The separation of aqueous and organic mixtures with thermal separation processes is an important and challenging task in the chemical industry. Rising prices for energy, stricter environmental regulations and the increasing demand for high purity chemicals are the main driving forces to find alternative solutions to common separation technologies such as distillation and absorption. These are mostly too energy consumptive and can show limited separation performance, especially when applied to close boiling or azeotropic mixtures. Pervaporation can overcome these thermodynamic limitations and requires less energy because only the separated components need to be evaporated. This separation technology is already well established for the production of anhydrous solvents, but not yet widely distributed in the chemical and petrochemical industry due to some crucial challenges, which are still to overcome. Besides the need of high selective membranes, the development of membrane modules adapted to the specific requirements of organoselective pervaporation needs more research effort. Furthermore, only few modelling and simulation tools are available, which hinders the distribution of this process in industrial scale. In this work, these issues are addressed in a combined approach. In close collaboration with our cooperation partners, a novel membrane module for organophilic pervaporation is developed. A novel technology to manufacture high selective polymeric pervaporation membranes is applied to produce a membrane for an industrially relevant organic-organic separation task. A three step modelling approach ranging from a shortcut and a discrete to a rigorous model is developed and implemented in a user interface. A hydrophilic and an organophilic membrane are characterised for the separation of a 2-butanol/water mixture in a wide range of feed temperature and feed concentration in order to establish a generally valid description of the membrane performances. This approach is implemented in the three developed models to simulate the novel membrane module in industrial scale. The simulations are compared to the results of pilot scale experiments conducted with the novel membrane module. Good agreement between simulated and experimental values is reached.

Le mémoire rapporte les travaux effectués pour l’élaboration de membranes asymétriques de type co-polyalcoxyéther-imide (PEI) afin d'obtenir des membranes polymères à haut flux, sélectives pour la séparation de molécules organiques à partir de mélanges aqueux par procédés membranaires. La séparation des mélanges liquides (i.e. toluène - heptane, eau - éthanol, soluté organique dilué en solution aqueuse) a été étudiée par pervaporation (PV) et par nanofiltration (NF) à l'aide de membranes PEI originales asymétriques comportant une peau dense autosupportée. Ces membranes ont été préparées dans des conditions expérimentales contrôlées à partir de solutions DMF-H2O de l'acide polyamique correspondant (APA) en relation avec le diagramme de phase ternaire ; après l’inversion de phase dans un bain d'eau, les membranes d’APA ont été cyclisées en imides par traitement thermique. Les propriétés physiques des membranes (IR, TGA) ont été caractérisées, et les morphologies correspondantes, enregistrées par SEM, ont été utilisées pour optimiser la préparation des membranes asymétriques pour améliorer les propriétés de séparation en ajustant l'épaisseur de la couche dense. Les performances obtenues en pervaporation et en nanofiltration ont été examinées à la lumière de l'influence de trois séries de paramètres, à savoir les paramètres d’élaboration des membranes (composition du collodion, température du bain d'inversion de phase), les conditions expérimentales de perméation (température, pression) et des propriétés moléculaires du soluté (masse molaire, rayon, polarité). Les résultats de pervaporation ont montré que des membranes asymétriques PEI à peau denses pouvaient bien être obtenues, donnant lieu à une sélectivité moléculaire en accord avec le modèle de solution-diffusion. Les résultats obtenus en NF pour des solutés organiques dilués dans l'eau (≈ 500 ppm) ont montré que le degré de rejet des solutés étaient fortement liés aux conditions d’élaboration des membranes PEI et des propriétés des solutés. Les valeurs de seuil de coupure moléculaire des membranes (MWCO) ont été déterminées avec une série de polyéthylène glycol (400 <MW (g/mole) < 6000) pour une pression appliquée allant jusqu'à 10 bar. Il a été montré que le seuil de coupure des membranes était compris entre 400 et 1000g/mol à 30°C. Il a également été constaté pour certaines membranes PEI que de grandes valeurs de flux de perméation associées à de bonnes sélectivités pouvaient être obtenues, conduisant à des performances intéressantes par rapport aux données de la littérature. Ainsi le développement de ces nouvelles membranes asymétriques copolyimides comprenant un bloc élastomère devrait permettre d’obtenir des membranes de hautes performances pour des applications dans les séparations liquide-liquide, en particulier pour les séparations de nanofiltration en milieu aqueux / The work aimed to prepare co-polyalkylether-imide (PEI) asymmetric membranes in order to get high flux water selective polymeric membranes suitable for the separation of organic molecules from aqueous mixtures by membrane processes. The separation of liquid mixtures (i.e. toluene – heptane, water – ethanol and low concentrated organic solute in aqueous solutions) was studied by pervaporation (PV) and by nanofiltration (NF) using homemade integrally skinned asymmetric PEI membranes. These membranes were prepared under controlled experimental conditions from DMF-H2O solutions of the corresponding polyamic acid (PAA) with respect to the ternary phase diagram; after the wet phase inversion in a water bath, the PAA membranes were imidized by thermal treatment. The membrane physical properties (IR, TGA) were characterized and the related morphologies, recorded by SEM, were used to optimize the asymmetric membrane preparation to improve the separation properties by tuning the thickness of the dense top layer. The performances of the pervaporation and nanofiltration separations were examined in the light of the influence of three sets of parameters, i.e. membrane elaboration parameters (dope composition, inversion bath temperature), experimental permeation conditions (temperature, applied pressure) and solute molecular properties (molecular weight, radius, polarity). The PV results showed that tight asymmetric PEI membranes could well be obtained, giving rise to a molecular selectivity in agreement with the solution-diffusion model. The NF results obtained with diluted organics in water (≈500ppm) have shown that the degree of rejection of the organic solutes was strongly linked to the PEI elaboration conditions and to the solute properties. The molecular cutoff values (MWCO) of the membranes were determined with a series of polyethyleneglycol (400 < Mw (g/mole) <6000) for an applied NF pressure up to 10 Bar; it was shown that the PEI membrane MWCO could be ranged between 400 and 1000g/mol at 30°C. It was also found with some PEI membranes that high permeation fluxes together with good separation selectivity could be obtained leading to interesting performances compared to literature data. Thus, it is expected that the development of these new asymmetric block copolyimide rubbery membranes might give rise to high performance membrane systems for applications in liquid-liquid separations, in particular in nanofiltration separations

