Phase tuning of multiple Andreev reflections of Dirac fermions and the Josephson supercurrent in Al-MoTe2-Al junctions

TL;DR

This study investigates the interplay between multiple Andreev reflections and Josephson supercurrent in high-transparency Al-MoTe2-Al junctions, revealing how coherent pair shuttling contributes to supercurrent formation at various voltages.

Contribution

It demonstrates the coexistence of MAR steps and Josephson supercurrent in Dirac semimetal junctions and quantifies the fraction of pairs forming the supercurrent.

Findings

MAR leads to staircase I-V profile in MoTe2 junctions.

Josephson supercurrent coexists with MAR steps and varies with phase.

The coherent fraction of pairs increases as voltage approaches zero.

Abstract

When a normal metal $N$ is sandwiched between two superconductors, the energy gaps in the latter act as walls that confine electrons in $N$ in a square-well potential. If the voltage $V$ across $N$ is finite, an electron injected into the well undergoes multiple Andreev reflections (MAR) until it gains enough energy to overcome the energy barrier. Because each reflection converts an electron to a hole (or vice versa), while creating (or destroying) a Cooper pair, the MAR process shuttles a stream of pairs across the junction. An interesting question is, given a finite $V$, what percentage of the shuttled pairs end up as a Josephson supercurrent? This fraction does not seem to have been measured. Here we show that, in high-transparency junctions based on the type II Dirac semimetal MoTe$_2$, the MAR leads to a stair-case profile in the current-voltage ($I$-$V$) response, corresponding to pairs shuttled incoherently by the $n^{th}$-order process. By varying the phase $\varphi$ across the junction, we demonstrate that a Josephson supercurrent ${\bf J}_{\rm s}\sim \sin\varphi$ co-exists with the MAR steps, even at large $V$. The observed linear increase in the amplitude of ${\bf J}_{\rm s}$ with $n$ (for small $n$) implies that ${\bf J}_{\rm s}$ originates from the population of pairs that are coherently shuttled. We infer that the MAR steps and the supercurrent are complementary aspects of the Andreev process. The experiment yields the percentage of shuttled pairs that form the supercurrent. At large $V$, the coherent fraction is initially linear in $n$. However, as $V\to 0$ ($n\gg 1$), almost all the pairs end up as the observed Josephson supercurrent.