The use of solar cells for energy production has indeed a bright future. Reduction of cost for fabrication along with increased efficiency are key features for a market boom, both achieved as a result of increased knowledge of the technology. Especially the thin film solar cell technology with absorbers made of Cu(In,Ga)Se2 (CIGS) is promising since it has proven high power conversion efficiency in combination with a true potential for low cost fabrication. In this thesis different recipes for fabrication of the Cu(In,Ga)Se2 absorber layer have been studied. The deposition technique used has been co-evaporation from elemental sources. For all depositions the substrate has been heated to a constant temperature of 500 ºC in order for the growing absorber to form a chalcopyrite phase, necessary for the photovoltaic functionality. The selenium has been evaporated such to always be in excess during depositions whereas the metal ratio Cu/(In+Ga) has been varied according to different recipes but always to be less than one at the end of the process. In the work emphasis has been on the radiative properties of the CIGS film during growth. The substrate heater has been temperature controlled to maintain the constant set temperature of the substrate, regardless of varying emitted power caused by changing surface emissivity. Depending on the growth conditions the emissivity of the growing film is changing, leading to a readable variation in the electrical power to the substrate heater. Since the thermal radiation from the substrate during growth has been of central focus, this has been studied in detail. For this reason the substrate has been treated as an optical stack composed of glass/Mo/Cu(In,Ga)Se2/CuxSe which determine the thermally radiated power by its emissivity. An optical model has been adopted to simulate the emissivity of the stack. In order to use the model, the optical constants for Cu(In,Ga)Se2 and CuxSe have been derived for the wavelength interval 2 μm to 20 μm. The simulation of the emissivity of the stack during CIGS growth agreed well with what has been seen for actual growth. Features of the OP-signal could hereby be explained as a result of film thickness of Cu(In,Ga)Se2 and CuxSe respectively. This is an important knowledge for an efficient fabrication in large scale.

A Mandarin speech recognition system for addresses based on MFCC, hidden Markov model (HMM) and Viterbi algorithm is proposed in this thesis. HMM is a doubly stochastic process describing the ways of pronunciation by recording the state transitions according to the time-varing properties of the speech signal. In order to simplify the system design and reduce the computational cost, the mono-syllable structure information in Mandarin is used by incorporating both mono-syllable recognizor and HMM for our system. For the speaker-dependent case, Mandarin address inputting can be accomplished within 60 seconds and 98% correct identification rate can be achieved in the laboratory environment.

A Mandarin speech recognition system for addresses in Taiwan, based on end-point detection, MFCC and HMM, is proposed and implemented in this thesis. It includes both phrase and monosyllable recognition tasks. For the phrase recognition part, we select the initial candidates before the final recognition stage to tremendously reduce the computational time. On the other side, for the monosyllable recognition part, we further refine the recognition details to improve the correct rate under easily confused circumstances. The final system can achieve 85% correct identification rate, and the address recognition can be completed within 2 seconds in the laboratory environment for speaker-dependent case.

A Mandarin phrase recognition system based on MFCC, LPC scaled excitation, vowel model, hidden Markov model (HMM) and Viterbi algorithm is proposed in this thesis. HMM, which is broadly used in speech recognition at present, is adopted in the main structure of recognition. In order to speed up the recognition time, we take advantage of stability of vowels in Mandarin and incorporate with vowel class recognition in our system. For the speaker-dependent case, a single Mandarin phrase recognition can be accomplished within 1 seconds on average in the laboratory environment.

Online process monitoring and feedback control are two widely researched aspects that can impact the performance of a myriad of process applications. Semiconductor manufacturing is one such application that due to the ever increasing demands placed on its quality and speed holds tremendous potentials for further research and development in the areas of monitoring and control. One of the key areas of semiconductor manufacturing that has received significant attention among researchers and practitioners in recent years is the online sensor based monitoring and feedback control of its nanoscale wafer fabrication process. Monitoring and feedback control strategies of nanomanufacturing processes often require a combination of monitoring using nonstationary and multiscale signals, and a robust feedback control using complex process models. It is also essential for the monitoring and feedback control strategies to possess stringent properties such as high speed of execution, low cost of operation, ease of implementation, high accuracy, and capability for online implementation. Due to the above requirement, a need is being felt to develop state-of-the-art sensor data processing algorithms that can perform far superior to those that are currently available both in the literature and commercially in the form of softwares.The contributions of this dissertation are three fold. It first focuses on the development of an efficient online scheme for process monitoring. The scheme combines the potentials of wavelet based multiresolution analysis and sequential probability ratio test to develop a very sensitive strategy to detect changes in nonstationary signals. Secondly, the dissertation presents a novel online feedback control scheme. The control problem is cast in the framework of probabilistic dynamic decision making, and the control scheme is built on the mathematical foundations of wavelet based multiresolution analysis, dynamic programming, and machine learning. Analysis of convergence of the control scheme is also presented. Finally, the monitoring and the control schemes are tested on a nanoscale manufacturing process (chemical mechanical planarization, CMP) used in silicon wafer fabrication. The results obtained from experimental data clearly indicate that the approaches developed outperform the existing approaches. The novelty of the research in this dissertation stems from the fact that they further the science of sensor based process monitoring and control by uniting sophisticated concepts from signal processing, statistics, stochastic processes, and artificial intelligence, and yet remain versatile to many real world process applications.

