Search

Renewable energy sources such as solar and wind are critical to combatting global warming. However, their intermittent energy generation requires the development of large-scale grid energy storage, in contrast to the on-demand generation of coal-based powerplants. Sodium-ion batteries offer a promising potential technology due to their similarity to the more mature lithium-ion battery chemistry and the higher abundance and lower sodium cost compared to lithium. Because sodium ions are larger than lithium ions, sodium-ion intercalation results in more drastic structural rearrangements. Of crucial importance to the development of sodium-ion batteries is an improved understanding of structural dynamics and ionic diffusion pathways. This research aims to improve the state of this understanding by tuning an existing thin-film model system to incorporate c-axis growth to form inclined NaxCoO2 epitaxial thin films that can be used for future interface studies, and ionic transport imaging.

The design of a gearbox is subject to multiple performance requirements that must be met. One such requirement is power density, a metric defined as power transmitted per gear volume or per weight. In aerospace applications, one method of reducing gearbox weight to increase power density has been removing material from gear blanks through the use of thin webs. This study adapts a representative spur gear pair design to investigate the effects of using thin-web gears. A deformable-body model of the gear pair is developed to perform quasi-static and dynamic analyses of the gear pair variations with solid and thin webs, subjected to errors causing the load distribution to skew axially. Under quasi-static conditions, the flexible web deflections are shown to ease some of the adverse effects of gear errors. To determine the dynamic conditions where flexible rim modes could be a factor, impact tests are performed along with a modal analysis using the model. The contributions of these rim modes to the overall dynamic behavior are shown to be modest.

Abstract: Ductile steel plate shear walls are an established lateral load resisting system. Past research indicates that cold-rolled infill panels less than 1 mm in thickness present one solution to an overstrength problem arising from selecting an infill panel thickness based on ease of welding and handling. This research program examines several possible welding procedures and joint geometry to connect the thin infill panel to the thick boundary elements. Primary welding parameters include short-circuiting gas metal arc welding process, electrode and shielding gas selection, heat input, and use of a chill strip. Four configurations of the infill panel-to-boundary element joint and two configurations of a lap splice joint between two sheets of thin steel in the infill panel were tested in monotonic tension and cyclic tension-compression. A quasi-static cyclic test of a single-storey moment resisting frame steel plate shear wall validated the use of one welding procedure and joint geometry.

This thesis reports investigations of the thermal modifications induced by ultra-short pulsed laser irradiation in thin films at fluences near and above ablation threshold. Specific work was aimed at understanding the role of heat dissipation in determining the nature and spatial extent of thermal processes induced by femtosecond laser irradiation. Thermal damage in the form of microstructural modifications, phase transitions, and/or chemical reactions was analyzed within, outside, and below areas of femtosecond laser irradiated surfaces. The thermal damage induced by femtosecond laser ablation of silicon thin foils was investigated by high resolution transmission electron microscopy. Extensive single-crystalline Si was observed at the ablated hole edges for irradiation near the ablation threshold. Higher fluence ablation produced hole edges exhibiting a strong amorphous content. The rate of lateral cooling following femtosecond laser irradiation is proposed to account for the resulting microstructure near femtosecond laser ablation. A nanoscale technique was developed for direct measurement of heat affected zones resulting from femtosecond laser ablation. The lateral extent of cobalt silicidation was measured by selected area diffraction and morphological investigations using transmission electron microscopy. Silicidation extending 3 mum from the edge of an ablated hole provided evidence of a zone where the local temperature reached ∼500°C. Evidence of Si recrystallization was observed within 1 mum of the ablated hole edge and validated by thermal modeling. Reactive multilayer films composed of Co/Al, Al/Pt, and Ti/Ni were irradiated by femtosecond and nanosecond pulsed lasers. Fluence thresholds for igniting self-propagating reactions were measured as a function of bilayer thickness. The resulting ignition thresholds versus bilayer thickness trends were nearly reciprocal that of reaction propagation speed versus bilayer thickness. The trend is indicative of a relationship between the ignition thresholds and the thermal damage necessary to achieve a critical volume of layer intermixing. Femtosecond laser ablation at lower fluences near ablation threshold showed evidence of single and multiple layer removal. Samples irradiated by nanosecond lasers of comparable fluence showed evidence of extensive intermixing with negligible material removal. Heat dissipation rates were observed to be an important factor for pulsed laser ignition of reactive multilayers based on numerical analysis and comparative ignition studies of three bimetallic systems.

