Finite-frequency noise in a topological superconducting wire

Stefano Valentini

NEST, Scuola Normale Superiore, and Istituto Nanoscienze-CNR, I-56126 Pisa, Italy

Michele Governale

MacDiarmid Institute for Advanced Materials and Nanotechnology, School of Chemical and Physical Sciences, Victoria University of

Wellington, P.O. Box 600, Wellington 6140, New Zealand

Rosario Fazio

NEST, Scuola Normale Superiore, and Istituto Nanoscienze-CNR, I-56126 Pisa, Italy

Fabio Taddei∗

NEST, Istituto Nanoscienze-CNR and Scuola Normale Superiore, I-56126 Pisa, Italy

Abstract

In this paper we study the finite-frequency current cross-correlations for a topological superconducting nanowire attachedto two terminals at one of its ends. Using an analytic 1D model we show that the presence of a Majorana bound stateyields vanishing cross-correlations for frequencies larger than twice the applied transport voltage, in contrast to what isfound for a zero-energy ordinary Andreev bound state. Zero cross-correlations at high frequency have been confirmedusing a more realistic tight-binding model for finite-width topological superconducting nanowires. Finite-temperatureeffects have also been investigated.

Contribution for the special issue of Physica E in memory of Markus Buttiker.

1. Introduction

One of the prototypical systems which host MajoranaBound States (MBS) is the Kitaev chain[1], a discretemodel for a one-dimensional p-wave superconductor. Sucha model can be realised in a semiconducting nanowire withstrong spin-orbit coupling by placing it in close proximityto a s-wave superconductor, thus inducing superconduc-tivity in the wire, and applying a strong magnetic fieldwhich leads to a large Zeeman splitting[2, 3]. With possi-ble solid-state realisations available, several experimentalstudies have gathered evidence compatible with the exis-tence of MBSs[4, 5, 6, 7, 8]. As a result, the quest foran unambiguous signature of the presence of a MBS hasbecome a priority and is stimulating a significant researcheffort[9, 10, 11, 12, 13]. Up to now only a few papers havefocused on the consequences of MBSs on the behaviour ofcurrent correlations [14, 15, 16, 17, 18, 19, 20, 21, 22, 23,24]. Very recently Haim et al. (Ref. [25]) have consideredthe spin-resolved current cross-correlations, finding thatthey are negative for a MBS in the case of correlationsbetween opposite spins.

∗Corresponding authorEmail address: fabio.taddei@sns.it (Fabio Taddei )

In this paper, we consider a semi-infinite topologicalsuperconducting wire attached to two normal terminalsat one end, as shown in Fig. 1. A bias voltage V is ap-plied to the two normal contacts (labelled 1 and 2), whilethe superconducting wire is grounded. We calculate thecross-correlations at finite frequency between the currentsI1 and I2 flowing in the two normal leads. Our main find-ing is that the cross-correlations at frequencies larger thantwice the voltage V vanish at zero temperature, when thesuperconducting wire is in the topological phase. On thecontrary, in the presence of an ordinary zero-energy An-dreev Bound State (ABS), which similarly to a MBS givesrise to a zero-bias peak in the differential conductance, thecross-correlations at high frequency are in general non-zeroand depend on the details of the system. The origin ofthis phenomenon can be attributed to the peculiar struc-ture of the energy-dependent scattering matrix for reflec-tions off a MBS. In addition, at zero-frequency the cross-correlations are always negative in the presence of a MBS,analogously to the case of spin-resolved spin-up/spin-downcross-correlations[25], but may be positive in the case ofan ABS. These results have been obtained with a simpleanalytical model whereby two normal leads are coupledeither to a Majorana state or a zero-energy level. Zero

Preprint submitted to Physica E March 7, 2018

arX

iv:1

508.

0783

2v1

[co

nd-m

at.m

es-h

all]

31

Aug

201

5

N N

S

1 2

3

V V

Figure 1: Schematic setup of the system. A superconducting (S)nanowire, which can be driven in and out of the topological phaseby an applied magnetic field, is contacted by means of a beam splitterto two normal (N) leads, labelled 1 and 2.

cross-correlations at high frequency are then confirmedusing a more realistic tight-binding model based on thesemiconducting-nanowire realisation of a one-dimensionalp-wave superconducting wire.

2. Formalism and methods

In this section we briefly review the scattering approachfor the finite-frequency current-current correlations in hy-brid superconducting systems [26]. The (non-symmetrized)current-current correlator between lead i and i′ is definedas

Sii′(t) = 〈Ii(t)Ii′(0)〉 − 〈Ii〉〈Ii′〉, (1)

where Ii(t) is the current1 operator at time t relative toterminal i and 〈· · · 〉 stands for the quantum-statistical av-erage. Taking the Fourier transform of Sii′(t) one obtainsthe finite-frequency correlator as

Sii′(ω) =

∫Sii′(t) e

iωt dt. (2)

Within the Landauer-Buttiker scattering approach[28, 30],the finite-frequency current-current correlator in a hybridsuperconducting system, calculated in the normal leads, is

1In this paper we consider only quasiparticles current and neglectthe role of displacement currents; the latter might induce correc-tions to the noise at high frequency in the case of strongly energy-dependent density of state [27, 28, 29]. This must be taken into ac-count in the analysis of actual experimental data. Such corrections,however, depend on the details of the system and their discussion isbeyond the scope of the present paper.

