Since strain changes the interatomic spacing of matter and alters electron and phonon dispersion, an applied strain ϵ can modify the thermal conductivity κ of a material. This thesis shows how the strain induced by heteroepitaxy is a passive mechanism to change κ in a thin film and how the film thickness is key to the functional form of κ(ϵ). Molecular Dynamics simulations of the physical vapor deposition and epitaxial growth of ZnTe thin films provide insights into the role of interfacial strain on the thermal conductivity of a deposited film. ZnTe films grown on a lattice mismatched CdTe substrate exhibit ~6% in-plane biaxial tension and ~7% out-of-plane uniaxial compression. In the T=700 K to 1100 K temperature range, the conductivities of strained ZnTe layers that are 5 unit cells thick decrease by ~ 35%, a result that is relevant to thermoelectric devices since strain can also enhance charge mobility and increase their overall efficiency. The resulting understanding of dκ/dT shows that strain engineering can also be used to create a thermal rectifier in a material that is partly strained and partly relaxed, like at the junction of an axial nanowire heterostructure. To better isolate the role of strain, the study is extended to free-standing ZnTe films with thicknesses between 116 Å to 1149 Å under the application of both uniform and biaxial strain between -3% to 3% at 300 K. Since the boundaries of the film are diffuse, κ becomes size dependent when the film thickness approaches the order of the mean free path of the phonons. As this thickness is decreased, the magnitude of κ decreases until boundary scattering dominates so that κ(ϵ) depends on v_g (ϵ). This conclusion is important as it can be generalized to other materials and potential functions; it suggests that if a film is thin enough for boundary scattering to dominate, then the behavior of κ(ϵ) can be predicted based on the bulk dispersion curve alone, which should greatly simplify strain-based device design. / Thesis / Doctor of Philosophy (PhD) / Since strain changes the interatomic spacing of matter and alters electron and phonon dispersion, an applied strain ϵ can modify the thermal conductivity κ of a material. This thesis shows how the strain induced by heteroepitaxy is a passive mechanism to change κ in a thin film and how the film thickness is key to the functional form of κ(ϵ). Molecular Dynamics simulations of the physical vapor deposition and epitaxial growth of ZnTe thin films provide insights into the role of interfacial strain on the thermal conductivity of a deposited film. The result is relevant to thermoelectric devices since strain can also enhance charge mobility and increase their overall efficiency. The resulting understanding of dκ/dT shows that strain engineering can also be used to create a thermal rectifier in a material that is partly strained and partly relaxed, like at the junction of an axial nanowire heterostructure.

An investigation of polysulfone-Al/Li alloy interaction involved single lap shear joints and wedge samples following an FPL etch, sulfuric acid anodization (SAA) and phosphoric acid anodization (PAA). The study of the Al/Li surfaces involved the determination of the elemental composition and morphological features of the pretreated adherend before bonding and following failure. When thermoplastic polysulfone (PSF) was either thermally pressed or primed onto the microporous surface, the PSF indeed penetrated into the porous oxide and thereby provided a mechanical means of adhesion. The wedge test results for the adherend pretreated by PAA and SAA were superior to those for the FPL etched adherend. The failure path for the FPL etched samples was at the adhesive/oxide interface whereas the failure path for the SAA and PAA samples was within the adhesive but with occasional divergence of the crack into the oxide. The porous oxides on Al/Li alloy formed after PAA and SAA treatment were shown to undergo dramatic changes in morphology on short term (<90 hrs) exposure to 71 C and 100% R.H. environment. The mechanism of failure was due to moisture which caused slight hydration of the Al/Li oxide and subsequent debonding of the PSF from the oxide layer. Lithium was not concentrated at the surface in the PAA treated Al/Li alloy as shown by AES depth profiling. The effect of lithium on the durability of the bonded alloy is considered minimal. Along these lines, cyclic loading, use of primers, and infrared spectroscopy studies have been carried out. The mode of moisture intrusion into the polysulfone-Al/Li oxide interphase region is discussed. Specifically, water molecules diffuse into the polysulfone rather than transporting along the interface. Moisture then attacks the oxide interface. Thin polysulfone coatings on pretreated aluminum surfaces were characterized utilizing dynamic mechanical thermal analysis (DMTA), and dielectric thermal analysis (DETA) to detect changes in the molecular motions and structural transitions in the polysulfone-aluminum interphase. The order of the loss peak temperature of the polysulfone is, PSF coating on a porous Al > PSF coating on a smooth Al > neat PSF film. The activation energy of relaxation is also lower for neat PSF when compared to the thin film cast onto a smooth Al or a porous PAA Al substrate. The loss peak temperature shift and the higher activation energy associated with the coated films can be explained by the entropy being reduced when the chains are laid down in two dimensions. Thus studies of polymer properties in the interphase region will contribute to the understanding of the adhesive-adherend interaction. / Ph. D.

