Observation of Andreev Reflection Enhanced Shot Noise

Abstract

We have experimentally investigated the quasiparticle shot noise in NbN/MgO/NbN superconductor -insulator -superconductor tunnel junctions. The observed shot noise is significantly larger than theoretically expected. We attribute this to the occurrence of multiple Andreev reflection processes in pinholes present in the MgO barrier. This mechanism causes the current to flow in large charge quanta (Andreev clusters), with a voltage dependent average value of m ≈ 1 + 2∆ eV times the electron charge. Because of this charge enhancement effect, the shot noise is increased by the factor m.PACS numbers: 72.70.+m, 74.50.+r, 74.80.Fp A dc current I flowing through a vacuum tube or a tunnel junction generates shot noise, time dependent fluctuations of the current due to the discreteness of charge carriers. Shot noise studies provide information on the nature of conduction not obtainable by conductance studies, e.g. the electric charge of carriers or the degree of correlation in the conducting system. For an uncorrelated system in which the electrons do not interact, the passage of carriers can be described by a Poisson distribution. The spectral density of current fluctuations S I then equals full shot noise: S I = 2qI = P Poisson for zero frequency and temperature 1 . The charge quantum q is normally the electron charge e.In superconductor -normal metal (SN) systems Andreev reflection occurs, causing an effective charge to be transferred of 2e. Due to this doubling of the charge, the shot noise in such a system is predicted to have a maximum of twice the Poisson noise 2-4 . More recently, giant shot noise in the supercurrent is predicted in a singlechannel superconductor -normal metal -superconductor (SNS) point contact 5 which is attributed to transport of large charge quanta (q ≫ e) at finite voltages caused by Multiple Andreev Reflection (MAR)6 . Observation of enhanced charge quanta in SN or SNS structures requires a combination of conductance and shot noise measurements. Despite extensive theoretical work, experimental results in this field are rare. A recent experiment 7 is performed on a NbN/c/Nb structure in which c is assumed to be a Nb constriction with a length of 7 nm and a diameter of 15 nm. At 9.5 K the structure acts like an NS interface but doubled shot noise is not observed. The predicted giant supercurrent shot noise is not observed either (at 4.2 K).From an applied point of view, shot noise in superconducting structures is of interest since this noise forms a major limitation to the sensitivity of NbN/MgO/NbN Superconductor-Insulator-Superconductor (SIS) THz radiation detectors 8 . For these reasons, we have investigated quasiparticle current transport and shot noise in an SNS structure in which the quasiparticle current is carried by MAR. Anticipating the experimental shot noise results presented in this paper we demonstrate that in a system in which multiple Andreev reflections occurs the quasiparticle shot noise at V < 2∆/e has a maximum value given by S I = (1 +

Shot noise in mesoscopic conductors

“…This probability obeys n∇T R = 0 with the diffusive boundary conditions at the surface; furthermore, it equals 0 and 1 provided the particle is headed to Ω L and Ω R , respectively. In the leading order T R = α R /(α L + α R ); however, this order does not contribute to the noise due to the sum rule (211). The subleading order is that for n pointing out of the left (right) contact, (248), we obtain for the Fano factor…”

“…An experimental observation of MAR-enhanced shot noise is reported by Dieleman et al [211]. They investigate noise in a NbN/MgO/NbN structure, where the two superconducting layers (NbN) are separated by an insulator.…”

“…This discrepancy with theory is not understood. [211]. The Fano factor (plotted here as an effective charge, black squares) is compared with the theoretical prediction (solid line).…”

Shot Noise in Mesoscopic Diffusive Andreev Wires

“…For a num-ber of generic conductors, it turns out that the suppression of the Fano factor is universal, i. e. it does not depend on details of the conductor like geometry or impurity concentration. In particular, a diffusive metal with elastic scattering leads to F N dif f = 1/3 [7,8], which is independent on the concrete shape of the conductor [9] and has been confirmed experimentally [10,11].…”

Current fluctuations in rough superconducting tunnel junctions

“…Indeed, in Ref. [22], the importance of pinholes was pointed out, but also Ref. [23] discusses the significance of pinholes in mesoscopic devices, e.g., for the Kondo effect.…”