Search

Introduction: There is evidence that sample treatment of blood-based biosamples may affect integral signals in nuclear magnetic resonance-based metabolomics. The presence of macromolecules in plasma/serum samples makes investigating low-molecular-weight metabolites challenging. It is particularly relevant in the targeted approach, in which absolute concentrations of selected metabolites are often quantified based on the area of integral signals. Since there are a few treatments of plasma/serum samples for quantitative analysis without a universally accepted method, this topic remains of interest for future research. Methods: In this work, targeted metabolomic profiling of 43 metabolites was performed on pooled plasma to compare four methodologies consisting of Carr-Purcell-Meiboom-Gill (CPMG) editing, ultrafiltration, protein precipitation with methanol, and glycerophospholipid solid-phase extraction (g-SPE) for phospholipid removal; prior to NMR metabolomics analysis. The effect of the sample treatments on the metabolite concentrations was evaluated using a permutation test of multiclass and pairwise Fisher scores. Results: Results showed that methanol precipitation and ultrafiltration had a higher number of metabolites with coefficient of variation (CV) values above 20%. G-SPE and CPMG editing demonstrated better precision for most of the metabolites analyzed. However, differential quantification performance between procedures were metabolite-dependent. For example, pairwise comparisons showed that methanol precipitation and CPMG editing were suitable for quantifying citrate, while g-SPE showed better results for 2-hydroxybutyrate and tryptophan. Discussion: There are alterations in the absolute concentration of various metabolites that are dependent on the procedure. Considering these alterations is essential before proceeding with the quantification of treatment-sensitive metabolites in biological samples for improving biomarker discovery and biological interpretations. The study demonstrated that g-SPE and CPMG editing are effective methods for removing proteins and phospholipids from plasma samples for quantitative NMR analysis of metabolites. However, careful consideration should be given to the specific metabolites of interest and their susceptibility to the sample treatment procedures. These findings contribute to the development of optimized sample preparation protocols for metabolomics studies using NMR spectroscopy

Conventionally, odours emitted by different sources present in wastewater treatment plants (WWTPs) are measured by dynamic olfactometry, where a human panel sniffs and analyzes air bags collected from the plant. Although the method is considered the gold standard, the process is costly, slow, and infrequent, which does not allow operators to quickly identify and respond to problems. To better monitor and map WWTP odour emissions, here we propose a small rotary-wing drone equipped with a lightweight (1.3-kg) electronic nose. The "sniffing drone" sucks in air via a ten-meter (33-foot) tube and delivers it to a sensor chamber where it is analyzed in real-time by an array of 21 gas sensors. From the sensor signals, machine learning (ML) algorithms predict the odour concentration that a human panel using the EN13725 methodology would report. To calibrate and validate the predictive models, the drone also carries a remotely controlled sampling device (compliant with EN13725:2022) to collect sample air in bags for post-flight dynamic olfactometry. The feasibility of the proposed system is assessed in a WWTP in Spain through several measurement campaigns covering diverse operating regimes of the plant and meteorological conditions. We demonstrate that training the ML algorithms with dynamic (transient) sensor signals measured in flight conditions leads to better performance than the traditional approach of using steady-state signals measured in the lab via controlled exposures to odour bags. The comparison of the electronic nose predictions with dynamic olfactometry measurements indicates a negligible bias between the two measurement techniques and 95 % limits of agreement within a factor of four. This apparently large disagreement, partly caused by the high uncertainty of olfactometric measurements (typically a factor of two), is more than offset by the immediacy of the predictions and the practical advantages of using a drone-based system.

Metal oxide (MOX) sensors are widely used for chemical sensing due to their low cost, miniaturization, low power consumption and durability. Yet, getting instantaneous measurements of fluctuating gas concentration in turbulent plumes is not possible due to their slow response time. In this paper, we show that the slow response of MOX sensors can be compensated by deconvolution, provided that an invertible, parametrized, sensor model is available. We consider a nonlinear, first-order dynamic model that is mathematically tractable for MOX identification and deconvolution. By transforming the sensor signal in the log-domain, the system becomes linear in the parameters and these can be estimated by the least-squares techniques. Moreover, we use the MOX diversity in a sensor array to avoid training with a supervised signal. The information provided by two (or more) sensors, exposed to the same flow but responding with different dynamics, is exploited to recover the ground truth signal (gas input). This approach is known as blind deconvolution. We demonstrate its efficiency on MOX sensors recorded in turbulent plumes. The reconstructed signal is similar to the one obtained with a fast photo-ionization detector (PID). The technique is thus relevant to track a fast-changing gas concentration with MOX sensors, resulting in a compensated response time comparable to that of a PID.

MEET (Motif Elements Estimation Toolkit) es un paquete en R que integra un conjunto de algoritmos para la detección computacional de los puntos de unión de los factores de transcripción (TFBS). El paquete en R MEET incluye cinco programas de búsqueda de motivos: MEME/MAST (Multiple Expectation-Maximization for Motif Elicitation), Q-residuals, MDscan (Motif Discovery scan), ITEME (Information Theory Elements for Motif Estimation) y Match. Además, permite al usuario trabajar con diferentes algoritmos de alineamiento múltiple: MUSCLE (Multiple Sequence Comparison by Log-Expectation), ClustalW y MEME. El paquete puede trabajar en dos modos diferentes, entrenamiento y detección. El modo entrenamiento permite escoger los parámetros óptimos del detector escogido. Y el modo detección permite, una vez escogidos los parámetros, analizar un genoma en busca de puntos de unión. Además, ambos modos pueden combinar los diferentes métodos de alineamiento y de detección, permitiendo al usuario un amplio abanico de posibilidades. Esta característica permite comparar los diferentes métodos computacionales al mismo nivel,sin realizar ningún agravio comparativo debido al alineamiento.

