Hydrogen-terminated surface cleaning techniques of silicon substrates were investigated by using x-ray photoelectron spectroscopy (XPS), secondary ion mass spectrometry (SIMS), and transmission electron microscopy (TEM). Either a 4% HF dip or an HF-terminated abbreviated Shiraki clean was used as the cleaning technique. Shiraki-cleaned samples were grown as control samples. XPS was used to measure the C, O, and F remaining on the surface at various stages of the cleaning/growth process, including after a 1 h bake at 200 °C prior to growth. XPS did not detect a significant difference in the adsorbate concentrations between the baked and unbaked samples. From SIMS, the lowest impurity concentrations at the epitaxial/substrate interface were achieved with the abbreviated Shiraki clean, approximately at the same levels as obtained with the standard Shiraki clean, 1.3×1013, 5.4×1012, 1.6×1010, and 4.2×1011/cm2 for C, O, F, and N, respectively. This was achieved without the 850 °C anneal required to desorb the oxide on the Shiraki–cleaned samples. The epitaxial structure chosen for the TEM study was a 30 nm Si buffer grown at 650 °C followed by a 60 nm Si0.7Ge0.3 quantum well and a 250 nm Si layer, both grown at 500 °C. The structures were observed both as-grown and after a 850 °C, 1/2 h anneal used to study defect propagation. There were no detectable differences in the as-grown structures using plan-view TEM. However when the samples were annealed, the HF-dipped specimen developed inclusions in the top Si layer which were not observed in the other samples. In contrast, the prebaked HF-dip-cleaned sample did not have these defects when annealed.
No takes yet. Share an insight, caveat, or question.
Thompson et al. (1993) studied this question.