Cette thèse présente une approche originale du procédé de distillation membranaire avec balayage gazeux pour la concentration des solutions thermosensibles (SGMD). Pour ce faire, un nouveau contacteur membranaire avec des fibres creuses métalliques a été conçu afin réaliser le procédé de distillation à basse température. La chaleur nécessaire au procédé est produite au niveau des fibres par effet Joule, plutôt qu'à partir de chaleur latente de la phase aqueuse. La génération localisée de la chaleur a comme conséquence une réduction du phénomène de polarisation de la température. Des fibres creuses en acier inoxydable ont été synthétisées avec les propriétés structurales appropriées et une bonne résistance mécanique. La surface des pores des fibres a été rendue hydrophobe par le dépôt d'une fine couche d'un élastomère. En outre, une nouvelle méthode « verte » a été développée pour fabriquer des fibres creuses en alumine et acier inoxydable. Cette méthode est basée sur la gélification ionique des bio-polymères et ne n'utilise pas des solvants nocifs. L'étude expérimentale détaillée du SGMD a permis de déterminer l'influence de différents paramètres opérationnels sur les performances du procédé. Il a été démontré que l'effet Joule permet d'améliorer le flux et l'efficacité de la séparation non seulement pour le SGMD mais aussi pour la pervaporation. / This thesis presents an original approach for the concentration of thermo-sensitive solutions: the Sweep Gas Membrane Distillation (SGMD) process. A new membrane contactor with metallic hollow fibers has been designed and allows the distillation process to be operational at low temperature. Heat is generated in the fibers by the Joule effect, rather than being supplied as latent heat in the liquid bulk. The localized generation of heat results in a reduction of temperature polarization phenomena. The stainless-steel hollow fiber membranes have been synthetized with appropriate structural properties and sufficient mechanical strength. The pore surface of the fibers has been made hydrophobic by the deposition of a thin layer of an elastomer. Moreover, a novel and green method is presented to fabricate alumina and stainless-steel hollow fibers. This method is based on ionic gelation of a biopolymer and completely avoids the use of harmful solvents. By a detailed experimental study of the SGMD the influence of different operational parameters on the process performance has been investigated. The improvements in the flux and the separation efficiency using Joule effect have been successfully demonstrated, even in the case of pervaporation.