Dentro del amplísimo campo de los sensores, están emergiendo con gran fuerza los dispositivos basados en tecnología de silicio. El uso de la tecnología microelectrónica para la producción de sensores permite obtener dispositivos de bajo coste, altas prestaciones y reducido tamaño.Además, la utilización del silicio como material base para la realización de sensores permite, no sin dificultades, la integración de circuitos de tratamiento de señal en el mismo chip. El silicio, por otro lado, es un material de excelentes propiedades mecánicas exhibiendo una gran variedad de mecanismos de transducción que permiten desarrollar una gran variedad de sensores: mecánicos, térmicos, magnéticos, químicos, etc. En particular la medida de magnitudes mecánicas es posible debido a variación de la resistividad del silicio cuando se encuentra sometido a esfuerzos mecánicos. Esta variación se conoce como el efecto piezorresistivo y fue descubierto por Smith en 1954.Las tecnologías para el micromecanizado de silicio monocristalino permiten la fabricación de microestructuras tales como ranuras, membranas, agujeros, barras de torsión o vigas. En concreto, los sensores de presión se basan en la fabricación de un diafragma delgado de silicio. En presencia de una presión diferencial, el diafragma deflecta dando lugar a la aparición de esfuerzos mecánicos en el material. Integrando un puente de Wheatstone en el diafragma y colocando las piezorresistencias de forma adecuada se obtiene un voltaje proporcional a la presión aplicada. Este diafragma se obtiene mediante el grabado anisótropo húmedo de silicio utilizando disoluciones de carácter básico.Con el objetivo de fabricar sensores de presión piezorresistivos en tecnología de silicio, se procedió a un exhaustivo estudio experimental del grabado anisótropo húmedo mediante una disolución de KOH, 2-propanol y H2O. El análisis de los resultados ha permitido escoger aquellas condiciones de trabajo óptimas que aseguran unas rugosidades residuales inferiores a 0.1 milimicra y una aceptable velocidad de grabado. A continuación se han fabricado diafragmas delgados de silicio mediante el control cronométrico del ataque. Para una medida no destructiva del grosor de los diafragmas se ha utilizado las interferencias presentes en un espectro de absorción infrarroja. Se ha diseñado un proceso tecnológico utilizando la tecnología microelectrónica estándar y el ataque anisótropo previamente estudiado para la fabricación de sensores de presión.El correcto funcionamiento de estos ha sido verificado y se han extraído los parámetros que definen su comportamiento.Para un mejor control dimensional del grosor del diafragma hemos estudiado el paro electroquímico en una unión p-n. Se ha comparado los resultados obtenidos en configuraciones con 2, 3 y 4 electrodos, estudiando además la influencia de la luz. La puesta a punto de esta técnica ha permitido la fabricación de membranas mediante capas epitaxiales y difusiones con una gran uniformidad y repetitividad, así como la fabricación de puentes de silicio monocristalino.Se ha hecho especial hincapié en la posibilidad de obtener membranas estructuradas verticalmente. Los mejores resultados se han obtenido en las configuraciones de 4 electrodos sin luz y en la de 3 electrodos bajo iluminación.La utilización de galgas de silicio implantado limita el rango de operación del sensor a temperaturas inferiores a 125°C. Temperaturas mayores pueden alcanzarse utilizando galgas de polisilicio aisladas del substrato por una capa de óxido de silicio. Sin embargo, la optimización de capas de polisilicio LPCVD en sensores lleva a la utilización de condiciones de depósito no estándar. Se ha procedido a una completa caracterización estructural y eléctrica de capas de polisilicio dopadas con boro obtenidas a diferentes temperaturas y presiones. Se ha observado como las muestras depositadas entre 580°C y 600°C, a bajas presiones de silano presentan una fuerte textura {110} y estructura columnar. Estas muestras presentan una resistividad más elevada que las depositadas en otras condiciones. En cambio, muestras depositadas en fase amorfa presentan tras un recocido posterior una estructura formada por granos equiaxiados y de mayor tamaño, que darán lugar a unos mayores coeficientes piezorresistivos. Por otro lado, el análisis eléctrico ha mostrado que es posible controlar el coeficiente térmico de las resistencias a través del nivel de dopaje. Utilizando las polisilicio optimizado tras el estudio estructural y eléctrico, se han fabricado sensores de presión con galgas de polisilicio que manteniendo niveles de sensibilidad altos han mostrado un correcto funcionamiento hasta temperaturas de 200°C con gran estabilidad de sus características.En la fabricación de sensores de presión de alta sensibilidad la disminución progresiva del grosor del diafragma da lugar a niveles de no-linealidad inaceptables. Este problema se solventa mediante el diseño de diafragmas que introducen zonas rígidas en el interior del diafragma que habitualmente tienen el grosor total de la oblea. Sin embargo, si las dimensiones del sensor se reducen la inclusión de zonas rígidas con ese grosor queda prohibida por la dimensión lateral de los muros {111}. Se ha analizado el comportamiento mecánico de diafragmas que incluyen zonas rígidas de grosor reducido. Se ha observado que una relación de grosores 5/20 es suficiente para una eficaz reducción de la no-linealidad manteniendo niveles de sensibilidad altos. Mediante un método gráfico basado en un diagrama no-linealidad fondo de escala, las prestaciones de cada estructura han sido comparadas con el diseño estándar basado en una membrana cuadrada de grosor uniforme.La medida invasiva de la presión sanguínea en el diagnóstico de enfermedades cardiovasculares impone severas restricciones en el diseño de sensores de presión. Su tamaño, sensibilidad y nolinealidad son parámetros críticos que deben ser optimizados en relación a un sensor convencional. Se ha desarrollado un sensor de presión de reducidas dimensiones para incorporarse en el extremo de un catéter 7F (diámetro exterior 2.3 mm.) Para obtener sensores de alta sensibilidad minimizando el error debido a la no-linealidad se han fabricado sensores de presión con diafragma no-uniforme utilizando la tecnología del paro electroquímico. La caracterización de los sensores fabricados ha demostrado que el diseño y la tecnología utilizada permite mejorar la respuesta de transductores caracterizados por diafragmas delgados y con área reducida, reduciendo la no-linealidad exhibida por los dispositivos., Silicon integrated circuit technology may be used in the fabrication of miniaturised, low-cost, high-performance mechanical sensors. Pressure sensors are based on a thin silicon diaphragm, fabricated using anisotropic etching, what deflects under a pressure between both sides. The integration of a Wheatstone bridge permits to obtain an analog output proportional to the pressure difference because of the piezoresistive effect.Implanted resistors limit the sensor operation range below 125°C, because of the leakage currents in the p-n junctions. Higher temperatures are allowed if polysilicon resistors, isolated from the substrate by a silicon dioxide layer, are used instead. However, the optimization of the LPCVD polysilicon layers for this particular application leads to non-standard deposition conditions and it is essential a previous structural and electrica1 characterization work. The fabricated sensors are very stable for temperatures up to 200°C.The invasive blood pressure measurement for cardiovascular diagnostics imposes strong requirements in the sensor design. Sensibility, size and non-linearity are critical parameters that have to be optimized. For high sensitivity designs, very thin diaphragms are used because of the reduced diaphragm area. They require an accurate control of the diaphragm thickness in order to attain good uniformity in the sensor characteristics. This is achieved using the electrochemical etch-stop, which has been thoroughly studied. Especial attention has been devoted to the fabrication of non-uniform diaphragms. The manufacturing of high sensitive devices just reduces the utilization of the electrochemical etch-stop for the inclusion of rigid zones in the diaphragm.In order to study whether this improves the sensor response an extensive analysis of different mechanical structures by the finite element method has been carried out. Finally, the devices have been fabricated and the improvement of the characteristics comparing with the standard flat diaphragm design has been demonstrated. An specially designed package has been designed and manufactured in order to mount the pressure sensor in the catheter tip.

In this article, we analyze the ability of the early olfactory system to detect and discriminate different odors by means of information theory measurements applied to olfactory bulb activity images. We have studied the role that the diversity and number of receptor neuron types play in encoding chemical information. Our results show that the olfactory receptors of the biological system are low correlated and present good coverage of the input space. The coding capacity of ensembles of olfactory receptors with the same receptive range is maximized when the receptors cover half of the odor input space - a configuration that corresponds to receptors that are not particularly selective. However, the ensemble's performance slightly increases when mixing uncorrelated receptors of different receptive ranges. Our results confirm that the low correlation between sensors could be more significant than the sensor selectivity for general purpose chemo-sensory systems, whether these are biological or biomimetic.

Low-cost tin oxide gas sensors are inherently nonspecific. In addition, they have several undesirable characteristics such as slow response, nonlinearities, and long-term drifts. This paper shows that the combination of a gas-sensor array together with self-organizing maps (SOM's) permit success in gas classification problems. The system is able to determine the gas present in an atmosphere with error rates lower than 3%. Correction of the sensor's drift with an adaptive SOM has also been investigated

Treballs Finals de Grau d'Enginyeria Biomèdica. Facultat de Medicina i Ciències de la Salut. Universitat de Barcelona. Curs: 2021-2022. Director: Albert Compte. Tutor: Santiago Marco., Neuroscience is a field in constant evolution, searching for better ways to understand human brains and behaviour. Challenged by the complexity and difficult accessibility of the brain, neuroscience must constantly be at the edge of scientific and technical developments, learning new ways to obtain and understand neural data. The approach of this project departs from one such novel findings, human single neuron signal recordings, an emerging technology with a great many applications in research fields but currently limited by the lack of extensive data and previous research. Invasive cell-level recording techniques have been extensively used since its inception in animal brain studies, where they have contributed to wide-ranging research and count with consistent and standardized procedures for its application and analysis. On the other hand, its potential use in theoretical human research is still underdeveloped and does not count with a comprehensive and systematic data analysis framework. However, human single neuron data has the potential to path the way for state-of-the-art research focusing on the behaviour of deep neural processes, such as learning and cognition. The main objective of the project is to adapt pre-existing single-neuron data-analysis software for animal studies to human recordings, creating a continuous dataflow from the obtention of the data to its ultimate analysis.