A new approach to the fabrication of polymeric thin films is presented. This approach, chemical vapor copolymerization (CVcP), has all the advantages of chemical vapor polymerization (CVP), including exceptional purity, highly conformal coatings, continuous films even when very thin, low stress, and low environmental impact. The range of film properties available by CVcP is much greater than by CVP. A specially modified deposition system was constructed to study deposition kinetics. A model was developed which allowed quantification of the order of initiation of paraxylylene (PX), which is the initiation system for all the work reported here. This model suggests a trimer diradical is the smallest stable diradical species formed by PX at room temperature, confirming thermodynamic predictions. This system also allows the calculation of reactivity ratios of PX with vinylic comonomers. A model is developed in which reactivity ratios can be determined if the following quantities are known: a) thickness vs. position of the final film; b) partial pressures of each reactive species entering the deposition chamber and c) composition vs. position in the final film. This model was tested yielded reasonable values for reactivity ratios. A polymeric film extremely low in refractive index (1.38-1.39 in the visible region) is presented. This film is formed by copolymerizing poly(parachloroxylylene), or PX-C, with perfluorooctyl methacrylate (PFOMA). The refractive index of the homopolymers, by contrast, are in the 1.60-1.68 range. Film growth rates are very low for this new material. Finally, a new deposition procedure is introduced in which a comonomer with a low vapor pressure is codeposited with PX. The reactor temperature is above the ceiling temperature of PX deposition, but at a temperature where the comonomer can condense. This makes the deposition environment extremely rich in comonomer, yielding films whose final properties are nearly identical to films composed entirely of the comonomer. The procedure is demonstrated using n-phenyl maleimide (NPMI) as a comonomer; films produced had thermal stabilities nearly matching those of poly(NPMI). This procedure has great promise for broadening the use of polymeric thin films in the microelectronics industry, as well as other fields. / Ph. D.

A systematic method was developed for the deposition of β-SiC thin films on Si(100) substrates in a hot wall reactor, using low pressure chemical vapor deposition (LPCVD). Due to poor adhesion resulting from lattice mismatch and difference in thermal expansion coefficients between the (SiC films and the Si(100) substrates, the feasibility of forming a SiC buffer layer on the Si(100) surface before beginning the chemical vapor deposition (CVD) process was investigated. The SiC buffer layers were formed with either a smooth or porous morphology. A nonporous Si(100) substrate with a 35Å thick SiC buffer layer was formed when the Si surface was heated at 1050°C in an atmosphere of C₂H₂ and H₂. A porous surface was obtained when the Si substrate was heated at 1000°C in C₂H₂ alone. The porous defects were correlated to the out—diffusion of Si in the carburizing process. On smooth Si(100) substrates, polycrystalline and stoichiometric β-SiC thin films with the (111) planes paralleling the Si(100) substrates were grown from a CH₃SiCl₃ (MTS)—H₂ mixture at 1050°C. At high H₂/MTS ratios and/or low deposition pressures, no etching on the Si substrates of the β-SiC films was observed, resulting in a smooth topography. Degradation in film morphology, changes in the preferred orientation, and etching of the Si substrates were observed at higher pressures, temperatures, and H₂/MTS ratios. The etching of the Si substrate was due to the out—diffusion of Si atoms from the substrate and the presence of Cl—containing radicals, which resulted from the decomposition of MTS molecules before arriving at the substrates. A model of the deposition mechanism is proposed which predicts the deposition rates in a hot wall CVD reactor and agrees very well with the experimental data. On the Si(100) substrate with a porous topography, epitaxial β-SiC(100) thin films were grown from MTS—H₂ at 1150°C. The crystallinity of the deposited films was influenced by the deposition time. With increasing deposition time, rotational β-SiC(100) crystals and polycrystalline β-SiC with a highly preferred orientation of (100) and/or (111) were obtained. At a lower temperature of 1100°C, poor morphology and polycrystalline β-SiC thin films were observed. Finally, a new approach to the calculation of the local equilibrium CVD phase diagrams, which represent the most stable phases above the substrates in a hot wall reactor, for SiC deposition from the MTS—H₂ gas mixture by coupling the depletion effects to the equilibrium thermodynamic computer code SOLGASMIX—PV. The calculated CVD phase diagrams were also compared with experimental and the literature data. Although the local equilibrium CVD phase diagrams predicted the deposition of single phase SiC better than established CVD phase diagrams, the experimental regions for depositing single phase SiC are larger than those calculated from local CVD phase diagrams. This may be because of the high linear velocity of the gas flux under low pressure and the polarity of the Si—containing intermediate species. / Ph. D.

