Time-dependent Andreev reflection

Dmitri V. Averin1, Gongqi Wang1, and Andrey S. Vasenko2,3 1Department of Physics and Astronomy, Stony Brook University, SUNY, Stony Brook, NY 11794-3800 2National Research University Higher School of Economics, 101000 Moscow, Russia 3I.E. Tamm Department of Theoretical Physics, P.N. Lebedev Physical Institute, Russian Academy of Sciences, 119991 Moscow, Russia

We extend the basic theory of Andreev reflection (AR) in a normal metal/superconductor junction to the situation with an arbitrary time-dependent bias voltage V (t) across the junction. The central element of the theory is the fact that the Fourier transform of the AR amplitude has a casual structure. As an example, the theory is used to describe the current response to short pulses of the bias voltage, which create coherent superposition of states with different energies. The current oscillates in time, with the gap frequency ∆/~, and also as a function of the pulse area R V (t)dt, with the period of the single- flux quantum e/h.

Andreev reflection (AR) [1] is the process of conversion of in a normal metal (N) into Cooper pairs in a superconductor (S) and vice versa, and represent the main mechanism of electron transport across an NS in- terface with large electron transparency. As a result, AR determines the basic transport characteristics of the NS junctions, including the linear conductance [2, 3], average current [4–6] and current noise [7, 8]; in junctions with very low transparency, one can observe individual AR FIG. 1: Sketch of (a) a short constriction between a normal metal transitions [9]. AR also gives rise to an enormous amount (N) and a superconductor (S) used as the basic model of an NS of various other transport phenomena. To give just a few junction, and (b) a quasiparticle scattering scheme in the junction that consists of the normal scattering with the scattering matrix examples, it is the basic mechanism of the supercurrent U (U ∗ for the holes), energy change of the incident flow in Josephson junctions [10–12] and, in the form of according to Eq. (3) due to the time-dependent bias voltage V (t), multiple Andreev reflections (MAR), determines all their and Andreev reflection at the NS interface with the amplitude A() transport characteristics at finite bias voltages: average (5). Also indicated schematically in (b) is the Fermi level in the current [13–19], current noise [20–23] and full statistics two electrodes. of charge transfer [24, 25]. AR produces thermoelectric effects in NS junctions [26–30], and plays an important R role in superconducting structures with other materials, total “magnetic flux” V (t)dt carried by the pulse with e.g., carbon nanotubes [31–33], graphene [34–37], topo- the period of the single-electron flux quantum h/e. logical insulators [38–41]. In all these situations, AR is We begin by outlining the derivation of our main re- typically considered under the conditions of the constant sults. We use the most basic model of an NS junction, a bias voltage, when the energy of the quasiparticles which small constriction between the normal and superconduct- determine the AR amplitude can be taken to be constant ing electrodes (Fig. 1a) with all transport modes charac- throughout the scattering process. However, in many sit- terized by one transparency D. For short constriction, uations, a more detailed theory of AR with an arbitrary it is possible to consider the constriction region itself time-dependent bias voltage V (t) is desirable. The pri- as normal, reducing the transport inside it to the mo- mary goal of this work