Treballs Finals de Grau d'Enginyeria Biomèdica. Facultat de Medicina i Ciències de la Salut. Universitat de Barcelona. Curs: 2020-2021. Director: Albert Compte. Tutor: Santiago Marco., EEG source localization is a non-invasive imaging technique developed to locate the anatomical sources recorded at the scalp during an EEG recording. The reconstruction of the sources can be computed by solving the so-called inverse problem. This is an ill-posed problem which aims in estimating the sources that fit the recorded measurements. There exist several powerful commercial and academic software packages that cover multiple methods on data processing, source localization, and statistical analysis. In this work, the open-source MNE-Python package was selected as the working environment used to address the challenge of characterizing and locating neural activation. This study provides a pipeline with practical steps on the EEG source localization technique. The results obtained in this project have been validated by experts in the Theoretical Neurobiology and Computational Neuroscience fields In this project, the EEG source localization has been computed over a group of encephalitic patients and a control group. The two groups had shown differences regarding the electrical activity in a working memory trial. This study aimed in localizing the anatomical brain regions that were responsible of the electrical differences. It has been observed that instants before the stimulus, the activated sites between control groups and encephalitic groups differ. In the case of the control group, the activated region was located at the frontal lobe of the left hemisphere. Whereas, in the case of the encephalitic group the activated region was located at the temporal lobe of the right hemisphere.

Treballs Finals de Grau d'Enginyeria Biomèdica. Facultat de Medicina i Ciències de la Salut. Universitat de Barcelona. Curs: 2020-2021. Directors: Rachel Lyne and Gos Micklem i Tutor: Santiago Marco, [spa] InterMine és un sistema del tipus magatzem de dades que permet crear grans bases de dades biològiques fàcilment des de fonts heterogènies en serveis web RESTful. El sistema proporciona eines estadístiques i d’anàlisis de dades potents. InterMine com a plataforma té una interfície web des d’on l’usuari pot fer cerques i incorpora recursos especialitzats com mètodes d’anàlisis d’enriquiment de conjunts de gens. InterMineR és una de les biblioteques informàtiques que dóna suport a la plataforma des de l’entorn de programació de R Studio. En aquest projecte, s’han implementat noves funcionalitats per permetre que l’usuari treballi amb col·leccions i conjunts de dades creats des de el servei web, anomenats llistes. Les noves funcions han estat incorporades i publicades en la nova versió del paquet a Bioconductor. L’objectiu principal d’aquest projecte ha estat la creació d’una interfície gràfica d’usuari (GUI) per fer cerques a les bases de dades d’InterMine i generar visualitzacions de xarxes Cytoscape. S’ha desenvolupat un programari executable en entorns d’escriptori. Aquest permet als usuaris reduir la complexitat de les dades i extreure’n significat amb fins de recerca. S’espera que permeti analitzar amb més profunditat les fonts d’InterMine i augmentar-ne la FAIR (dades trobables, accessibles, interoperables i reutilitzables en anglès)., [eng] InterMine is a biological data warehousing system for creating large biological databases of heterogenous data sources easily in RESTful web services. It provides powerful statistical and analysis tools. InterMine as a platform has a web interface in which user can run searches and incorporates specialized resources such as enrichment statistics. InterMineR is one of the software ’client’ libraries that interfaces the biological databases built using the InterMine platform with a programming environment, in this case R. In this project, new functionalities have been implemented to enable the user to work with stored collections of data and saved results sets created from a web-service, called lists. The new functions have been merged and published in the new Bioconductor version of the package. The main aim of this project is to create a graphical user interface (GUI) to query the InterMine databases and generating Cytoscape network visualizations. We developed a software that runs on desktop environment. It allows users to reduce the complexity of the data and extract meaning from it. It will allow data in InterMines to be further analysed - increasing its FAIRness.

[eng] In some types of fire, namely, smoldering fires or involving polymers without flame, gases and volatiles appear before smoke is released. Most of the fatalities registered for fires, are caused due to the intoxication of the building occupants over the burns. Nowadays, conventional fire detectors are based on the detection of smoke or airborne particles. In smoldering fires situations, conventional fire detectors triggers the alarm after the release of toxic emissions. The early emission of gas in fires opens the possibility to build fire alarm systems with shorter response times than widespread smoke-based detectors. Actually, the sensitivity of gas sensors to combustion products has been proved for many years. However, already early works remarked the challenge of providing reliable fire detection using chemical sensors. As gas sensors are not specific, they can be calibrated to detect large variety of fire signatures. But, at the same time, they are also potentially sensitive to any activity that releases volatiles when being performed. Cross-sensitivity to water vapor and other chemical compounds make gas-based fire alarm systems prone to false positives. For that reason, the development of reliable and robust fire detectors based on gas sensors relies in pattern recognition and Machine Learning algorithms to discriminate fire from nuisance sensor signatures. The presented PhD. Thesis explore the role of pattern recognition algorithms for fire detection using detectors based exclusively in chemical sensors. Two prototypes based on different types of gas sensors were designed. The sensor selection was performed to be sensitive to combustion products and to capture other volatiles that may help to discriminate fire and nuisances. Machine Learning algorithms for the prediction of fire were trained using standard fire tests stablished in EU norm 54. Additionally to those test experiments that may induce false alarms were also performed. Two approaches of machine learning algorithms were explore. The first prediction algorithms is based on Partial Least Squares Discriminant Analysis and the second set of algorithms are based on Support Vector Machines. Additionally, two new methodologies for cost reduction are presented. The first methodology build fire detection algorithms using the combination of Standard fire test and a reduced version of those experiments. The reduced version were performed in a small chamber. The smaller setup allows the performance of experiments in a shorter period of time. In consequence, the number of experiments to test the models increase and also the robustness of the prediction algorithms. The second methodology built general calibration models using replicates of the same sensor array. The use of different units rejects the variance between sensor arrays and allows the construction of general calibration models. The use of a single model to calibrate sensor arrays systems allows the mass production and resulting in the reduction of costs production., [cat] Les alarmes convencionals d'incendis es basen en la detecció de fums. Tanmateix, els incendis solen emetre molts volàtils abans d'emetre fum. Altres grups de recerca ja han proposat sistemes detectors d'incendis basats en sensors químics, que poden proporcionar una resposta més ràpida, però segueixen sent propensos a falses alarmes davant d'interferències. Les tècniques de reconeixement de patrons poden ser útils per mitigar aquesta limitació. En aquesta tesi, es desenvolupen dos detectors d’incendis basats exclusivament en sensors de gas, de diverses tecnologies, que proporcionen una alarma d’incendi basada en algorismes d’aprenentatge automàtic. Els detectors van ser exposats a incendis estandarditzats i a diverses interferències. La tesi presenta dos enfocaments diferents pel reconeixement de patrons: el primer es basa en una anàlisi discriminant de mínims quadrats parcials, PLS-DA, i el segon es basa en una màquina de vectors de suport, SVM. Els resultats confirmen la capacitat de detectar incendis a una fase inicial del seu desenvolupament i el rebuig de la majoria de les interferències. A més, es presenten dues metodologies per a la reducció dels costos de calibratge d'agrupacions de sensors de gas per la detecció d'incendis, tenint present que els experiments per avaluar els detectors es fan en una sala d'incendis estàndard i són molt llargs i costosos. La primera metodologia proposada combina dades procedents d'una sala d'incendis estàndard i dades d'experiments fets a petita escala, més ràpids i menys costosos. Els resultats mostren que el rendiment dels models de predicció pot millorar amb la fusió de dades. La segona metodologia de reducció de costos compensa la necessitat de models de calibratge individuals per a cada matriu de sensors (a causa de la variabilitat del sensor) rebutjant la variabilitat del sensor i proporcionant models generals de calibratge.