Gas metal arc welding (GMAW) is the most commonly used arc welding method in industry for joining steels and aluminum alloys. But due to the mathematical difficulties associated with the free surface motion of the molten droplet and the weld pool, the process is not well understood and the development of new welding procedures in the manufacturing industry highly depends on expensive, time-consuming and experience-based trial and error. In this dissertation, numerical methods are developed to overcome the difficulties and to simulate the metal transfer and weld pool development in the GMAW of sheet metals. The simulations are validated by experiments and used to study an industrial welding process. A numerical procedure is first developed to model the free surface motion in fusion welding processes. Thermal and electromagnetic models are integrated with the fluid models. Recommendations are made on the selection and improvement of publicly available numerical algorithms, while alternative methods are also reviewed. A model combining the enthalpy, effective-viscosity and volume-of-fluid methods is then developed to simulate the metal transfer process in globular, spray and short-circuiting transfer modes. The model not only describes the influence of gravity, electromagnetic force and surface tension on droplet profile and transfer frequency, but also models the nonisothermal phenomena such as heat transfer and phase change. The melting front motion, the droplet detachment and oscillation, the satellite formation and the fluid convection within the droplet are analyzed. It has been found that the taper formation in spray transfer is closely related to the heat input on the unmelted portion of the welding wire, and the taper formation affects the globular-spray transition by decelerating the transfer process. Experiments with a high-speed motion analyzer validate the simulation results. The model is then extended to simulate the initiation, development and solidification of the weld pool, with consideration of the droplet impingement on the pool surface. The characteristics of physical variables in the weld pool are analyzed. Burn-through of thin sheet metals and penetration of a multi-layered workpiece are also simulated. Experiments are used to verify the predicted weld penetration. Finally, numerical simulation is used to analyze an industrial welding process---the gas metal arc spot welding of multi-layered workpieces. The traditional spot welding and plug welding methods are simulated. It is shown that the plug method can ensure a stable short-circuiting transfer through a pre-made hole. Both the simulation and tensile-shear tests show that the plug method improves the penetration consistency by ensuring the effective joint diameter.

This doctoral thesis presents the development of a systematic method of automatic generation of fabrication processes (or process flows) for thin film devices starting from schematics of the device structures. This new top-down design methodology combines formal mathematical flow construction methods with a set of library-specific available resources to generate flows compatible with a particular laboratory. Because this methodology combines laboratory resource libraries with a logical description of thin film device structure and generates a set of sequential fabrication processing instructions, this procedure is referred to as process compilation, in analogy to the procedure used for compilation of computer programs. Basically, the method developed uses a partially ordered set (poset) representation of the final device structure which describes the order between its various components expressed in the form of a directed graph. Each of these components are essentially fabricated one at a time in a sequential fashion. If the directed graph is acyclic, the sequence in which these components are fabricated is determined from the poset linear extensions, and the component sequence is finally expanded into the corresponding process flow. This graph-theoretic process flow construction method is powerful enough to formally prove the existence and multiplicity of flows thus creating a design space ${\cal D}$ suitable for optimization. The cardinality $\Vert{\cal D}\Vert$ for a device with N components can be large with a worst case $\Vert{\cal D}\Vert\le(N-1)$! yielding in general a combinatorial explosion of solutions. The number of solutions is hence controlled through a-priori estimates of $\Vert{\cal D}\Vert$ and condensation (i.e., reduction) of the device component graph. The mathematical method has been implemented in a set of algorithms that are parts of the software tool MISTIC (Michigan Synthesis Tools for Integrated Circuits). MISTIC is a planar process compiler that generates process flows for thin film devices from schematics of their structures. The algorithms also include the capability of grading the process flows based on the expected device yield and some empirical factors. The MISTIC software uses a lab-specific database of process recipes and materials to produce process flows for a specific set of laboratory resources and process statistics that help to choose the most suitable process flow in a comparative manner. Currently the process compiler consists of five modules, viz., the graphical device editor, the database and the database editor, the compiler, and the process viewer constituting a complete design environment. The program has been implemented with approximately 213,800 lines of C code that utilize the X11/Motif library. The compiler in its current version accepts devices with Manhattan-like geometries over a multiplicity of one dimensional slices of the device, hence all calculations are inherently one-dimensional in nature. The compilation procedure has been successfully tested with several conventional integrated circuit devices, e.g., DIODE, CMOS and BICMOS, etc. It has also been successfully applied to Micro-Electro Mechanical System (MEMS) devices such as accelerometer, micro-bridge, micro-motor structures with or without on chip circuits. In each case the compiler has generated a set of process flows which included the established process for that device along with several alternative processes.