Polyimides, due to their high thermal stability, excellent chemical stability, useful mechanical properties, and extremely high electrical resistivity, are utilized in aerospace, electronic, and specialty consumer markets. However, in some applications, lower electrical resistivity is preferable. Toward this goal, polyimide films have been modified with metal salts and metal complexes. Depending upon processing conditions, the films contain ionic species uniformly distributed throughout their bulk, or are highly anisotropic structures containing near-surface metal or metal oxide. Evaluation of solvent cast films by a variety of analytical techniques has been used to develop structure-property-process correlations in cobalt chloride modified polyimides. To date, no interaction between the additive and the matrix has been indicated by ultraviolet, visible, or infrared spectroscopy, though by differential scanning calorimetry and a specialized thermogravimetry technique, some interaction is implied. Elemental analysis has verified that polyimides having a metal oxide surface also have residual bulk metal ion content. A major controller of the bulk resistivity of metal ion modified polyimide films was found to be the polymer glass transition temperature; a lower glass transition temperature resulted in lower electrical resistivity at a particular temperature. Central to this research work was the design and construction of a sensitive and reliable electrical resistivity measurement system. The system was also found useful for probing polymer contamination and molecular motion. Further, the electrical measurements indicated that uniformity between samples was poor. Modification of an inert gas oven, allowing processing in controlled atmosphere, proved that film surface conductivity is critically influenced by humidity. A model has been proposed that supports both the d.c. electrical properties and surface spectroscopic data. The model and details in the ceramic literature were the primary factors in pursuing a specific codoped polyimide system. With the codoped system, the electrical resistivity and activation energy for conduction, compared with either of the singly doped polyimide films, were predictably and favorably reduced. / Ph. D.

Yes / Reverse roll coating is probably the most widely used coating operation, yet its full potential has not been exploited as it is shown in this paper which considers operation with a negative gap. We demonstrate through a wide range of experimental data that such operation can yield very thin and stable films with no ribbing or cascade instabilities when low viscosity fluids are used. Typically, stable film thickness less than 5μm can be obtained at speeds up to 150 m/min when a rubber roller is used at -100 μm gap with fluids of viscosity in the range 10-200 mPa.s. These film thicknesses can be made to decrease further down to 1 or 2 microns with a judicious choice of speed ratios (applicator to metering roller) and rubber hardness. Such new findings make this simple coating method an attractive roll to roll technique for application in the newer coating technologies, such as in the production of solar cells and plastic electronics. The data obtained in this study have been underpinned by a model based on the classical lubrication theory, well developed for such flow situations. Essentially it is shown that the film thickness non dimensionalised with respect to the set negative gap is controlled through a single parameter, the elasticity number Ne which combines all the operating parameters. Of course, this flow problem has complexities, particularly at high speed ratios and at zero gap so the data obtained here can serve as a basis for more comprehensive modelling of this classical fluid mechanic problem. / Films R&D Centre of Toyobo Co. Ltd., Otsu, Japan and the Thin Films Research Group of the University of Bradford, UK.

Yes / Tensioned web-roll coating is widely used but has surprisingly received little research attention. Here, a new semi-empirical model that predicts film transfer from applicator roller to web is developed and tested against data collected from a pilot coating line. The film transfer is found to vary linearly with web to applicator speed ratio S. Flow stability investigations revealed three types of defects: rivulets, air entrainment due to dynamic wetting failure and cascade, occurring at different values of S and capillary number Ca. Rivulets occurred at Ca< 0.4 and S> 0.71-0.81, air entrainment at Ca>0.4 and S>0.71-0.83 and cascades at S>1.1 for Ca up to 6. Web speeds at which dynamic wetting failure occurred were, for the same Ca, comparatively higher than those that occur in dip coating. The data show that such hydrodynamic assistance is due to the coating bead being confined, more so with increasing web wrap angle β. / The authors acknowledge the support of the Films R&D Centre of Toyobo Co. Ltd., Otsu, Japan and of the Thin Liquid Films Research Group of the University of Bradford, UK.