is to develop such a theory. tion of independent quasiparticles. We assume that the Physically, the main novel feature of the time- bias voltage V (t) across the constriction is arbitrary, but dependent bias is the creation of coherent quantum su- varies on the time scale set by the energy gap ∆ of the perposition of the quasiparticle states with different en- superconductor and other small energies in the problem, ergies. As a result, the junction produces an oscillatory e.g., temperature T ; i.e., its characteristic frequencies are arXiv:2007.07483v1 [cond-mat.mes-hall] 15 Jul 2020 current response, the magnitude of which is sensitive to smaller than the microscopic energy scales set by the the coherence properties of these superpositions. For in- Fermi energy and by the traversal time of the barrier stance, as shown below, the oscillating current generated that determines the transparency D. In this regime, we by the AR processes is more stable to thermal averaging can neglect the effect of the bias voltage on D, and also than that produced by the quasiparticle tunneling, since use the quasiclassical approximation for the quaiparticle AR always involve two quasiparticles with vanishing to- motion through the constriction. This general approach tal energy. Also, in the limit of short pulses of the bias is similar to the one used to describe the time-dependent voltage, the quantum superposition of quasiparticle ener- transport in normal conductors – see, e.g., [42]. gies leads to the current oscillations as a function of the One starts by accounting for the effect of the bias volt- 2 age V (t) on the quasiparticle motion between the two For the holes incident from the normal electrode, the AR electrodes. The voltage creates the electric field E(t, x) amplitude is the same, while the scattering matrix is U ∗. localized in the constriction, V (t) = R dxE(t, x), where Solving this scattering scheme for the electrons and the x is the coordinate along the constriction. (The assump- holes, taking the standard average over their equilibrium tion of the relatively low frequencies of V (t) made above energy distributions at temperature T , and combining also implies that E(t, x) is quasistatic from the point of the electron and the hole contributions to the current, view of electromagnetism.) Describing this field through we get the total current in the constriction: the vector potential A(t, x), E = −∂A/∂t, and solving eN Z Z I(t) = dωdνW ∗(ω)W (ν)ei(ω−ν)t d[f( − ω) the time-dependent Scr¨odingerequation with A(t, x) in 2π the quasiclassical approximation, we see that the ampli- h A∗() A( + ν − ω) tude ψ (t) of the wavefunction of an electron crossing the 2 e −f( + ν)] · 1 + D ∗ 2 2 constriction from the normal metal to the superconduc- 1 − [A ()] R 1 − A ( + ν − ω)R 1 − [A∗()]2 1 − A2( + ν − ω) i tor acquires the A(t, x)-dependent phase: −R , (6) 1 − [A∗()]2R 1 − A2( + ν − ω)R ie Z ψe(t) → exp{ dxA(t, x)}ψe(t) where N is the number of the spin-degenerate trans- ~ Z t port modes in the constriction. In principle, this ex- −ie 0 0 −iφ(t) = exp{ dt V (t )}ψe(t) = e ψe(t) , (1) pression can be used directly to calculate the current in ~ the NS junction. For instance, for constant bias voltage, where the phase φ is defined by the relation φ˙ = e V (t). W (ω) = δ(ω − eV/~), and Eq. (6) reduces to the well- ~ Electrons passing through the constriction in the op- known expression [4], which describes the evolution of the posite direction accumulate the phase of the opposite dc IV curves from the regime of the tunnel to the bal- sign. The phases acquired by the holes