An external magnetic field of a few kG is used to alter significantly the radiative properties of atmospheric pressure plasmas generated by the capacitive discharge vaporization of thin Ag films. The magnetic field is generated by the plasma current in a large inductor surrounding the plasma. Peak magnetic field strength of 2.7 kG is obtained with a peak discharge current of 2.7 kA. The magnetic field is oriented normal to the electric field in the plasma. This orientation results in a drift motion of the plasma normal to the plane established by the electric and magnetic field vectors. The drift motion of the plasma is directed toward the plastic substrate on which the Ag film and a powder or solution residue sample are originally located. This increases plasma-sample interaction. Time integrated, spatially resolved photographic spectra and spatially resolved photoelectric radiation profiles show that with the magnetic field, the continuum background radiation is confined to a relatively narrow region (0.5-1 cm) near the substrate surface. This results in significantly larger analyte line-to-background intensity ratios in the region 1-2 cm above the substrate surface. Particle-size studies show a reduction in particle-size dependent radiation when a 30 (mu)F, 6 kV, 105 (mu)H discharge is used with a magnetic field. Particle-size independent emission for particles under 30 (mu)m in diameter is observed under these conditions after the second current half-cycle (0.34 ms) of the discharge. Faster vaporization rates of the sample are also evident with the magnetic field. Delayed integration of radiation techniques are used to generate analytical curves for the direct determination of V, Pb, Cr and Mn in seven NBS solid powder standard reference materials. Excellent results are obtained in the NBS analyses with an average relative error of 10.8% for the 15 determinations. A single set of discharge conditions is used for all analyses. Readily available salts are used for aqueous standard solutions. No matrix matching is required and the only sample pretreatment is grinding to reduce sample particle size to under 30 (mu)m in diameter.

This thesis work has involved studying epitaxial growth of single crystal silicon thin films using a unique approach. Pulsed Supersonic Jet Epitaxy utilizes high kinetic energy ($\approx$1.6 eV) jets of a disilane $\rm(Si\sb2H\sb6$)--hydrogen mixture incident on a silicon substrate to conduct growth. The necessary activation energy for epitaxy is provided by the inherent energy of the precursors, precluding the need for a high substrate temperature. We have successfully demonstrated film growth of single crystal silicon (without external activation) at temperatures as low as 400$\sp\circ$C. This technique has potential applications in processes requiring low thermal budgets, including abrupt-interface multilayer film growth and flat panel display manufacturing with glass substrates. We have been able to accurately control the film thickness with sub-Angstrom resolution by effectively utilizing the self limiting nature of this hydrogen desorption limited system. An order of magnitude increase in sticking coefficients $\rm(S\approx0.3$ at T = 450$\sp\circ$C) is observed compared to conventional gas source molecular beam epitaxy. This technique has also simplified the surface kinetics and we have been able to develop a fundamental reaction scheme for a pulsed system to predict growth rate dependencies on various parameters. Silicon epitaxy from pulsed jets has also been analyzed using a Monte Carlo simulation to study the effects of high kinetic energy jets on the surface morphology of the growing epilayer. We have been able to predict temperature independent atomic layer epitaxy for incident energy values exceeding the hydrogen desorption activation energy. We have successfully applied this technique for in situ n-type doping of silicon using supersonic jets of phosphine (PH$\sb3).$ The high flux, kinetic energy and low growth temperature associated with this technique have enabled us to obtain active carrier concentrations up to $\rm5\times10\sp{19}\ cm\sp{-3}$ in silicon films (with no subsequent annealing) at substrate temperatures of 550$\sp\circ$C, hitherto only possible with ion implantation. The thin films exhibit uniformity in doping levels with Hall mobilities of 90 cm$\rm\sp2V\sp{-1}s\sp{-1}$ at the highest doping concentration, comparable to bulk layers. This method also provides good control of doping level by variation of substrate temperature and pulse parameters.

The objective of this research has been to develop methods for modifying the surfaces of neural prosthetic devices to enhance biocompatibility. Also central to this work was the characterization of the processes used to modify the surfaces, the resulting macroscopic and microscopic structure, and the relevant physical properties of the new surface. The application required a coating that could attract and adhere cells, mediate the stiffness mismatch between the device and tissue, and facilitate signal transport from the device to tissue. The materials chosen for use as surface modifiers were genetically engineered polypeptides that combine biofunctional sequences with structural segments, creating a processable bioadhesive agent. An electric field mediated deposition process was used to create thin coatings on the devices from these protein polymers. Varying the process parameters was found to exert controllable changes on the morphology, and porous thin films with a range of structures were fabricated. This deposition process was combined with lithographic techniques to generate high-fidelity patterned surfaces. It was anticipated that the surface structure of these films could augment their biochemical composition and facilitate cell adhesion. A Fourier Transform-based method of explicitly quantifying the surface topography was employed to evaluate the effects of process parameters on topography. The mechanical properties of the coatings were examined to determine a suitable morphology for joining the mechanically dissimilar device and tissue. Fibrous coatings composed of randomly oriented filaments exhibited a stiffness gradient while under compression. The films were compliant near the tissue and stiffer near the device. The biological performance of these films was assayed and the films were seen to be potent cellular adhesives. The coatings were also found to be capable of delivering biologically-relevant molecules in vitro.