given by[26]

Sii′(ω) =e2

2h

∑α,α′

β,β′

∑σ,σ′

τ,τ ′

∑j,j′

sign(β)sign(β′) (3)

×∫ +∞

−∞dE Aαjσα′j′σ′(β,E,E + ~ω, τ, i)

×Aα′j′σ′

αjσ (β′, E + ~ω,E, τ ′, i′)× fαi(E) [1− fα′i′(E + ~ω)] ,

where the indices α, α′, β, β′ ∈ {±1} indicate electrons(+1) and holes (-1) in Nambu space, σ, σ′, τ, τ ′ refer tothe spin-projection quantum number and i, i′, j, j′ labelthe leads. The Fermi distribution function in the normallead i for a α-like particle at temperature T and voltageVi is given by

fαi(E) =

[1 + exp

(E − αeVikBT

)]−1

. (4)

Moreover, in Eq. (3) we have defined

Aαjσα′j′σ′(β,E,E′, τ, i) = δαβδστδjiδα′βδσ′τδj′i

− [siτjσβα (E)]?siτj′σ′

βα′ (E′), (5)

where siτjσβα (E) is the scattering amplitude at energy Efor a α-like particle with spin σ injected from lead j tobe reflected as a β-like particle with spin τ in lead i. Inthe rest of this paper, we shall focus on the symmetrizednoise, defined as SSii′(ω) = Sii′(ω) + Sii′(−ω), since thisis the quantity that is measured by a classical detector[31]. We shall furthermore assume that the two normalterminals are kept at the same voltage (V1 = V2 = V ).

3. Analytic 1D model

The system depicted in Fig. 1 can be modelled in asimple way by composing[30] the scattering matrix sM ofa normal lead coupled to a Majorana state with the scat-tering matrix sbs of a 3-leg beam splitter, which describesthe connection to terminals 1 and 2. The matrix sM canbe calculated from the Hamiltonian describing a normallead coupled to a Majorana state (see Ref. [25])

H =∑k,σ

εkψ†kσψkσ + iγ

∑k,σ

(tσψkσ + tσψ

†kσ

), (6)

where the first term describes the lead, ψ†kσ being the cre-ation operator of a spin-σ particle with momentum k andenergy εk, and the second term the coupling to the lo-calised Majorana fermion γ. Without loss of generality,we assume the coupling parameters t↑ and t↓ to be real.Using Eq. (6) the scattering matrix in Nambu space takesthe following form

sM =

(rMee rM

eh

rMhe rM

hh

). (7)

2

Here rMαβ are matrices, in spin space, of reflection ampli-

tudes given by

rMee =

(1 00 1

)+

1

iE − Γ

(Γ↑

√Γ↑Γ↓√

Γ↑Γ↓ Γ↓

)(8)

and

rMhe =

1

iE − Γ

(Γ↑

√Γ↑Γ↓√

Γ↑Γ↓ Γ↓

), (9)

where Γ↑ = 2πν0|t↑|2, Γ↓ = 2πν0|t↓|2, Γ = Γ↑ + Γ↓ withν0 being the density of states of the normal leads, whilerMeh and rM

hh are determined by particle-hole symmetryrMα,β(E) = [rM

−α,−β(−E)]?.Assuming no spin mixing to occur in the beam-splitter,

the block of the scattering matrix sbs for spin-σ electronsis a 3 × 3 unitary matrix which can be parameterized asfollows[32]

sbseσ =

cσ λ1σ λ2σ

λ1σ a1σ bσλ2σ bσ a2σ

. (10)

Here λ1(2)σ is the scattering amplitude for an electron withspin σ injected from lead 1(2) to be transmitted in 3 (withλ2

1σ + λ22σ ≤ 1), cσ = −s1

√1− λ2

1σ − λ22σ (with s1 = ±1)

is the scattering amplitude for an electron to be reflectedback in lead 3, bσ = −λ1σλ2σ(s2 + cσ)(λ2

1σ +λ22σ)−1 (with

s2 = ±1) is the amplitude for an electron to be transmittedfrom lead 1 to 2, and a1σ = (s2λ

22σ − cσλ2

1σ)(λ21σ + λ2

2σ)−1

[a2σ = (s2λ21σ − cσλ2

2σ)(λ21σ + λ2

2σ)−1] is the amplitude foran electron to be reflected back in lead 1(2). Hereafter,if not otherwise stated, we fix s1 = −1 and s2 = +1.The 3 × 3 block of the scattering matrix sbs for spin-σholes is related to the one for electrons by the electron-holesymmetry relation sbs

hσ = sbs?eσ , while the Andreev blocks

are zero. The total scattering matrix sbs has dimensions12× 12.

By substituting the scattering matrix obtained fromthe composition of sbs and sM into Eqs. (3) and (5) onegets the cross-correlator SS12(ω) plotted as a dashed redline in Fig. 2 for zero temperature and full spin degener-acy. Fig. 2 shows that SS12 is negative at ω = 0 and there-after decreases exhibiting a minimum around ~ω = eV .Remarkably, the noise vanishes at frequencies higher than2eV . While the occurrence of a minimum is not a genericfeature, we find that SS12(0) ≤ 0 and that SS12(ω) = 0 for~ω > 2eV independently of the choice of parameters Γσ,λ1σ and λ2σ, hence these two characteristic features persisteven when the spin degeneracy is removed. As an exam-ple, the solid blue line in Fig. 2 shows the cross-correlatorSS12 for Γ↑ = 0.1 eV , Γ↓ = 0.2 eV , λ1↑ = 0.8, λ2↑ = 0.6,λ1↓ = 0.6, λ2↓ = 0.8.