are switched listic junction with increasing quasiparticle transparency in comparison to those for the electrons. Physically, D. For general time-dependent voltage, however, it is these phase factors describe the acceleration/deceleration more convenient to transform Eq. (6) explicitly into the of the quasiparticles as they move between the junction time domain. electrodes. In the quasiclassical approximation, the par- As the first step in this direction, we separate the term ticle energies change, while the change of their velocity in Eq. (6) that does not decay at large ||. The magnitude is small and is neglected. of this term in the integral over  is 1−R = D and it gives R Next, we introduce the Fourier components of the ac- D d[f(−ω)−f(+ν)] = D(ω +ν), meaning that this cumulated phase: contribution in Eq. (6) corresponds to the normal-state Z current IN (t) in the junction: e−iφ(t) = dωW (ω)e−iωt, (2) eND Z I (t) = dωdν(ω + ν)W ∗(ω)W (ν)ei(ω−ν)t N 2π in close analogy to what is done in the “Werthamer the- eND  ∂  ory” of the time-dependent properties of the Josephson = − i eiφ(t) e−iφ(t) = GV (t) . (7) π ∂t tunnel junctions [43, 44]. Then, the electron acceleration 2 process (1) has the following form in terms of the energy Here G = e ND/π~ is the normal-state conductance. After the separation of the normal-state part, the cur- components a() of the wavefunction ψe(t): rent can be expressed as Z a() → dωW (ω)a( − ω) . (3) eN Z h Z W (ω)A( + ω) 2 I = I + df() D2 dωe−iωt N 2π 1 − A2( + ω)R The quasiparticle acceleration/deceleration described Z 2 2 2 −iωt W (ω)A ( + ω) above should be combined with the standard quasipar- +RD dωe + 2DR 1 − A2( + ω)R ticle scattering scheme in the NS junction as illustrated n Z W (ω)A2( + ω) 2o i schematically in Fig. 1b. For electrons incident from the ·Re eiφ(t) dωe−iωt − ... , (8) normal electrode on the constriction at energy , the scat- 1 − A2( + ω)R tering process consists of the normal barrier scattering where the ellipsis denotes the subtracted identical terms characterized by the scattering matrix U, in which A( + ω) is replaced everywhere with A∗( − ω). In this expression, we can transform the AR amplitude  r , t  U = , r0 = −t0r∗/t∗, (4) into the time domain. More precisely, we introduce the t0 , r0 two response functions that enter Eq. (8): 2 0 2 2 0 2 with |t| = |t | = D and |r| = |r | = R, and Andreev i Z A(x) K(τ) = dxe−ixτ , (9) reflection at the NS interface with the amplitude: 2π 1 − A2(x)R √ x − sgn(x) x2 − 1, |x| > 1 , 1 Z A2(x) A() = √ x = /∆. (5) L(τ) = dxe−ixτ , (10) x − i 1 − x2, |x| < 1 , 2π 1 − A2(x)R 3 where τ is the time normalized to the gap frequency, τ = where Φ is the total area under the voltage pulse, which t∆/~, and the prefactors are chosen for later convenience. has the meaning of the magnetic flux carried by this An important property of the AR amplitude A(x) (5) pulse, and φV ≡ eΦ/h is the magnitude of this flux in is that it can be viewed as the reduction to the real axis units of the single-electron flux quantum. Equations (7) √of the function of the complex variable z: A(z) = z − and (14) show that the current in this case is z2 − 1. The function A(z) is analytic on the whole z plane except for the cut on the [−1, 1] interval of the real I(t) = GΦδ(t) + Θ(t)I0(t) sin(2πφV ) , (15) axis, and A(x) (5) on this interval is the value of A(z) where qualitatively, the current I (t) oscillates and de- on the upper (Imz > 0) branch of the cut. This property 0 cays on the time scale /∆. The magnitude of this oscil- implies that the AR amplitude in the time domain has a ~ latory response