Download files
Abstract
In this thesis we investigate electron transport between ultra-thin, atomically abrupt, buried, nanoscale gaps in silicon, formed between coplanar source and drain leads consisting of highly phosphorus doped silicon. The leads are patterned using scanning probe lithography, dosed with phosphine gas and encapsulated with silicon using low temperature molecular beam epitaxy. These unique tunnel gaps have leads with low sheet resistivities of <1 kohms, are atomically abrupt in all three dimensions, and are scaled down to sub-20 nm dimensions.
The overall tunneling resistance of these devices depends exponentially on the aspect ratio (lead width/gap separation) such that, by engineering the gap dimensions, we control the resistance over seven orders of magnitude from <10 kohms to >500 Gohms. This relationship allows us to predictively engineer gap resistance with device geometry. We also demonstrate an asymmetric barrier
shape, forming an atomic-scale diode with clear rectifying behaviour. We explore six commonly used models to estimate the barrier heights in these nanoscale gaps (rectangular and parabolic barrier models, transition voltage, Fowler-Nordheim field emission and thermionic emission), and where appropriate, adapt these models to include the complex silicon bandstructure. Although these models provide reasonable estimates, no single model describes the conductance over the complete range of gaps studied, and highlight the need for more accurate models that self-consistently account for both tunneling within the electrostatic potential of the exact lead geometry and Thomas-Fermi screening.
We develop three different methods of top-gating these devices using a Schottky barrier, a native oxide, or a silicon dioxide dielectric grown at low temperature. The low temperature oxide exhibits the largest gating range, and the overall tunnel gap
resistance can be gated by three orders of magnitude. Finally, we investigate the possibility of engineering the dopants between the source and drain leads. We confirm that we can order dopants with sub-10 nm precision in the channel and demonstrate their overall effect on device conductance.
This thesis shows that across a statistical number of devices, we can engineer the lead geometry, the tunneling resistance, the barrier height, and the level of disorder, between atomically abrupt source/drain leads.