This thesis describes direct measurements of equilibrium adsorption of ions in thin (< 100 nm) aqueous films. Adsorption in thin films is important because it is through adsorption that the stability of colloidal suspensions is frequently tuned. The vast majority of measurements of adsorption to date have been to a single interfaces, whereas the subject of this thesis is adsorption in a thin film between two interfaces. There are two isolated interfaces when particles in a suspension are far apart, but during the collision, a thin film forms between the particles, and the properties of the thin film determines the stability of the colloid. Thus, adsorption in the thin film determines the stability of the colloidal dispersion. There is a distinct gap in the scientific literature concerning adsorption in thin films mainly because there is no technique for measuring such adsorption. To fill this gap in knowledge, I first developed of a technique to directly measure adsorption in thin films, and then applied this technique to explore the behavior of co-ions near charged interfaces as a function of bulk solution composition and the thickness of the film. The adsorption behavior of fluorescein, a di-anion, to negatively charged silica interfaces was studied in dilute electrolytes. The focus was on the effect of the electrostatic screening length, or Debye-length. The separation was measured using interference microscopy and the adsorption of fluorescein was measured using fluorescence microscopy. The Debye-length was altered by variation of the background salt (NaCl) concentration in dilute (<1 M) solution. The surface excess of adsorption for fluorescein was shown to depend on both the Debye-length and the separation distance between two interfaces. Increasing the Debye-length from 4 nm to 21 nm increased the plateau surface excess at large separations, and decreasing the separation lead to a monotonically decreasing surface excess. The surface excess varied over a range that scaled with the Debye-length. The results were compared to solution of the Poisson-Boltzmann model and good agreement was found between the model and the experiment. The effect of background salt concentration on fluorescein adsorption was also studied in concentrated electrolytes (2.5 – 10 M) for various monovalent salts (LiCl, NaCl and CsCl). The results showed that the fitted electrostatic screening length showed an opposite trend to predictions from Poisson-Boltzmann, with the screening-length increasing with increasing salt concentration. That is, the Debye-length prediction was quantitatively incorrect and predicts the incorrect trend. For example, in 10 M LiCl where the Debye-length is 0.1 nm, and therefore colloidal chemists would traditionally predict that double-layer forces are negligible, my results show that the actual decay length is about 10 nm, which is about the same as in 10-3 M LiCl solution. The rate of increase of screening-length as a function of concentration was also an ion specific effect. In addition, the results show that there is an inversion of the surface charge in concentrated salt solution. The original device on which all the above measurements were made had two limitations: (1) the maximum film thickness was 50 nm and (2) the film was asymmetric, which hampered calculation of the surface excess and increased the number of degrees of freedom in modeling of the adsorption. In the last part of my thesis, I describe development of a symmetric sample which (1) enables measurement of films up to 1 µm, (2) simplifies modeling of the optics by eliminating optical interference of the fluorescence excitation, and reduces the number of parameters when comparing to models. / Doctor of Philosophy / This thesis aims to understand the behavior of electrically charged molecules and atoms in thin nanometer scale (< 100 nm) liquid films subject to confinement between two charged interfaces. This situation frequently arises in colloidal suspensions, which consist of tiny sub-microscopic particles (colloid), droplets and large molecules dispersed in a second continuous medium. The stability of these suspensions, i.e. whether the colloidal materials agglomerate and sediment out of the suspension or remain stably suspended, depends on the surface forces between their interfaces during collision events, which frequently arise due to Brownian motion. As the fluid between particles thins as they approach each other during these collision events, the behavior of the dissolved molecules can be significantly different than when they are far apart due to the presence two interacting interfaces. Typically the dissolved molecules are used to tune the surface forces and understanding their behavior in confinement is relevant to a colloid scientist whose aim is to tune the behavior of the suspension. In the first part of this work, a technique is developed that serves as the static analogue to colloidal objects colliding with each other. The equilibrium behavior of a negatively charged fluorescent ion is measured as a function of film thickness and background salt concentration between two negatively charged interfaces. The Poisson-Boltzmann model predicts that with decreased salt concentration, there is a greater magnitude of depletion of the fluorescent ion at large separations and the characteristic length over which there is a change in the magnitude of depletion increases. Good agreement is found between the model and the experiment validating the technique developed and providing the first direct observation of molecular behavior subject to confinement as a function of solution composition. This effect of background salt type and concentration was tested for concentrated electrolytes as well. The experimental results showed an opposite trend to predictions from the Poisson-Boltzmann model. The fluorescent ion was now adsorbed to negatively charged interfaces indicating that the negatively charged interfaces were now positively charged. The magnitude of adsorption at large separations and characteristic length over which the magnitude of adsorption changes was a function of the salt concentration and the ion type. Finally, improvements were made to the original device to overcome limitations with the original device. The limitations were that (1) the maximum film thickness was 50 nm and (2) the interfaces were asymmetric which complicated theoretical calculations of the equilibrium behavior of the ions. In the last part of my thesis, I develop a sample which (1) enables measurements of films up to 1 µm and (2) simplifies the optical modeling necessary in the first two sections of this thesis.