[eng] Chemical source localization (CSL) by autonomous robots has been a topic of research since the early 1990s and still today remains elusive beyond simple scenarios. It has numerous potential applications, such as the localization of toxic emissions, malodors, gas leaks and hazardous substances in general, without risking human lives. An intuitive CSL approach is to mimic the known chemo-orientation behaviour of some flying insects, such as moths and mosquitos, which effectively use odor plumes for mating and foraging. However, terrestrial robots are too slow to perform insect-like movements and the response time and limit of detection (LOD) of current odor sensors for key compounds of biological relevance for plume navigation is orders of magnitude higher than in biological chemoreceptors. Instead of using a slow terrestrial robot equipped with complex instrumentation, in this thesis we address the CSL problem with a nano-drone, i.e. a miniaturized aerial robot, equipped with a simple metal oxide semiconductor (MOX) sensor. Improving key specifications of MOX sensors for this application is one of the core parts of this thesis. Specifically, we introduce novel signal processing methods for estimating and optimizing the LOD, reducing the power consumption and improving the response time. The LOD is a key figure of merit of an odor sensor but difficult to estimate in non-linear devices (e.g. MOX sensors) or sensory systems with multivariate output (e.g. sensor arrays or temperature-modulated MOX sensors). We propose a univariate LOD optimization method based on linearized calibration models and a multivariate approach based on orthogonal partial least squares (O-PLS). To improve the response time, we use high-frequency features extracted from the MOX signal derivative, known as ’bouts’ and optimize the algorithm for changing wind conditions and real-time operation. A novel setup consisting on a 3D grid of MOX sensors is proposed for 3D feature selection and real-time visualization of the gas distribution. Two map-based CSL strategies, one based on the instantaneous response and the other one based on bouts, are finally evaluated using the nano-drone in experiments performed in a large indoor environment (160 m2) where a chemical source is placed in challenging positions. The experimental results demonstrate that the proposed platform can be used to quickly (, [spa] Esta tesis representa un paso adelante en el uso de sensores semiconductores de óxido de estaño (MOX) en diversas aplicaciones, desde cuantificación de bajas concentraciones de monóxido de carbono en presencia de interferencias químicas en un entorno controlado; hasta medidas dinámicas con robots para localización de fuentes químicas en entornos no estructurados. El objetivo principal de la tesis es desarrollar un nano-drone que sea capaz de localizar una fuente química en entornos “indoor” de grandes dimensiones. Los sensores MOX son la opción preferible para nano-robots debido a su reducido tamaño y bajo peso, pero también presentan importantes limitaciones como la no- linealidad de la respuesta, baja selectividad, bajo ancho de banda y alto consumo. Uno de los objetivos específicos de esta tesis es mitigar estas limitaciones mediante técnicas de procesado de señal. En primer lugar, desarrollamos métodos estadísticos univariantes y multivariantes para calcular el límite de detección (LOD) en sensores no lineales. Pese a que el LOD es una figura de mérito fundamental, los sensores MOX presentan características incompatibles con las fórmulas para el cálculo del LOD definidas por la International Union of Pure and Applied Chemistry (IUPAC). El elevado consumo de los sensores MOX reduciría el tiempo de vuelo de un nano- drone considerablemente. Para alcanzar un compromiso entre consumo y estabilidad, en esta tesis se exploran métodos de alimentación de bajo consumo basados en operación bajo demanda y operación “duty-cycling”. El bajo ancho de banda de los sensores MOX dificulta la localización de fuentes químicas en entornos reales de grandes dimensiones donde las sustancias químicas de interés se propagan turbulentamente. En esta tesis se proponen mejoras del tiempo de respuesta basadas en la fabricación de micro-cámaras de gas y extracción de características dinámicas de la respuesta. Para el desarrollo de algoritmos de localización de fuentes químicas en entornos tridimensionales (3D), es necesario estudiar la dispersión de gases en 3D. Para ello, en esta tesis desarrollamos por primera vez una red 3D de sensores MOX para analizar patrones de dispersión de gas producidos por una fuga de gas. Finalmente, los algoritmos desarrollados son evaluados en un nano-drone equipado con un sensor MOX encargado de localizar una fuente química en una sala de grandes dimensiones (160 m2).

[cat] Una de les principals causes d’accidents d’automòbil és l’estat inadequat del conductor. Diversos estudis situen entre el 16% i el 30% el nombre d'accidents de trànsit relacionats amb el cansament, fatiga, somnolència i estats similars. Per aquest motiu una de les prioritats actuals de grups de recerca i empreses automotrius és el desenvolupament de sistemes capaços de detectar somnolència i/o estats inadequats en conductors. Els diversos mètodes existents per detectar somnolència poden agrupar-se en quatre grans grups: subjectius, fisiològics, visuals i de conducció. Els primers, basats en avaluacions subjectives, presenten una baixa fiabilitat si el conductor és el seu propi avaluador i el seu ús en vehicle no és viable si l’avaluador és extern. Els sistemes basats en informació fisiològica tenen una major fiabilitat però requereixen l’ús de dispositius específics per part del conductor (casc amb elèctrodes, sensors d’activitat cardíaca, banda respiratòria, etc.). Aquest fet és molest pels usuaris i, mentre no es desenvolupi un sensor sense contacte fiable, limita la utilització d’aquests sistemes a conductors professionals d’alt risc (conductors de llargues distàncies, de mercaderies perilloses, etc.). El tercer grup, en canvi, és totalment transparent a l’usuari i analitza, a partir del processat d’imatges, l’atenció del conductor en funció del seu comportament ocular. En condicions de bona il·luminació presenten molts bons resultats, mentre que les il·luminacions irregulars els són problemàtiques, així com l’ús d’ulleres de sol. Finalment, el quart grup de detectors està format per aquells que es basen en l’anàlisi de la conducció del vehicle, els quals són també transparents per l’usuari. La majoria d’aquests detectors estan desenvolupats en simuladors, sense tenir en compte la complexitat d’un entorn real. En aquesta tesi és presentat un sistema per a vehicles reals que estima l’estat del conductor a partir d’una avaluació de la qualitat de la conducció i de la seva interacció amb el vehicle. Partim de la hipòtesi que aquesta qualitat es degrada amb la fatiga i la somnolència. El sistema estudia els senyals disponibles a la xarxa de comunicacions del vehicle (bus CAN) i, opcionalment, les dades proporcionades per una càmera frontal amb un programari específic per la detecció de línies de carril. Per validar la nostra hipòtesi inicial s’han realitzat assajos en simulador amb subjectes privats del son. Les dades obtingudes han estat analitzades i en elles s’han identificat les característiques de la conducció més significatives en somnolència. A continuació s’han dut a terme diferents assajos en vehicle real, en els que s’ha pogut observar un important increment de la complexitat de l’entorn. Per això, en el desenvolupament del detector hem hagut de centrar gran part dels esforços en evitar les falses alarmes. A part dels tres grans mòduls que generen penalitzacions (basats en la posició en el carril, els moviments de volant i les correccions brusques), es presenten a la tesi diversos mòduls centrats en identificar determinades situacions i condicions en la conducció a partir dels que es modula dinàmicament la sensibilitat del sistema. Els resultats obtinguts mostren la possibilitat de detectar estats no aptes mitjançant el control dinàmic del vehicle, sempre i quant aquests es vegin reflectits en la conducció. El sistema final pot treballar exclusivament amb les dades del CAN, tot i que l’adició d’una càmera frontal incrementa la qualitat de la informació d’entrada del sistema. Està optimitzat per treballar en microcontroladors de vehicle, complint amb restriccions temporals i computacionals així com satisfent les principals normatives automotrius., [eng] A main cause of vehicle accidents is the driver's condition. Several studies estimate that drowsiness and similar states cause around 16-30% of all traffic accidents. The existing drowsiness detection methods can be divided into four main groups: subjective, physiological, visual and driving behavioral. The first and second methods are not feasible for non professional divers due to their low reliability and/or need of devices attached to the driver. The third method can be already found in vehicles, but has the drawback of depending highly on lighting conditions. Finally, the fourth methods, based on the driving behavior analysis, have the advantage of being transparent and non-invasive to the user and also not depending highly on external conditions. Most of these detectors are developed in simulators, without considering the real environment's complexity. This thesis presents a system for real vehicles to estimate the driver's state based on an evaluation of his/her driving quality and interaction with the vehicle. We assume that this quality deteriorates with fatigue and drowsiness. The system analyses data available on the vehicle's bus and optionally data provided by a lane detection frontal camera. To validate our initial hypothesis simulator tests have been done with sleep-deprived drivers. The obtained data were analyzed to obtain the main drowsy driving characteristics. Afterwards, different real vehicle tests were carried out, where a significant increase of the environment's complexity appeared. Therefore, our development had a focus on false alarms prevention. Besides the major modules that generate penalties (based on the vehicle's position, steering movements and abrupt corrections), we presented other modules focused on identifying specific situations and conditions that modulate dynamically the system's sensitivity. The results show the feasibility of detecting inattentive states using the vehicle's dynamic control. The system is optimized to work in a vehicle's microcontroller, fulfilling automotive's standards and constraints.

