Une portion importante du cycle du carbone se situe au niveau des échanges gazeux entre l'eau et l'atmosphère. L'importance de ces flux est accentuée par l'enjeu des changements climatiques et de la gestion des gaz à effet de serre. Il reste cependant plusieurs aspects de ce processus qui sont mal compris et plusieurs biais persistent dans la méthodologie. Dans le but d'améliorer les techniques d'échantillonnage, la possibilité d'utiliser la turbulence pour prédire la vitesse d'échange gazeux (k) a été explorée et l'hypothèse que la chambre flottante engendre de la turbulence artificielle a aussi été testée. De plus, dans le but d'unifier les diverses relations entre le vent et k, plusieurs variables facilement mesurables combinées à la vitesse du vent ont été testés. Les données de cette étude ont été échantillonnées d'une part dans le réservoir hydroélectrique d'Eastmain-1 et d'autres parts dans 11 lacs en Estrie, Québec. La vitesse d'échange gazeux à été mesuré in situ à l'aide d'une chambre flottante. Plusieurs variables météorologiques dont la turbulence de l'eau ainsi que quelques variables limnologiques ont aussi été mesurés. Un modèle robuste a été élaboré en utilisant la turbulence de l'eau à l'intérieur de la surface d'échantillonnage de la chambre pour expliquer k. Il a aussi été démontré la chambre flottante surestime k et cela est due à la turbulence à l'intérieur de la surface d'échantillonnage créée par celle-ci. Le rapport de surestimation peut atteinte dix fois la valeur réelle et ensuite diminue plus la turbulence naturelle du système augmente. Finalement, il a été montré que l'ajout de l'aire du système aux vitesses de vents dans une régression multiple améliore grandement la prédiction de k dans une variété de systèmes aquatiques différents. En apportant de meilleurs outils de mesure et d'estimation des vitesses d'échanges gazeux, cette étude permettra d'améliorer la précision des estimations des émissions de gaz à effet de serre provenant des milieux aquatiques terrestres. ______________________________________________________________________________ MOTS-CLÉS DE L’AUTEUR : échanges gazeux, dioxyde de carbone, interface air-eau, turbulence, chambre flottante

The large-scale production of monoclonal antibodies (MAb) by mammalian cells in batch and fed-batch culture systems is limited by the unwanted decline in cell viability and reduced productivity that may result from changes in culture conditions. Therefore, it becomes imperative to gain an in-depth knowledge of the factors affecting cell growth and cell viability that in turn determine the antibody production. An attempt has been made to obtain an overall model that predicts the behaviour of both batch and fed-batch systems as a function of the extra-cellular nutrient/metabolite concentrations. Such model formulation will aid in identifying and eventually controlling the dominant factors in play to optimize monoclonal antibody (MAb) production in the future. Murine hybridoma 130-8F producing anti-F-glycoprotein monoclonal antibody was grown in D-MEM medium (Gibco 12100) with 2% FBS. A systematic approach based on Metabolic Flux Analysis (MFA) was applied for the calculation of intracellular fluxes for metabolites from available extracellular concentration values. Based on the set of identified significant fluxes (from MFA), the original metabolic network was reduced to a set of significant reactions. The reactions in the reduced metabolic network were then combined to yield a set of macro-reactions obeying Monod kinetics. Half saturation constants were fixed empirically to avoid computational difficulties that parameter estimation for an over-parameterized system of equations would cause. Using Quadratic Programming, the proposed Dynamic Model was calibrated and model prediction was carried out individually for batch and fed-batch runs. Flux distribution for batch and fedbatch modes were compared to determine whether the same model structure could be applied to both the feeding profiles. Correlation analysis was performed to formulate a Biomass Model for predicting cell concentration and viability as a function of the extracellular metabolite concentrations in batch and fed-batch experiments. Quadratic Programming was applied once again for estimation of growth and death coefficients in the equations for viable and dead cell predictions. The prediction accuracy of these model equations was tested by using experimental data from additional runs. Further, the Dynamic Model was integrated with the Biomass Model to get an Integrated Model capable of predicting concentration values for substrates, extracellular metabolites, and viable and dead cell concentration by utilizing only starting concentrations as input. It was found that even though the set of significant fluxes was the same for batch and fedbatch operations, the order of these fluxes was different between the two systems. There was a gradual metabolic shift in the fed-batch system with time indicating that under conditions of nutrient limitation, the available energy is channeled towards maintenance rather than growth. Also, available literature with regard to cell kinetics during fed-batch operation suggests that under nutrient limited conditions, the cells move from a viable, non-apoptotic state to a viable apoptotic state. This is believed to lead to variations in antibody production rates and might explain inaccurate predictions for MAb obtained from the model proposed in the current work. As a result more detailed analysis of the system and in particular, the switch from non-apoptotic to apoptotic state is required. As a continuation of efforts to study the system in-depth, fluorescence imaging is currently being applied as a tool to capture the changes in cell morphology along the course of experimental batch and fed-batch runs. These experiments maybe able to elucidate the transition from non-apoptotic to apoptotic cells and this information maybe used in the future to improve the accuracy of the existing mathematical model.