is modulated periodically, with the period clear casual structure, vanishing for τ < 0: of the single-electron flux quantum, by the area Φ of the i Z i I bias voltage pulse. dxe−ixτ A(x) = Θ(τ) dze−izτ A(z) 2π 2π C Quantitatively, the current I0(t) is Z 1 2 p 2 J1(τ) 2 Z t Z 0 00 = Θ(τ) dx 1 − x cos xτ = Θ(τ) , (11) 2eNT ∆nD ∆ 0 dt π 0 τ I0(t) = 2 dt 0 00 ~ ~ 0 −∞ sinh[πT (t − t )/~] where C is the contour going clock-wise around the ·[K(t − t0)K(t − t00) − RL(t − t0)L(t − t00)] branch cut, and J is the Bessel function. One can see di- 1 Z 0 L(t − t0) o rectly that the denominator in the functions (9) and (10) 0 +DR dt 0 .(16) does not add any poles to these functions, and therefore, −∞ sinh[πT (t − t )/~] they have the same analytical properties as the AR am- Equation (16) can be evaluated explicitly at large tem- plitude A(z). From this, we obtain directly the following peratures, T  ∆. In this limit, different terms in (16) expressions: have two different behaviors as functions of temperature. √ 2(1 + R) Z 1 1 − x2 cos xτ The last term, which corresponds to the quasiparticle K(τ) = Θ(τ) dx , (12) π (1 + R)2 − 4x2R tunneling in the tunnel-junction limit, decays exponen- 0 √ tially with T : 4 Z 1 x 1 − x2 sin xτ L(τ) = −Θ(τ) dx 2 2 (qp) 4G∆ π 0 (1 + R) − 4x R −πT t/~ I0 (t) = RL(t)e . (17) 2 ∂K(τ) e = , τ > 0 . (13) (1 + R) ∂τ This temperature dependence reflects the fact that the quasiparticles tunnel at different energies, and thermal Taking the inverse Fourier transform to express the averaging of the partial amplitudes at different energies amplitudes W (ω) in terms of φ(t), and using the func- results in the exponential decay. In contrast to this, the tions (12) and (13) in Eq. (8), we obtain our main general other two terms in Eq. (16) decay only as 1/T , as a re- result for the current in the NS junction driven by an ar- sult of all AR processes having total zero energy, and bitrary time-dependent bias voltage, expressed directly therefore, thermal averaging reducing only the probabil- in the time domain: ity of the incident quasiparticle to have the initial energy 2 Z t 0 00 eNT ∆nD ∆ 0 00 sin[φ(t ) − φ(t )] within the range of ∆. Indeed, in the limit T  ∆, the I(t) = IN + 2 dt dt 0 00 ~ ~ −∞ sinh[πT (t − t )/~] first two terms in Eq. (16) give at times t larger than ~/T , when the exponentially decaying quasiparticle con- ·[K(t − t0)K(t − t00) − RL(t − t0)L(t − t00)] tribution (17) can be neglected: Z t sin[φ(t) − φ(t0)] o +2DR dt0 L(t − t0) . (14) πG∆2D sinh[πT (t − t0)/ ] I (t) = [K2(t) − RL2(t)]. (18) −∞ ~ 0 2eT Equation (14) for the time-dependent NS current can Equations (14) – (18) are valid for junctions with arbi- be used to calculate the current under many different bias trary quasiparticle transparency D. They can be simpli- conditions. For intance, it shows that the main qualita- fied further in the two limits of ballistic and the tunnel tive feature of the junction response to the voltage that junction. For the ballistic junction with D = 1, the ker- varies rapidly on the time scale /∆ is interference of the ~ nel K(τ) (9) is given directly by the Fourier transform quasiparticle reflection from the gap edges which leads (11) of the AR amplitude, and the total current is: to oscillations of the current in time with the gap fre- quency, and also to oscillations in magnitude with the Z t 0 00 n πT 0 00 sin[φ(t ) − φ(t )] applied voltage. Consider the simplest model of the volt- I(t) = G V + dt dt 0 00 e −∞ sinh[πT (t − t )/~] age pulse that is infinitely short on the gap time scale: J [(t − t0)∆/ ]J [(t − t00)∆/ ]o · 1 ~ 1 ~ . (19) V (t) = Φδ(t) , φ(t) = 2πφV Θ(t) , (t − t0)(t − t00) 4

