Title:
|
Band-gap engineering of Germanium monolithic light sources using tensile strain and n-type doping
|
Band-gap engineering of bulk germanium (Ge) holds the potential for realizing a laser source, permitting full integration of monolithic circuitry on CMOS platforms. Techniques rely mainly on tensile strain and n-type doping. In this thesis, we focus on studying diffusion-based phosphorus (P) doping of Ge using spin-on dopants (SOD), and tensile strain engineering using freestanding micro-electro-mechanical systems (MEMS)-like structures. Process development of a reliable SOD recipe was conducted using furnace and rapid-thermal annealing, and successful doping up to 2.5 x 10¹⁹cm⁻³ was achieved,resulting in approximately 10 enhancement in direct-gap emission. A transition in Ge direct-gap photoluminescence (PL) behaviour is observed upon doping, from being quadratically dependent on excitation power to linear. We have also demonstrated that the limited doping concentration of P in Ge using SOD is not source limited, but more probably related to the diffusion mechanism. The other part of the project concentrated on Ge strain engineering. Previous works reported high tensile strain values based on freestanding MEMS-like structures made of Ge, yet without embedding an optical cavity (until recently). In this project, we realize this combination by fabricating Ge micro-disks as an optical cavity on top of freestanding SiO2 structures, utilizing Ge-on- Insulator wafers (GOI). 3D computer simulations were used to understand and optimize the devices, in terms of strain and optical performance. Raman spectroscopy and PL measurements confirmed simulation results showing higher tensile strain for beams with shorter lengths, with a maximum uniaxial strain of 1.3%. Splitting of light and heavy hole energy bands was observed by PL as the strain increases, agreeing with theoretical models. Direct-gap sharp-peak whispering-gallery modes (WGMs) were confined in 3 m disks with a maximum quality-factor of 200. Two loss mechanisms could be distinguished, red-shift of the absorption edge, and free-carrier absorption. In order to avoid these excitation-related losses, higher strain values combined with heavy n-type doping are required. A possible implementation using the same GOI platform is proposed for future work.
|