[eng] Electronic noses (e-noses) have been utilized during the past three decades as general purpose instruments for chemical sensing. These instruments are inspired by natural olfactory systems, where fine odour discrimination is performed without the necessity for highly specialized receptors. Instead, odour information is extracted in these systems using arrays of broadly tuned receptors organized in a convergent pathway. Such a sensing architecture allows combining the responses of the array of receptors, giving rise to particular representations of the different odour stimuli. The key advantage provided by this approach is that odour representation is more efficient and robust when the encoding is performed by the population of receptors than by any of its individual elements (hyper-acuity). A population of receptors obtains its maximum performance in encoding odour stimulus features when it balances the benefits of sensory diversity and redundancy. By sensor diversity we understand the number of different receptor types responsible for enhancing the variability of the array response to a collection of odours. Likewise, by sensor redundancy we refer to the average number of receptor replicates on a population. The role of sensor redundancy accounts for the robustness to receptor damage and noise exhibited by the odour stimuli representation. This variety of odour receptor types along with its outstanding number of receptors is characteristic of natural olfactory systems. Though, traditional electronic noses tend to exhibit a limited number of sensor units with very much correlated responses to odour stimuli. Several strategies to enhance odour representation in gas sensor arrays are based on boosting sensor diversity and redundancy. However, it has not been until recently that large arrays of cross-selective have become technologically available. In this dissertation, we have developed one of these new generation arrays to investigate the advantages odour stimuli representation through population coding in artificial olfaction. In particular, we proposed to build a chemical sensing system based on an array of metal oxide (MOX) gas sensors, and endowed with a high a degree of sensor diversity and redundancy. We proposed the use this bio-inspired sensing architecture alongside statistical pattern recognition techniques to cope with some of the unsolved problems in machine olfaction (robustness to sensor damage, feature selection, and calibration transfer). The main contributions of this work were the following: We defined functionally sensor diversity and redundancy. These definitions were based on the clustering of the array features according to their similitude when responding to an odour dataset. We compared the different manner how natural and artificial olfactory systems encode for odour information using simple sensors models. We found that natural olfactory system principally encoded odour information in terms of odour quality, whereas that artificial ones in terms of odour quantity. Also, we studied the effect of sensor noise on odour concentration encoding. We proposed to decrease the contribution of the sensor noise by means of the redundant sensor feature averaging and sensor array optimization. These strategies were effective in case of independent sensor noise, but not for removing common sources of sensor noise. Similarly, we detected the importance of sensor failure dependency on the odour discrimination capabilities of a sensor. We found that this sensor fault distribution across had to be independent of the sensor type to prevent a dramatic worsening on the array’s predictive performance. In addition to this, we proposed an update of a feature selection method including a dimensionality reduction stage so as to take into account the redundant information provided by the sensor array. Finally, we performed instrument standardization between temperature modulated sensor arrays to correct global shifts of temperature. A method to categorize the quality of the calibration transfer based on the bias-variance trade-off was presented., [spa] La nariz electrónica (e-nose) ha sido utilizada durante las últimas tres décadas como instrumento de propósito general para la detección química. Este instrumento está inspirado en los sistemas olfativos naturales, donde la discriminación de olores se realiza eficientemente sin la necesidad de receptores altamente especializados. La ventaja clave proporcionada por esta aproximación es que la representación de los olores es más eficiente y robusta cuando la codificación del olor es llevada a cabo por una población de receptores, pues esta supera la calidad de cualquiera realizada por sus elementos individuales. Una población de receptores obtiene su máximo rendimiento en la codificación de las características de un estímulo odorífero cuando se equilibran los beneficios de la diversidad y redundancia sensoriales. Lamentablemente, las narices electrónicas tradicionales tienden a exhibir un número limitado de sensores con respuestas muy correlacionas ante diferentes conjuntos de estímulos odoríferos. Sin embargo, no ha sido hasta hace relativamente poco que la creación grandes matrices de sensores con selectividades cruzadas ha sido tecnológicamente posibles. El objetivo de esta tesis es el desarrollo de una de estas matrices de nueva generación para investigar las ventajas de la representación de los estímulos odoríferos través de una codificación poblacional soportada por la diversidad y redundancia sensoriales. En particular, hemos construido un sistema de detección química basado en una matriz de sensores de gas de óxido metálico (MOX), y dotada de un alto grado de diversidad y redundancia sensoriales. Hemos utilizado esta arquitectura de detección química bioinspirada junto técnicas de reconocimiento de patrones estadísticas para hacer frente a algunos de los problemas sin resolver en olfacción artificial, a saber, la robustez al fallo de sensores, la selección de características, y la transferencia de calibración entre diferentes narices electrónicas.

Gene expression is a complex and highly regulated process. Most of the regulation is controlled by short DNA sequences that can be bound by some proteins called transcription factors (TF). Binding to these sites, the transcription factors, can start the transcription of mRNA, stop it, or just control the amount of mRNA produced. The DNA binding sites of these transcription factors have some specific characteristics: (1) They are short sequences (2) They can be located anywhere in the genome and (3) they are degenerated, which means that some mutations in the binding site sequence do not alter its functionality. These characteristics made impossible to look for a specific sequences in a specific region and, create the need to model the binding sites in order to detect them. Due to the importance of gene expression in the study of cell differentiation and its implication in some genetic diseases, many computational models and experimental processes to model binding site motifs and then find them into a genome have appeared. The computational models can be divided into two main groups: motif discovery methods which try to find binding sites within a set of co-regulated sequences without previous knowledge and motif search methods which use previous known sites to create a model and then try to locate binding sequences fitting this model. Most of the algorithms for binding site detection (both discovery and search) are based on Position weight matrices (PWM), which are matrices of frequencies of each nucleotide in each position, and assume that positions are independent. Some others take into account interdependences, but they need many sequences to be trained and high computational times. The focus of this thesis is to use the conversion from symbolical to numerical DNA and the previous knowledge of binding site sequences in order to construct models for DNA motifs. In this context, known multivariate signal processing techniques can be the ideal tools to construct models which can take into account interdependences without needing a large number of sequences or a high computational time. To characterize the transcriptions factors, the relationships TF-protein were studied, showing that most transcription factors regulate the expression of 5-10 genes and at the same time most proteins are regulated by more than 1 TF. The study of interdependences between positions showed that more than 90% of the binding sites have significant interdependences, but that the percentage of interdependences is not enough to classify TF according to structure. The conversion of DNA motif matrices into numerical matrices allows the use ofl Component Analysis (PCA) to model the binding sites which captures the information of the interdependences into the covariance, a second order statistics. Using the hypothesis that the binding sites will fit better to the PCA model than genomic, sequences, the Q-residuals can be used to detect binding sites within the genome. When compared to PWM the Q-residuals detector performs as least as well, and the improvement of detection is significantly correlated to the percentage of positions with interdependences. The disadvantage of these PCa models is that they are difficult to interpret. Converting the DNA symbolical matrix into a DNA numerical cube allows the calculation PARAFAC models which are easier to interpret. Since PARAFAC models have unique solutions, their scores can be combined with the PARAFAC Q-residuals in order to construct a quadratic detector that also performs better than PSSM models. When the numerical detectors are compared to detectors that take into account interdependences, they perform better when there are not many sequences available, but there are more sensitive to the number of positions.