portional to D2. In both of this contributions, one can calculate the kernels K,L taking D = 0, i.e. R = 1 in Eqs. (12) and (13) to get 1 1 K(τ) = Θ(τ)J (τ) ,L(τ) = − Θ(τ)J (τ) , (20) 2 0 2 1 where J’s are Bessel functions. Equation (18) for the large-temperature oscillatory AR current is simplified then accordingly. The single-particle current is described by the last term in Eq. (14) and explicitly is:

Z t 0 n πT ∆ 0 sin[φ(t) − φ(t )] IT (t) = G V (t) − dt 0 e~ −∞ sinh[πT (t − t )/~] 0 o ·J1[(t − t )∆/~] . (21)

One can also see directly that expansion in D of the gen- FIG. 2: Oscillating current in a ballistic NS junction driven by the 2 δ-function pulse of the bias voltage, as a function of time t for sev- eral form of this term does not have the D part, and eral temperatures T . The oscillations are produced by interference therefore, the AR part of the current is given directly by 2 of the quasiparticle Andreev-reflection from the gap edges and are the D term in Eq. (14). With the current kernels (20), suppressed as 1/T at large temperatures. the time-dependent AR current in the tunnel limit is:

2 2 Z t 0 00 eNT D ∆ 0 00 sin[φ(t ) − φ(t )] IAR(t) = 3 dt dt 0 00 4~ −∞ sinh[πT (t − t )/~] 0 00 0 00 ·[J0(τ − τ )J0(τ − τ ) − J1(τ − τ )J1(τ − τ )]. (22) Finally, an important point to check is how the ideal- ized δ-function limit of the voltage pulses is approached by the pulses of finite time width. To do this, we adopt the shape of the bias voltage pulse: 1 Φ/t 0 t/t0 V (t) = , φ(t) = 4φV arctan e , (23) π cosh t/t0 related to the one that can be produced by switching Josephson junctions in the context of superconductor electronics. An example of the current in the NS tunnel junction calculated in the quasiparticle approximation (21) for this pulse shape at low temperatures is shown FIG. 3: Current in an NS tunnel junction as a function of time t in Fig. 3. This Figure shows that the main qualitative driven by the two voltage pulses (23) with two different amplitudes prediction of the δ-function approximation, flux modu- and the same characteristic time width t0. The curves illustrate lation of the interference component of the NS current, the pulse flux control of the interference current which manifests is reproduced by the pulses of the-not-extremely small itself through the oscillating “tail” of the current response to the short voltage pulse. The oscillation amplitude is modulated by duration t0 = 0.1~/∆. The work at Stony Brook was supported by the IARPA the flux φV carried by the pulse according to Eq. (15). In the plotted curves, the oscillations are suppressed for the larger pulse Supertools program through the Synopsys and Hypres with φV = 0.5 in comparison to the smaller pulse with φV = 0.25. subcontracts. The authors would like to acknowledge useful discussions with the participants of this program.

In this case, the amplitude I0(t) (15) can be calculated numerically from Eq. (19) and is shown in Fig. 2 for sev- eral values of the temperature T . Outside of a small [1] A.F. Andreev, “The thermal conductivity of the interme- range t ∼ ~/T near t = 0, where I0 goes to zero as t ln t, the lowest curve in Fig. 2 agrees with Eq. (18), which for diate state in superconductors”, Zh. Eksp. Teor. Fiz. 46, 1823 (1964) [Sov. Phys. JETP 19, 1228 (1964)]. D = 1 simplifies to I (t) = πG[ J (t∆/ )/t]2/(2eT ). 0 ~ 1 ~ [2] A. Kastalsky, A.W. Kleinsasser, L.H. Greene, R. Bhat, In the tunnel limit D  1, one can separate the single- F.P. Milliken, and J.P. Harbison, “Observation of pair particle contribution IT to the current, which is propor- currents in superconductore- contacts”, tional to D and the Andreev-reflection current IAR pro- Phys. Rev. Lett. 67, 3026 (1991). 5

