and serves as the reference.

Figure 6. Current-phase relation of the output supercurrent J2J_2 in the three-terminal structure with two energy levels εA\varepsilon_A and εB\varepsilon_B in the presence of the inter-level Coulomb repulsion U′U'. The level A is kept at half-filling by setting εA/Δ=−UA/2=−1.5\varepsilon_A/\Delta=-U_A/2=-1.5 (UA/Δ=3U_A/\Delta=3), while UB/Δ=2U_B/\Delta=2, t/Δ=0.6t/\Delta=0.6 and β=0.4\beta=0.4. The level separation decreases from left to right as εB/Δ=3\varepsilon_B/\Delta=3 (leftmost column) towards εB/Δ=0\varepsilon_B/\Delta=0 (rightmost column), whereas U′U' is fixed within each row and increases downwards from U′/Δ=0U'/\Delta=0 [(a1)-(a4)] to U′/Δ=1.5U'/\Delta=1.5 [(d1)-(d4)]. Each panel shows J2(φ2)J_2(\varphi_2) in the open regime (black) as well as in the closed regime at g1Bg_{1B}-g3Bg_{3B} asymmetry values α∈{0,0.2,0.4}\alpha\in\{0,0.2,0.4\} (blue, orange, red). The uppermost row reproduces the β=0.4\beta=0.4 results of Figure 5. Increasing U′U' shifts level B upwards and thereby restores the current-phase relation characteristic of well-separated levels, recovering the 0-π\pi transition and suppressing the leakage in the closed regime.

Reading Figure 6 along its columns reveals a systematic trend. In the first column, where the two levels are well separated, level B is only marginally occupied making U′U' negligible for the system. Increasing U′U' in such a scenario thus enhances the critical current in the open regime only slightly. Also the shifts of the 0-π\pi transition to marginally higher φ2\varphi_2 are present. However, the already severely suppressed leakage currents in the closed regime are further diminished. Towards the right-hand columns, where the U′=0U'=0 results of the uppermost row show a distorted open-regime current-phase relation and a sizable leakage, the effect of U′U' becomes pronounced. Moderate values restore a current-phase relation resembling that of well-separated levels: the 0-π\pi jump reappears, the critical current in the open regime is recovered, and the closed-regime leakage is suppressed for all values of α\alpha considered.

This behavior is readily understood. Since level A is kept at half-filling, ⟨nA⟩≈1\langle n_A\rangle\approx1 and the inter-level repulsion shifts level B upwards by approximately U′U', so that U′U' effectively restores part of the level separation that Sec. VI identified as protective. The progressive shift of the 0-π\pi transition with increasing U′U', most visible in the right-hand columns, is a direct signature of this mechanism. Within the parameter range examined, the inter-level repulsion therefore acts in favor of the transistor operation, which suggests that it needs neither to be suppressed nor compensated in an experimental realization.

Design Proposal in 2DEG Platforms

The theoretical requirements identified in the preceding sections translate directly into a concrete device architecture that is realizable with existing nanofabrication techniques. Following various studies of multi-terminal Josephson junctions that utilized the combination of electrostatic gateability of superconductivity in hybrid materials [39, 40, 51], we thus consider a planar superconductor/semiconductor (SC/SE) heterostructure in which epitaxial Al is grown in situ on a shallow InAs quantum well hosting a two-dimensional electron gas (2DEG) as shown in Figure 1. Crucially, the 2DEG carrier density and thus all tunneling barriers are continuously tunable by electrostatic gates, enabling the precise barrier control that the resonance condition (21) demands.