Due to recent progress, higher-order chemical instrumentation provides large amounts of data which need automated processing in order to extract relevant information. In most cases, the raw signals or spectra are too complex for manual analysis. The ability to detect, identify and quantitate chemical substances in gas phase in field operations is required in a huge number of applications. Among them, I would like to highlight the need for chemical sensing on diverse humanitarian, safety and security applications. In these cases, it becomes extremely important to continuously monitor the environments where chemicals are spread in order to be ready to act when abnormal events are discovered. In most critical scenarios the sample can not just be taken to the laboratory and analyzed, since an immediate answer is needed. In some other scenarios, the exploration of the area must be performed because the localization of the gas source or material of interest is unknown. This exploration can be performed using multiple mobile sensors in order to localize the chemical source or material. Different sensing technologies have been successfully used to detect and identify different chemical substances (gases or volatile compounds). These compounds could be either toxic or hazardous, or they can be signatures of the materials to be detected, for instance, explosives or drugs. Among these technologies, mobility based analyzers provide fast responses with high sensitivity. However, IMS instruments are not exempt of problems. Typically, they provide moderate selectivity, appearing overlapped peaks in the spectra. Moreover, the presence of humidity makes peaks wider, thus worsening the resolving power and the resolution. Furthermore, the response of IMS is non-linear as substance concentration increases and more than one peak can appear in the spectra due to the same compound. In the present thesis, these problems are addressed and applications using an Ion Mobility Spectrometer (IMS) and a Differential Mobility Analyzer (DMA) are shown. It is demonstrated that multivariate data analysis tools are more effective when dealing with these technologies. For the first time, multivariate data analysis tools have been applied to a novel DMA. It is shown that DMA could be established as a good instrumentation for the detection of explosives and the detection and quantitation of VOCs. Furthermore, Multivariate curve resolution Alternating Least Squares (MCR-ALS) is shown to be suitable to analyze IMS spectra qualitatively when interfering chemicals appear in the spectra and even when their behaviour is non-linear. Partial Least Squares (PLS) methods are demonstrated to work properly for the quantitative analysis of these signals; from this analysis the chemical concentrations of the target substances are obtained. It is also demonstrated in this thesis that the quantitative measurements from these sensors can be integrated in a gas source localization algorithm in order to improve the localization of the source in those scenarios where it is required. It is shown that the new proposal works significantly better in cases where the source strength is weak. This is illustrated presenting results from simulations generated under realistic conditions. Moreover, real-world data were obtained using a mobile robot mounting a photo ionization detector (PID). Experiments were carried out under forced ventilation and turbulences in indoors and outdoors environments. The results obtained validate the simulation results and confirm that the new localization algorithm can effectively operate in real environments., Debido a los progresos recientes, la instrumentación química genera mayores volúmenes de datos los cuales requieren de un procesado automático con la finalidad de extraer la información relevante, ya que un análisis manual no suele ser viable debido a la elevada complejidad de los datos. La habilidad de detectar, identificar y cuantificar sustancias químicas en fase gas en operaciones de campo es requerida en un gran número de aplicaciones. Entre ellas, aplicaciones humanitarias y de seguridad. En estos casos, la monitorización continua de los entornos es extremadamente importante, ya que se debe estar alerta de eventos anormales. En los escenarios más críticos, debe realizarse una exploración del área porque la posición de la fuente de gas de interés es desconocida. Esta exploración puede realizarse usando múltiples robots. Diferentes tecnologías de sensores se han aplicado con éxito a la detección e identificación de diferentes sustancias químicas (gases o compuestos volátiles). Estos compuestos pueden ser tóxicos, peligrosos, o precursores de explosivos o drogas. De entre estas tecnologías, los analizadores basados en movilidad iónica (IMS) proporcionan rápidas respuestas con gran sensibilidad. Sin embargo, estos instrumentos no están exentos de problemas. Típicamente, proporcionan una moderada selectividad, apareciendo picos solapados en los espectros. Además, la presencia de humedad provoca que los picos se ensanchen, así empeorando la resolución. Además, la respuesta de IMS es no lineal al incrementar la concentración y es posible que más de un pico debido al mismo compuesto aparezca en el espectro. En la presente tesis se trata con estos problemas y se demuestra que las herramientas de análisis de datos multivariantes son más efectivas que las herramientas típicas univariantes al tratar con tecnologías de movilidad iónica (IMS y DMA), especialmente para el análisis cualitativo y cuantitativo de sus espectros. Además, se demuestra que las medidas cuantitativas pueden integrarse de manera efectiva en un algoritmo de localización de fuentes químicas. Los resultados obtenidos (simulaciones realistas y datos reales) muestran que el algoritmo desarrollado durante la tesis puede funcionar especialmente bien en situaciones en las que la potencia de emisión de la fuente a detectar sea débil.

This PhD thesis introduces basic concepts of smart chemical sensors design, which are afterwards applied to a particular application: the analysis of natural gas. The thesis addresses thus two sets of objective, a first set of objectives related to the conceptual design of a smart chemical sensor using smart sensor standards: - The design of an optimal smart chemical sensor architecture - The novel combination in a working prototype of the highly complementary smart sensor standards IEEE-1451 and BS-7986 A second set of objectives is directly related to the selected application. Natural gas quality control. Natural gas is an energy source of major importance in the world energy supply, its quality control is increasingly important due to its origin-dependent properties and the progressive liberalization of the energy market. The objectives related to this application are: - To solve the natural gas quality analysis problem by using a lower cost approach taking advantage of MEMS technology, smart sensor features, and embedded intelligent signal processing. - To select suitable sensing technologies and associated signal processing. An overall goal addressed by the PhD Thesis is in the end the reporting of a working smart sensor prototype implementing all the smart sensor features, MEMS based natural gas analysis and advanced signal processing as a demonstration of a novel low-cost and high speed natural gas analyzer. The thesis covers this research along 7 chapters, introducing the concepts and application in chapters 1 and 2, the objectives in chapter 3, the simulation of a proposed MEMS sensor approach in chapter 4, the description of the advanced signal processing approach adopted in chapter 5, the description of the electronics and engineering of the smart natural gas analyzer prototype in chapter 6, and finally the conclusions of the work in chapter 7., La tesis introduce conceptos básicos sobre el diseño de sensores químicos inteligentes, en particular presenta los estándares propuestos IEEE-1451 y BS-7986, y elabora una propuesta para el diseño óptimo de dichos sensores químicos inteligentes. Se implementa la propuesta de diseño para una aplicación concreta, el análisis de gas natural. Además de la aplicación de los conceptos sobre sensores químicos inteligentes se pretende además diseñar un analizador compacto, rápido y de bajo coste, para ello se estudia el uso de un microsensor termoeéctrico como sensor principal del analizador. Una vez probada su viabilidad se implementan ambos conceptos (sensores inteligentes y microsensor termoeléctrico) en un prototipo funcional validado en laboratorio. Como resultado se obtiene una propuesta para el diseño de sensores químicos inteligentes basada en estándares, y por otro lado se presenta un nuevo analizador de gas natural, más rápido y compacto que los existentes. Los resultados obtenidos originan diversas publicaciones en revistas así como dos patentes de método y sistema.