[3] B.J. van Wees, P. de Vries, P. Magn´ee, and T. (1996). M. Klapwijk, “Excess conductance of superconductore- [21] Y. Naveh and D.V. Averin, “Nonequilibrium current semiconductor interfaces due to phase conjugation be- noise in mesoscopic disoredered SNS junctions”, Phys. tween electrons and holes”, Phys. Rev. Lett. 69, 510 Rev. Lett. 82, 4090 (1999). (1992). [22] X. Jehl, P. Payet-Burin, C. Baraduc, R. Calemczuk, and [4] G.E. Blonder, M. Tinkham, and T.M. Klapwijk, “Tran- M. Sanquer, “Andreev reflection enhanced in sition from metallic to tunneling regimes in supercon- mesoscopic SNS junctions, Phys. Rev. Lett. 83, 1660 ducting microconstrictions: Excess current, charge im- (1999). balance, and supercurrent conversion”, Phys. Rev. B 25, [23] Y. Ronen, Y. Cohen, J.H. Kang, A. Haim, M.T. Rieder, 4515 (1982). M. Heiblum, D. Mahalu, and H. Shtrikman, “Charge of a [5] A.F. Volkov, A.V. Zaitsev, and T.M. Klapwijk, “Prox- quasiparticle in a superconductor”, Proc. Nat. Acad. Sci. imity effect under nonequilibrium condictions in double- 113, 1743 (2016). barrier superconducting junctions”, Physica C 210, 21 [24] J.C. Cuevas and W. Belzig, “Full counting statistics (1993). of multiple Andreev reflections”, Phys. Rev. Lett. 91, [6] F.W.J. Hekking and Yu.V. Nazarov, “Interference of 187001 (2003). two electrons entering a superconductor”, Phys. Rev. [25] G. Johansson, P. Samuelsson, and A. Ingerman, “Full Lett. 71, 1625 (1993); “Subgap conductivity of a counting statistics of multiple Andreev reflections”, Phys. superconductor-normal metal tunnel interface”, Phys. Rev. Lett. 91, 187002 (2003). Rev. B 49, 6847 (1994). [26] A. Bardas and D.V. Averin, “Peltier effect in NS micro- [7] B.A. Muzykantskii and D.E. Khmelnitskii, “Quantum contacts”, Phys. Rev. B 52, 12873 (1995). shot noise in a normal-metal-superconductor point con- [27] A. Brinkman, A.A. Golubov, H. Rogalla, F.K. Wil- tact”, Phys. Rev. B 50, 3982 (1994). helm, and M.Yu. Kupriyanov, “Microscopic nonequilib- [8] A.V. Galaktionov and A.D. Zaikin, “Shot noise and rium theory of double-barrier Josephson junctions” Phys. of Andreev reflection”, Phys. Rev. B Rev. B 68, 224513 (2003). 80, 174527 (2009). [28] S. Rajauria, P. Gandit, T. Fournier, F.W.J. Hekking, [9] V.F. Maisi, O.-P. Saira, Yu.A. Pashkin, J.S. Tsai, D.V. B. Pannetier, and H. Courtois, Andreev current-induced Averin, and J.P. Pekola, “Real-time observation of dis- heating in a hybrid superconducting tunnel junction, crete Andreev tunneling events”, Phys. Rev. Lett. 106, Phys. Rev. Lett. 100, 207002 (2008). 217003 (2011). [29] A.S. Vasenko, E.V. Bezuglyi, H. Courtois, and F.W.J. [10] I.O. Kulik and A.N. Omelyanchuk, “Josephson effect in Hekking, Electron cooling by diffusive normal metal- superconducting microbridges: microscopic theory”, Sov. superconductor tunnel junctions, Phys. Rev. B 81, J. Low Temp. Phys. 4, 142 (1978). 094513 (2010). [11] A. Furusaki and M. Tsukada, “DC Josephson effect and [30] J.P. Pekola, J.V. Koski, and D.V. Averin, “Refrigerator Andreev reflection”, Sol. State. Commun. 