This study applies a new morphometric technique, the method of thin-plate splines (Bookstein, 1991) to the problem of describing and quantifying cranial shape in European and Near Eastern Neandertals. This method begins with a geometry that includes all information that could be gained from ratios of linear measurements or angles measured using the same landmark set. Shape difference is modelled as a Cartesian deformation and decomposed by its partial warps, yielding multiple geometrically orthogonal components of shape difference. I compared European and Near Eastern Neandertal cranial shape to recent Norwegian, Australian, Eskimo, Ugandan, and Polynesian samples, as well as to prehistoric samples from the French Mesolithic and European Upper Paleolithic periods. I also included the Middle Pleistocene crania Petralona and Broken Hill, usually attributed either to Homo erectus or early archaic Homo sapiens. Twenty landmarks were digitized from photographs or scale drawings of crania in norma lateralis. The mean landmark locations for each sample were calculated to a glabella-inion baseline using shape coordinates. The Norwegian mean is the reference form to which all other mean forms and individuals were compared. Results show that the Neandertals do not differ significantly from more recent humans in 14 of the 18 features defined by the partial warp analysis. In the four features where the Neandertals differ, there is overlap with the post-Neandertal human samples in every case. No autapomorphic features of lateral cranial shape were found for the Neandertals. For two warps, the Neandertals share the more modern state, whereas Petralona and Broken Hill are similar to one another and different from the later samples. The La Chapelle cranium differs from the other Neandertals in important ways and should not be used to typify Neandertal craniofacial morphology. Evidently, Neandertals are not as different from more modern humans as is commonly reported. Neandertal autapomorphies proposed in the literature are examined, and previous characterizations of Neandertal facial form are discussed in the light of these results.

In this study, the potential of a novel hot-wire technique for silicon thin film deposition and etching has been investigated. For silicon thin film deposition, molecular hydrogen was passed over a heated tungsten filament and catalytically dissociated into atomic hydrogen; source gas disilane was fed into the reactor bypassing the filament to form film forming species by reacting with atomic hydrogen. By using this method, good quality amorphous, microcrystalline, polycrystalline, and epitaxial silicon thin films have been deposited at low substrate temperatures and relatively high deposition rates. Particularly, polycrystalline silicon with a crystalline volume fraction of 98% has been deposited at a substrate temperature of 310$\sp\circ$C and a deposition rate of 110 A/min. Hydrogenated amorphous silicon has been deposited at a high growth rate of $\sim$1500 A/min. In addition, polycrystalline silicon has been selectively deposited on molybdenum or silicon over silicon dioxide, silicon nitride or Corning 7059 glass, or on silicon over molybdenum and these substrates at a substrate temperature of 300$\sp\circ$C and a deposition rate of $\sim$100 A/min with excellent selectivity on features of size as small as 1 $\mu$m. These results are attributed to a high concentration of atomic hydrogen present in the process and suggest that catalytic dissociation of molecular hydrogen by a heated filament is an efficient method to produce atomic hydrogen. A major function that atomic hydrogen has in achieving these results is etching of silicon during deposition. In this study, it has been found that a-Si:H can be etched at $\sim$60A/sec by atomic hydrogen. In addition, it has been found that etching rate increases with the increase in filament temperature, hydrogen pressure, and flow rate, and decreases with the increase in substrate temperature. Etching rate is sensitive to oxygen contamination on silicon surface, which may also cause surface roughening during etching. There are etching selectivities between intrinsic and boron-doped amorphous silicon, and poly-Si.

Recent advances in microelectronics and silicon processing have been exploited to fabricate miniaturized chemical sensors. Although the capability of chemical sensing technology has grown steadily, it has been outpaced by the increasing demands for more reliable, inexpensive, and selective sensors. The diversity of applications requires the deployment of different sensing materials that have rich interfacial chemistry. However, several promising sensor materials are often incompatible with silicon micromachining and their deposition requires complicated masking steps. The new approach described here is to first micromachine a generic, instrumented, conductometric, microelectronic sensor platform that is fully functional except for the front-end sensing element. This generic platform contains a thin dielectric membrane, an integrated boron-doped silicon heater, and conductance electrodes. The membrane has low thermal mass and excellent thermal isolation. A proprietary selected-area chemical vapor deposition (SACVD) process in a cold-wall reactor at low pressures was then used to achieve maskless, self-lithographic deposition of thin films. The temperature-programmable integrated microheater initiates localized thermal decomposition/reaction of suitable CVD precursors confined to a small heated area (500 $\mu$m in diameter), and this creates the active sensing element. Platinum and titania (TiO$\sb2$) films were deposited from pyrolysis of organometallic precursors, tetrakistrifluorophosphine platinum Pt(PF$\sb3)\sb4$ and titanium tetraisopropoxide Ti(OCH(CH$\sb3)\sb2\rbrack \sb4$, respectively. Deposition of gold metal films from chlorotriethylphosphine gold $\rm (C\sb2H\sb5)\sb3$PAuCl precursor was also attempted but without success. The conductance electrodes permit in situ monitoring of film growth. The as-deposited films were characterized in situ by conductance measurements and optical microscopy and ex situ by electron microscopy and spectroscopy methods. Devices equipped with platinum CVD films showed little chemical response to carbon monoxide and propylene pulses, but exhibited a small response to oxygen and a significant response to hydrogen in vacuum. Finally, a finite difference FLUENT model for low pressure CVD of platinum films was developed and implemented. The process model predicts growth rate, thermal boundary layer profile and an insight into the reaction mechanism.

