We measure the conductance variations of submicrometer inversion-layer segments in silicon devices, systematically changing the length, width, inelastic diffusion length, gate voltage, magnetic field, and temperature. Results agree with the theory of universal conductance fluctuations, demonstrating that random quantum interference causes rms conductance changes AG^^^e^/h in each phase-coherent subunit of each segment. The random quantum interference is extremely sensitive to change of a single scatterer.PACS numbers: 72.20. My, 72.20.Dp, 73.40.Qv It is now well understood that low-temperature conductance changes due to weak localization (coherent backscattering) are an example of electron quantum interference in disordered metals.^ In recent theoretical investigations,^"^ a new type of sample-specific random quantum interference of a surprisingly universal character is asserted to affect to varying degrees all conductance measurements on disordered metals. In this Letter, we experimentally confirm key predictions of this theory of ''universal conductance fluctuations."' Random quantum interference arises from the scattering of electron waves by the particular disordered configuration of scatterers ("impurities") in each specimen. A single specimen should exhibit different conductances G corresponding to changes of magnetic field (phase) and electron energy (wavelength) sufficient to rerandomize the interference. Macroscopically similar specimens-having the same dimensions, electron density, and average density of scatterers-should show a similar range of specimento-specimen variation due to microscopic differences of configuration. In both cases, for phase-coherent specimens, the theory predicts an rms variation 8G with a universal magnitude of approximately e^/h = (25.8 kfi)"^ For specimens consisting of A^ phase-coherent subunits, it predicts that the fractional fluctuations are smaller by A^^^^ because of selfaveraging.Two manifestations of this random interference are the aperiodic magnetoresistance fluctuations observed in small metal wires^*^^ and narrow Si inversion layers,^^'^^ and the periodic A/^ Aharonov-Bohm oscillations observed in metal^^"^^ and quantum-well^^ rings. Prior comparisons between theory and the aperiodic phenomena have been based on one or two devices with very small fractional effects. In the present Letter, we systematically confirm a broad range of key predictions of the universal conductance fluctuations theory, including systems in which the fractional effect is of order unity.We have fabricated dozens of Si inversion-layer segments of various lengths and widths in the range 0.04-1.0 fxm and have measured the conductance of each at a variety of gate voltages, magnetic fields, and temperatures. Our devices are /^channel metaloxide-silicon field-effect transistors (MOSFETs) with multiple contacts and narrow channel segments. The device structure and fabrication is described in Mankiewich.^^ Each device contains a narrow channel of width W with several side branches that are used ...

We present the results of a model-system study of the competition between superconductivity and disorder in narrow superconducting wires. As one moves from the two-dimensional regime toward the onedimensional limit, large and systematic reductions in the superconducting transition temperature are obtained. The observed behavior extrapolates to the total destruction of superconductivity in the disordered one-dimensional limit. Our findings are in clear disagreement with a recent theoretical treatment. In addition, the superconducting fluctuations appear to be modified by disorder for the narrowest samples.PACS numbers: 74.40.+k, 74.70.Kn, 74.70.Mq It is now well understood that increasing disorder leads to Anderson localization and the related enhancement of the Coulomb interactions. 1 ' 2 Such disorderinduced enhancement of the Coulomb interaction inherently competes with conventional superconductivity. In fact, it has been experimentally observed in homogeneous two-and three-dimensional systems that superconductivity is far more sensitive to the presence of disorder than are the normal-state properties. 3-8 Furthermore, since the effects of localization and increased Coulomb interactions upon the normal-state properties increase with decreasing dimensionality, the relevance of disorder upon superconductivity should become very large for few-dimensional systems. However, there exist serious discrepancies in our present understanding that require elucidation. The importance of a more complete understanding of the consequences of disorder cannot be underestimated. For example, the newly discovered high-temperature superconducting oxide systems are few-dimensional in nature and possess relatively large resistivities. Thus disorder may play an important role in their superconductivity. Finally, this work argues strongly that the effects of disorder should not be ignored in the important class of quasi-ID superconductors, including the organic systems.In the weakly localized region, where the precursor effects of localization may be treated with perturbation theory, it is predicted for homogeneous systems that the reduction in the superconducting transition temperature, AT C , normalized to the T c o of the otherwise identical material, behaves as 9 AT c /T c0~-(h/4x 2 E F T)*>-1 .Here T> is the dimensionality, E? is the Fermi energy, h is Planck's constant, and r is the elastic-scattering time.Note that perturbation theory breaks down in ID. Experimentally, the competition between superconductivity and disorder has been most closely studied in 2D, 4 " 81011 where the observed effects are large and in general agreement with theoretical predictions. 12,13 In the more technologically important 3D case, the understanding has been hampered by several problems. For 3D systems in the region of weak disorder, it is experimentally difficult to distinguish nonlocalization disorder-induced effects upon T c , e.g., the broadening of structure in the density of states and alterations of the phonon structure. Also, among the th...

Fiory et al. Reply: We agree with the Comment byMishonov that the Cooper-pair mass in a clean superconductor does not vary strongly with temperature [1], as implied by our data when a two-fluid correction is taken into account [2]. Accordingly, we present data for m* in Fig. 1, computed anew, as described herein.The two-fluid expression for the kinetic inductance of a film is L(T)=m*/e 2 N s = (N toi /N s )L(0), where L(0) = m*/e 2 7V t ot is the kinetic inductance of the condensate at r=0, Af/v and Ns are normal and superfluid areal particle densities, respectively, and Nxot^N/v+Ns.Quasiparticle excitations produce negligible dissipation at low frequency, except for the fluctuation regime near T c [3], so NN is not directly measurable. Thus, an electrostatic charge density, £(? =e<57V to t, applied to the surface of a film, induces a modulation in kinetic inductance, 8L.The interpretation presented by Mishonov assumes that the ratio Ns/N {oi is not modulated [1]. However, the small modulation we observe in the superconducting transition, denoted by £7V, implies a correction to 8L may be important as T-+T c [2]. Therefore, we propose to express the modulation in L(T) in the following form:

A vortex-unbinding mechanism for dissipation in epitaxial thin-film microstrips of YBa,Cu30, is established for zero and small magnetic fields. Excellent self-consistent quantitative agreement is found with the two-dimensional Ginzburg-Landau Coulomb gas theory at low currents above the Kosterlitz-Thouless temperature Tt, , and with a simple nucleation theory for three-dimensional vortex rings at higher currents and lower temperatures.