L'apprentissage statistique propose un vaste ensemble de techniques capables de construire des modèles prédictifs à partir d'observations passées. Ces techniques ont montré leurs capacités à traiter des volumétries importantes de données sur des problèmes réels. Cependant, de nouvelles applications génèrent de plus en plus de données qui sont seulement visibles sous la forme d'un flux et doivent être traitées séquentiellement. Parmi ces applications on citera : la gestion de réseaux de télécommunications, la modélisation des utilisateurs au sein d'un réseau social, le web mining. L'un des défis techniques est de concevoir des algorithmes permettant l'apprentissage avec les nouvelles contraintes imposées par les flux de données. Nous proposons d'abord ce problème en proposant de nouvelles techniques de résumé de flux de données dans le cadre de l'apprentissage supervisé. Notre méthode est constituée de deux niveaux. Le premier niveau utilise des techniques incrémentales de résumé en-ligne pour les flux qui prennent en compte les ressources mémoire et processeur et possèdent des garanties en termes d'erreur. Le second niveau utilise les résumés de faible taille, issus du premier niveau, pour construire le résumé final à l'aide d'une méthode supervisée performante hors-ligne. Ces résumés constituent un prétraitement qui nous permet de proposer de nouvelles versions du classifieur bayésien naïf et des arbres de décision fonctionnant en-ligne sur flux de données. Les flux de données peuvent ne pas être stationnaires mais comporter des changements de concept. Nous proposons aussi une nouvelle technique pour détecter ces changements et mettre à jour nos classifieurs.

The large-scale production of monoclonal antibodies (MAb) by mammalian cells in batch and fed-batch culture systems is limited by the unwanted decline in cell viability and reduced productivity that may result from changes in culture conditions. Therefore, it becomes imperative to gain an in-depth knowledge of the factors affecting cell growth and cell viability that in turn determine the antibody production. An attempt has been made to obtain an overall model that predicts the behaviour of both batch and fed-batch systems as a function of the extra-cellular nutrient/metabolite concentrations. Such model formulation will aid in identifying and eventually controlling the dominant factors in play to optimize monoclonal antibody (MAb) production in the future. Murine hybridoma 130-8F producing anti-F-glycoprotein monoclonal antibody was grown in D-MEM medium (Gibco 12100) with 2% FBS. A systematic approach based on Metabolic Flux Analysis (MFA) was applied for the calculation of intracellular fluxes for metabolites from available extracellular concentration values. Based on the set of identified significant fluxes (from MFA), the original metabolic network was reduced to a set of significant reactions. The reactions in the reduced metabolic network were then combined to yield a set of macro-reactions obeying Monod kinetics. Half saturation constants were fixed empirically to avoid computational difficulties that parameter estimation for an over-parameterized system of equations would cause. Using Quadratic Programming, the proposed Dynamic Model was calibrated and model prediction was carried out individually for batch and fed-batch runs. Flux distribution for batch and fedbatch modes were compared to determine whether the same model structure could be applied to both the feeding profiles. Correlation analysis was performed to formulate a Biomass Model for predicting cell concentration and viability as a function of the extracellular metabolite concentrations in batch and fed-batch experiments. Quadratic Programming was applied once again for estimation of growth and death coefficients in the equations for viable and dead cell predictions. The prediction accuracy of these model equations was tested by using experimental data from additional runs. Further, the Dynamic Model was integrated with the Biomass Model to get an Integrated Model capable of predicting concentration values for substrates, extracellular metabolites, and viable and dead cell concentration by utilizing only starting concentrations as input. It was found that even though the set of significant fluxes was the same for batch and fedbatch operations, the order of these fluxes was different between the two systems. There was a gradual metabolic shift in the fed-batch system with time indicating that under conditions of nutrient limitation, the available energy is channeled towards maintenance rather than growth. Also, available literature with regard to cell kinetics during fed-batch operation suggests that under nutrient limited conditions, the cells move from a viable, non-apoptotic state to a viable apoptotic state. This is believed to lead to variations in antibody production rates and might explain inaccurate predictions for MAb obtained from the model proposed in the current work. As a result more detailed analysis of the system and in particular, the switch from non-apoptotic to apoptotic state is required. As a continuation of efforts to study the system in-depth, fluorescence imaging is currently being applied as a tool to capture the changes in cell morphology along the course of experimental batch and fed-batch runs. These experiments maybe able to elucidate the transition from non-apoptotic to apoptotic cells and this information maybe used in the future to improve the accuracy of the existing mathematical model.