The proposed device geometry, as shown in Figure 1, consists of an SC layer (typically Al) that is selectively removed to expose three arms of the junction. Electrostatic top gates, separated from the heterostructure surface by a thin oxide dielectric (e.g. Al2_2O3_3 or HfOx_x deposited by atomic layer deposition), locally deplete the 2DEG beneath each junction to define three independently tunable tunneling barriers and confine the central region into a gate-defined QD. Such a fully top-down approach, illustrated schematically in Figure 1(b), has been demonstrated for two- and three-terminal planar Josephson junctions on the same Al/InAs platform in Refs. [39, 40, 51, 71], including independent gate control over each terminal-to-dot barrier down to the single-mode or pinch-off regime. Magnetic flux threading the SC loop is controlled by an on-chip flux line or a small external coil, as routinely done in these geometries [39].

The main requirement for the ideal SC transistor mechanism to hold is the presence of a single dominant level in the central region: the resonance condition γ1=γ3\gamma_1 = \gamma_3 must hold for the transport-active level, and all other levels must be energetically remote or strongly suppressed at the output barrier. In the Al/InAs platform, this is achieved by increasing the output gate voltage to approach pinch-off of terminal 2. This simultaneously lowers γ2\gamma_2, which keeps the device inside the region of elevated switching ratio identified in Sec. IV, and suppresses the leakage contribution of higher-lying levels as demonstrated in § 7. The loop barriers γ1\gamma_1 and γ3\gamma_3 are then tuned symmetrically to satisfy (21), while the total hybridization is kept at intermediate strength Γ≈Δ\Gamma\approx\Delta, where the resonance is surrounded by the broadest region of significant transistor yield.

achieved by threading half a flux quantum through the control loop while tuning the barriers to γ3=(1−γ2)/2=γ1\gamma_3 = (1-\gamma_2)/2 = \gamma_1, i.e. so that both arms of the loop couple to the QD with equal tunneling strengths. At this point the two loop phasors cancel, the geometric factor χ\chi becomes independent of the output phase, and the output supercurrent vanishes identically while the loop sustains a purely circulating supercurrent. The device thus operates as a symmetric SQUID at half-flux quantum. Since the condition involves only the geometric current factor χ2\chi^2, it is independent of the Coulomb interaction UU, the level position εd\varepsilon_d, the total hybridization Γ\Gamma, and temperature, all of which enter solely through the prefactor JT(χ)\mathcal{J}_T(\chi).

Interactions, however, are significant for practical operations where barrier tuning is not ideal. NRG calculations across the full γ2\gamma_2-γ3\gamma_3 parameter space show that the resonance line is present for any parameters, but only at intermediate hybridization Γ≈Δ\Gamma\approx\Delta it becomes embedded in a broad region of elevated RR. Both weaker and stronger coupling reduce this region to a thin isolated line, so that intermediate Γ\Gamma constitutes the optimal operating regime. The structures surrounding the resonance originate in the singlet-doublet QPTs encoded in JT(χ)\mathcal{J}_T(\chi): elevated ratios require phase diagrams in which singlet islands are well separated by doublet regions, whereas striped phase diagrams yield R≈1R \approx1. Finite temperature smooths the underlying discontinuity and suppresses the supercurrent amplitude, but does so comparably in the open and closed regimes, leaving the switching ratio essentially intact up to T/Δ=0.08T/\Delta= 0.08. The associated reduction of transmitted supercurrents nevertheless sets a practical lower bound on the operating temperature.

Realistic devices host additional transport-relevant levels in the QD region, and since the gates act on barriers rather than on the actual orbitals, a configuration that symmetrizes the loop couplings of the transport-dominant level generically leaves those of the next level asymmetric. Leakage through the closed device is therefore unavoidable. As expected, leakage is negligible as long as the second level remains well separated in energy as modeled by a ZBW approach. Crucially, raising the output tunneling barrier suppresses it considerably. Benignly, such an operating mode is also beneficial for the purely single-level scenario where lower values of γ2\gamma_2 show stable resonance width and improve the accessibility of the SC transistor effect. Inter-level Coulomb repulsion was found to act in the same favorable direction, as it shifts the parasitic level upwards and thereby restores part of the protective level separation.

