Temperature dependence of resistivity and Hall coefficient in strongly disordered NbN thin films

Abstract

We report the temperature dependence of resistivity ͑͒ and Hall coefficient ͑R H ͒ in the normal state of homogeneously disordered epitaxial NbN thin films with k F l ϳ 1.68-10.12. The superconducting transition temperature ͑T c ͒ of these films varies from 2.7 to 16.8 K. While our least disordered film displays usual metallic behavior, for all the films with k F l Յ 8.13, both d dT and dR H dT are negative up to 285 K. We observe that R H ͑T͒ varies linearly with ͑T͒ for all the films and ͓ R H ͑T͒−R H ͑285 K͒ R H ͑285 K͒ ͔ = ␥͓ ͑T͒−͑285 K͒ ͑285 K͒ ͔, where ␥ = 0.68Ϯ 0.11. Measurements performed on a 2-nm-thick Be film show similar behavior with ␥ = 0.69. This behavior is inconsistent with existing theories of localization and e-e interactions in a disordered metal.

Quantum critical magnetotransport at a continuous metal-insulator transition

“…is in the range of 0.5 − 0.7 near the Mott QCP, again in good qualitative accord with the value of 0.69 from Hall data [8]. In Panel (d), we show clear scaling of the experimentally extracted log(σ (c) xy /σxy(T )) in very good accord with theory for the same sample set used for Panel (a).…”

“…Recent work on NbN [8] most clearly reveals ill-understood signatures of localization incompatible with weak localization predictions. In NbN, the effect of intentional charge disorder is to cause a random variation in the local atomic potential, which increases as k F l is reduced by increasing the dis- , in excellent accord with theory [11].…”

“…We have reanalyzed Chand et al's data on NbN in light of the above results to test how our strong coupling view performs relative to data. To make meaningful contact with data on NbN, we make a reasonable assumption that increasing U/t in the FKM corresponds to decreasing k F l, since the scattering strength should increase with U/t, reducing k F l to O(1) [8] near the MIT. We find: as shown in Fig.…”

“…Specifically, while both σ xx (n) ≃ (n c − n) ν and σ xy (n) ≃ (n c − n) ν ′ turned out to be critical at the MIT, and the ratio ν ′ /ν ≃ 1 in stark contrast to the value of 2 predicted at the Anderson MIT [7]. More recent work on intentionally disordered NbN [8], wherein the system is driven across k F l ≃ O(1) (here, k F is the Fermi wave-vector and l the mean-free path, and k F l >> 1 describes a good metal, while k F l ≤ 1.0 describes a bad metal without welldefined electronic quasiparticles), shows clear signatures of an unusual type of localization at odds with theoretical predictions if one insists on an Anderson disorder-driven MIT: (i) ρ xx (T ) ≃ C + AR H (T ), both increasing with reduction in T over a wide range of k F l far before the MIT occurs (in NbN, this is pre-empted by a superconductorinsulator transition (SIT) [8] at very low T near the critical (k F l) c , and (ii) ∆R H /R H ≃ 0.69(∆ρ xx (T )/ρ xx ), widely different from ∆R H /R H ≃ 2.0(∆ρ xx (T )/ρ xx ) expected to hold in WL theory [9] (k F l >> 1). These anomalies in both ρ xx and R H (T ) are inexplicable within WL views (where R H is T -independent and non-critical at the MIT), and point toward a fundamentally new mechanism at work.…”

Correlated Conductance Fluctuations Close to the Berezinskii-Kosterlitz-Thouless Transition in Ultrathin NbN Films

“…The shaded region corresponds to the BKT fluctuation regime dominated by interacting vortices. The samples used in our experiments were grown on lattice matched MgO substrate using RF magnetron sputtering (details refer [20,24,25]). Contacts (Cr/Au) were defined by wire masking for four-probe measurements.…”