Mechanisms for directed self-assembly of heteroepitaxial Ge/Si quantum dots with deterministic placement and sub-23nm spacing on SiC nanotemplates

TL;DR

This paper presents a method for creating ordered arrays of Ge quantum dots on SiC nanotemplates with sub-23nm spacing, enabling precise control for quantum computing applications.

Contribution

It introduces a templating technique using electron-beam irradiation and annealing to achieve deterministic placement and spacing of quantum dots at the nanoscale.

Findings

Ordered Ge quantum dot arrays with 22.5 nm spacing achieved

Excess carbon reduces Ge surface diffusion and coarsening

Controlling thermal budget preserves pattern fidelity

Abstract

Artificially ordered Ge quantum dot (QD) arrays, where confined carriers can interact via exchange coupling, may create unique functionalities such as cluster qubits and spintronic bandgap systems. Development of such arrays for quantum computing requires fine control over QD size and spatial arrangement on the sub-35 nm length scale. We employ fine-probe electron-beam irradiation to locally decompose ambient hydrocarbons onto a bare Si (001) surface. These carbonaceous patterns are annealed in UHV, forming ordered arrays of nanoscale SiC precipitates that serve as templates for subsequent Ge quantum dot self-assembly during heteroepitaxy. This templating approach has so far produced interdot spacings down to 22.5 nm, and smaller spacings should be possible. We investigate the templated feature evolution during UHV processing to identify key mechanisms that must be controlled in order to preserve pattern fidelity and reduce broadening of the quantum dot size distribution. A key finding is that the presence of a small background of excess carbon reduces Ge surface diffusion, thereby suppressing coarsening to relatively high temperatures. In fact, coarsening of the carbonaceous nanodot template prior to conversion to SiC can be a more important contributor to size dispersion, and must be avoided through control of thermal budget.