The magnetic anisotropy of ultrathin magnetic Co/Cu (111) and Co/Cu (001) films was investigated. A comparison of two experimental techniques, Ferromagnetic Resonance (FMR) and Magneto-Optic Kerr Effect (MOKE), revealed surprising differences in the observed in-plane magnetic anisotropy. For the Co/Cu (111), FMR observed two resonance signatures: one exhibiting 6-fold symmetry and another having uniaxial symmetry in the plane of the film. However, MOKE measurements displayed only the uniaxial symmetry plus strong singularities when the external magnetic field was aligned along the magnetic hard axis. Similar results were seen for the Co/Cu (001) films, in which FMR observed the anticipated 4-fold symmetry, while MOKE observed the 4-fold symmetry plus singularities along the magnetic hard axes. These differences highlight the role of domain dynamics and indicate that the singularities are an inherent signature of the magnetization reorientation. The dynamics of the magnetization reorientation were investigated through the use of a simple theoretical model which predicted the shape of the hysteresis curve. The model is based on a free-energy minimization approach in which spin reorientation kinetics are included via coherent rotation with the additional provision that the magnetization can jump between local minima. The field at which the magnetization is allowed to jump is determined by the competition between the energy gained by spin reorientation verses the energy cost of nucleating and propagating magnetic domains. This introduces the possibility of a latched state when the magnetization is saturated along a magnetic hard axis, leading to marked singularities in the coercive field and remanent moment. These singularities emerge as a general consequence of the kinetics imposed by the constrained geometry in ultrathin films.

Microfabricated gas sensors utilizing ultrathin platinum transducing films offer great promise for meeting current challenges to chemical gas sensing. With the research efforts presented in this thesis, considerable advancements have been made toward understanding and optimizing such sensors. The transducing films consist of evaporated, ultrathin (35-100 A) platinum (Pt) films deposited onto metallic titanium (Ti) layers or alumina (Al$\rm\sb2O\sb3)$ layers. Pt/Ti films are studied in microfabricated gas sensing structures and on macrosamples. Titanium may participate in the response of the Pt/Ti films depending upon its oxidation state. Pt/Al$\rm\sb2O\sb3$ thin films are investigated on macrosamples as alternatives to Pt/TiO$\rm\sb{x}$ films used in the microchemical gas sensor. Alumina was chosen as a thermally stable and chemically inactive support. Stability and gas sensitivity of the transducing films are optimized with thermal pretreatments in oxidizing or reducing atmospheres. A range of compositions and microstructures are obtained from one Pt/Ti film. Films pretreated above 600$\sp\circ$C in O$\sb2$ are the most sensitive of the film types investigated. Oxidation of a Pt/Ti thin film at 600$\sp\circ$C results in the preparation of a mixed Pt/TiO$\rm\sb{2-x}$ film with a discontinuous microstructure. Resistance measurements, photoelectron spectroscopies and temperature programmed desorption are used to identify the surface chemical processes responsible for the electrical response of Pt/TiO$\rm\sb{2-x}$ sensing films. Surface reactions involving chemisorbed oxygen play a dominant role in the transduction mechanism of ultrathin platinum films. The films exhibit 10-100% resistance increases upon exposure to O$\sb2$ in the $1\times10\sp{-8}$ to $1\times10\sp{-5}$ Torr range. Films that are pre-exposed to oxygen are sensitive to reducing gases such as hydrogen, propylene and benzene. Resistance decreases in the 20-60% range are observed with reducing gas exposure in the $10\sp{-6}$ to $10\sp{-5}$ Torr range. The mechanistic studies are concluded with temperature programmed reaction spectroscopy (TPRS) experiments involving benzene and propylene on an oxygen saturated platinum foil and 100 A platinum-alumina film. The hydrocarbons react with oxygen at characteristic temperatures to produce CO$\sb2$ and H$\sb2$O. The reaction proceeds with an initial hydrogen abstraction followed by oxidation of the dehydrogenated hydrocarbon intermediates. The dominant parts of the oxidation reactions are the same for the foil and film indicating that alumina does not play a role in the catalytic behavior of platinum-alumina thin films.