In contrast to the MBS case, the behaviour of SS12(ω)is different in the presence of an ordinary zero-energy An-dreev bound state (ABS). This can be realised in a Zeeman-split resonant level strongly proximized by an s-wave su-perconductor, which we describe through the following

0.0 0.5 1.0 1.5 2.0 2.5 3.0

hω/eV

−0.20

−0.15

−0.10

−0.05

0.00

0.05

SS 12[e

3V/h

]

non-degeneratedegenerate

Figure 2: Symmetrized current cross correlator SS12(ω) as a functionof frequency in the presence of a MBS for a spin-degenerate system(dashed red curve) and for the case of broken spin-degeneracy (solidblue curve). For the spin degenerate case the values of the systemparameters are: Γσ = 0.05 eV , λ1σ = 0.5 , λ2σ = 0.6 , s1 = −1and s2 = +1. For the case of broken spin degeneracy the values ofthe system parameters are: Γ↑ = 0.1 eV , Γ↓ = 0.2 eV , λ1↑ = 0.8,λ2↑ = 0.6, λ1↓ = 0.6, λ2↓ = 0.8.

reflection-amplitude matrices in spin space [25]

rAee =

1

iE − Γ/2

(iE +

Γ↑−Γ↓2 0

0 iE − Γ↑−Γ↓2

), (11)

rAhe =

√Γ↑Γ↓

iE − Γ/2

(0 11 0

), (12)

where, rAeh and rA

hh are determined as usual by electron-hole symmetry rA

α,β(E) = [rA−α,−β(−E)]?. We find that

SS12(ω) is in general finite even at high frequencies. As anexample we plot in Fig. 3 the cross correlator as a functionof frequency for a case of broken spin degeneracy.

As we shall show below, the vanishing cross-correlatorin the presence of MBS is a result of the peculiar struc-tural properties of the energy-dependent scattering matrix.This is easier to understand by abandoning the beam split-ter and by noticing that the index σ in the Hamiltonian(6) can be thought of as identifying a terminal insteadof a spin projection. Such a Hamiltonian would describea Majorana state γ separately coupled to two differentsingle-channel, i.e. fully spin-polarized, normal terminalswithout explicitly needing an additional beam splitter. Inthis case, the reflection amplitudes in Eqs. (8) and (9) canthen be rewritten in the normal-terminal space as

rMee =

(1 00 1

)+

1

iE − Γ

(Γ1

√Γ1Γ2√

Γ1Γ2 Γ2

)(13)

and

rMhe =

1

iE − Γ

(Γ1

√Γ1Γ2√

Γ1Γ2 Γ2

), (14)

3

0.0 0.5 1.0 1.5 2.0 2.5 3.0

hω/eV

−0.30

−0.25

−0.20

−0.15

−0.10

−0.05

0.00

0.05

SS 12[e

3V/h

]

Figure 3: Symmetrized current cross correlator SS12(ω) as a functionof frequency for an ABS, for the following values of the system param-eters: Γ↑ = 0.1eV , Γ↓ = 0.2eV , λ1↑ = λ2↓ = 0.7, λ2↑ = λ1↓ = 0.4,s1 = −1 and s2 = +1.

where now Γi is the coupling strength to lead i and Γ =Γ1 + Γ2. The cross-correlator can now be calculated bysubstituting the scattering matrix (7), with (13) and (14),in Eq. (3). Assuming for the sake of simplicity Γ1 = Γ2 =Γ/2 one obtains

SS12(ω) =

{2e2

hΓ2

|~ω| lnΓ2+(|eV |−|~ω|)2

Γ2+(eV )2 if |~ω| < 2|eV |0 if |~ω| ≥ 2|eV |.

(15)We have explicitly checked that SS12(ω) = 0 vanishes for|~ω| > |2eV | also when Γ1 6= Γ2. Incidentally the autocor-relations for this case are given in Appendix B. As shownin Appendix A, the reason why the high-frequency cor-relator vanishes at zero temperature, is the fact that the2× 2 matrices rM

ee and rMhe are related by

rMee = I + rM

he, (16)

where I is the 2 × 2 unity matrix. At finite tempera-ture T , the cross-correlator is exponentially suppressed for|~ω| > 2|eV | as long as kBT � 2|eV | and the frequencyis such that |~ω| − 2|eV | � kBT . For the sake of com-pleteness, it is important to notice that a MBS located atthe end of a superconducting wire can be coupled to twonormal leads through a 3-leg structure which in generaldoes not ensure that the two leads are only coupled viathe MBS. Indeed, for the 3-leg beam splitter modelled bythe scattering matrix in Eq. (10) the two leads are notcoupled separately due to the transmission coefficients bσ.Nevertheless, it turns out that when s2 = +1 the rela-tion (16) holds for any choice of the other parameters and,as shown in Fig. 2, the cross-correlations vanish at highfrequency. This is however not the case when s2 = −1.

