No Arabic abstract
We study conductance fluctuations in a two-dimensional electron gas as a function of chemical potential (or gate voltage) from the strongly insulating to the metallic regime. Power spectra of the fluctuations decay with two distinct exponents (1/v_l and 1/v_h). For conductivity $sigmasim 0.1 e^{2}/h$, we find a third exponent (1/v_i) in the shortest samples, and non-monotonic dependence of v_i and v_l on sigma. We study the dependence of v_i, v_l, v_h, and the variances of corresponding fluctuations on sigma, sample size, and temperature. The anomalies near $sigmasimeq 0.1 e^{2}/h$ indicate that the dielectric response and screening length are critically behaved, i.e. that Coulomb correlations dominate the physics.
Measurements of conductance $G$ on short, wide, high-mobility Si-MOSFETs reveal both a two-dimensional metal-insulator transition (MIT) at moderate temperatures (1 $<~ T <$ 4~K) and mesoscopic fluctuations of the conductance at low temperatures ($T~ <$ 1~K). Both were studied as a function of chemical potential (carrier concentration $n_s$) controlled by gate voltage ($V_g$) and magnetic field $B$ near the MIT. Fourier analysis of the low temperature fluctuations reveals several fluctuation scales in $V_g$ that vary non-monotonically near the MIT. At higher temperatures, $G(V_g,B)$ is similar to large FETs and exhibits a MIT. All of the observations support the suggestion that the MIT is driven by Coulomb interactions among the carriers.
We explore the scaling description for a two-dimensional metal-insulator transition (MIT) of electrons in silicon. Near the MIT, $beta_{T}/p = (-1/p)d(ln g)/d(ln T)$ is universal (with $p$, a sample dependent exponent, determined separately; $g$--conductance, $T$--temperature). We obtain the characteristic temperatures $T_0$ and $T_1$ demarking respectively the quantum critical region and the regime of validity of single parameter scaling in the metallic phase, and show that $T_1$ vanishes as the transition is approached. For $T<T_1$, the scaling of the data requires a second parameter. Moreover, all of the data can be described with two-parameter scaling at all densities -- even far from the transition.
We perform combined resistivity and compressibility studies of two-dimensional hole and electron systems which show the apparent metal-insulator transition - a crossover in the sign of dR/dT with changing density. No thermodynamic anomalies have been detected in the crossover region. Instead, despite a ten-fold difference in r_s, the compressibility of both electrons and holes is well described by the theory of nonlinear screening of the random potential. We show that the resistivity exhibits a scaling behavior near the percolation threshold found from analysis of the compressibility. Notably, the percolation transition occurs at a much lower density than the crossover.
Reports of metallic behavior in two-dimensional (2D) systems such as high mobility metal-oxide field effect transistors, insulating oxide interfaces, graphene, and MoS2 have challenged the well-known prediction of Abrahams, et al. that all 2D systems must be insulating. The existence of a metallic state for such a wide range of 2D systems thus reveals a wide gap in our understanding of 2D transport that has become more important as research in 2D systems expands. A key to understanding the 2D metallic state is the metal-insulator transition (MIT). In this report, we demonstrate the existence of a disorder induced MIT in functionalized graphene, a model 2D system. Magneto-transport measurements show that weak-localization overwhelmingly drives the transition, in contradiction to theoretical assumptions that enhanced electron-electron interactions dominate. These results provide the first detailed picture of the nature of the transition from the metallic to insulating states of a 2D system.
We introduce an elementary model for the electrostatic self-consistent potential in a two-dimensional electron gas. By considering the perpendicular degree of freedom arising from the electron tunneling out of the system plane, we predict a threshold carrier density above which this effect is relevant. The predicted value agrees remarkably well with the onset for the insulator to quasi-metallic transition recently observed in several experiments in SiO2-Si and AlGaAs-GaAs heterojunctions.