| Abstract Scope |
Recent advances in synthesizing high-quality Ge<SUB>1-x</SUB>Sn<SUB>x</SUB> alloys have generated much interest due to the ability to highly engineer a column-IV alloy with favorable predicted properties, including high carrier mobility (>10<SUP>5</SUP> cm<SUP>2</SUP>/V∙s) [1], tunable lattice constant (for straining subsequent epitaxial layers, such as MOSFET channels and Ge layers) [2-4], and a direct-bandgap crossover point [2,5,6]. However, many of these properties require a Sn composition that is many times larger than the ~1% solid solubility in Ge. Sn alloying also increases the lattice constant of the alloy, making growth on a fixed lattice increasingly difficult with higher Sn percentages. In order to overcome these issues, we grow our Ge<SUB>1-x</SUB>Sn<SUB>x</SUB> alloys on GaAs (100) wafers with lattice-relaxed InGaAs buffer layers using low-temperature molecular beam epitaxy to control the degree of strain. While the main applications of these alloys are in CMOS-compatible devices on Si, our InGaAs buffers provide us with an incredible amount of freedom in both controlling the epitaxial lattice constant while maintaining high distinguishability in optical measurements due to its large bandgap difference with Ge<SUB>1-x</SUB>Sn<SUB>x</SUB>. InGaAs buffers of up to 20% In are mostly relaxed and have been shown to effectively control the strain of epitaxial layers [7]. Our Ge<SUB>1-x</SUB>Sn<SUB>x</SUB> layers are grown at low temperatures of around 200<SUP>o</SUP>C and with thicknesses of ~300nm. After growth, we perform an oxide strip using both HCl and HF immediately followed by a rapid thermal annealing at 550<SUP>o</SUP>C - 650<SUP>o</SUP>C for 30s under a forming gas ambient with the goal of passivating surface states and reducing point defects. After annealing, our samples show Sn compositions of up to ~4.5% Sn as determined by x-ray photoemission measurements. Room-temperature photoluminescence spectra were taken on our samples using a doubled Nd:YAG laser at 532nm using a standard lock-in technique. Due to the predicted shrinking of the bandgap with Sn alloying, we use a thermoelectrically-cooled extended InGaAs detector with a 3dB rolloff at 2.5μm. Our samples show that as we increase the Sn percentage, the direct bandgap of Ge shifts from 0.8eV down to around 0.55eV with the addition of Sn, nearing the predictions of the direct-gap crossover point (which can be as low as around 6.5% Sn). In our presentation, we will further discuss these results and also show how tensile strain can be used to further achieve direct bandgap. |