As multistep, multilayer processing in semiconductor industry becomes more complex, the role of cleaning solutions and etching chemistries are becoming important in enhancing yield and in reducing defects. This thesis demonstrates successful formulations that exhibit copper and tungsten compatibility, and are capable of Inter Layer Dielectric ILD) cleaning and selective Ti etching. The corrosion behavior of electrochemically deposited copper thin films in deareated and non-dearated cleaning solution containing hydrofluoric acid HF) has been investigated. Potentiodynamic polarization experiments were carried out to determine active, active-passive, passive, and transpassive regions. Corrosion rates were calculated from tafel slopes. ICP-MS and potentiodynamic methods yielded comparable Cu dissolution rates. Interestingly, the presence of hydrogen peroxide in the cleaning solution led to more than an order of magnitude suppression of copper dissolution rate. We ascribe this phenomenon to the formation of interfacial CuO which dissolves at slower rate in dilute HF. A kinetic scheme involving cathodic reduction of oxygen and anodic oxidation of Cu0 and Cu＋1 is proposed. It was determined that the reaction order kinetics is first order with respect to both HF and oxygen concentrations. The learnings from copper corrosion studies were leveraged to develop a wet etch/clean formulation for selective titanium etching. The introduction of titanium hard-mask HM) for dual damascene patterning of copper interconnects created a unique application in selective wet etch chemistry. A formulation that addresses the selectivity requirements was not available and was developed during the course of this dissertation. This chemical formulation selectively strips Ti HM film and removes post plasma etch polymer/residue while suppressing the etch rate of tungsten, copper, silicon oxide, silicon carbide, silicon nitride, and carbon doped silicon oxide. Ti etching selectivity exceeding three orders of magnitude was realized. Surprisingly, it exploits the use of HF, a chemical well known for its SiO2 etching ability, along with a silicon precursor to protect SiO2. The ability to selectively etch the Ti HM without impacting key transistor/interconnect components has enabled advanced process technology nodes of today and beyond. This environmentally friendly formulation is now employed in production of advanced high-performance microprocessors and produced in a 3000 gallon reactor.

Electric power systems exhibit low frequency oscillations associated with dynamics known as electromechanical modes. A mode is described by the frequency, damping, and shape of the oscillation. The mode shape defines the amplitude and phasing of the oscillation throughout the system. Knowledge of the electromechanical modal properties of a power system is of great importance to its safe and reliable operation. If the damping of a particular mode is allowed to become too low, the oscillation of the mode may grow out of control and cause a system wide outage like the one observed by the Western Electricity Coordinating Council in 1996. Therefore, accurate estimates of the electromechanical modes are required. In the past, the modes were estimated through the creation and maintenance of detailed models. When linearized, an eigenanalysis of the state matrix associated with the system provides the complete modal information. The accuracy of the estimated modes, however, is dependent on the accuracy of the model, which for the 1996 outage proved to be inadequate for the conditions that led to the event. In the years since, several mode estimation schemes based on measured power system data have been developed using modern system identification techniques. These methods benefit from the fact that a detailed system model is not required and they can also serve to validate and, if necessary, update the detailed system models. This dissertation presents two new methods for mode estimation from measured data. The first uses transfer functions constructed between pairs of system outputs to estimate the mode shape. The second examines the elements of the multichannel system transfer function to estimate the modal frequency, damping, and shape. Both methods benefit from the fact that they may be implemented using any one of a number of available system identification techniques. Typically the accuracy of the mode estimates is assessed using bootstrapping. Here a more efficient method of bootstrapping based on a derived asymptotic parameter distribution is presented. Each of the new methods performance is verified using both simulated and measured data.