Thin film studies of the melittin-lipid interaction were performed to determine if lipid structure, composition and fluidity influence the topographical placement and conformation of melittin in monolayers and Langmuir-Blodgett (LB) multilayers. Monolayers of melittin and the interaction of melittin with various fatty acids, alcohols, dimethyloctadecylchlorosilane and DPPC were studied by surface pressure ((pi)) and potential ((DELTA)V) measurements, and their LB multilayers analyzed by infrared spectroscopy. Topographically different, pH dependent melittin conformations were present at the surface of PO(,4) solutions. Penetration into monolayers displayed large lipid and pH dependent variations. Lipid structure and fluidity affected the rate and extent of incorporation, and the degree of binding of the protein to the monolayers. LB transfer of melittin-lipid monolayers only on emersion of hydrophobic substrates through the monolayer (Z type) was common. Melittin interfered with the deposition of fatty acid monolayers onto hydrophobic substrates under conditions where the protein-free films could be regularly transferred. The nature of the protein lipid interaction, and composition of the LB assemblies were not simply related to changes in (DELTA)V and (pi). The ability to transfer melittin during LB assembly and the infrared spectra of the lipid protein films depended on the proximity of the LB substrate surface and lipid type. Amide absorptions characteristic of antiparallel pleated sheets, (alpha)-helices and random coils were observed. The results suggested that melittin may fold via a type I (beta)-turn, although type III may also be present. Melittin's effect on the lipid CH stretching vibrations was analogous to that found by increasing the interfacial pressure of the protein free monolayer during transfer. Orthorhombic crystalline splitting in the fatty acid (delta)CH(,2) absorptions was present at all interfacial pressures, but disappeared upon collapse. Ionic modification of the membrane components may have occurred as a result of the protein-lipid interaction. In conclusion, lipid composition, fluidity and the characteristics of the LB process all strongly influence melittin's topographical placement and conformation in monolayers and LB multilayers.

This work combines the ultrafast temporal resolution of femtosecond laser spectroscopy with the subwavelength imaging capability of Near-field Scanning Optical Microscopy (NSOM)$\sp2.$ By combining these two techniques, we have demonstrated the potential for making highly localized time resolved spectroscopic measurements in GaAs thin films or devices. This thesis provides an introduction to the principles of subwavelength imaging through NSOM, some details of the construction and operation of this instrument, and the results of femtosecond time resolved measurements made with the instrument. The equal pulse correlation technique (EPC)$\sp3,$ was combined with NSOM to achieve femtosecond time resolution. Using a Ti:Sapphire mode locked laser, the time resolution was approximately 80 fs in all the measurements discussed. Compensation for material dispersion in the single mode fiber was achieved by pre-chirping the pulses using a grating pair$\sp4.$ Cross-correlation by upconversion was used to verify the temporal width of the pulses emitted from the NSOM tip. Spatially resolved (spatial resolution $\approx$200nm) EPC signals at various time delays on a femtosecond time scale ($\approx$90 fs intervals) were recorded in GaAs thin films. In addition to subwavelength spatial resolution in the plane of the sample, the NSOM in combination with EPC shows a very short depth of field (estimated at 200 nm), verified by our measurements in GaAs films. Measurements in GaAs at wavelengths between 800 nm and 840 nm are consistent with the picture that carrier transport plays a role in the transient absorption of a material when using this technique. Measurements in GaAs compared with measurements in LT-GaAs also suggest this. ftn$\sp2$Betzig, Science 251, 1468 (1991). $\sp3$Tang, Phys. Rev. Lett. 51 (9), (1983). $\sp4$Treacy, IEEE J. Quant. Elec. QE-5 (9), (1969).