A novel photon flux monitor has been designed and tested for use at the Duke University High Intensity Gamma Source, where the photon beam produced is essentially mono-energetic but it is not tagged. Direct counting of the number of photons using a high-efficiency detector is not possible because of the high photon fluxes expected. Therefore, a direct counting detector with a low, accurately known efficiency was required.<p> The photon flux monitor based on a five scintillator paddle system detects recoil electrons and positrons from photoelectric, Compton and pair-production processes. It has been designed to be insensitive to gain and detector threshold changes and to be usable for photon energies above 5 MeV. It has been calibrated using direct counting with a NaI detector and its efficiency has been shown to be well predicted by a GEANT4 simulation.<p> Results of measurements, calibration and calculations required to characterize the 5-paddle photon flux monitor are presented. The photon flux monitor has met its design specifications of being able to determine the number of photons incident on it during the live time of a measurement to within a systematic error of 2%.<p> A paper based on the work for this thesis has been published in the Nuclear Instruments and Methods in Physics Research Journal.

Dissolved organic carbon (DOC) is one of the largest pools of carbon in the ocean, and is of the same size as the carbon dioxide in the atmosphere. Estuaries connecting the land and the ocean are one of the most important DOC sources to the ocean, and play an important role in the global carbon cycle. Because of their complex chemical, physical, geological and biological properties, estuaries have become rich ecological environment. In this study, we investigated the seasonal distributions of DOC in the Taiwan Strait (TS) and Taiwanese rivers, aiming to understand the distributions and variations of DOC in different seasons. The results show that DOC concentrations are generally the highest in the upper estuary, and then decrease downstream due to mixing with the low DOC seawater. The process of river flow constantly accumulates terrestrial material, and the DOC shows positive correlations with Chl. a, CH4 and BOD (Biochemical Oxygen Demand), suggesting that biological activities and pollutions could be sources of DOC in the estuary. The DOC concentrations (salinity<1) varied in dry (Nov.-Apr.) and wet (May-Oct.) seasons with ranges of 42-1185 £gM (mean=245¡Ó254£gM; n=32) and 18-565 £gM (mean=183¡Ó151£gM; n=24), respectively. The total DOC flux of 25 rivers is 87.8 Gg C/yr, which can be translated to the fluxes of all rivers in Taiwan to be 101.9 Gg C/yr. The amount of DOC flux in Taiwan is only about 0.07% of the tropical area, but the per unit area flux (3.92 gC /m2 /yr) is almost twice those of the tropical rivers (2.13 gC /m2 /yr). In Taiwan, the population density and land use are higher than the world average. Consequently, the impacts of the environment by human activities reveal the utmost export of DOC, and need further investigation. Next, in the TS, the DOC shows significant negative correlations with Sigma-T, and the distributions of DOC are mainly controlled by physical mixing in both winter and summer. In the western TS, DOC concentration is relatively high, compared to the eastern part, and is because of low temperature and salinity, but high DOC coastal China current flowing from north to south. DOC concentration decreases with increasing depth owing to the intrusion at depth by the Kuroshio, which contains relatively low DOC. In winter, the import of coastal China current brings more nutrients from north to south, and supports the growth of bacteria which depletes the DOC and oxygen. As the result, DOC decomposition rate is higher in winter than in summer. The TS¡¦s DOC fluxes in summer (northern TS: 3.85¡Ñ1012mol C/yr¡Fsouthern TS: 3.75¡Ñ1012mol C/yr) are higher than in winter (northern TS: 3.69¡Ñ1012mol C/yr¡Fsouthern TS: 2.84¡Ñ1012mol C/yr). Main differences are due to the prevailing southwest monsoon winds in summer transporting more water from the South China Sea to the TS, and the river discharge brings more terrigenous organic matters into the TS. Therefore, the DOC export in summer is higher than in winter.