[spa] Actualmente el control de la concentración de etileno en las cámaras de almacenamiento de fruta proporciona información de gran interés sobre su estado. Por un lado, la fruta produce pequeñas concentraciones de etileno que indican su estado de maduración, y por otro lado, se añade etileno para acelerar el proceso de maduración. En esta Tesis, se ha desarrollado un espectrómetro óptico multicanal para medir de forma contínua la concentración de etileno en las cámaras frigoríficas. Se ha diseñado y fabricado componentes ópticos y la correspondiente electrónica de amplificación y procesado de señal. El resultado es un espectrómetro de gases basado en una aqrquitectura NDIR. En la tendencia actual de integración y miniturización, se ha diseñado un novedoso detector IR como componente del sistyema NDIR. El detector consta de una matriz 2x2 de termopilas, filtros ópticos para seleccionar las bandas de absorción y una matriz 2x2 de lentes de Fresnel fabricadas en un mismo sustrato silíceo. La etapa electrónica está formada por una parte de pre-amplificación analógica y otra parte digital basada en un lock-in. El instrumento se ha testeado a nivel de laboratorio y simulando condiciones reales. El resultado es que el espectrómetro es capaz de detectar 30 ppm de etileno, que es una concentración de amoníaco, que puede ser útil para detectar una posible fuga del sistema de refrigeración.

Projecte Final de Carrera d'Enginyeria Electrònica, Universitat de Barcelona, Facultat de Física. Director: Santiago Marco Colás. Any: 2007, L’objectiu del present treball és el disseny i realització d’un sistema lock-in capaç de recuperar el senyal procedent d’una termopila, que es troba integrada en un sistema òptic de detecció de gasos. Una termopila és un dispositiu electrònic que converteix energia tèrmica en energia elèctrica, valent-se de l’efecte Seebeck. El voltatge de sortida és proporcional a la diferència de temperatures entre les dues juncions de la termopila. Així doncs, el paràmetre de la termopila que estima la quantitat de radiació a l’entrada d’aquesta és la tensió en contínua de sortida. La radiació infraroja que absorbeix un gas concret depèn de la concentració d’aquest gas. Així, il·luminant la termopila amb radiació infraroja i avaluant el nivell de tensió d’aquesta a la sortida, es pot tenir una estimació de la concentració de gas present. El diagrama de blocs de la figura 1.1 mostra l’esquema de l’aplicació de l’amplificador lock-in.

En aquest treball s'ha establert una metodologia d'anàlisi i caracterització del comportament tèrmic, mecànic i termomecànic d'estructures micromecanitzades per a sensors de gas a través de la combinació de simulacions numèriques i tècniques de caracterització de microsistemes (mesures electro-tèrmiques, termografia, nanoindentació, AFM, XRD, microscopia confocal, Auger).L'estudi del comportament tèrmic de les estructures micromecanitzades ha permès obtenir les característiques bàsiques que controlaran el comportament del sensor, com són el consum en potència, la distribució de temperatura i el temps de resposta del substrat. L'anàlisi termomecànic ha consistit en determinar els esforços residuals en cada estructura així com l'estudi de la deformació dels diferents dissenys per a diferents temperatures de treball. S'han identificat diferents mecanismes de degradació en els materials que formen els sensors i s'ha obtingut el comportament termomecànic fins la ruptura del sensor. Tant en l'estudi tèrmic com en el termomecànic, la interacció entre la capa sensora i el substrat micromecanitzat així com l'influencia del material sensor en el comportament global del dispositiu han estat aspectes investigats.El treball inclou, a més, la caracterització termomecànica del Pt-Ti emprat en estructures micromecanitzades a través de la utilització de mètodes de Nanoindentació, Microscopia de Forces Atòmiques (AFM), Difracció de Raigs X (XRD) i espectroscopia Auger. També inclou el desenvolupament d'una metodologia per predir la fatiga tèrmica en microsistemes basada en la combinació dels models elasto-plàstics de metalls en capa prima (Alumini, Pt-Ti) amb simulacions numèriques. Finalment, de la metodologia d'anàlisi electro-termo-mecànic que s'ha dut a terme, es poden obtenir regles de disseny per la implementació de microsistemes que treballin en diferents règims de temperatura i en concret, directament aplicables al disseny i fabricació d'estructures micromecanitzades per a sensors de gas, This work presents a complete thermomechanical study of different micromachined gas sensor substrates based on closed and suspended membrane microstructures. The work has been carried out combining coupled electro-thermo-mechanical three-dimensional finite element method simulations with different experimental techniques such as those used in Microsystems characterization (thermo-electrical, thermography, AFM, XRD, confocal microscopy, Auger..). The performances predicted by simulations, such as the power consumption, the temperature distribution, the time response, the membrane deflection during operation and the preferential failure sites in the micromachined substrates have been confirmed by experience. The work includes the thermo-mechanical characterization of Pt-Ti thin films used in the structures using Nanoindentation, AFM, XRD and Auger spectroscopy. Additionally, a methodology to predict the thermal fatigue in microsystems, which combines experimental thin metal elasto-plastic models (Al, Pt-Ti) and coupled thermo-mechanical FEM simulations, has been developed.The good agreement between simulations and experimental results validates the numerical models, and allows us to consider the adaptability of the analyzed designs as micromachined substrates for integrated gas sensors.Keywords: MEMS, Microsystems, gas sensors, thermal fatigue, Al, Pt-Ti, FEM., En este trabajo se ha establecido una metodología de análisis y caracterización térmica y termomecánica de estructuras micromecanizadas en silicio para aplicaciones en sensores de gas. Esta investigación ha combinado simulaciones numéricas mediante el método de los elementos finitos con técnicas experimentales de caracterización utilizadas en el campo de los microsistemas (medidas electro-térmicas, termografía, AFM, XRD, microscopia confocal, Auger).El estudio térmico de dichas estructuras ha permitido obtener su consumo en potencia, la distribución de temperaturas, la dinámica térmica, así como ha permitido fijar con precisión las propiedades térmicas de los materiales típicamente utilizados en la tecnología de los microsistemas. El estudio mecánico ha permitido obtener los esfuerzos residuales inducidos por los procesos de fabricación. Además, se ha obtenido la deformación de las estructuras a diferentes temperaturas de trabajo hasta la ruptura total de las membranas. Durante las altas temperaturas de trabajo se han detectado y analizado diferentes mecanismos de degradación en los materiales. El trabajo incluye además, la caracterización termo-mecánica del Pt-Ti depositado por sputtering, ampliamente utilizado en microsensores de gas, mediante el empleo de técnicas de Nanoindentación, Microscopia AFM, Difracción de Rayos X (XRD) y espectrocopia Auger. También presenta el desarrollo de una metodología para la predicción de la fatiga térmica en microsistemas, que se basa en la combinación de modelos elasto-plásticos para metales en capa delgada con simulaciones numéricas.Finalmente, de la metodología de análisis termo-mecánico que se ha llevado a cabo, se pueden obtener reglas de diseño para microsistemas que trabajen a diferentes temperaturas, y en concreto directamente aplicables al diseño y fabricación de estructuras micromecanizadas para sensores de gas. Palabras clave: MEMS, microsistemas, sensores de gas, fatiga térmica, Al, Pt-Ti, MEF.

