This thesis concerns the topic of molecular beam epitaxial (MBE) growth on Si, and presents work related to different aspects of this central topic. We have experimentally investigated the electronic properties of structurally near-perfect, metal-semiconductor interfaces involving metallic, transition-metal silicide films, grown epitaxially on Si by MBE. In particular, measurements of the Schottky barrier height of single-crystal, metal-semiconductor structures has been performed.In addition, inelastic electron tunneling spectroscopy has been used to look for elementary excitations of single-crystal, metal-semiconductor interfaces. Other work presented here concerns the building of a "home-made" Si MBE growth apparatus. Lastly, we present theoretical work on a topic unrelated to Si MBE; namely, the calculation of vibrational modes of point-defects in semiconductors.
In Chapter 2, the Schottky barrier heights of high quality, type-A and type-B NiSi2/Si structures are measured, with the use of photoresponse and forward I-V techniques. This work is motivated by a recent controversy over the barrier heights of type-A and type-B structures. Our principal finding is that a substantial difference (greater than 0.1 eV) is observed in Schottky barrier height between the type-A and type-B oriented films. For type-A epitaxy, both measurement techniques consistently yield a Schottky barrier height φBn of 0.62 ± 0.01 eV. For the case of type-B, the I-V measurement yields a barrier height of 0.69 eV, in apparent disagreement with the photoresponse value (0.77 eV). Interpretation of the type-B photoresponse data is complicated by an unusual curvature in the plot of √photoresponse vs. photon energy. It is shown that both the detailed shape of the type-B photoresponse data, and the apparent inconsistency between photoresponse- and I-V-determined Schottky barrier height values can be quantitatively explained in terms of a phenomenological model, in which we represent the type-B interface as a mixture of regions of high and low Schottky barrier height (φhi and φlo), electrically in parallel with each other. In terms of this model, the type-B interface is explained as a mixed-barrier structure, having 91% areal coverage of φhi = 0.81 eV, and 9% of φlo = 0.64 eV.
In Chapter 3, the first electron tunneling spectroscopy experiments on single-crystal, type-A and type-B NiSi2 films are described. For comparison, single-crystal CoSi2 films, and polycrystalline Au films, all grown on similar, degenerate, n-type Si substrates, are also studied. The purpose of this study is to see if crystalline perfection of the interface results in additional or enhanced structure in the electron tunnel spectrum, due to rigorous k||-conservation. This is not observed. Instead, we observe direct evidence of bulk, Si, k-conserving phonons assisting the transport of electrons through the Schottky barrier. Our electron tunneling spectra appear to to be independent of the identity, crystalline perfection, and crystallographic orientation of the metal layer. No enhancements or new features in our tunneling spectra due to the single-crystallinity of the silicide films are evident, suggesting that, the observed inelastic transport is dominated by bulklike processes excited within the tunneling barrier rather than by processes intrinsically related to the metal-semiconductor interface.
In Chapter 4, the design and construction of a Si/silicide MBE growth system is presented. The original contribution in this work is the design of the apparatus: both of the overall system, and of its many detailed components. In particular, the design of this system reflects a number of issues which are of serious concern in state-of-the-art, Si MBE: the problem of unintentional p-type doping, the concern over the flaking of Si inside the growth chamber, and the accurate monitoring and control of substrate temperature. Most of the key assemblies, including the substrate heater, load-lock mechanism, sample holder, cryoshrouds, and shutter mechanism were designed and fabricated at Caltech.Our main result is the successful operation of this system. Preliminary silicide growth attempts demonstrate the ability of the system to perform closed-loop, rate-controlled depositions onto a heated Si substrate.
Finally, Chapter 5 involves the theoretical calculation of vibrational modes of point-defects in semiconductors. An understanding of such modes is useful in providing structural information about point-defects. Often, vibrational modes are observable by optical techniques. In our work, three distinct defect systems are investigated: substitutional oxygen on a phosphorus site in GaP (GaP:Op), substitutional and interstitial iron in Si (Si:Fe), and substitutional carbon on an arsenic site in GaAs (GaAs:CAs). Our theoretical method involves modeling the defect, e.g., in terms of masses and springs. In the calculation, the masses are known, but the spring constants near the defect are unknown, and varied as parameters. This requires a comparison with experimental data. For the GaP:Op defect, we predict and fit two main defect modes through the adjustment of parameters, and in the process, find detailed agreement between theory and experiment for several additional modes, also predicted by our calculation, but not involved in the fit. Our calculations provide the first coherent framework to account for details from many different optical measurements on the GaP:Op defect, within a single, theoretical description. We have also calculated the low-energy defect modes associated with Fe substitutional and interstitial defects in Si, to determine whether a frequency shift is expected upon the isotopic substitution, 54Fe → 56Fe. The calculation predicts the existence of a single, low-energy mode, and that a frequency shift should occur, in disagreement with an experimental result. Our calculation demonstrates that the observed mode cannot therefore be due to a defect phonon. Finally, the effect of the natural isotopic variation of Ga in GaAs on the infrared-active defect modes of the GaAs:CAs, system is investigated. The calculated results show detailed agreement with infrared absorption data with respect to defect-mode splittings and relative intensities, and establish conclusively the identity of the impurity substitutional site as an As site.