Abstract This thesis is mainly aimed to the study the basic physical properties of different flux, and the combination with semiconductor packaging of 0.6mm Sn96.5/Ag3.0/Cu0.5 (SAC305) big solder ball, and 0.3mm Sn98.5/Ag1.0/Cu0.5 (SAC105) small solder ball. Big solder ball was proceeded with shear stress test, pull ball test, and tray drop test. However small solder ball was proceeded with shear stress test, pull ball test, zone shear solder balls test and board level drop test. After experiments, samples of different flux, the effect of solder strength and mechanical properties were received and discussed. And failure modes were observed by high-power microscope, SEM and OM. The experimental result shows that the flux doesn¡¦t have direct relationship with the point of welding strong and weak, the point of welding strong and weak is determined by the metal¡¦s characteristics. However, the less of residual flux is, the higher the solder balls shearing force as well as pulling force are. As for the more of residual more of flux, after aged processing, some solder balls have the phenomenon of missing from the IMC layer. And the difference of the residual flux will affect the various failure modes of zone shear solder balls. From board level drop test, the majority of failure modes is pad peeling at the corner of test board side for all test flux, and the remaining failure modes exhibit solder fractured and IMC layer broken. Therefore, the judgement of the residual flux is a very important key factor for the semiconductor packaging. It doesn¡¦t have the direct correlation with the reliability of final products. Keywords: Lead Free, Solder Ball, Flux, Semiconductor Packaging, Tests.

Previous studies have shown that the continental shelf and slope area off northeastern Taiwan is a region where the material exchange processes between the East China Sea and open ocean are most rigorous under the influences of the Kuroshio and the monsoons. In order to understand the topographic effects on the particulate flux and the transport path of the settling particles in this region, this study has conducted a series of field measurements using time-series sediment traps and current meters as tools. The temporal and spatial variations on the current, the particulate flux and the particle-size distribution as well as the activity and flux of 210 Pb and 210 Po are presented in this thesis. The biological productivity and the runoff of main rivers from Taiwan were not correlated with the particulate fluxes in our study area. However, the variations of the particulate flux in the canyon and on the slope areas were clearly affected by the tidal currents, showing good positive correlation with their amplitudes. The particulate, 210 Pb and 210 Po fluxes all showed an increase toward the ocean bottom but a decrease away from the shore, suggesting that the transports of particle and particle reactive elements were strongly influenced by the lateral transport process in the study area. Highest fluxes were observed in the Mien-Hua Canyon(MHC) and the lower slope southwest of the canyon. Intermediate values were observed in the South Okinawa Trough and the North Mien-Hua Canyon(NMHC). The intercanyon slope and the shallow traps in the NMHC had lowest values. The particulates and particle reactive elements were transported out of the shelf or upper slope area essentially through the MHC toward the slope southwest of the canyon. The NMHC and the intercanyon area were not major pathways for these materials. The fluxes of particle, 210 Pb and 210 Po in the study area were generally at least an order of magnitude greater than those observed from other continental margins in the world, except for the lowest values at the intercanyon station and the shallow water in the NMHC. Thus the continental margin off northeastern Taiwan is one of the most important areas for material exchanges with open oceans in the world. The trapped particulates were mainly silt with variable sand fractions. High sand contents were observed in the MHC and the lower slope southwest of the canyon, the intermediate values were in the outlet of the MHC and the NMHC, and the lowest in the South Okinawa Trough. This shows that the sand content decreases awayfrom the shore. On the contrary, silt and clay fractions in percent increase away from the shore. Under the influence of tidal currents, the settling sand particles in the MHC and its adjacent lower slope were laterally transported from resuspended sandy sediments of the continental shelf and upper slope area adjoining the head of the MHC. The fluxes of sand, silt and clay all decreased toward the South