This dissertation approaches high resolution scatterometry from a new perspective. Three related general topics are addressed: high resolution sigma 0 imaging, wind estimation from high resolution sigma0 images over the ocean, and high resolution wind estimation directly from the scatterometer measurements. Theories of each topic are developed, and previous approaches are generalized and formalized. Improved processing algorithms for these theories are developed, implemented for particular scatterometers, and analyzed. Specific results and contributions are noted below. The sigma0 imaging problem is approached as the inversion of a noisy aperture-filtered sampling operation–extending the current theory to deal explicitly with noise. A maximum aposteriori MAP) reconstruction estimator is developed to regularize the problem and deal appropriately with noise. The method is applied to the SeaWinds scatterometer and the Advanced Scatterometer ASCAT). The MAP approach produces high resolution sigma 0 images without introducing the ad-hoc processing steps employed in previous methods. An ultra high resolution UHR) wind product has been previously developed and shown to produce valuable high resolution information, but the theory has not been formalized. This dissertation develops the UHR sampling model and noise model, and explicitly states the implicit assumptions involved. Improved UHR wind retrieval methods are also developed. The developments in the sigma0 imaging problem are extended to deal with the nonlinearities involved in wind field estimation. A MAP wind field reconstruction estimator is developed and implemented for the SeaWinds scatterometer. MAP wind reconstruction produces a wind field estimate that is consistent with the conventional product, but with higher resolution. The MAP reconstruction estimates have a resolution similar to the UHR estimates, but with less noise. A hurricane wind model is applied to obtain an informative prior used in MAP estimation, which reduces noise and ameliorates ambiguity selection and rain contamination. Keywords: scatterometry, image reconstruction, irregular sampling, aperture-filtered sampling, wind, ambiguity selection, hurricane, maximum aposteriori estimation, inverse problems

As CMOS technology scales down, process variation introduces significant uncertainty in power and performance to VLSI circuits and significantly affects their reliability. If this uncertainty is not properly handled, it may become the bottleneck of CMOS technology improvement. This dissertation proposes novel techniques to model, analyze, and optimize power and performance of FPGAs and ASICs considering process variation. This dissertation focuses on two aspects: 1) Process and architecture concurrent optimization for FPGAs； 2) Statistical timing modeling and analysis. To perform process and architecture concurrent optimization, an efficient and accurate FPGA power, delay, and variation evaluator, Ptrace, is proposed. With Ptrace, we present the first in-depth study on device and FPGA architecture co-optimization to minimize power, delay, area, and variation considering hundreds of device and architecture combinations. Furthermore, to enable early stage process and architecture co-optimization without stable device models, we develop transistor level and circuit level power, delay, and reliability models and incorporate them with Ptrace. With the extended Ptrace, we perform architecture and process parameters concurrent optimization for FPGA power, delay, variation, and reliability. To perform statistical timing modeling and analysis, we first present an efficient and accurate statistical static timing analysis SSTA) flow for non-linear cell delay model with non-Gaussian variation sources. All operations in this flow are performed by analytical equations without any time consuming numerical approach. Then, to further improve the efficiency and accuracy of statistical timing analysis, we develop a new die-level spatial variation model which accurately models the across-wafer variation. Besides modeling spatial variation, mean and variance uncertainty introduced by limited number of samples is another problem in SSTA. To solve this problem, we evaluate the confidence for statistical analysis and estimate the guardband value to ensure a target confidence. To the best of our knowledge, this dissertation is the first novel study of device, process, and architecture concurrent co-optimization for FPGA power, delay, variation, and reliability； and is the first work to model across-wafer variation at die-level and to consider confidence guardband in statistical analysis.

This dissertation investigates the application of multiple hypothesis testing procedures also known as multiple comparison problems), especially the control of the False Discovery Rate FDR), to several signal detection problems including distributed detection in wireless sensor networks, target detection in radars and Bayesian hypothesis testing with uncertain priors. The objective of the distributed detection problem is to design the local sensor decision rules and the fusion rule such that the system-level detection performance is optimized. Under the conditions that the local sensor performance metrics are unknown and the sensor and target locations are random, design of the optimum decision rules is an open and challenging research problem. In this dissertation, we propose a novel detection framework where the local sensor decision rules are obtained by controlling the FDR and the fusion rule is a randomized decision rule. The proposed approach is shown to provide substantial detection performance improvement over the present state of the art. In radar systems, the objective is to detect the presence of a target in clutter and noise. Conventional detection strategies involve hypothesis tests on each test cell at a pre-defined probability of false alarm. In this dissertation, we propose a novel approach where hypothesis tests are performed simultaneously on a number of test cells in a surveillance area SA) while controlling the FDR. Our approach thus proposes a shift from the conventional control of a cell-based statistic to the control of a region-based statistic for radar target detection. The proposed approach shows adaptivity to the unknown target density and provides improved detection performance in target-rich environments. However, control of a region-based statistic has several limitations, especially at very low and very high target densities. Hence, in this dissertation, we propose several solutions based on the concepts of decision fusion as well as algorithm fusion to overcome these limitations. We demonstrate that the proposed detection framework can distinguish between target-rich and target-starved regions and provides a more efficient and robust system design. In the Bayesian hypothesis testing problem, the objective is to minimize the average misclassification error while conducting multiple binary hypothesis tests with identical but unknown priors. It is further assumed that the prior is not a deterministic value but is a random variable with known density function. Under this problem setting, we propose a FDR based detection approach and demonstrate that the detection performance of the FDR based approach is close to that of a near-optimal Expectation-Maximization EM) based approach while requiring significantly less computation. Thus, the proposed approach is particularly suitable for Bayesian detection problems involving multiple binary hypothesis tests under real-time performance requirements.