Taken together, these results define a compact set of design rules for a gate-tunable device manifesting flux-controlled SC switching behavior. First, the two loop barriers shall be matched by gate tuning, while the total hybridization is kept at intermediate strength Γ≈Δ\Gamma\approx\Delta. Second, the output barrier should be raised relative to the tunneling between the loop terminals and the QD region. Both requirements are within reach of existing nanofabrication for shallow Al/InAs 2DEG platforms, making this

Conclusions

We have demonstrated the working principle of a three-terminal SC transistor switch with an embedded QD level that operates on purely geometric principles. Ideal switching with a diverging ratio R=Jon/Joff→∞R = J_{\mathrm{on}}/J_{\mathrm{off}} \to\infty is geometric switching principle a viable route toward SC transistor elements for quantum circuits and cryogenic classical electronics.

Acknowledgments

The authors would like to thank Tomáš Novotný for helpful discussions. DN was supported by Grant No. 23-05263K of the Czech Science Foundation of the Czech Ministry of Education, Youth and Sports. KW was supported by the European Union and the Ministry of Education, Youth and Sports of the Czech Republic (Project: MSCA Fellowships CZ FZU IV - CZ.02.01.01/00/22 010/0013442). FK acknowledges support by Lumina Quaeruntur fellowship LQ100102602 of the Czech Academy of Sciences. WB acknowledges support by the DAAD PPP (Projekt-ID: 57753337) and the Deutsche Forschungsgemeinschaft (DFG; German Research Foundation) by SFB 1432 (Project No. 425217212). PZ acknowledges the support by the Czech Republic-Germany Mobility programme (ID:8J25DE001) and by the TERAFIT project - CZ.02.01.01/00/22_008/0004594.

Data Availability

The data that support the findings of this article are openly available [72].