Figure 4: Setup of the structure for the numerical tight-bindingmodel. A semiconducting nanowire of width w with strong spin-orbitcoupling and proximized by an s-wave superconductor (vertical blueregion 3) is attached to a normal semiconducting nanowire (red hor-izontal regions 1 and 2). Two barriers (black stripes) in proximityof the junction are used to reduce the direct coupling between thetwo normal leads. An in-plane magnetic field B, which can affecteither both nanowires or just the superconducting one, controls thetopological phase transition.

4. Numerical results

The results obtained with the minimal model reportedin the previous section still hold for a more realistic situa-tion. For this purpose, we consider a system based on semi-conducting nanowires with a strong spin-orbit coupling. Inthe presence of a Zeeman field and an induced s-wave su-perconducting order parameter, a strongly spin-orbit cou-pled nanowire hosts MBS at its ends when the parametersare properly tuned (see Refs. [33, 34, 35]). The system wesimulate is described in details in Fig. 4: a superconduct-ing nanowire (vertical) is attached to the side of a normalnanowire (horizontal). Two barriers (black stripes) are in-cluded in the normal nanowire in order to suppress thedirect coupling between terminals 1 and 2 and make theMBS couple separately to the two normal reservoirs, atleast to some extent. The tight-binding Hamiltonian ofthe superconducting nanowire reads

HS = −t∑〈i,j〉,σ

c†i,σcj,σ + (εS − µ)∑i,σ

c†i,σci,σ

+ iλSO

∑〈i,j〉,σ,σ′

(ν′ijσxσσ′ − νijσyσσ′)c†i,σcj,σ′ (17)

+B∑i,σ,σ′

σxσσ′c†i,σci,σ′ +

∑i

[∆ c†i,↑c

†i,↓ + H.c.

],

where t is the hopping energy, εS = 4t is a uniform on-siteenergy which sets the zero of energy, λSO is the Rashbaspin-orbit (SO) coupling strength, B is the Zeeman energyin the wire, ∆ is the induced superconducting pairing, σi

are spin-1/2 Pauli matrices, νij = x · dij , and ν′ij = y ·dij with dij = (ri − rj)/|ri − rj | being the unit vectorconnecting site j to site i. The Hamiltonian of the normal

4

nanowire reads

HN = −t∑〈i,j〉,σ

c†i,σcj,σ + (εN − µ)∑i,σ

c†i,σci,σ

+B′∑i,σ,σ′

σxσσ′c†i,σci,σ′ , (18)

and the barriers are implemented by means of the followingterm:

Hb = −t

(ηL − 1)

(bL)∑〈i,j〉,σ

c†i,σcj,σ + (ηR − 1)

(bR)∑〈i,j〉,σ

c†i,σcj,σ

+ηD

(bD)∑〈i,j〉,σ

c†i,σcj,σ

+ H.c., (19)

where the superscripts (bL), (bR) and (bD) in the sums in-dicate that the sites i, j are at the interfaces of the left (L)and right (R) barriers and between the horizontal wire andthe vertical one (D), respectively. The parameters ηL, ηRand ηD are the strengths of the coupling to the left, rightand bottom regions, respectively. Lead 3 is grounded whileleads 1 and 2 are kept at the same voltage V . In Eq. (18)εN is a uniform on-site energy and B′ is the Zeeman energyin the normal wire, which we choose to be either 0 or thesame as the Zeeman energy B of the superconducting one.In the following, we set ηD = 1, ηL = ηR = 0.1, µ = 0,λSO = 0.1t, ∆ = 0.1t and nanowire’s width w = 10 sites.Moreover, if not otherwise stated, we fix eV = 3× 10−4t.

Let us first consider the case when εN = εS and B′ =B, so that the normal nanowire supports a single, spin-polarised open channel. We have checked that, in thetopological phase (B = 0.2t), the Andreev and normal re-flection matrices [rMhe(E) and rMee (E)] are very well approx-imated by the analytical expressions (13) and (14), withΓ1 ' Γ2 ' 1.625 × 10−5t ' 0.054eV . In Fig. 5 the cross-correlator SS12(ω) is plotted for different temperatures. Atzero temperature (solid blue line) the behaviour is qualita-tively the same as the one for the analytical model plottedin Fig. 2. At finite temperatures the dip at ~ω ' eVis suppressed, while the noise at high frequencies remainsexponentially small in ~ω/kBT .

As discussed at the end of Sec. 3, a three-leg junc-tion gives rise in general also to a direct coupling betweenthe two normal terminals. This coupling can be controlledby the two barriers in the normal nanowire which tunethe transparencies ηL and ηR in Eq. (19). We studiedthe influence of the direct coupling on the cross-correlatorSS12(ω) and plotted in Fig. 6 the cross-correlator for dif-ferent values of the transparencies ηL and ηR which, forsimplicity, we assume equal, that is ηL = ηR = η. Be-sides a general increase of the noise magnitude due to thebroadening of the scattering probabilities peaks, one cannotice that for higher transparencies SS12(ω) tends towardsa linear behavior for small frequencies. For intermediatevalues of transparencies, the value of the cross-correlator

0.0 0.5 1.0 1.5 2.0 2.5 3.0

hω/eV

−0.030

−0.025

−0.020

−0.015

−0.010

−0.005

0.000

SS 12[e

3V/h

]

kBT/eV00.050.10.2

Figure 5: Symmetrized noise SS12(ω) calculated for the realisticsystem in the topological phase for different temperatures, with asingle open channel in the normal nanowire. The parameters fornormal and superconducting nanowires are µ = 0, λSO = 0.1t,B = B′ = 0.2t, w = 10 sites, and ∆ = 0.1t.

at high frequencies becomes non-negligible. As expected,the direct coupling of the normal leads renders the effectof the MBS less visible.

