This thesis investigates the mixing and dispersion of indoor air pollutants under a variety of conditions using standard experimental methods. It also extensively tests and improves a novel technique for measuring contaminant concentrations that has the potential for more rapid, non-intrusive measurements with higher spatial resolution than previously possible. Experiments conducted in a sealed room support the hypothesis that the mixing time of an instantaneously released tracer gas is inversely proportional to the cube root of the mechanical power transferred to the room air. One table-top and several room-scale experiments are performed to test the concept of employing optical remote sensing (ORS) and computed tomography (CT) to measure steady-state gas concentrations in a horizontal plane. Various remote sensing instruments, scanning geometries and reconstruction algorithms are employed. Reconstructed concentration distributions based on existing iterative CT techniques contain a high degree of unrealistic spatial variability and do not agree well with simultaneously gathered point-sample data.

Novel unsaturated polymers have been synthesized and studied as precursors to silicon carbide and third order nonlinear optical materials. X ray structures were obtained. Kinetic and mechanistic studies of the unique thermal isomerization of dimethylenedisilacyclobutane to a carbene were conducted.

The magnetic properties related to the following high temperature superconductors were measured utilizing a Faraday magnetometer: BaCuO{sub 2+x}, La{sub 2} CuO{sub 4}, Sr{sub 2} RhO{sub 4}, Sr{sub 2} VO{sub 4}, and Sr{sub 2} CuO{sub 3}. Neutron diffraction, magnetic susceptibility, and heat capacity measurements are discussed.

The fundamental processes associated with a photochemical reaction are described with reference to experimental properties of azabenzenes. Consideration of both excitation and relaxation processes led to presentation of the symmetry propagator, a unifying principle which maps system fluctuations (perturbations acting on an initial state) with dissipations (transitions to different states), thus directing the energy flow along competing reactive and nonreactive pathways. A coherent picture of relaxation processes including chemical reactions was constructed with the aid of spectroscopic data. Pyrazine (1,4 diazine) possesses vibronically active modes which provide an efficient mechanism for internal conversion to the first excited singlet state, where other promoting modes of the correct symmetry induce both intersystem crossing to the triplet manifold, isomerization through diaza-benzvalene, and chemical reactions through cycloreversion of dewar pyrazine to yield HCN plus an azete. At higher energies simple H atom loss and internal conversion become more predominant, leading to ring opening followed by elimination of methylene nitrile and ground state reaction products. Efficiency of chemical transformations as dissipation mechanisms versus competing fluorescence, phosphorescence and radiationless relaxation was mapped from near ultraviolet to far ultraviolet by photodissociation quantum yields into reaction channels characterized by molecular beam photofragment translational spectroscopy. A reaction path model …

A parallel program consists of sets of concurrent and sequential tasks. Often, a reduction (such as array sum) sequentially combines values produced by a parallel computation. Because reductions occur so frequently in otherwise parallel programs, they are good candidates for optimization. Since reductions may introduce dependencies, most languages separate computation and reduction. The Sisal functional language is unique in that reduction is a natural consequence of loop expressions; the parallelism is implicit in the language. Unfortunately, the original language supports only seven reduction operations. To generalize these expressions, the Sisal 90 definition adds user-defined reductions at the language level. Applicable optimizations depend upon the mathematical properties of the reduction. Compilation and execution speed, synchronization overhead, memory use and maximum size influence the final implementation. This paper (1) Defines reduction syntax and compares with traditional concurrent methods; (2) Defines classes of reduction operations; (3) Develops analysis of classes for optimized concurrency; (4) Incorporates reductions into Sisal 1.2 and Sisal 90; (5) Evaluates performance and size of the implementations.

Electrical conductance has been measured in-situ in two dimensions in the Ag/Si(111) system as a function of incident adatom flux rate with a 4-probe method. A conductance study in a 3-D conical structure was also made using field emission. For the 2-D study, the origin of conduction is still unclear, as transport by percolating Ag clusters and conduction through the substrate lvia electrons from the film have both been suggested. Experiments varying the flux rate were conducted to decide between the two. Smoother films are expected at lower growth rates which would result in faster drops in the 4-probe voltage; however the 4-probe voltage vs deposition time for various flux rates collapse into a universal curve which indicates that the morphology is not relevant and supports through the substrate. In the 3-D conductance study, a single, lateral micromachined W protrusion on a silica substrate is examined to identify the factors controlling emission in micromachined structures. The I-V characteristics and emission pattern indicate that miniprotrusions of a few hundred Angstroms, much smaller than the nominal radius of the tip, exist on the tip and are responsible for the emission. Adsorption-desorption events from the background environment are the cause of large fluctuations …

Excellent separations of metal ions can be obtained very quickly by capillary electrophoresis provided a weak complexing reagent is incorporated into the electrolyte to alter the effective mobilities of the sample ions. Indirect photometric detection is possible by also adding a UV-sensitive ion to the electrolyte. Separations are described using phthalate, tartrate, lactate or hydroxyisobutyrate as the complexing reagent. A separation of twenty-seven metal ions was achieved in only 6 min using a lactate system. A mechanism for the separation of lanthanides is proposed for the hydroxyisobutyrate system.