References

  1. [1]M. Lundstrom, Moore’s law forever?, Science 299, 210 (2003).pubmed.ncbi.nlm.nih.gov/12522237
  2. [2]M. T. Bohr and I. A. Young, CMOS scaling trends and beyond, IEEE Micro 37, 20 (2017).DOI
  3. [3]S. J. Paukka, K. Das, R. Kalra, A. Moini, Y. Yang, M. Trainer, A. Bousquet, C. Cantaloube, N. Dick, G. C. Gardner, M. J. Manfra, and D. J. Reilly, A cryogenic CMOS chip for generating control signals for multiple qubits, Nat. Electron. 4, 64 (2021).
  4. [4]G. De Simoni, F. Paolucci, P. Solinas, E. Strambini, and F. Giazotto, Metallic supercurrent field-effect transistor, Nat. Nanotechnol. 13, 802 (2018).DOI
  5. [5]F. Paolucci, G. De Simoni, E. Strambini, P. Solinas, and F. Giazotto, Ultra-efficient superconducting Dayem bridge field-effect transistor, Nano Lett. 18, 4195 (2018).DOI
  6. [6]G. De Simoni, F. Paolucci, C. Puglia, and F. Giazotto, Josephson field-effect transistors based on all-metallic Al/Cu/Al proximity nanojunctions, ACS Nano 13, 7871 (2019).DOI
  7. [7]G. De Simoni, C. Puglia, and F. Giazotto, Niobium Dayem nano-bridge Josephson gate-controlled transistors, Appl. Phys. Lett. 116, 242601 (2020).DOI
  8. [8]C. Puglia, G. De Simoni, and F. Giazotto, Electrostatic control of phase slips in Ti Josephson nanotransistors, Phys. Rev. Appl. 13, 054026 (2020).DOI
  9. [9]M. Rocci, G. De Simoni, C. Puglia, D. D. Esposti, E. Strambini, V. Zannier, L. Sorba, and F. Giazotto, Gate-controlled suspended titanium nanobridge supercurrent transistor, ACS Nano 14, 12621 (2020).DOI
  10. [10]M. F. Ritter, A. Fuhrer, D. Z. Haxell, S. Hart, P. Gumann, H. Riel, and F. Nichele, A superconducting switch actuated by injection of high-energy electrons, Nat. Commun. 12, 1266 (2021).
  11. [11]M. Rocci, D. Suri, A. Kamra, G. Vilela, Y. Takamura, N. M. Nemes, J. L. Martinez, M. G. Hernandez, and J. S. Moodera, Large enhancement of critical current in superconducting devices by gate voltage, Nano Lett. 21, 216 (2021).DOI
  12. [12]L. Bours, M. T. M. Mercaldo, M. Cuoco, E. Strambini, and F. Giazotto, Unveiling mechanisms of electric field effects on superconductors by a magnetic field response, Phys. Rev. Res. 2, 033353 (2020).
  13. [13]L. Chirolli, T. Cea, and F. Giazotto, Impact of electro-static slips in layered crystalline BCS superconductors, Phys. Rev. Res. 3, 023135 (2021).DOI
  14. [14]A. Amoretti, D. K. Brattan, N. Magnoli, L. Martinoia, I. Matthaiakis, and P. Solinas, Destroying superconductivity in thin films with an electric field, Phys. Rev. Res. 4, 033211 (2022).DOI
  15. [15]S. Chakraborty, D. Nikolić, J. C. Cuevas, F. Giazotto, A. Di Bernardo, E. Scheer, M. Cuoco, and W. Belzig, Microscopic theory of supercurrent suppression by gate-controlled surface depairing, Phys. Rev. B 108, 184508 (2023).DOI
  16. [16]A. Zaccone and V. M. Fomin, Theory of superconductivity in thin films under an external electric field, Phys. Rev. B 109, 144520 (2024).DOI
  17. [17]L. Ruf, C. Puglia, T. Elalaily, G. De Simoni, F. Joint, M. Berke, J. Koch, A. Iorio, S. Khorshidian, P. Makk, S. Gasparinetti, S. Csonka, W. Belzig, M. Cuoco, F. Giazotto, E. Scheer, and A. Di Bernardo, Gate control of superconducting current: Mechanisms, parameters, and technological potential, Appl. Phys. Rev. 11, 041314 (2024).
  18. [18]G. J. Dolan, Offset masks for lift-off photoprocessing, Appl. Phys. Lett. 31, 337 (1977).DOI
  19. [19]K. K. Likharev, Superconducting weak links, Rev. Mod. Phys. 51, 101 (1979).DOI