We also investigated the role of multiple (spin-) chan-nels in the normal nanowire by setting εN = 3.8t andB′ = 0, while all the other parameters are the same as inthe previous case. With such a choice of parameters thereare two (opposite-spin) channels for each normal terminal.The cross-correlator SS12(ω) in such a regime at zero tem-perature is almost equal to the T = 0 curve in Fig. 5, i.e.,SS12(0) < 0 and SS12(ω) ' 0 for |~ω| > 2|eV |.

We shall now study the case of an ordinary ABS occur-ring at zero energy and verify that it yields non-vanishingcross-correlations for |~ω| > 2|eV |. First we note that theoccurrence of a zero-energy ABS implies a peak in the An-dreev reflection probability only if at least 2 open channelsare present in the normal nanowire (otherwise because ofthe Beri degeneracy, the Andreev reflection probability iseither 1, for the topological phase, or 0, for the trivialphase). To do so, we fix the parameters of the normalnanowire as follows: εN = 3.8t and B′ = 0. In order to bein the trivial phase we also put B = 0.1t. Furthermore, weadd a gate voltage Vg and a magnetic field B′′ in the regionbetween the barriers in order to tune the Andreev reflec-tion probability to have just a single peak at the Fermienergy and nothing more structured in the whole subgapenergy range. This is realised by an additional term in theHamiltonian:

HG = −Vg∑i,σ

c†i,σci,σ +B′′∑i,σ,σ′

σzσσ′c†i,σci,σ′ . (20)

The cross-correlator SS12(ω) in the presence of a zeroenergy ABS is shown in Fig. 7. The main differences withrespect to the MBS are the increasing behaviour at very

5

0.0 0.5 1.0 1.5 2.0 2.5 3.0

hω/eV

−0.9

−0.8

−0.7

−0.6

−0.5

−0.4

−0.3

−0.2

−0.1

0.0

SS 12[e

3V/h

]

η0.10.20.30.41

×30

Figure 6: Symmetrized noise SS12(ω) at zero temperature calculatedfor the realistic system in the topological phase for different trans-parencies η, with a single open channel in the normal nanowire. Theparameters for normal and superconducting nanowires are µ = 0,λSO = 0.1t, B = B′ = 0.2t, w = 10 sites, and ∆ = 0.1t.

small frequencies and the evident non-zero noise at highfrequencies.

5. Conclusions

We have presented a comprehensive study of the fi-nite frequency current cross-correlations for a MBS cou-pled to two different normal reservoirs. By introducinga simple model for a MBS coupled to normal degrees offreedom combined with a three-leg beam-splitter, we haveobtained all finite frequency current correlations by meansof the scattering approach to mesoscopic transport. Wefind that the topological regime is characterised by van-ishing cross correlations for |~ω| > 2|eV | and negativecross correlations at zero frequency. The cancellation ofhigh-frequency fluctuations is due to the particular rela-tion between normal and Andreev scattering that is ful-filled in the MBS case. The absence of this feature in thenoise spectrum would rule out the MBS as the origin ofthe vanishing cross-correlation. We have then verified thatall results are still valid for a more realistic tight-bindingmodel for finite-width nanowires.

Acknowledgements

This work has been supported by the EU project “Ther-miQ”, by MIUR-PRIN “Collective quantum phenomena:from strongly correlated systems to quantum simulators”,by the EU project COST Action MP1209 “Thermody-namics in the quantum regime”, and by the EU projectCOST Action MP1201 “NanoSC”. M.G. acknowledgesthe hospitality of Scuola Normale Superiore, Pisa. R.Facknowledges the Oxford Martin School for support andthe Clarendon Laboratory for hospitality during the com-pletion of the work.

0.0 0.5 1.0 1.5 2.0 2.5

hω/eV

−0.0030

−0.0025

−0.0020

−0.0015

−0.0010

−0.0005

0.0000

SS 12[e

3V/h

]

Figure 7: Symmetrized noise SS12(ω) calculated for the realistic sys-tem in the trivial phase with a zero energy ABS at zero temperature,with two open channels in the normal nanowire. The parameters forthe normal nanowire are µ = 0, εN = 3.8t, λSO = 0, B′ = 0, w = 10sites, while for the superconducting nanowire are µ = 0, εS = 4t,λSO = 0.1t, w = 10 sites, ∆ = 0.1t, B = 0.1t and for the gate areVg = 0.25147785t and B′′ = 0.2t.

Appendix A. Vanishing cross-correlations for |~ω| >|2eV |

In this Appendix we prove that the scattering matri-ces in Eqs. (13) and (14) yield vanishing cross-correlationsfor frequencies |~ω| > |2eV |. Here we actually consider ascattering matrix that generalises the relation of Eq. (16),namely including additional phases, and can thus be parametrizedas

sijαβ(E) = δijδαβ + sij+−(E) ei[1−sign(α)]φi

2 ei[1+sign(β)]φj

2 ,(A.1)

where φi and φj are energy-independent phases. 2 In thisAppendix, i and j are collective indices for transport chan-nels which represent a pair of degrees of freedom, namelyterminal number and spin.