Cellulose and hemicelluloses have attracted increasing interest as renewable biopolymers because of their abundance. Furthermore, the recognition of biomass as a sustainable and renewable source of biofuels has driven research into the assembly and disassembly of polymers within plant cell walls. Cellulose thin films are useful in the study of interactions between cellulose and hemicelluloses, and quartz crystal microbalances with dissipation monitoring (QCM-D), surface plasmon resonance (SPR) and atomic force microscopy (AFM) are widely used to investigate polymer adsorption/desorption at liquid/solid interfaces. In this study, smooth trimethylsilyl cellulose (TMSC) films were spincoated onto gold QCM-D sensors and hydrolyzed into ultrathin cellulose films upon exposure to aqueous HCl vapor. The adsorption of arabinoglucuronoxylan (AGX) and arabinoxylan (AX) onto these cellulose surfaces was studied. The effects of structure, molar mass and ionic strength of the solution were considered. Increasing ionic strength increased AGX and AX adsorption onto cellulose. While AGX showed greater adsorption onto cellulose than AX by QCM-D, the trend was reversed in SPR experiments. The combination of QCM-D and SPR data showed a greater amount of water was trapped within the AX films. Both adsorbed AGX and AX films were subsequently visualized by AFM. Images from AFM showed AGX and AX adsorbed as aggregates from water, while AGX and AX adsorbed from CaCl2 yielded smaller xylan particles with more numerous globular structures on the cellulose surfaces. Images from AFM of xylan films on bare gold surfaces also showed layers of uniform aggregates that were consistent with AX and AGX aggregation in solution. / Master of Science

In SiC/Ti composites, the physical-chemical behaviour at the interface at fabrication temperatures, is important. Ideally, only a thin bonding film is desired after fabrication with zero growth during the actual service conditions. In this study, two mechanisms which lower the growth rate of silicides about the SiC/Ti interface were investigated. Planar composites that incorporated Mo₅Si₃ or Al films, were compared with the uncoated SiC/Ti standards to investigate the "barrier" and "rejection" mechanisms respectively. Also, SiC/Ti₃Al/Ti composites were used to study the effectiveness of the two mechanisms combined, with the Ti₃Al acting as a barrier and with Ti(Al) retarding the reaction by rejection. Samples were vacuum-annealed at 875°C for various times and examined by X-ray diffraction techniques. The results from the SiC/Al/Ti and SiC/Ti₃Al/Ti composites indicated that silicide formation would be retarded in the presence of an Al saturated Ti matrix. Also, the formation of a Ti₃Al layer near the SiC/Ti interface serves as an effective barrier to diffusion of Si atoms. In the SiC/Mo₅Si₃/Ti composites however, the dissociation of the Mo₅Si₃ phase formed during sample preparation results in additional silicide formation at the interfaces. If stoichiometric Mo₅Si₃ films are sputter-deposited, they could be used as effective barrier. The inter-diffusion of Al films, sputter deposited on Ti substrates, at elevated temperatures have been analyzed. Samples were vacuum annealed at 635°C and 900°C for various times and then examined using X-ray diffraction. In each case, intensity bands from alpha-Ti and diffraction lines from an intermetallic compound have been observed. For subsequent annealing times, the integrated intensities from the compound decreased slightly, indicating that the compound partially decomposed and released Al into the alpha-Ti lattice. This resulted in an alpha-Ti diffracted intensity band that results from a range of compositions. Intensity bands from Ti(101) were used to obtain composition profiles for Al in alpha-Ti, by using X-ray diffraction techniques and computer simulations. In each case, mass conservation of Al atoms was used to determine the interface motion. It was assumed that the Al profiles extend continuously into the_disordered Ti₃Al and Ti₂Al composition regions. This assumption was verified by preparing a Ti₂Al (disordered) powder standard and obtaining accurate lattice parameters. An iterative solution was used to determine composition dependent diffusion coefficients of one-dimensional zones, for two-phase systems. Diffusion zones extending from a few tenths to several microns have been examined. Al diffusion coefficients in alpha-Ti and the Ti₃Al phase have been determined. Activation energies were calculated from the diffusion coefficients at 635°C and 900°C and compared with those obtained from a melting point correlation. Lower activation energies from the present investigation indicated that there was some grain boundary diffusion at these intermediate temperatures. / Ph. D.