The initial growth process of ${\mathrm{Si}}_{\text{1−x−y}}{\mathrm{Ge}}_x{\mathrm{C}}_y$ thin films on Si(001) surfaces is examined by scanning tunneling microscopy. The surface morphology of the film critically depends on the C fraction in the film. Evidence is presented on an atomic scale that the epitaxial growth of ${\mathrm{Si}}_{\text{1−x−y}}{\mathrm{Ge}}_x{\mathrm{C}}_y$ films with large C fractions is dominated by phase separation between Si–C and Si–Ge, concomitant with C condensation on the surface of the growing films. We find that the addition of a thin (1–2 ML) SiGe interlayer between the ${\mathrm{Si}}_{\text{1−x−y}}{\mathrm{Ge}}_x{\mathrm{C}}_y$ film and the Si substrate drastically improves the film structure, leading to a planar morphology even with large C fractions present in the film.

As the amount of data that people handle increases, the conventional Neumann-type computer architecture is reaching its limits. Therefore, research on hardware implementation of machine learning systems is being actively conducted. In this paper, we have implemented and evaluated neuromorphic hardware that realizes human brain neurons and synapses using oxide semiconductor of amorphous In-Ga-Zn-O (a-IGZO) and a cellular neural network. It was confirmed how variations of initial resistance and deterioration rate of the oxide semiconductor affect operation accuracy of the neuromorphic hardware. Furthermore, we clarified that an activation function suitable for the hardware implementation is a ReLU function.