Let us consider the case of different-channel cross-correlation.The two channels (labelled 1 and 2 for simplicity) can beeither in the same lead with different spins or in differentleads, in which case the spin is not important. Eq. (3)becomes

S12(ω) =e2

2h

∫ +∞

−∞dE∑α,α′

β,β′

∑l,k

sign(β)sign(β′) (A.2)

× {δαβδl1δα′βδk1 − [s1lβα(E)]?s1k

βα′(E + ~ω)}× {δα′β′δk2δαβ′δl2 − [s2k

β′α′(E + ~ω)]?s2lβ′α(E)}

× fα(E) [1− fα′(E + ~ω)] .

2Particle-hole symmetry implies φi + φj = −2Arg[sij+−(0)].

6

Let us consider the quantity

Bαα′(ω) =∑l,kβ,β′

sign(β)sign(β′) (A.3)

× {δαβδl1δα′βδk1 − [s1lβα(E)]?s1k

βα′(E + ~ω)}× {δα′β′δk2δαβ′δl2 − [s2k

β′α′(E + ~ω)]?s2lβ′α(E)},

which is composed by 4 terms.The first term

B(1)αα′(ω) =

∑l,kβ,β′

sign(β)sign(β′)δαβδl1δα′βδk1δα′β′δk2δαβ′δl2

(A.4)is identically zero because of terms like δl1δl2 = 0.

The second term is

B(2)αα′(ω) = −

∑l,kβ,β′

sign(β)sign(β′)δαβδl1δα′βδk1 (A.5)

× [s2kβ′α′(E + ~ω)]?s2l

β′α(E)

= −δαα′sign(α){[s21+α(E + ~ω)]?s21

+α(E)

− [s21−α(E + ~ω)]?s21

−α(E)}.

The condition of Eq. (A.1) implies that B(2)αα′(ω) = 0.

The third term is

B(3)αα′(ω) = −

∑l,kβ,β′

sign(β)sign(β′)δα′β′δk2δαβ′δl2 (A.6)

× [s1lβα(E)]?s1k

βα′(E + ~ω)

= −δαα′sign(α){[s12+α(E)]?s12

+α(E + ~ω)

− [s12−α(E)]?s12

−α(E + ~ω)}.

As for the second term, the condition of Eq. (A.1) implies

that B(3)αα′(ω) = 0.

The forth term is

B(4)αα′(ω) =

∑l,kβ,β′

sign(β)sign(β′)[s1lβα(E)]? (A.7)

× s1kβα′(E + ~ω)[s2k

β′α′(E + ~ω)]?s2lβ′α(E)

=∑l,k

{[s1l+α(E)]?s1k

+α′(E + ~ω)− [s1l−α(E)]?s1k

−α′(E + ~ω)}

× {[s2k+α′(E + ~ω)]?s2l

+α(E)− [s2k−α′(E + ~ω)]?s2l

−α(E)}.

Because of Eq. (A.1), the only nonzero contributions to

B(4)αα′(ω) are the ones with l = 1 and k = 2 or l = 2 and

k = 1, and Eq. (A.7) simplifies to

B(4)αα′(ω) = sign(α)sign(α′)[(s12

+−(E)s21+−(E + ~ω) ei(φ1+φ2))?

+ s21+−(E)s12

+−(E + ~ω) ei(φ1+φ2)]. (A.8)

Notice that B(4)αα′(ω) does not depend on α and α′ but for

the sign, thus Eq. (A.2) reduces to

S12(ω) =e2

2h

∫ +∞

−∞dE[(s12

+−(E)s21+−(E + ~ω) ei(φ1+φ2))?

+ s21+−(E)s12

+−(E + ~ω) ei(φ1+φ2)] (A.9)

×∑α,α′

sign(α)sign(α′)fα(E) [1− fα′(E + ~ω)] .

It is possible to show that for T = 0 and ~ω > 2eV or~ω < −2eV the following expression holds:

F (E,ω) ≡∑α,α′

sign(α)sign(α′)fα(E) [1− fα′(E + ~ω)] = 0,

(A.10)thus the (symmetrized and non-symmetrized) cross-correlatoris zero for frequencies higher than 2eV .

At finite temperature T , the previous quantity does notvanish and evaluates to

F (E,ω) = −1

4sech[

E − eV2kBT

] sech[E + eV

2kBT] sinh[

eV

kBT]2

(A.11)

× sech[E − eV + ~ω

2kBT] sech[

E + eV + ~ω2kBT

],

which is exponentially vanishing, for ~ω > 2eV , as long askBT � 2eV and the frequency is such that ~ω − 2eV �kBT .