78, 299 (1991). based on the Coulomb barrier for single-electron tunnel- [12] C.W.J. Beenakker and H. van Houten, Josephson cur- ing”, Phys. Rev. B 89, 081309(R) (2014). rent through a superconducting quantum point contact [31] M.R. Buitelaar, W. Belzig, T. Nussbaumer, B. Babi´c, shorter than the coherence length, Phys. Rev. Lett. 66, C. Bruder, and C. Sch¨onenberger, “Multiple Andreev re- 3056 (1991). flections in a carbon nanotube quantum dot”, Phys. Rev. [13] T.M. Klapwijk, G.E. Blonder, and M. Tinkham, “Ex- Lett. 91, 057005 (2003). planantion of sub-harmonic energy-gap structure in su- [32] P. Jarillo-Herrero, J.A. van Dam, and L.P. Kouwenhoven, perconducting contacts” Physica B+C 109-110, 1657 “Quantum supercurrent transistors in carbon nanotubes”, (1982). Nature 439, 953 (2006). [14] G.B. Arnold, “Superconducting tunneling without the [33] T. Hata, R. Delagrange, T. Arakawa, S. Lee, R. De- tunneling Hamiltonian. II. Subgap harmonic structure” block, H. Bouchiat, K. Kobayashi, and M. Ferrier, “En- J. Low Temp. Phys. 68, 1, (1987). hanced shot noise of multiple Andreev reflections in a car- [15] U. Gunsenheimer and A.D. Zaikin, “Ballistic charge bon nanotube quantum dot in SU(2) and SU(4) Kondo transport in superconducting weak links”, Phys. Rev. B regimes”, Phys. Rev. Lett. 121, 247703 (2018). 50, 6317 (1994). [34] C.W.J. Beenakker, “Colloquium: Andreev reflection and [16] E.N. Bratus, V.S. Shumeiko, and G. Wendin, “Theory Klein tunneling in graphene”, Rev. Mod. Phys. 80, 1337 of subharmonic gap structure in superconducting meso- (2008). scopic tunnel contacts”, Phys. Rev. Lett. 74, 2110 (1995). [35] Xu Du, I. Skachko, and E.Y. Andrei, “Josephson current [17] D.V. Averin and A. Bardas, “AC Josephson effect in and multiple Andreev reflections in graphene SNS junc- a single quantum channel”, Phys. Rev. Lett. 75, 1831 tions”, Phys. Rev. B 77, 184507 (2008). (1995). [36] N. Mizuno, B. Nielsen, and Xu Du, “Ballistic-like su- [18] J.C. Cuevas, A. Martin-Rodero, and A. Levy Yeyati, percurrent in suspended graphene Josephson weak links”, “Hamiltonian approach to the transport properties of su- Nat. Commun. 4, 2716 (2013). perconducting quantum point contacts”, Phys. Rev. B 54, [37] D.K. Efetov, L.Wang, C. Handschin, K.B. Efetov, J. 7366 (1996). Shuang, R. Cava, T. Taniguchi, K. Watanabe, J. Hone, [19] E. Scheer, W. Belzig, Y. Naveh, M.H. Devoret, D. Esteve, C.R. Dean, and P. Kim, “Specular interband Andreev re- and C. Urbina, “Proximity effect and multiple Andreev flections at van der Waals interfaces between graphene reflections in gold atomic contacts”, Phys. Rev. Lett. 86, and NbSe2”, Nat. Phys. 12, 328 (2016). 284 (2001). [38] D.M. Badiane, M. Houzet, and J.S. Meyer, “Nonequilib- [20] D.V. Averin and H.T. Imam, “Supercurrent noise in rium Josephson effect through helical edge states”, Phys. quantum point contacts”, Phys. Rev. Lett. 76, 3814 Rev. Lett. 107, 177002 (2011). 6

[39] A.D.K. Finck, C. Kurter, Y.S. Hor, and D.J.V. Harlin- 122, 097003 (2019). gen, “Phase coherence and Andreev reflection in topolog- [42] M.V. Moskalets, “Scattering matrix approach to non- ical insulator devices”, Phys. Rev. X 4, 041022 (2014). stationary quantum transport”, (Imperial College Press, [40] L.A. Jauregui, M. Kayyalha, A. Kazakov, I. Miotkowski, 2012). L.P. Rokhinson, and Y.P. Chen, “Gate-tunable su- [43] N.R. Werthamer, “Nonlinear self-coupling of Josephson percurrent and multiple Andreev reflections in a radiation in superconducting tunnel junctions”, Phys. superconductor- topological insulator nanoribbon- Rev. 147, 255 (1966). superconductor hybrid device”, Appl. Phys. Lett. 112, [44] A.I. Larkin and Yu. N. Ovchinnikov, “Tunnel effect be- 093105 (2018). tween superconductors in an alternating field”, Zh. Eksp. [41] T. Jonckheere, J. Rech, A. Zazunov, R. Egger, A. Levy Teor. Fiz. 51, 1535 (1966) [Sov. Phys. JETP 24, 1035 Yeyati, and T. Martin, “Giant shot noise from Majorana (1967)]. zero modes in topological trijunctions”, Phys. Rev. Lett.