From surface dimer orientations to bonds at the GaP/Si(100) heterointerface

Despite intense research in III-V-on-silicon-heteroepitaxy since the 1980ies, ultra-high efficiency opto-electronic devices are yet to be realized. Pseudomorphic growth of GaP films on Si(100) is an adequate model system to study the polar-on-non-polar heterointerface which still is a major source of defects. Generally, in situ control of growth processes is vital in order to understand and, finally, avoid defect formation. While metalorganic vapor phase epitaxy is of big interest regarding device production at large scale, the existence of a process gas limits in situ access to electron-based surface science techniques. Reflection anisotropy spectroscopy (RAS), however, is applicable in vapor phase. Dimerized (100) surfaces of cubic crystals usually exhibit characteristic RA spectra, as known for both monohydride-terminated Si(100) and for P-rich GaP(100). Rotation of an anisotropic structure by 90° flips the sign of its RA spectrum so that dimer orientations are controllable in situ during surface preparation. This is essential, in particular, directly before III-V nucleation since process parameters strongly affect the surface formation. Moreover, the polarity of the GaP film determines the orientation of the P-dimers at the Prich GaP/Si(100) surface. Knowing both the polarity of the GaP film and the dimer orientation at the Si(100) substrate prior to nucleation, a simplistic model published by Beyer et al. allows to estimate whether Si-Ga or Si-P bonds are preferred at the heterointerface. Our findings are in favor of Si-P bonds which we demonstrate here for both preferentially A-type and B-type Si(100) 2° substrates.