Appendix B. Auto-correlations

In this Appendix we consider the auto-correlations forthe model of Sec. 3 in the case where the beam splitter isabsent (or equivalently when the scattering matrix sbs

eσ ofEq. (10) is characterized by λ1σ = 1 and λ2σ = 0). In thiscase the auto-correlator S11(ω) for a MBS is calculated bysubstituting Eq. (7) in (3). Assuming t↑ = t↓ one gets:

S11(ω) =e2Γ2

2h

∫ +∞

−∞dE

1

(E2 + Γ2)[(E + ~ω)2 + Γ2]×(B.1)

{[4Γ2 + (~ω)2](F++ + F−−) + (2E + ~ω)2(F+− + F−+)}

whereFαβ = fα(E)[1− fβ(E + ~ω)]. (B.2)

We can discuss some limiting situations for the sym-metrized noise, defined as SS(ω) = S11(ω) +S11(−ω), andassuming eV > 0. At zero temperature Eq. (B.1) gives

SS(ω) =2e2

hΓ

[Arctan

(~ω + eV

Γ

)+ Arctan

(eV

Γ

)+

Γ

~ωln

Γ2 + (~ω − eV )2

(eV )2 + Γ2

], (B.3)

7

for 0 < ~ω < 2eV , and

SS(ω) =2e2

hΓ

[Arctan

(~ω + eV

Γ

)+ Arctan

(~ω − eV

Γ

)],

(B.4)for ~ω > 2eV .

At eV = 0 one finds

SS(ω) =4e2

hΓArctan

( |~ω|Γ

), (B.5)

whereas for eV 6= 0, but ω = 0 one finds

SS(0) =4e2

hΓ

(Arctan

eV

Γ− ΓeV

Γ2 + (eV )2

). (B.6)

The small-frequency expansion (~ω � eV or ~ω � Γ) ofEq. (B.3) reads

SS(ω) ' SS(0) +4e2

h

Γ4

(Γ2 + (eV )2)2~ω (B.7)

+2e2

h

Γ2eV (3Γ2 − 5(eV )2)

3(Γ2 + (eV )2)3(~ω)2.

Note that the quadratic term in ~ω contributes positivelyto the noise only if Γ >

√5/3eV . Now, by assuming that

Γ� eV , i.e. in the limit ~ω � Γ� eV , Eq. (B.7) becomes

SS(ω) ' SS(0) +4e2

h

(Γ

eV

)4

~ω, (B.8)

meaning that SS(ω) increases linearly with frequency. Onthe other hand, by assuming that Γ � ~ω, i. e. in thelimit Γ� ~ω � eV , Eq. (B.7) becomes

SS(ω) ' SS(0)− e2

h

10

3

(Γ

eV

)2(~ω)2

eV, (B.9)

meaning that SS(ω) decreases quadratically with frequency.We notice that, for an ABS, Equations (B.3) and (B.4)

still hold after replacing ω with 2ω and eV with 2eV .

References

References

[1] A. Y. Kitaev, Unpaired majorana fermions in quantum wires,Phys. Usp. 44 (10S) (2001) 131–136. doi:10.1070/1063-7869/

44/10s/s29.[2] R. M. Lutchyn, J. D. Sau, S. Das Sarma, Majorana fermions and

a topological phase transition in semiconductor-superconductorheterostructures, Phys. Rev. Lett. 105 (2010) 077001. doi:10.

1103/PhysRevLett.105.077001.[3] Y. Oreg, G. Refael, F. von Oppen, Helical liquids and majorana

bound states in quantum wires, Phys. Rev. Lett. 105 (17) (2010)177002. doi:10.1103/physrevlett.105.177002.

[4] V. Mourik, K. Zuo, S. M. Frolov, S. R. Plissard, E. P. A. M.Bakkers, L. P. Kouwenhoven, Signatures of majorana fermionsin hybrid superconductor-semiconductor nanowire devices, Sci-ence 336 (6084) (2012) 1003–1007. doi:10.1126/science.

1222360.

[5] A. Das, Y. Ronen, Y. Most, Y. Oreg, M. Heiblum, H. Shtrik-man, Zero-bias peaks and splitting in an al–InAs nanowire topo-logical superconductor as a signature of majorana fermions, Na-ture Phys. 8 (12) (2012) 887–895. doi:10.1038/nphys2479.

[6] M. T. Deng, C. L. Yu, G. Y. Huang, M. Larsson, P. Caroff,H. Q. Xu, Anomalous zero-bias conductance peak in a nb–insbnanowire–nb hybrid device, Nano Lett. 12 (12) (2012) 6414–6419. doi:10.1021/nl303758w.

[7] A. D. K. Finck, D. J. V. Harlingen, P. K. Mohseni, K. Jung,X. Li, Anomalous modulation of a zero-bias peak in a hy-brid nanowire-superconductor device, Phys. Rev. Lett. 110 (12)(2013) 126406. doi:10.1103/physrevlett.110.126406.

[8] H. O. H. Churchill, V. Fatemi, K. Grove-Rasmussen, M. T.Deng, P. Caroff, H. Q. Xu, C. M. Marcus, Superconductor-nanowire devices from tunneling to the multichannel regime:Zero-bias oscillations and magnetoconductance crossover, Phys.Rev. B 87 (24) (2013) 241401. doi:10.1103/physrevb.87.

241401.[9] J. Alicea, New directions in the pursuit of majorana fermions

in solid state systems, Rep. Prog. Phys. 75 (7) (2012) 076501.doi:10.1088/0034-4885/75/7/076501.

[10] M. Leijnse, K. Flensberg, Introduction to topological super-conductivity and majorana fermions, Semicond. Sci. Technol.27 (12) (2012) 124003. doi:10.1088/0268-1242/27/12/124003.

[11] C. W. J. Beenakker, Search for majorana fermions in super-conductors, Annu. Rev. Condens. Matter Phys. 4 (1) (2013)113–136. doi:10.1146/annurev-conmatphys-030212-184337.