Ce travail de thèse porte sur la réalisation de MEMS organiques dans un dispositif original, le microévaporateur, couplant la technique MIMIC (Micromolding in Capillaries) à la pervaporation microfluidique. Il est expliqué comment le phénomène de pervaporation peut être utilisé pour concentrer des solutions polymériques diluées jusqu'à l'obtention de matériaux composites dans des géométries de dimensions typiques 25 μm x 100 μm x 10 mm. Il a été montré qu'il est possible d'établir des modèles décrivant cette croissance en excellent accord avec l'expérience et l'étude de l'influence de différents paramètres (concentration, géométrie...) sur la croissance a alors permis de prédire les vitesses de croissance des matériaux composites. Deux systèmes ont été réalisés à partir de ces derniers, associés à deux effets : l'effet bilame thermique et l'effet piezorésistif mettant en avant une preuve de concept d'une nouvelle voie de fabrication des MEMS organiques : la voie microfluidique. Un dispositif plus complexe comprenant également des vannes microfluidiques a permis de programmer des matériaux à gradients de composition dans la longueur de divers matériaux allant des cristaux colloïdaux aux matériaux polymères. / This work deals with the making of organic MEMS within an original device, the microevaporator, coupling the MIMIC technique (Micromolding in Capillaries) and microfluidic pervaporation. It is shown how the pervaporation phenomenon can be used to concentrate polymeric diluted solutions until we obtain composite materials into geometries with typical dimensions about 25 μm x 100 μm x 10 mm. We showed that it is possible to establish models which describe this growth in excellent agreement with experiments and the study of the influence of different parameters (concentration, geometry...) upon the growth thus allowed us to predict the growth velocities of those composite materials. Two systems have been made associated to two effects : bimetallic strip effect and piezoresistive effect in order to demonstrate a new proof of concept of a new way to make organic MEMS using microfluidics. A more complex device including microfluidic valves allowed us to encode materials with a gradient of composition within their largest dimension from colloidal cristals to polymeric materials.

Ce travail de thèse porte sur le développement technologique d’outils miniaturiséspour l’exploration de diagrammes de phase de fluides complexes (dispersions colloïdales,solutions de polymères ou tensioactifs, etc). Les outils élaborés permettent dedéterminer des diagrammes de phase par une approche continue à l’aide de la microfluidique.Ils sont basés sur deux types de procédés membranaires différents : la pervaporation(mécanisme d’évaporation de solvant) et la dialyse (mécanisme d’échangesosmotiques). En s’appuyant sur le processus de pervaporation, il a été montré théoriquementet expérimentalement qu’il existe une géométrie pour laquelle le séchageconfiné est homogène. Il est donc possible de construire des diagrammes de phase demélanges à plusieurs composants de l’échelle moléculaire aux colloïdes. Une étudeconsacrée à la compréhension de la complexité du séchage des nanoparticules de silicecommerciales dans un canal microfluidique de type microévaporateur a été miseen place. La cinétique de concentration des particules est décrite jusqu’à la formationd’un état dense ainsi que les divers phénomènes liés au séchage comme l’existenced’une transition de phase dans un système colloïdal, l’apparition de fractures ou la délaminationdu matériau dense. Un nouvel outil microfluidique intégrant une membranede type dialyse offre la possibilité de contrôler les échanges osmotiques à l’échelle dunanolitre. Le protocole de fabrication ainsi que le dimensionnement de la géométriesont présentés. Grâce à cet outil, il est possible de mesurer des pressions osmotiquesde dispersions colloïdales. / This work deals with the technological development of miniaturized tools for theexploration of the phase diagram of complex fluids (colloidal dispersions, solutions ofpolymers or surfactants, etc). The microfluidic tools we elaborated make it possibleto determine phase diagrams of a series of formulations of complex fluids by consumingonly minute amounts of samples. These devices exploit two types of membraneprocesses to concentrate the chemical species : pervaporation (solvent evaporationthrough a dense membrane) and dialysis (osmotic exchanges through a membrane).Concerning the case of pervaporation, we demonstrated theoretically and experimentallythat a specific microfluidic design exists for which concentration fields of chemicalspecies remain spatially homogeneous along the kinetic path followed withinthe phase diagram. Then, it enables to obtain phase diagrams of multi-componentsmixtures from molecular compounds up to colloids, at the nanolitre scale. We reporta study concerning the understanding of the drying process of commercial silica nanoparticlesusing a dedicated microfluidic experiment involving pervaporation. Wepresent the kinetics of the concentration of the particles within the channel up to theformation of a dense colloidal packed bed which invades the channel at a controlledrate. We developed an original microfluidic tool integrating a dialysis membranewhich makes it possible to control osmotic exchanges at the nanoliter scale. We reportthe protocol of microfabrication of this chip and its specific geometry.We presentpreliminary results showing that this tool can be used to measure osmotic pressures ofcolloidal suspensions.