Monte Carlo simulations of a proposed a-Si:H-based current-integrating gamma camera were performed. The analysis showed that the intrinsic resolution of such a camera was 1 {approximately} 2.5 mm, which is somewhat better than that of a conventional gamma camera, and that the greater blurring, due to the detection of scattered {gamma}-rays, could be reduced considerably by image restoration techniques. This proposed gamma camera would be useful for imaging shallow organs such as the thyroid. Prototype charge-storage a-Si:H pixel detectors for such a camera were designed, constructed and tested. The detectors could store signal charge as long as 5 min at {minus}26C. The thermal generation current in reverse biased a-Si:H p-i-n photodetectors was investigated, and the Poole-Frenkel effect was found to be the most significant source of the thermal generation current. Based on the Poole-Frenkel effect, voltage- and time-dependent thermal generation current was modeled. Using the model, the operating conditions of the proposed a-Si:H gamma camera, such as the operating temperature, the operating bias and the {gamma}-scan period, could be predicted. The transient photoconductive gain mechanism in various a-Si:H devices was investigated for applications in digital radiography. Using the a-Si:H photoconductors in n-i-n configuration in pixel arrays, enhancement in signal collection …

Mo{sub 5}Si{sub 3} shows promise as a high temperature creep resistant material. The high temperature oxidation resistance of Mo{sub 5}Si{sub 3} has been found to be poor, however, limiting its use in oxidizing atmospheres. Undoped Mo{sub 5}Si{sub 3} exhibits mass loss in the temperature range 800{degrees}-1200{degrees}C due to volatilization of molybdenum oxide, indicating that the silica scale does not provide a passivating layer. The addition of boron results in protective scale formation and parabolic oxidation kinetics in the temperature range of 1050{degrees}-1300{degrees}C. The oxidation rate of Mo{sub 5}Si{sub 3} was decreased by 5 orders of magnitude at 1200{degrees}C by doping with less than two weight percent boron. Boron doping eliminates catastrophic {open_quote}pest{close_quote} oxidation at 800{degrees}C. The mechanism for improved oxidation resistance of boron doped Mo{sub 5}Si{sub 3} is due to scale modification by boron.

This thesis describes a new implementation of the implicitly parallel functional programming language SISAL, for massively parallel processor supercomputers. The Optimizing SISAL Compiler (OSC), developed at Lawrence Livermore National Laboratory, was originally designed for shared-memory multiprocessor machines and has been adapted to distributed-memory architectures. OSC has been relatively portable between shared-memory architectures, because they are architecturally similar, and OSC generates portable C code. However, distributed-memory architectures are not standardized -- each has a different programming model. Distributed-memory SISAL depends on a layer of software that provides a portable, distributed, shared-memory abstraction. This layer is provided by Split-C, a dialect of the C programming language developed at U.C. Berkeley, which has demonstrated good performance on distributed-memory architectures. Split-C provides important capabilities for good performance: support for program-specific distributed data structures, and split-phase memory operations. Distributed data structures help achieve good memory locality, while split-phase memory operations help tolerate the longer communication latencies inherent in distributed-memory architectures. The distributed-memory SISAL compiler and run-time system takes advantage of these capabilities. The results of these efforts is a compiler that runs identically on the Thinking Machines Connection Machine (CM-5), and the Meiko Computing Surface (CS-2).

This dissertation describes a new algorithm for simulating low frequency, kinetic phenomena in plasmas. DArwin Direct Implicit Particle-in-Cell (DADIPIC), as its name implies, is a combination of the Darwin and direct implicit methods. One of the difficulties in simulating plasmas lies in the enormous disparity between the fundamental scale lengths of a plasma and the scale lengths of the phenomena of interest. The objective is to create models which can ignore the fundamental constraints without eliminating relevant plasma properties. Over the past twenty years several PIC methods have been investigated for overcoming the constraints on explicit electrodynamic PIC. These models eliminate selected high frequency plasma phenomena while retaining kinetic phenomena at low frequency. This dissertation shows that the combination of Darwin and Direct Implicit allows them to operate better than they have been shown to operate in the past. Through the Darwin method the hyperbolic Maxwell`s equations are reformulated into a set of elliptic equations. Propagating light waves do not exist in the formulation so the Courant constraint on the time step is eliminated. The Direct Implicit method is applied only to the electrostatic field with the result that electrostatic plasma oscillations do not have to be resolved for stability. …