The application and study of iterative message-passing decoders has exploded in recent years, due to their amazing efficiency and near-optimal performance. Much of the analysis of these decoders relies on a heuristic link between the local nature of these algorithms and certain graph structures, called graph covers, that are locally indistinguishable. The precise relationship between graph covers and computation trees, which Wiberg proved can be used to exactly model the behavior of iterative message-passing decoders, remains unclear. The focus of this dissertation is to further explore the relationship between graph covers and computation trees, and their related pseudocodewords, so that the plethora of results on graph covers may be more readily applied to computation trees, and hence to the analysis of iterative message-passing decoding algorithms. We show that every graph cover pseudocodeword gives rise to a computation tree pseudocodeword and that, conversely, every computation tree pseudocodeword does indeed arise from a graph cover pseudocodeword. Although these results strengthen the relationship between these different types of pseudocodewords, it is clear that more study is needed, as we show that there is a single graph cover pseudocodeword that simultaneously gives rise to every computation tree pseudocodeword. We also present a completely graphical characterization of certain graph cover pseudocodewords which are known to cause errors in graph cover decoding of cycle codes.

In this thesis, we present a study of two problems relevant to sensor networks. We first study a new form of sensor networks where the parameters of interest are the inter-node time-varying channels) and then study the problem of determining the optimal location of a mobile relay node to optimize a certain cost function for the sensor network. Networks which collect information about the channels are becoming relevant in environmental monitoring systems, like underwater tsunami detecting networks. We first present two examples to drive home the point that joint channel estimation and communication can outperform training based schemes. First, the binary symmetric channel is examined； an achievable capacity-distortion trade-off is derived for both joint and time-orthogonal protocols. For the flat fading, additive, white, Gaussian noise channel, a novel joint communication and estimation scheme using low correlation sequences is presented. It is observed that in most situations, joint communication and estimation performs better than a scheme where communication and estimation are performed individually, furthermore, the gains of joint communication and estimation over individual communication and estimation can be significant as the distortion tolerance increases. It is also observed that even a slight tolerance to errors in the channel parameters close to the theoretical lower bounds yield significant improvements in the rate at which reliable communication is achievable. We then formulate a joint communication and estimation problem: simultaneous communication over a one-hop noisy channel and estimation of certain channel parameters. The trade-off between the achievable rate and distortion in estimating the channel parameters is then quantified for a variety of channels. Finally, we formulate the information theoretic problem corresponding to maximizing the achievable rate while simultaneously meeting the distortion constraint for channel estimation at the destination. The capacity for this problem is evaluated and the theory applied to examples to highlight the results presented. These results are then extended from the one-hop scenario to the multiple access channel MAC) and the two-hop relay—though we only present achievable and outer-bounds for the latter network. It is also noted that these bounds coincide for particular networks and one particular class of these networks is presented. Each ideas for the theoretical results presented are bolstered with examples as appropriate. More general networks are then studied where the only objective at the destination is to minimize the end-to-end distortion of collecting channel estimates for all inter-node channels. Two particular protocol classes: amplify-and-forward and encode-and-forward are analyzed in this study. First we show that asymptotically in SNR, amplify-and-forward can outperform encode-and-forward and in fact can achieve the maximum possible distortion diversity order of unity. Second, we compare two topologies, a linear network and a tree network operating with orthogonal access, and conclude that fewer hops are beneficial to achieve better end-to-end distortion performance. We derive lower and upper bounds on distortion for both protocols and both networks, which can be used to optimize finite SNR performance. Finally, we study the problem of determining the optimal location of a mobile relay node in a sensor network to optimized some end-to-end metric. Three such metrics are considered: distortion, delay and energy. For each of these cases when we want to optimize the end-to-end metric in the estimation of a certain phenomenon, the problem is reduced to some problems which have already been solved and the results are applied to gain intuition about the results. We also study problems with multiple cost constraints—like minimizing end-to-end distortion while imposing an upper bound on the delay across the network. It is observed that the mobility of the relay node gives us additional degrees of freedom to optimize over while simultaneously making the optimization problem hard as the number of nodes in the network increases.