Hydrogenated amorphous silicon (a-Si:H) thin-film transistors (TFTs) have been widely used in active-matrix liquid-crystal displays (AMLCDs) because they can be deposited over a larger area at low temperatures. However, there are two main disadvantages of a-Si:H TFTs: First, the low field-effect mobility of a-Si:H TFTs limits their application to high-definition and large-area AMLCDs. Second, the low deposition-rate of a-Si:H reduces the throughput of AMLCD manufacturing. Thus, there is an increasing need for a-Si:H TFT having high field-effect mobility, and made by high deposition-rate materials. For this purpose, a high performance a-Si:H TFT based on high deposition-rate materials was fabricated. This a-Si:H TFT has a high field-effect mobility (1.55 cm2/Vs), a low threshold voltage (1.2 V), and a sharp subthreshold slope (0.34 V/decade). Under A.C. bias-temperature-stressing testing, this TFT also shows good stability compared with low-rate deposited a-Si:H TFT. In order to further improve performance, we need to determine what kind of parameters limit a-Si:H TFT performance. For this purpose, an a-Si:H TFT model based on two-dimensional simulation was developed. The simulation program is based on the finite-element method and resolving Poisson's and continuity equations for electrons and holes. Thus the simulation is very suitable and flexible for a-Si:H TFT having high density of states and a thin semiconductor layer. Based on a two-dimensional a-Si:H TFT model, the influence of the density of states and source/drain series resistances on a-Si:H TFTs were analyzed. The simulation results indicate that the conduction-band-tail states mainly affect the a-Si:H TFT field-effect mobility and the non-linearity of current-voltage characteristics. When deep-gap state densities are larger than a certain critical value, they will mainly influence the threshold voltage. This critical value depends on the a-Si:H film thickness, and the thickness and dielectric constant of the gate-insulator. A new technique based on a gated-four-probe TFT structure was introduced for determining the intrinsic performance of a-Si:H TFT without any influence from series resistances. This new structure can also be used to accelerate the optimization process. Based on the simulation for a-Si:H TFT and gated-four-probe TFT structures, for a 10-$\mu$m-channel-length a-Si:H TFT, the source/drain series resistances cause the field-effect mobility to decrease by 43.8%. Of the total source/drain series resistances, contact resistances contribute 70%, and resistances associated with intrinsic a-Si:H and source/drain-gate overlap occupy 15% each. Thus, reduction of contact resistances is the key to achieving short-channel-length high performance a-Si:H TFTs.

The reaction mechanisms of nitrogen containing compounds on semiconductor surfaces and the structure of nitrogen containing insulating compounds have been characterized. The reactions of ammonia and hydrazine with silicon and gallium arsenide surfaces have been studied. Amorphous silicon nitride was grown through cyclic dosing and thermal decomposition of ammonia or hydrazine on the Si(100)-(2x1) surface. Hydrazine was found to nitride the surface more efficiently than ammonia. On the GaAs(100)-c(8x2) surface, the adsorption process and decomposition mechanism for ammonia and hydrazine adsorption were studied using temperature programmed desorption (TPD), x-ray photoelectron spectroscopy (XPS), and high resolution electron energy loss spectroscopy (HREELS). For ammonia adsorption on the GaAs(100)-c(8x2) surface, decomposition to NH$\sb2$ was observed at moderate temperatures (130-250 K) by the HREELS deformation mode at 1476 cm$\sp{-1}$. Hydrazine adsorbs on the GaAs(100)-c(8x2) surface in a side on fashion and decomposition occurs through both N$\sb2$H$\rm{\sb{y}}$ and NH$\rm{\sb{x}}$ (y = 1, 2, 3; x = 1, 2) surface intermediates to give ammonia, hydrogen, and nitrogen gases. Above 350 K, only NH$\rm\sb{x}$, species are present on the surface. Nitrogen containing insulating materials have been characterized by ultra-soft x-ray absorption spectroscopy (USXAS). The core level spectra for imidazole (IM), 4,5-dicyanoimidazole (DCIM), s-triazine (TZ), and hexacarbonitrile-triimidazo-triazine (HTT) above the carbon and nitrogen K edges are reported. The spectra show sensitivity to core level shifts, especially for IM at the nitrogen edge. Substitution of electron withdrawing groups, such as cyano groups, results in a loss of sensitivity to these core level shifts. Complex spectra, such as HTT, are interpreted using the spectra of smaller model compounds (IM, DCIM, TZ). Linear combinations of the model compound spectra can be used to simulate the spectra for HTT and make accurate structural predictions. The limitations of this building block approach to spectral modeling of large molecules with extended $\pi$-systems can largely be circumvented by starting with building blocks, i.e., model compounds, which already contain delocalized $\pi$-systems.

Phase formation was studied in Ni$\sp{+}$ and Xe$\sp{+}$ ion irradiated multilayer and coevaporated Ni-20 at.% Al films. Irradiation resulted in the formation of metastable supersaturated $\gamma$ and HCP phases in both film types. Strong irradiation-induced textures were observed in the multilayers that were less pronounced in the coevaporated films. The texture in the multilayers was attributed to the presence of a slight as-evaporated texture combined with the enhanced atomic mobility due to the heat-of-mixing released during irradiation. Post-irradiation annealing of multilayers transformed the metastable phases to a two phase $\gamma$ + $\gamma\sp\prime$ microstructure. The formation of $\gamma\sp\prime$ proceeded by a traditional diffusional growth mechanism resulting in small ($$ 0. Observed ion-induced grain growth in coevaporated alloy films was successfully modeled as a thermal spike phenomenon in which the driving force due to grain boundary curvature affects atom migration within spikes. The model showed that the ion-induced grain boundary mobility is proportional to the thermal spike-related parameter $\rm{F\sbsp{D}{2}\over\Delta H\sbsp{coh}{3}}$.

