Toward Direct Bandgap Germanium Optoelectronics

Germanium was one of the first semiconductors of promise, serving as the basis for the first transistor, a key enabler of the Information Age, which has fundamentally transformed society. Ge has since been supplanted by silicon (Si), which has superior electronic and chemical attributes, and gallium arsenide (GaAs), which possesses superior properties for light-matter interaction (optoelectronics). Ge is most widely applied today as a chemically compatible means of boosting performance in mature Si technology. But Ge may also address a growing need for tightly integrated optoelectronic functions on Si.

The key to excellent optoelectronic performance in Ge lies in interacting with the direct bandgap in the conduction band Γ valley, which is only 140 meV above the fundamental bandedge. This has been exploited by other groups to realize state-of-the-art photonic components such as photodetectors and modulators, but no Ge devices with a truly direct bandgap have been demonstrated to date.

To this end, it would be highly advantageous if Ge could be transformed into a direct bandgap semiconductor, such that the behavior of most electrons in the material is strongly Γ in character. This confers efficient radiative recombination (light emission) and large absorption coefficients, giving rise to large gains under population inversion in laser applications. Two methods for shifting the bandstructure are investigated in this thesis: deposition of stress liners to induce tensile strain, and alloying Ge with dilute amounts (<1%) of carbon. In the first case, tensile strain induced by highly compressive films modifies the band structure, potentially dropping the Γ valley below the L valley. In the second case, carbon sitting on Ge lattice sites induces strong local perturbations leading to an effect called band anticrossing, which strongly repels the Γ valley downward, again potentially yielding a direct bandgap.

Ge is also valued for long carrier lifetimes. However, low rates of recombination require high quality material and interfaces. Though bulk purity has been refined to acceptable levels for substrates, defects at surfaces are inevitably a challenge. This work shows that by changing the stoichiometry of surface oxide passivation, drastic improvements in light emission from Ge are realized. Furthermore, it is found that for optoelectronics, GeO interfaces are superior to GeO₂, but Ge⁺³ states are entirely undesirable.