This research presents hardware and software solutions to many of the problems facing biological solid state nuclear magnetic resonance ssNMR) spectroscopy at high fields. The low-E 750 MHz magic angle spinning MAS) probe was designed, constructed, and thoroughly characterized. Under normal operating conditions, a proton hydrogen, isotope weight 1) RF field nutation rate of 93 kHz and homogeneity 810 degrees/90 degrees) of 93% can be obtained with a sample length of 8.4 mm corresponding to a volume of 80 uL. With a higher power amplifier, we should be able to exceed 110 kHz decoupling fields based on bench measurements. Carbon isotope weight 13) RF field nutation rates greater than 70 kHz with a homogeneity 810 degrees/90 degrees) of 70% are routinely observed for this sample length； the carbon RF homogeneity can be increased to 89% with a 6.7 mm sample length. Under full proton decoupling for long periods of time, sample heating due to the high RF field is minimal even for samples containing physiological levels of salt. We have not noticed any sample degradation in heat sensitive samples after extensive experimentation. The power handling characteristics, RF fields, and homogeneities make this an ideal probe for applying the full range of MAS solid state NMR experiments, including sequences which use extended periods of continuous RF pulsing on both channels, to biological samples which are inherently dilute. A system for optimizing pulse sequences for ssNMR was also developed, demonstrated, and is running. This system was demonstrated on the two standard pulse sequences used to test pulse optimization systems: the inversion experiment and the refocusing experiment. In both cases, pulse sequences were derived which had a wider bandwidth than existing pulse sequences and had extremely good agreement between experiment and simulation. These pulse sequences should be useful in maintaining high signal strength and phase coherence in future research. The methods of optimization and verification allow them to be easily extended to more complex situations in future research. The combination of the new probe and the method for optimizing pulse sequences for use at higher fields opens many opportunities for new research on biological solids.

In this thesis, we study model problems that capture the essence of dipolar re-coupling in magic angle spinning NMR (nuclear magnetic resonance) spectroscopy. We also address the problem of ensemble control where elements of an ensemble show variation in parameters that govern the system dynamics and make control and measurement difficult. We provide explicit methods to overcome and compensate dispersions across the ensemble using composite pulse sequences, phase modulation, and adiabatic passage. The methodologies developed are used to design novel pulse sequences that make heteronuclear and homonuclear recoupling experiments broadband and insensitive to rf inhomogeneity and anisotropic interactions. We finally incorporate these experiments into multidimensional solid-state NMR experiments that use powder dephased antiphase coherence (gamma preparation) to encode chemical shift information. This allows for the complete transfer of transverse magnetization giving improvements in sensitivity and resolution. Experimental confirmation of these results are obtained from a variety of molecular systems.

We develop statistical methods for the analysis of nonstationary time series and apply them to a variety of problems arising in speech signal processing. Information-carrying natural sound signals such as speech exhibit a degree of controlled nonstationarity in that their statistical properties vary slowly over time. Faithfully modeling these temporal variations is extremely valuable for a wide range of applications and can be accomplished by relying on well-understood acoustic models of speech production, which motivate many of the methods developed in this thesis. First, we make a number of contributions to the classical problem of formant tracking, in which vocal tract resonances are estimated under the assumption of their invariance on the 15-30 ms scale. Next, we relax this piecewise-stationarity constraint and model the temporal dynamics of the vocal tract using time-varying autoregressive TVAR) models. We develop their algebraic and geometric properties, introduce several new estimators, and use TVAR models to develop a hypothesis test to detect the presence of vocal tract variation in speech waveform data. We study its asymptotic properties, and illustrate its practical efficacy by detecting vocal tract changes across different timescales of speech dynamics. Next, we explore how standard fixed-resolution short-time Fourier representations may be generalized in order to adapt to the time-frequency structure of a speech signal. To this end, we introduce a family of adaptive, linear time-frequency representations termed superposition frames and show that they are invertible, numerically-stable, and admit fast overlap-add reconstruction akin to standard short-time Fourier techniques. The general construction proceeds via a local signal-adaptive modification of a Gabor frame. Two signal-dependent schemes for selecting an appropriate superposition frame for signal analysis are given, and the framework is illustrated in the context of speech enhancement. Finally, we introduce a joint model of the vocal tract and the source waveform in order to take into account its quasi-periodic temporal variations during voicing. We incorporate an estimate of the source waveform into the traditional linear prediction framework via nonparametric wavelet regression； the resultant semi-parametric model is applied to various speech analysis problems including formant and source-harmonics-to-noise ratio estimation, inverse filtering, and voicing detection.