Destruction of superconductivity through phase fluctuations in ultrathin a -MoGe films

Abstract

We investigate the evolution of superconductivity with decreasing film thickness in ultrathin amorphous MoGe (a-MoGe) films using a combination of sub-Kelvin scanning tunneling spectroscopy, magnetic penetration depth measurements and magneto-transport measurements. We observe that superconductivity is strongly affected by quantum and classical phase fluctuations for thickness below 5 nm. The superfluid density is strongly suppressed by quantum phase fluctuations at low temperatures and evolves towards a linear-T dependence at higher temperatures. This is associated with a rapid decrease in the superconducting transition temperature, Tc, and the emergence of a pronounced pseudogap above Tc. These observations suggest that at strong disorder the destruction of superconductivity follows a Bosonic route where the global superconducting state is destroyed by phase fluctuations even though the pairing amplitude remains finite.In the late fifties, Anderson 1 predicted that in an s-wave superconductor the attractive pairing interaction forming Cooper pairs would remain largely unaffected by the presence of non-magnetic impurities. This has been loosely interpreted to imply that the superconducting transition temperature, Tc will also not be strongly sensitive to disorder scattering. However, later experiments showed that this is valid only in the limit of weak disorder: In the presence of strong disorder, Tc gets gradually suppressed 2,3 and eventually the material is driven into a non-superconducting state. The mechanism driving the transition from a superconductor to an insulator or a metal has been a subject of considerable debate. In principle, the suppression of Tc with increase in disorder can happen from two origins. The first mechanism is through the loss of effective screening with increase in disorder that weakens the attractive pairing interaction, and thus suppresses the mean field transition temperature 4,5 . The second mechanism results from the decrease in superfluid density, ns, induced by disorder scattering, which renders the superconductor susceptible to phase fluctuations 6,7 . When ns is small, the phase coherent superconducting state can get destroyed due to strong phase fluctuations even when the pairing amplitude remains finite 8,9,10 .The superconductor to non-superconductor transition driven by these two mechanisms are often classified as the Fermionic and Bosonic routes respectively 11 . In the Fermionic route, the pairing attraction drops to zero at a critical disorder where superconductivity is destroyed. This non-superconducting state is either a bad metal or an Anderson insulator. In the Bosonic mechanism the pairing interaction remains finite and therefore signatures of Cooper pairing continue to survive even in the non-superconducting state.Experimentally, this manifests as a persistence of the superconducting gap in the electronic excitation spectrum, known as the pseudogap, even after the global superconducting state is destroyed 12,13,14,15 .However, recent studies indicate that th...

Stiffness and coherence length measurements of ultra-thin superconductor, and implications to layered superconductors

“…Comparing our results with different measurements of thin films such as δ − NbN 9 or MoGe 10,11 , we find a similar film-thickness dependency of T c and penetration depth λ order of magnitude. However, while in our and Ref.…”

“…We speculate that this is not a material issue but rather a fundamental difference between experimental techniques. We also find that our value of ξ in MoSi is on the same order of magnitude as determined from the upper critical field H c2 in MoSi 12 and MoGe 10,11 . It should be pointed out that in H c2 measurements there is some arbitrariness since it is determined as the field at which the resistivity drops by 50%.…”

“…In Ref. 11 ξ decreases with decreasing thickness only for d < 5 nm, and is ascribed to a transition between Fermionic and Bosinic SC. The difference between different ξ measurements could also be due to the difference between global and local superconductivity as detected by the different techniques.…”

Superfluid Density in Conventional Superconductors: From Clean to Strongly Disordered

“…TABLE I. Experimental data of Tc, ∆(0), λ(0), n, and normal-state resistivity ρ N , mean free path and effective mas m * of an electron obtained from a number of experiments [8][9][10]13,14,[21][22][23][24][25][26][27][28][29][30] In this section, we analyze some of the published experimental data of n s (T ) which are extracted from the measured penetration depth using London's formula 2 : 11); blue and green curves respectively represent the contributions of 1st and 2nd terms of Eq. ( 11).…”

“…While the clean BCS limit can only be reached in specially prepared very clean single crystals, frequently available polycrystalline and thin film superconductors are in the opposite limit, i.e., dirty limit where, τ ∆ 0 / . In such a situation, n s (0) n. n s /n can be intuitively estimated based on the oscillator sum rule [38][39][40] that gets the result correct within a factor of order unity; here we outline this derivation and compare with the accurate expression of n s that has already been derived microscopically in this paper (10) and originally by Abrikosov and Gorkov 4 in the linear response theory.…”