[12] T. D. Stanescu, S. Tewari, Majorana fermions in semiconduc-tor nanowires: fundamentals, modeling, and experiment, J.Phys.: Condens. Matter 25 (23) (2013) 233201. doi:10.1088/

0953-8984/25/23/233201.[13] S. Das Sarma, M. Freedman, C. Nayak, Majorana Zero Modes

and Topological Quantum Computation, ArXiv e-printsarXiv:1501.02813.

[14] A. Golub, B. Horovitz, Shot noise in a majorana fermion chain,Phys. Rev. B 83 (15) (2011) 153415. doi:10.1103/physrevb.

83.153415.[15] B. H. Wu, J. C. Cao, Tunneling transport through supercon-

ducting wires with majorana bound states, Phys. Rev. B 85 (8)(2012) 085415. doi:10.1103/physrevb.85.085415.

[16] H.-F. Lu, H.-Z. Lu, S.-Q. Shen, Nonlocal noise cross correlationmediated by entangled majorana fermions, Phys. Rev. B 86 (7)(2012) 075318. doi:10.1103/physrevb.86.075318.

[17] P. Wang, Y. Cao, M. Gong, G. Xiong, X.-Q. Li, Cross-correlations mediated by majorana bound states, Europhys.Lett. 103 (5) (2013) 57016. doi:10.1209/0295-5075/103/57016.

[18] J. Liu, F.-C. Zhang, K. T. Law, Majorana fermion induced non-local current correlations in spin-orbit coupled superconduct-ing wires, Phys. Rev. B 88 (6) (2013) 064509. doi:10.1103/

physrevb.88.064509.[19] B. Zocher, B. Rosenow, Modulation of majorana-induced cur-

rent cross-correlations by quantum dots, Phys. Rev. Lett.111 (3) (2013) 036802. doi:10.1103/physrevlett.111.036802.

[20] H.-F. Lu, H.-Z. Lu, S.-Q. Shen, Current noise cross correla-tion mediated by majorana bound states, Phys. Rev. B 90 (19)(2014) 195404. doi:10.1103/physrevb.90.195404.

[21] H. Soller, A. Komnik, Charge transfer statistics of transportthrough majorana bound states, Physica E 63 (2014) 99–104.doi:10.1016/j.physe.2014.05.020.

[22] J. Li, T. Yu, H.-Q. Lin, J. Q. You, Probing the non-locality ofmajorana fermions via quantum correlations, Sci. Rep. 4 (2014)4930. doi:10.1038/srep04930.

[23] N. V. Gnezdilov, B. van Heck, M. Diez, J. A. Hutasoit, C. W. J.Beenakker, Topologically protected charge transfer along theedge of a chiral p-wave superconductor, ArXiv e-printsarXiv:1505.06744.

[24] J. Ulrich, F. Hassler, Majorana-assisted nonlocal electron trans-port through a floating topological superconductor, ArXiv e-printsarXiv:1506.01390.

[25] A. Haim, E. Berg, F. von Oppen, Y. Oreg, Signatures of ma-jorana zero modes in spin-resolved current correlations, Phys.

8

Rev. Lett. 114 (16) (2015) 166406. doi:10.1103/physrevlett.

114.166406.[26] M. P. Anantram, S. Datta, Current fluctuations in mesoscopic

systems with andreev scattering, Phys. Rev. B 53 (24) (1996)16390–16402. doi:10.1103/physrevb.53.16390.

[27] M. Buttiker, Role of scattering amplitudes in frequency-dependent current fluctuations in small conductors, PhysicalReview B 45 (7) (1992) 3807–3810. doi:10.1103/PhysRevB.45.3807.

[28] Y. Blanter, M. Buttiker, Shot noise in mesoscopic conduc-tors, Phys. Rep. 336 (1-2) (2000) 1–166. doi:10.1016/

s0370-1573(99)00123-4.[29] M. H. Pedersen, S. A. van Langen, M. Buttiker, Charge fluc-

tuations in quantum point contacts and chaotic cavities in thepresence of transport, Phys. Rev. B 57 (3) (1998) 1838–1846.doi:10.1103/PhysRevB.57.1838.

[30] S. Datta, Electronic Transport in Mesoscopic Systems, Cam-bridge University Press, Cambridge, United Kingdom, 1997.

[31] Y. Nazarov, Y. Blanter, Quantum Transport: Introduction toNanoscience, Cambridge University Press, Cambridge, UnitedKingdom, 2009.

[32] M. Buttiker, Y. Imry, M. Y. Azbel, Quantum oscillations inone-dimensional normal-metal rings, Phys. Rev. A 30 (4) (1984)1982–1989. doi:10.1103/physreva.30.1982.

[33] R. M. Lutchyn, T. D. Stanescu, S. Das Sarma, Search for ma-jorana fermions in multiband semiconducting nanowires, Phys.Rev. Lett. 106 (12) (2011) 127001. doi:10.1103/physrevlett.

106.127001.[34] T. D. Stanescu, R. M. Lutchyn, S. Das Sarma, Majorana

fermions in semiconductor nanowires, Phys. Rev. B 84 (14)(2011) 144522. doi:10.1103/physrevb.84.144522.

[35] M. Gibertini, F. Taddei, M. Polini, R. Fazio, Local densityof states in metal-topological superconductor hybrid systems,Phys. Rev. B 85 (14) (2012) 144525. doi:10.1103/physrevb.

85.144525.

9