  20. [20]H. Takayanagi and T. Kawakami, Superconducting proximity effect in the native inversion layer on InAs, Phys. Rev. Lett. 54, 2449 (1985).pubmed.ncbi.nlm.nih.gov/10031345
  21. [21]J. Nitta, T. Akazaki, H. Takayanagi, and K. Arai, Transport properties in an InAs-inverted-channel In₀.₅₂Al₀.₄₈As/In₀.₅₃Ga₀.₄₇As heterostructure coupled superconducting junction, Phys. Rev. B 46, 14286 (1992).pubmed.ncbi.nlm.nih.gov/10003520
  22. [22]T. Akazaki, H. Takayanagi, J. Nitta, and T. Enoki, A Josephson field effect transistor using an InAs-inserted-channel In₀.₅₂Al₀.₄₈As/In₀.₅₃Ga₀.₄₇As inverted modulation-doped structure, Appl. Phys. Lett. 68, 418 (1996).DOI
  23. [23]A. Y. Kasumov, R. Deblock, M. Kociak, B. Reulet, H. Bouchiat, I. I. Khodos, Y. B. Gorbatov, V. T. Volkov, C. Journet, and M. Burghard, Supercurrents through single-walled carbon nanotubes, Science 284, 1508 (1999)pubmed.ncbi.nlm.nih.gov/10348734
  24. [24]Y.-J. Doh, J. A. van Dam, A. L. Roest, E. P. A. M. Bakkers, L. P. Kouwenhoven, and S. De Franceschi, Tunable superconducting current through semiconductor nanowires, Science 309, 272 (2005).
  25. [25]P. Jarillo-Herrero, J. A. van Dam, and L. P. Kouwenhoven, Quantum superconducting transistors in carbon nanotubes, Nature 439, 953 (2006).
  26. [26]J.-P. Cleuziou, W. Wernsdorfer, V. Bouchiat, T. Ondarçuhu, and M. Monthioux, Carbon nanotube superconducting quantum interference device, Nat. Nanotechnol. 1, 53 (2006).
  27. [27]J.-P. Cleuziou, W. Wernsdorfer, S. Andersgassen, M. Florens, V. Bouchiat, T. Ondarçuhu, and M. Monthioux, Gate-tuned high frequency response of carbon nanotube Josephson junctions, Phys. Rev. Lett. 99, 117001 (2007).DOI
  28. [28]H. I. Jørgensen, K. Grove-Rasmussen, T. Novotný, K. Flensberg, and P. E. Lindelof, Electron transport in single-wall carbon nanotube weak links in the Fabry–Pérot regime, Phys. Rev. Lett. 96, 207003 (2006).pubmed.ncbi.nlm.nih.gov/16803198
  29. [29]J. Shabani, M. Kjaergaard, H. J. Suominen, Y. Kim, F. Nichele, K. Pakrouski, T. Stanekevic, R. M. Lutchyn, P. Krogstrup, R. Feidenhans’l, S. Kraemer, C. Nayak, M. Troyer, C. M. Marcus, and C. J. Palmstrøm, Two-dimensional epitaxial superconductor–semiconductor heterostructures: A platform for topological superconducting networks, Phys. Rev. B 93, 155402 (2016).DOI
  30. [30]M. Kjaergaard, F. Nichele, H. J. Suominen, M. P. Nowak, M. Wimmer, A. R. Akhmerov, J. A. Folk, K. Flensberg, J. Shabani, C. J. Palmstrøm, and C. M. Marcus, Quantized conductance doubling and hard gap in a two-dimensional semiconductor–superconductor heterostructure, Nat. Commun. 7, 12841 (2016).DOI
  31. [31]C. T. Ke, C. M. Moehle, F. K. de Vries, C. Thomas, S. Metti, C. R. Guinn, R. Kallaher, M. Lodari, G. Scappucci, T. Wang, R. E. Diaz, G. C. Gardner, M. J. Manfra, and S. Goswami, Ballistic superconductivity and tunable π-junctions in InSb quantum wells, Nat. Commun. 10, 3764 (2019).DOI
  32. [32]E. Bocquillon, R. S. Deacon, J. Wiedenmann, P. Leubner, T. M. Klapwijk, C. Brüne, K. Ishibashi, H. Buhmann, and L. W. Molenkamp, Gapless Andreev bound states in the quantum spin Hall insulator HgTe, Nat. Nanotechnol. 12, 137 (2017).DOI
  33. [33]F. Vigneau, R. Mizokuchi, D. C. Zauz, X. Huang, S. Tan, R. Maurand, S. Frolov, A. Sammak, G. Scappucci, F. Lefloch, and S. De Franceschi, Germanium quantum-well Josephson field-effect transistors and inverters, Nano Lett. 19, 1023 (2019).DOI
  34. [34]M. Luethi, K. Laubscher, S. Bosco, D. Loss, and J. Klinovaja, Planar Josephson junctions in germanium: Effect of cubic spin-orbit interaction, Phys. Rev. B 107, 053435 (2023).DOI
