Synthesis of High Sn Content Ge1-x-ySixSny (0.1 < y < 0.22) Semiconductors on Si for MWIR Direct Band Gap Applications.

Abstract

A systematic effort has been described to grow ternary Ge1-x-ySixSny semiconductors on silicon with high Sn concentrations spanning the 9.5-21.2% range. The ultimate goal is not only to produce direct band gap materials well into the infrared region of the spectrum but also to approach a critical concentration (yc) for which further additions of Si would decrease─rather than increase─the band gap. This counterintuitive behavior is expected as a result of the giant bowing parameter in the compositional dependence of the band gap associated with the presence of Si-Sn pairs. The growth approach in this study was based on a chemical vacuum deposition method that uses Si4H10, Ge3H8, and SnD4 or SnH4 as the sources of Si, Ge, and Sn, respectively. A fixed Si concentration near x = 0.05-0.07 was chosen to focus the exploration of the compositional space. A first family of samples was grown of Ge-buffered Si substrates. For Sn concentrations y < 0.12, it was found that the samples relaxed their mismatch strain in situ during growth, resulting in high Sn content films that had relatively low levels of strain and exhibited photoluminescence signals that demonstrated direct band gap behavior for the first time. The device potential of these materials was also demonstrated by fabricating a prototype photodiode with low dark currents. The optical studies suggest that the above-mentioned critical concentration is close to yc = 0.2. As the growth temperature was lowered in an effort to reach such values, Sn concentrations as high as y = 0.15 were obtained, but the films grew fully strained with compressive levels as high as 1.7%. To increase the Sn concentration beyond y = 0.15, a new strategy was adopted, in which the Ge buffer layer was eliminated, and the ternary alloy was grown directly on Si. The much higher lattice mismatch between the Ge1-x-ySixSny layer and the Si substrate caused strain relaxation right at the film/substrate interface, and the subsequent films grew with much lower levels of strain. This made it possible to lower the growth temperatures even further and achieve a comprehensive series of strained relaxed samples with tunable Sn concentrations as high as y = 0.21 (and beyond). The latter represent the highest Sn contents in crystalline Ge1-x-ySixSny attained to date and reach the desired yc = 0.2 range. The synthesized films exhibited significant thickness, allowing a thorough determination of composition, crystallinity, morphology, and bonding properties, indicating the formation of single-phase single-crystal alloys with random cubic structures. Further work will focus on optimizing the latter samples to explore the optical and electronic properties.