No Arabic abstract
Despite their ubiquity in nanoscale electronic devices, the physics of tunnel barriers has not been developed to the extent necessary for the engineering of devices in the few-electron regime. This problem is of urgent interest, as this is the precise regime into which current, extreme-scale electronics fall. Here, we propose theoretically and validate experimentally a compact model for multi-electrode tunnel barriers, suitable for design-rules-based engineering of tunnel junctions in quantum devices. We perform transport spectroscopy at $T=4$ K, extracting effective barrier heights and widths for a wide range of biases, using an efficient Landauer-Buttiker tunneling model to perform the analysis. We find that the barrier height shows several regimes of voltage dependence, either linear or approximately exponential. The exponential dependence approximately correlates with the formation of an electron channel below an electrode. Effects on transport threshold, such as metal-insulator-transition and lateral confinement are non-negligible and included. We compare these results to semi-classical solutions of Poissons equation and find them to agree qualitatively. Finally, we characterize the sensitivity of a tunnel barrier that is raised or lowered without an electrode being directly above the barrier region.
We present transport measurements of silicon MOS split gate structures with and without Sb implants. We observe classical point contact (PC) behavior that is free of any pronounced unintentional resonances at liquid He temperatures. The implanted device has resonances superposed on the point contact transport indicative of transport through the Sb donors. We fit the differential conductance to a rectangular tunnel barrier model with a linear barrier height dependence on source-drain voltage and non-linear dependence on gate bias. Effects such as Fowler-Nordheim (FN) tunneling and image charge barrier lowering (ICBL) are considered. Barrier heights and widths are estimated for the entire range of relevant biases. The barrier heights at the locations of some of the resonances for the implanted tunnel barrier are between 15-20 meV, which are consistent with transport through shallow partially hybridized Sb donors. The dependence of width and barrier height on gate voltage is found to be linear over a wide range of gate bias in the split gate geometry but deviates considerably when the barrier becomes large and is not described completely by standard 1D models such as FN or ICBL effects.
We report scanning tunneling microscopy studies of individual adatoms deposited on an InSb(110) surface. The adatoms can be reproducibly dropped off from the STM tip by voltage pulses, and impact tunneling into the surface by up to ~100x. The spatial extent and magnitude of the tunneling effect are widely tunable by imaging conditions such as bias voltage, set current and photoillumination. We attribute the effect to occupation of a (+/0) charge transition level, and switching of the associated adatom-induced band bending. The effect in STM topographic images is well reproduced by transport modeling of filling and emptying rates as a function of the tip position. STM atomic contrast and tunneling spectra are in good agreement with density functional theory calculations for In adatoms. The adatom ionization effect can extend to distances greater than 50 nm away, which we attribute to the low concentration and low binding energy of the residual donors in the undoped InSb crystal. These studies demonstrate how individual atoms can be used to sensitively control current flow in nanoscale devices.
We report the results of an analysis, based on a straightforward quantum-mechanical model, of shot noise suppression in a structure containing cascaded tunneling barriers. Our results exhibit a behavior that is in sharp contrast with existing semiclassical models for this particular type of structure, which predict a limit of 1/3 for the Fano factor as the number of barriers is increased. The origin of this discrepancy is investigated and attributed to the presence of localization on the length scale of the mean free path, as a consequence of the strictly 1-dimensional nature of disorder, which does not create mode mixing, while no localization appears in common semiclassical models. We expect localization to be indeed present in practical situations with prevalent 1-D disorder, and the existing experimental evidence appears to be consistent with such a prediction.
We have observed a negative differential conductance with singular gate and source-drain bias dependences in a phosphorus-doped silicon quantum dot. Its origin is discussed within the framework of weak localization. By measuring the current-voltage characteristics at different temperatures as well as simulating the tunneling rates dependences on energy, we demonstrate that the presence of shallow energy defects together with an enhancement of localization satisfactory explain our observations. Effects observed in magnetic fields are also discussed.
Van der Waals heterostrucutures allow for novel devices such as two-dimensional-to-two-dimensional tunnel devices, exemplified by interlayer tunnel FETs. These devices employ channel/tunnel-barrier/channel geometries. However, during layer-by-layer exfoliation of these multi-layer materials, rotational misalignment is the norm and may substantially affect device characteristics. In this work, by using density functional theory methods, we consider a reduction in tunneling due to weakened coupling across the rotationally misaligned interface between the channel layers and the tunnel barrier. As a prototypical system, we simulate the effects of rotational misalignment of the tunnel barrier layer between aligned channel layers in a graphene/hBN/graphene system. We find that rotational misalignment between the channel layers and the tunnel barrier in this van der Waals heterostructure can significantly reduce coupling between the channels by reducing, specifically, coupling across the interface between the channels and the tunnel barrier. This weakened coupling in graphene/hBN/graphene with hBN misalignment may be relevant to all such van der Waals heterostructures.