  35. [35]A. Leblanc, C. Tangchingchai, Z. Sadre Momtaz, E. Kiyooka, J.-M. Hartmann, F. Gustavo, J.-L. Thomassin, B. Brun, V. Schmitt, S. Zihlmann, R. Maurand, É. Dumur, S. De Franceschi, and F. Lefloch, Gate- and flux-tunable sin(2φ) Josephson element with planar-Ge junctions, Nat. Commun. 16, 1010 (2025).DOI
  36. [36]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, 6991 (2018).DOI
  37. [37]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
  38. [38]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, 1039 (2019).DOI
  39. [39]M. Coraiola, D. Z. Haxell, D. Sabonis, H. Weisbrich, A. E. Svetogorov, M. Hinderling, S. C. ten Kate, E. Cheah, F. Krizek, R. Schott, W. Wegscheider, J. C. Cuevas, W. Belzig, and F. Nichele, Phase-engineering the Andreev band structure of a three-terminal Josephson junction, Nat. Commun. 14, 6784 (2023).
  40. [40]M. Coraiola, A. E. Svetogorov, D. Z. Haxell, D. Sabonis, M. Hinderling, S. C. ten Kate, E. Cheah, F. Krizek, R. Schott, W. Wegscheider, J. C. Cuevas, W. Belzig, and F. Nichele, Flux-tunable Josephson diode effect in a hybrid four-terminal Josephson junction, ACS Nano 18, 9221 (2024).
  41. [41]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, and F. Nichele, Exploring the energy spectrum of a four-terminal Josephson junction: Toward topological Andreev band structures, Phys. Rev. X 15, 031066 (2025).DOI
  42. [42]T. Jonckheere, J. Rech, T. Martin, B. Douçot, D. Feinberg, and R. Mélin, Multipair dc Josephson resonances in a biased all-superconducting bijunction, Phys. Rev. B 87, 214501 (2013).DOI
  43. [43]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).pubmed.ncbi.nlm.nih.gov/21770667
  44. [44]R. Mélin and D. Feinberg, Quantum interferometer for quartets in superconducting three-terminal Josephson junctions, Phys. Rev. B 107, L161405 (2023).DOI
  45. [45]F. Zhang, A. S. Rashid, M. T. Ahari, W. Zhang, K. M. Ananthanarayanan, R. Xiao, G. J. de Coster, M. J. Gilbert, N. Samarth, and M. Kayyalah, Andreev processes in mesoscopic multiterminal graphene Josephson junctions, Phys. Rev. B 107, L140503 (2023).DOI
  46. [46]R.-P. Riwar, M. Houzet, J. S. Meyer, and Y. V. Nazarov, Multi-terminal Josephson junctions as topological matter, Nat. Commun. 7, 11167 (2016).
  47. [47]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
  48. [48]E. Strambini, S. D’Ambrosio, F. Vischi, F. S. Bergeret, Y. V. Nazarov, and F. Giazotto, The ω-SQUIPT as a tool to phase-engineer Josephson materials, Nat. Nanotechnol. 11, 1055 (2016).DOI
  49. [49]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).
  50. [50]H. Weisbrich, R. L. Klees, G. Rastelli, and W. Belzig, Second Chern number and non-Abelian Berry phase in topological superconducting systems, PRX Quantum 2, 010310 (2021).DOI
  51. [51]M. Coraiola, D. Z. Haxell, D. Sabonis, M. Hinderling, S. C. ten Kate, E. Cheah, F. Krizek, R. Schott, W. Wegscheider, and F. Nichele, Spin-degeneracy breaking and parity transitions in three-terminal Josephson junctions, Phys. Rev. X 14, 031024 (2024).
  52. [52]J. A. van Dam, Y. V. Nazarov, E. P. A. M. Bakker, S. De Franceschi, and L. P. Kouwenhoven, Supercurrent reversal in quantum dots, Nature 442, 667 (2006).pubmed.ncbi.nlm.nih.gov/16900196
  53. [53]H. I. Jørgensen, T. Novotný, K. Grove-Rasmussen, K. Flensberg, and P. E. Lindelof, Critical current 0–π transition in designed Josephson quantum dot junctions, Nano Lett. 7, 2441 (2007).
