No abstract

A previous paper reported that when GaP epilayers are grown on Si substrates, an As-stabilized surface made by AsH3 preflow before growth prevents many defects from generating at the GaP/Si interface, and that consequently, the crystalline quality of GaP epilayers is markedly improved. This letter describes the AsH3 preflow effect on the initial stages of GaP epitaxial growth. The relative ease with which As or P atoms are absorbed onto Si surfaces is observed using X-ray photoelectron spectroscopy. Although preflow before GaP growth causes As or P atoms to absorb onto Si surfaces, As atoms are absorbed more easily than P atoms. The initial stages of GaP epitaxial growth on Si substrates with and without AsH3 preflow are investigated using high-resolution scanning electron microscopy and transmission electron microscopy. This confirms that AsH3 preflow suppresses island growth, allowing GaP epilayers to grow two-dimensionally, and also reduces the GaP-Si interfacial energy.

GaP epilayers are grown on Si substrates after AsH3 preflow. Electron beam induced current observation and double-crystal x-ray diffraction show that the AsH3 preflow drastically improves crystalline quality of GaP epilayers. The full width at half-maximum of the (400) reflection obtained from 4.8 μm GaP is as small as 115 arcseconds. Secondary ion mass spectroscopy shows that As atoms accumulate at the GaP/Si interfaces, playing an important role in preventing Si outdiffusion into the GaP epilayers.

The dependence of the critical layer thickness hc on mole fraction x of Si1−xGex/Si heterostructures is determined by direct observations of misfit dislocations by using the electron beam induced current (EBIC) technique and transmission electron microscopy (TEM). The EBIC images of the Si1−xGex/Si interfaces show square-grid patterns running in the 〈110〉 directions. These patterns are identified by TEM observation to be misfit dislocations generated at the Si1−xGex/Si interface. The dependence of hc on x is compared with the results reported by R. People and J. C. Bean [Appl. Phys. Lett. 47, 322 (1985) and 49, 229 (1986)]. The comparison reveals that inconsistency exists between them, especially in the range of x<0.3. This inconsistency is considered to arise from the difference in the techniques adopted to determine the hc’s.

An energy balance theory for predicting the critical thickness of the Si1-x Ge x /Si heterostructure is derived based on an experimentally identified dislocation generation mechanism. The theory is in close agreement with experimental results. The critical thickness predicted by this theory is about four times that by the mechanical balance theory of Matthews and Blakeslee.