Okinawa Trough but increased toward the bottom. This particle-size distribution pattern also reflected the effect of lateral transport process in the study area. Variations of the 210 Pb activity in this study area were generally correlated inversely with the particulate flux and the sand content, indicating the effect of dilution of the mass flux and the sand content, but were positively correlated with the silt and clay contents because they were major 210 Pb scavengers. No apparent correlations of 210 Po with the mass flux and the three particle-size fractions were observed, suggesting that the variation of 210 Po activity was not related to the mass fluxes or the three particle-size fractions. High 210 Pb activities were found to be associated with high contents of POC, Mn and Fe. Correlations of 210 Pb with both Mn and Fe indicate that the oxide and hydroxide precipitates of Mn and Fe supplied from sediments on the continental shelf, upper slope area and submarine hydrothermal activities might serve as important mechanisms for enhanced 210 Pb scavenging. The relationship between the 210 Pb activity and the POC content might reflect either the settling mode of the fine particles incorporated with fecal pellets or aggregates, or sharing the same source with fine particles. The variatio ns of 210 Po activity were not correlated with Mn, Fe, major components or biogenic components, implying that the association of 210 Po with biogenic components might be masked by the enormous terrigenous materials. Except for the upper-trap samples collected at the intercanyon slope and the NMHC, the 210 Po activity was lower than the 210 Pb activity with 210 Po/210 Pb ratios less than unity. Based on comparisons of 210 Pb and 210 Po with relevant parameters and references of past studies, we can propose several possible mechanisms to explain why the ratios were less then unity: 1. 210 Pb was preferentially scavenged by advecting and settling particulates of terrigenous origin. 210 Po, known to be biophilic, might remain associated with the suspended organic particulates which could be dissolved or not settling. Moreover, the oxide and hydroxide precipitates of Mn and Fe in the study area would further enhance the scavenging of 210 Pb. 2. The particulates in suspension were favorable for differentiation between 210 Pb and 210 Ponuclides while those settled in sediments would allow them to grow toward the secular equilibrium. Therefore, the trapped particulates might be in frequent suspension and short in deposited sediments during their lateral transport. 3. The 210 Pb supplied from the East Asia aerosol in large quantity was accumulated in water column abd carried by the Kuroshio to the study area. The intrusion and eddies of Kuroshio off northeastern Taiwan could bring large amount of 210 Pb into the region of continental shelf and slope, allowing particulates to scavenge 210 Pb more effectively from the surrounding water. The upper trap in the NMHC and that in the intercanyon site yield lowest particulate flux but highest POC content. With less terrigenous material and more biogenic components in these samples, 210 Po/210 Pb ratios were greater than unity and variations of 210 Pb activity were hardly correlated with those of POC contents but were inversely correlated with those of Mn contents. However, the 210 Po activities of upper trap at the intercanyon site were well positively correlated with POC and PIC contents, indicating effects of the fresh biogenic particles. Based on the mass balance calculations of particulate and 210 Pb fluxes, current data and other related data gathered in this study, we may infer the sedimentation dynamics and transport path of the settling particles below 300m depth in the study area. Under the influence of tidal oscillation, terrigenous sediments on the shelf and the shelf break as well as the upper slope would be resuspended and carried into the MHC and transported further away. Most of the sand particles would deposit in the canyon while part of the silt and clay would be transported to the South Okinawa Trough directly and the rest to the lower slope southwest of the MHC. The particulate flux at the lower slope southwest of MHC might also have a portion of particulates swept from the upper slope by the Kuroshio eddies. As the NMHC and the intercanyon sites are located in the central area of the Kuroshio eddies, the particulates from the continental shelf and upper slope area would be blocked, resulting in very low particulate fluxes. High particulate fluxes observed in the lower NMHC trap should largely be due to resuspension of the sediments from the banks of the canyon.