The quantification of stress and strain in thin films has been an important matter for many years because of the impact these parameters can have on thin film properties. Thus, many experimental techniques have been developed to measure the amount of stress and strain in thin films. It would be informative to know the entire three-dimensional strain tensor as a function of x-ray penetration depth into thin films, because it has been shown that microstructural features such as grain size, preferred orientation and void density change during film growth. Thus, the first goal of this research project was to develop a technique to yield the entire three-dimensional strain tensor as a function of x-ray penetration depth into thin films. This technique is referred to as high-resolution x-ray diffraction (HRXRD), and was initially used to analyze two random and polycrystalline Mo films with thicknesses of 50 and 100 nm. The second goal was to exploit this technique to study the interplay between microstructure and strain state, so Mo films over a large thickness regime (170 nm to 1.60 $\mu$m), which possess pronounced growth textures, were studied. Large normal strains and stresses were identified near the free surface of the Mo films, and it was proposed that the surface roughness played an integral role in the strain and stress behavior. The third goal was to put traditional thin film stress measurement techniques into context with the improved capabilities of HRXRD. Thus, a systematic study of the average stresses in the aforementioned Mo films was determined with the sin$\sp2\Psi,$ HRXRD, and a substrate curvature technique. The final goal was to use the HRXRD technique to determine the strain tensor in each layer of a compositionally modulated material, a W/Ta/W trilayer. It was found that the in-plane strains were nominally constant throughout the film, but there was a differential in the normal strains due to Poisson contraction.

Positron Annihilation Lifetime Spectroscopy (PALS) and the availability of low energy positron beams permitted the investigation of the surface and interfacial regions of thin polymer films attached to silicon substrates. Unlike traditional measurements of the thermal expansion which track a 1-dimensional change in film thickness, PALS measures an expansion in the average hole volume which is an intrinsic quantity within the polymer itself. A previous dissertation determined that the hole volume expansion in polystyrene (PS) mimicked the thermal expansion behavior. This work determined that for thin PS films the thermal expansion coefficient $\rm(\beta\sb{r})$ in the rubber phase and the glass transition temperature (T$\rm\sb{g})$ decreased with decreasing film thickness. This behavior is modeled using a film comprised of 3 layers: a dead layer with no transition within the measurement range and $\rm\beta\sim\beta\sb{g}$ that extends 50 A from the film/substrate interface, a 20 A thick free surface layer with a $\rm T\sb{g}\sim75\sp\circ C,$ and a intermediate film having normal bulk properties that makes up the remainder of the film thickness. No clear explanation for the dead layer exists at present, but it is believed to result from the combined effects of a reduced entanglement density at a hard wall and some strong force (perhaps electrostatic) at the film/substrate interface. It is also determined that the standard positron implantation model (Makhovian) appears to be adequate for depth profile work in polymer films and the beam energy dependence of the ortho-positronium (o-Ps) formation fraction fits the Spur electron capture model for o-Ps formation within the polymer.

Investigation of the properties of $\beta$-FeSi$\sb2$ for evaluation of possible applications in the automotive industry has been performed. Before $\beta$-FeSi$\sb2$ can be utilized many fundamental questions concerning growth techniques and resulting electrical and optical properties need to be addressed. This work presents relationships between changes in the optical, electrical, and magnetic properties with differences in film microstructure brought about by small variations in growth conditions. Since each of the many growth techniques employed thus far have particular advantages as well as inherent drawbacks, the use of new growth techniques and their subsequent evaluation are important steps toward the implementation of $\beta$-FeSi$\sb2$ as a useful engineering material. We employed pulsed laser deposition to produce thin films of $\beta$-FeSi$\sb2$ on silicon (111) surfaces. The success of this technique is primarily due to congruent evaporation which preserves the stoichiometry of multicomponent targets. Using transmission electron diffraction we report that this technique produces polycrystalline epitaxial films with the smallest geometric misfit known for this material system. Results of optical transmission measurements and calculations reveal the existence of high defect densities that produce band tails with an apparent energy width greater than the measured difference between the indirect and direct transition energy levels. Results show that the defect density is influenced by growth induced variations in the microstructure. This evidence suggests that a high defect density precludes a correct assignment of the intrinsic value as well as of the nature of the energy gap. We also report that additional insight into the nature of the defects and their influence on the film properties is achieved through examination of the magnetization. The temperature dependence of the resistivity is consistent with Mott type variable range hopping conduction for a 3D disordered system. Calculations of the localization length, the optimum hopping distance, and the critical saturation field have produce reasonable values considering the physics of hopping conduction and the sample geometry. We have found that magnetoresistance follows a low-field dependence proportional to H$\sp2$ while the high-field dependence is proportional to H$\sp{1/2}.$ The onset of negative magnetoresistance for all samples occurs at $\sim$20K despite differences in microstructure, and defect density.