Submitted by Maike Costa (maiksebas@gmail.com) on 2017-06-21T14:26:33Z No. of bitstreams: 1 arquivototal.pdf: 3690977 bytes, checksum: 752560aa5c7d78968c32cb55f0778788 (MD5) / Made available in DSpace on 2017-06-21T14:26:33Z (GMT). No. of bitstreams: 1 arquivototal.pdf: 3690977 bytes, checksum: 752560aa5c7d78968c32cb55f0778788 (MD5) Previous issue date: 2016-02-29 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Total hardness and Total alkalinity are important physico-chemical parameters for the evaluation of water quality and are determined by volumetric analytical methods. These methods have difficult to detect the endpoint of the titration due to the difficult of viewing the color transition inherent to each of them. To circumvent this problem, here is proposed a new automatic method for the detection of the titration end point for the determination of total hardness and total alkalinity in mineral water samples. The proposed flow-batch titrator consists of a peristaltic pump, five three-way solenoid valves, a magnetic stirrer, an electronic actuator, an Arduino MEGA 2560TM board, a mixing chamber and a webcam. The webcam records the digital movie (DM) during the addition of the titrant towards mixing chamber, also recording the color variations resulting from chemical reactions between titrant and sample within chamber. While the DM is recorded, it is decompiled into frames ordered sequentially at a constant rate of 30 frames per second (FPS). The first frame is used as a reference to define the region of interest (RI) of 48 × 50 pixels and the R channel values, which are used to calculate the Pearson's correlation coefficient (r) values. r is calculated between the R values of the initial frame and each subsequent frame. The titration curves are plotted in real time using the values of r (ordinate axis) and the total opening time of the valve titrant (abscissa axis). The end point is estimated by the second derivative method. A software written in ActionScript 3.0 language manages all analytical steps and data treatment in real time. The feasibility of the method was attested by its application for the analysis of natural water samples. Results were compared with classical titration and did not present statistically significant differences when the paired ttest at the 95% confidence level was applied. The proposed method is able to process about 71 samples per hour, and its precision was confirmed by overall relative standard deviation (RSD) values, always lower than the 2,4% for total hardness and 1,4% for total alkalinity. / A dureza total e a alcalinidade total são importantes parâmetros físico-químicos para avaliação da qualidade de águas e são determinados por métodos volumétricos de análise. Estes métodos apresentam difícil detecção do ponto final da titulação devido à dificuldade de visualização das transições de cores inerentes a cada um deles. Para contornar este problema, foi proposta neste trabalho uma nova metodologia automática para a detecção do ponto final nas determinações de dureza total e alcalinidade total em águas. O titulador em fluxo-batelada proposto é composto de uma bomba peristáltica, cinco válvulas solenoides de três vias, um agitador magnético, um acionador de válvulas, uma placa Arduíno MEGA 2560TM, uma câmara de mistura e uma webcam. O programa de gerenciamento e controle do titulador foi escrito em linguagem ActionScript 3.0. A webcam grava o filme digital durante a adição do titulante na câmara de mistura, registrando as variações de cor decorrentes das reações químicas entre titulante e amostra no interior de câmara. Enquanto o filme é gravado, este é decomposto em quadros ordenados sequencialmente a uma taxa constante de 30 quadros por segundo (FPS). O primeiro quadro é utilizado como referência para definir uma região de interesse (RI) com 48 x 50 pixels, na qual seus valores R, G e B são utilizados para calcular os valores de coeficiente de correlação de Pearson (r). O valor de r é calculado entre os valores de R do quadro inicial e de cada quadro subsequente. As curvas de titulação são obtidas em tempo real usando os valores de r (ordenadas) e o tempo total de abertura da válvula de titulante (abscissas). O ponto final é estimado pelo método de segunda derivada. O método foi aplicado na análise de águas minerais e os resultados foram comparados com a titulação clássica, não apresentando diferenças estatisticamente significativas com aplicação do teste t pareado a 95% de confiança. O método proposto foi capaz de processar até 71 amostras por hora e a sua precisão foi confirmada pelos valores de desvio padrão relativos (DPR) globais, sempre inferiores as 2,4% para dureza total e 1,4% para alcalinidade total.

Patch antennas are fundamental elements in many microwave communications systems. However, patch antennas receive/transmit signals over a very narrow bandwidth (typically a maximum of 3% bandwidth). Design modifications directed toward bandwidth expansion generally yield 10% to 40% bandwidth. The series varactor tuned patch antenna configuration was the bandwidth enhancement method explored in this research; this configuration is implemented by dividing a patch antenna into multiple sections and placing varactors across the resultant gaps. In addition to yielding a large bandwidth, the configuration has a number of ancillary benefits, including straightforward integration and design flexibility. Through the research represented by this work, the properties of the series varactor tuned patch antenna, herein referred to as the Fragmented Patch Antenna (or FPA), were explored and optimized. As a result, an innovative patch antenna was produced that yielded 63.4% frequency tuning bandwidth and covered a frequency range between 2.8 and 5.4 GHz. The wide bandwidth was achieved through a detailed parametric study. The products of this study were the discovery of multiple tuning resonances that were used to expand the tuning bandwidth and the understanding/documentation of the significance of specific antenna dimensions. Measurement results were obtained through the fabrication of a prototype antenna using semiconductor varactors. In the second research phase, the construction of capacitors using the tunable permittivity material Barium Strontium Titanate (BST) was investigated. Using this material in conjunction with nano-fabrication techniques, varactors were developed that had good estimated performance characteristics and were considered appropriate for integration into adaptive microwave circuitry, such as the tunable antenna system. The varactors were constructed by using Focused Ion Beam (FIB) milling to create a nano-scale capacitive gap in a transmission line. A combination of end-point current detection (EPD) and cross-section scanning electron (SEM) and ion beam (FIB) microscope images were used to optimize the milling procedure. The future extensions of this work include the integration of the BST varactors with the antenna design; the configuration of the developed BST varactors lends itself to a straightforward integration with the FPA antenna.