The context of this thesis is the design of CMOS low power RFIC's suitable for direct conversion architectures. Our starting point is the analysis of the characteristics of the integrated passive components (inductors and mainly transformers) from a circuit designer perspective. Then, the achieved understanding of these components is exploited in order to optimize the performance of some of the building blocks of a RF front-end. This thesis is divided in three main chapters. Chapter 1 is dedicated to the analysis of integrated transformers. It starts with a revision of the state of the art of the integrated inductors in order to set up the basis for the analysis of the monolithic transformers. Due to their influence on the component quality factor, a special insight is dedicated to the analysis of the eddy currents. Then, we introduce the integrated transformers and revise the topologies used in their implementation and their characteristics and electrical equivalent models. However, it is important to remark that the goal of this thesis is not the modeling of the component. Instead, we will use EM simulators to reproduce their behavior and point out some physical mechanisms present in integrated transformers that have not been properly considered in the equivalent electrical models presented in the literature so far. In particular, we will study eddy currents in integrated transformers and will demonstrate their dependence on the operating mode of the transformer as well as on the loads connected to the primary and the secondary. As we will see this phenomenon has a direct influence on the component model as well as on the component optimization. Thus, different optimization procedures should be applied to minimize the component losses depending on the application. Finally, this chapter finishes with the design of a double balanced mixed that uses two differential transformers to increase the isolation between ports. It is demonstrated that a differential driving minimizes the effects of the parasitic capacitances and increases noticeably this isolation. It is also shown that the etching of the silicon underneath of the integrated transformers that reduces these parasitic capacitances also increases this isolation (even if a single-ended excitation is applied). Chapter 2 investigates the design of oscillators having low phase noise and large tunning range. After identifying the quality factor of the integrated inductors as one of the main factors limiting the oscillator phase noise, a novel transformer-based (parallel) resonator is introduced. This new topology overcomes the performance of the common inductor-based resonators in terms of tuning range and effective quality factor. A detailed description of the resonator is then performed in order understand its potentially and limitations. Thus, its properties are firstly studied from an electrical circuit perspective using the equivalent models discussed in the previous chapter. Then, it is realized the analysis in the EM domain in order to compare its performance with the standard inductor-based resonator. Once again, eddy currents are identified as the main factor limiting the effective quality factor of the transformer-based resonator. Thanks to the already achieved understanding of integrated transformers, the layout optimization method proposed by López-Villegas [35] for integrated inductors can be extended to the optimization of the transformer (when working in common-mode). The extension of the parallel design to N-resonators is also shortly discussed. Finally, the proposed resonator is used in a low phase-noise 1.7 GHz CMOS VCO.Chapter 3 presents a new topology of quadrature oscillator based on the differential coupling at the second harmonic of two differential oscillators. As in the previous design, an integrated transformer plays a relevant role in this circuit being used to establish the coupling between oscillators. In order to obtain a basic understanding of the phenomena involved in the generation of quadrature signals using the proposed method and set up the design procedure several concepts must be revised. In particular, the theory on forced (or injected) and coupled oscillators is discussed and applied to the proposed quadrature oscillator. Finally, to show the feasibility and potentiality of this methodology, different designs in integrated and hybrid implementations are presented.

El presente trabajo se enmarca dentro de las tareas centradas en el campo de modelización y procesado de señal, desarrolladas en el Departament d’Electrònica de la Facultad de Física de la Universitat de Barcelona. En su concepción de nivel más amplio, el trabajo demuestra la posibilidad de trasladar y aplicar con eficiencia las técnicas de procesado y modelización dinámica no lineal al ámbito de los sistemas de medida de gases. Son relativamente abundantes en la literatura otras aproximaciones más simples a esta temática y en este ámbito. Así, es posible encontrar en la literatura trabajos enfocados hacia la modelización estática-lineal, estática-no lineal y dinámica-lineal. Sin embargo, no existían antes de este trabajo aportaciones bajo el enfoque dinámico y no lineal en la modelización de sistemas de medida gaseosos. Por otra parte, el trabajo aborda además el problema desde el punto de vista de la multidimensionalidad junto con la restricción del uso de un reducido número de datos; factores éstos que aportan dificultad adicional a la vez que valor añadido a la solución. En lo concreto, el trabajo se centra en el análisis y modelización del comportamiento de un sistema de medida de gases basado en sensores QMB frente a cambios rápidos de dos concentraciones en una mezcla binaria de gases (tolueno y octano) en aire sintético y bajo régimen de altas concentraciones (miles de ppm). Frente a este problema, la calibración estática es incapaz de aportar una solución con un error tolerable. Por primera vez se ha abordado, frente a la solución estática, la modelización dinámica y no lineal de sistemas analizadores de gases, ofreciendo esta aproximación resultados óptimos. Los experimentos diseñados, orientados hacia la confección de los modelos, fueron fundamentalmente dos: 1) Sistema sometido a concentraciones en forma de escalón 2) Sistema sometido a señales de concentración en forma de ruido blanco con densidad de probabilidad gausiana. Los modelos generados son también de dos tipos: 1) Modelos directos: desde la entrada hacia la salida 2) Modelos inversos: desde la salida hacia la entrada Para la resolución del problema se han probado diferentes estimaciones, como estimadores de correlación, estimaciones paramétricas, series de Volterra, series de Wiener, modelos de cajas y redes neuronales. De entre todos ellos, los modelos basados en series de Wiener y en redes neuronales han demostrado una mayor eficiencia. Se demuestra que estos modelos son capaces de explotar la información del sistema estudiado hasta el límite de la resolución experimental, tanto en la modelización directa como en la inversa. Aparte de la tarea estrictamente relacionada con la modelización, se ha iniciado con este trabajo la construcción de una estación de test de sensores de gas que en la actualidad y después de diversas aportaciones del grupo de trabajo, es un complejo sistema de medición y test capaz de realizar una gran variedad de pruebas relacionadas con gases, dotando al departamento de una línea de trabajo nueva, potente y bien equipada., The work demonstrates the possibility of efficiently transferring and applying nonlinear dynamic processing and modeling techniques to the field of gas measurement systems. It is possible to find works focused on static-linear, static-non-linear and dynamic-linear modeling. However, prior to this work there were no contributions under the dynamic and non-linear approach in the modeling of gas measurement systems. On the other hand, the work also addresses the problem from the point of view of the curse of dimensionality: multidimensional problems with scarcity of data. The work focuses on the analysis and modeling of the behavior of a gas measurement system based on QMB sensors against rapid changes of two concentrations in a binary mixture of gases (toluene and octane) in synthetic air and under regime of high concentrations (thousands of ppm). The static calibration is unable to provide a solution with a tolerable error. The dynamic and nonlinear modeling of gas analyzer systems has been performed with optimal results. The experiments designed, oriented towards the preparation of the models, were fundamentally two: 1) System with step-like gas concentration inputs 2) System excited with gas input signals whose concentration values follow a white noise with Gaussian probability density. The generated models are also of two types: 1) Direct models: from the input to the output 2) Inverse models: from the output to the input Different approaches have been tested, such as correlation estimators, parametric estimates, Volterra series, Wiener series, box models and neural networks. Models based on Wiener series and on neural networks have demonstrated a greater efficiency. It is shown that these models are able to extract the information of the studied system until experimental resolution limit, in both direct and inverse modeling. Apart from the task related to gas sensor modeling, the development of a test station for gas sensors has been addressed. This task has been developed by the research group and it is capable to perform a wide variety of gas experiments, providing the department with a new, powerful and well-equipped line of work.