Based on the design concepts of linear induction motors, a non-contacting steel plate conveyance system for steel mill application has been constructed. To reduce the noise and friction from conventional roller conveyance system, the designed system is aimed to simultaneously provide adequate lift, propulsive, and guide forces to the steel plate. At first, the preliminary understandings of the characteristics of lift force have been gained through the simple magnetic circuit analyses, and together with other mechanical concepts develop the laboratory prototype. Then, through three-dimensional finite element analyses and state model developments, the system¡¦s static and quasi-dynamic/dynamic operational characteristics are investigated. Finally, the validity of this system has been verified by experimental measurement. Thus, the analyses and results of the experiment clearly show that the designed non-contacting steel plate conveyance system is certainly feasible.

This study reports the first sediment trap mooring results obtained under the SIBEX program (South China Sea Integrated Biogeochemical Experiments). Two strings of sediment traps were deployed respectively at M1 located to the southwest of Taiwan, and M2 in the northern basin of the South China Sea (SCS). Box cores were also taken at several sites to the south of M2. The main purposes are to measure settling particulate fluxes at various depths for the studies of temporal and spatial variations of the particulate flux and 210Pb activity. The box cores were used to determine the sedimentation rates. These are to enhance our understanding of the characteristics of the particulate distribution and the roles the particulate matter plays in the biogeochemical processes in the SCS. Particulate fluxes measured from different depths at M1 and M2 generally increase with depth. In temporal variation, M1 has higher amplitudes than M2, with highest amplitudes at 948 m where highest flux (2025 mg/m2/d) was observed. The particulate flux at 948 m has higher values than at 248 m, probably due to lateral transport. At M2, the particulate flux at 240 m has higher values with greater amplitudes than at greater depths, i.e. 1240 m, 2240 m and 3240 m, where their particulate fluxes show a synchronous trend with small amplitudes in temporal variation. The time-averaged particulate flux for each trap ranges from 199 to 554 mg/m2/d, consistently higher than previous observations (76~104 mg/m2/d). However, our values are comparable to the mean particulate flux (280 mg/m2/d) estimated from terrigenous inputs. The apparent changes in particulate flux in the SCS over the past ten years warrants further investigations. The temporal variations of Pb-210 show a synchronous trend and a rapid increase with depth as observed at M1 and M2. This rapid increase of Pb-210 with depth reflects effective scavenging by sinking particulates although particulate concentrations are low. The loss on ignition (L.O.I.) at M1 and M2 was inversely correlated with Pb-210, indicating that Pb-210 was expelled from biogenic particulates or organic matter. The particulate fluxes around 240 m at M1 and M2 were generally positively correlated with the L.O.I., suggesting that the high particulate fluxes are probably contributed by biogenic particulates or organic matter. The sedimentation rates as determined from excess Pb-210 profiles range from 9.01~23.13 cm/100yr. These rapid sedimentation rates reflect the effect of mixing and additional sediments accumulated through lateral transport. The surface layers of these cores were subject to perturbation and mixing, based on the penetration depths of the excess Pb-210. The Pb-210 flux estimated from the inventory of excess Pb-210 in the sediments is much greater than that observed from the traps, suggesting that additional Pb-210 has been accumulated via lateral transport and slumping of nearby surface sediments. Based on the mean particulate flux observed, one may roughly estimate the corresponding sedimentation rate of about 10 cm/ka, which is an order of magnitude lower than those determined by the excess Pb-210 method. Because the upper layers of the sediments have been strongly disturbed and mixed with the additional sediments accumulated through lateral transport, the sedimentation rates as determined by the excess Pb-210 method are probably over-estimated by a factor of 10 to 20.