STM Characterization of Interface Roughness in InAlAs/InGaAs Superlattices Grown by MOCVD
MetadataShow full item record
We employ cross–sectional scanning tunneling microscopy to obtain an accurate statistical representation of the as–grown heterojunctions delineating the quantum wells and quantum barriers in InAlAs / InGaAs strain–balanced superlattices grown by MOCVD intimately related to quantum cascade lasers. Small deviations from the presumed planarity of these interfaces are believed to have profound implications for device performance, with energy–level broadening and carrier scattering as immediate, unintended consequences. The aluminum–rich barrier layers in these structures present a significant challenge for present–day vacuum technology. Painstaking efforts are required to maintain a suitably–pristine habitat where freshly–exposed, aluminum–rich surfaces remain clean over the many days needed to conclude representative STM surveys. We describe the development of carefully–constructed navigation protocols minimizing the image distortion inherent to STM piezo–electric raster mechanisms. These advances are used to implement a novel, reciprocal–space technique (analogous to Bragg’s law in x–ray diffraction) that yields local distance metrics insensitive to STM raster non–idealities. The method's accuracy is demonstrated by way of local period measurements that agree (to within hundredths of monolayers) with high–resolution x-ray diffraction. We develop robust image processing algorithms that incorporate statistical criteria to reproducibly identify the interfaces separating quantum wells from barriers in cleavage–exposed cross–section. The heterojunction profiles obtained this way provide an experimentally–accessible avenue for delineating the confinement potential's spatial boundaries that appears logically consistent and physically reasonable. We conduct a systematic analysis of interface roughness from the viewpoint of fluctuations about an experimentally–determined profile mean. Subtleties central to a correct understanding of roughness variance and its uncertainty in the face of correlated fluctuations are addressed, and the ensuing discussion illustrated with, and corroborated by, numerical simulations. Heterojunction growth order and growth–plane anisotropy are conclusively established as distinguishing physical characteristics. We turn, finally, to the power spectra of these interface fluctuations to ascertain the correlation lengths and functional forms governing their respective spatial–frequency dependencies. The available data are consistent with a universal power spectrum for MOCVD roughness that is isotropic, independent of heterojunction growth order, and predominantly exponential in nature.
Lopez Cruz, Federico (2015). STM Characterization of Interface Roughness in InAlAs/InGaAs Superlattices Grown by MOCVD. Doctoral dissertation, Texas A&M University. Available electronically from