  54. [54]P. Zalom, M. Žonda, and T. Novotný, Hidden symmetry in interacting-quantum-dot-based multimterminal Josephson junctions, Phys. Rev. Lett. 132, 126505 (2024).DOI
  55. [55]R. C. Jaklevic, J. Lambe, A. H. Silver, and J. E. Mercereau, Quantum interference effects in Josephson tunneling, Phys. Rev. Lett. 12, 159 (1964).
  56. [56]C. D. Tesche and J. Clarke, dc SQUID: Noise and optimization, J. Low Temp. Phys. 29, 301 (1977).
  57. [57]T. A. Fulton, L. N. Dunkleberger, and R. C. Dynes, Quantum interference properties of double Josephson junctions, Phys. Rev. B 6, 855 (1972).DOI
  58. [58]J. Koch, T. M. Yu, J. Gambetta, A. A. Houck, D. I. Schuster, J. Majer, A. Blais, M. H. Devoret, S. M. Girvin, and R. J. Schoelkopf, Charge-insensitive qubit design derived from the Cooper pair box, Phys. Rev. A 76, 042319 (2007).DOI
  59. [59]P. Zalom, D. Rolih, and R. Žitko, Andreev bound state spectroscopy of a quantum-dot-based Aharonov–Bohm interferometer with superconducting terminals, Phys. Rev. B 113, 075130 (2026).DOI
  60. [60]In 2DEG systems, this level can be selected as the last one to hold electrons before the QD region is fully depleted electrostatically.
  61. [61]P. Zalom, G. Diniz, D. C. Ohnmacht, and W. Belzig, Symmetry-induced Weyl nodes in interacting multi-terminal Josephson junctions (2026), arXiv:2608.21200 [cond-mat.mes-hall].arxiv.org/abs/2608.21200
  62. [62]K. Satori, H. Shiba, O. Sakai, and Y. Shimizu, Numerical renormalization group study of magnetic impurities in superconductors, J. Phys. Soc. Jpn. 61, 3239 (1992).
  63. [63]T. Yoshioka and Y. Ohashi, Numerical renormalization group studies on single impurity Anderson model in superconductivity: A unified treatment of magnetic, non-magnetic impurities, and resonance scattering, J. Phys. Soc. Jpn. 69, 1812 (2000).DOI
  64. [64]P. Zalom, Rigorous Wilsonian renormalization group for impurity models with a spectral gap, Phys. Rev. B 108, 195123 (2023).DOI
  65. [65]F. Siano and R. Egger, Josephson current through a nanoscale magnetic quantum dot, Phys. Rev. Lett. 93, 047002 (2004).DOI
  66. [66]D. J. Luitz and F. F. Assaad, Weak-coupling continuous-time quantum Monte Carlo study of the single impurity and periodic Anderson models with s-wave superconducting baths, Phys. Rev. B 81, 024509 (2010).DOI
  67. [67]D. J. Luitz, F. F. Assaad, T. Novotný, C. Karrasch, and V. Meden, Understanding the Josephson current through a Kondo-correlated quantum dot, Phys. Rev. Lett. 108, 227001 (2012).DOI
  68. [68]R. Žitko, NRG Ljubljana (8f90ac4) (2021), https://doi.org/10.5281/zenodo.4841076.
  69. [69]D. E. Logan, A. P. Tucker, and M. R. Galpin, Common non-Fermi liquid phases in quantum impurity physics, Phys. Rev. B 90, 075150 (2014).DOI
  70. [70]K. Wrześniewski and I. Weymann, Kondo physics in double quantum dot based Cooper pair splitters, Phys. Rev. B 96, 195409 (2017).DOI
  71. [71]G. V. Graziano, M. Gupta, M. Pendharkar, J. T. Dong, C. P. Dempsey, C. Palmström, and V. S. Pribiag, Selective control of conductance modes in multi-terminal Josephson junctions, Nat. Commun. 13, 5933 (2022).DOI
  72. [72]D. Novák, K. Wrześniewski, F. Krizek, W. Belzig, and P. Zalom, Data for the manuscript “Gate-tunable supercurrent switching in a three-terminal quantum dot transistor” (2026), https://doi.org/10.5281/zenodo.23983552.

Paper details

vlnc:W_3cmgx3cr
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 8

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