Ce travail de thèse présente la conception et la réalisation d'un dispositif microfluidique en PDMS (Polydiméthylsiloxane) pour exploration des diagrammes de phase. Le microsystème est basé sur le principe de pervaporation (évaporation à travers une membrane) et comporte des microchambres indépendantes de 5nL de solution dont on fait varier la concentration au court du temps. Il est possible de concentrer jusqu'à l'observation de transitions de phases (démixtion, nucléation, cristallisation, ...). Nous avons montré que la pervaporation est une piste intéressante pour l'exploration de diagrammes de phases. En parallèle, nous avons développé un outil original de mesure in situ en temps réel de la concentration, paramètre essentiel du criblage. Cet outil, basé sur le principe de réfractométrie, a pour avantage d'être non intrusif et ne requiert aucune modification particulière du microsystème. La méthode consiste à utiliser les parois des microcanaux comme éléments optiques. Nous montrons que cette méthode permet de mesurer un coefficient de diffusion et un rapport de viscosité dans une jonction en T sans ajout de traceurs ni utilisation de la fluorescence. Nous avons utilisé cette méthode de mesure de la concentration lors d'expériences sur des systèmes modèles (solutions ioniques, surfactants, polymères, protéines, ...).

Doctor of Philosophy / Department of Chemical Engineering / Mary E. Rezac / Reversible organic reactions, such as esterification, transesterification, and acetalisation, have enjoyed numerous laboratory uses and industrial applications since they are convenient means to synthesize esters and ketals. Reversible organic reactions are limited by thermodynamic equilibrium and often do not proceed to completion. High yields for these equilibrium driven reactions can be obtained either by adding a large excess of one of the reactants, which results a reactant(s)/product(s) mixture requiring a separation, or by the selective removal of by-products. Conventional removal techniques including distillation, adsorption, and absorption have drawbacks in terms of efficiency as well as reactor design. Pervaporation membrane reactors are promising systems for these reactions since they have simpler designs, and are more energy efficient compared to conventional downstream separation techniques. This project created a general protocol that can guide one to carry out experiments and collect necessary data for transferring membrane reactor design concepts to the construction of industrial-scale membrane reactors for organic synthesis. Demonstration of this protocol was achieved by (1) experimental evaluation of membrane reactor performance, (2) modeling, and (3) scale-up. The capability of membranes for water/organic separations and organic/organic separations during reversible reactions was investigated. Our results indicated that enhanced membrane reactors selectively removed the by-product water and methanol from reaction mixtures and achieved high conversions for all investigated reactions. Second, modeling and simulation of pervaporation membrane reactor performance for reversible reactions were carried out. The simulated performance agrees well with experimental data. Using the developed model, the effects of permeate pressure and membrane selectivity on membrane reactor yield were examined. Finally, a scale-up on transesterification membrane reactors was carried out. The membrane modules investigated included a bench-scale flat sheet membrane, a bench-scale hollow fiber membrane module, and a pilot-scale hollow fiber membrane module. A 100% conversion was obtained by the selective methanol removal. It is found that with high methanol selectivity membranes, the reaction time to achieve a given conversion continuously decreases with increasing the methanol removal capacity of the reactor system. However, this is a highly nonlinear relationship.

Zeolite membrane synthesis is conventionally carried out in batch systems. Recently, several attempts have been performed to synthesize zeolite membranes in flow systems which can allow preparation of membranes with large specific surface areas. Membranes synthesized in the recirculating flow system had comparable N2/SF6 and n- C4H10/i-C4H10 ideal selectivities with the membranes prepared in the batch system, indicating that good quality membranes can be produced by this method. The objective of this study is to separate organic/water mixtures by pervaporation by using MFI type membranes synthesized in the flow system. Effect of number of synthesis steps and synthesis method on the separation factor and flux was investigated. Membranes were synthesized from clear solutions with a molar composition of 80SiO2:16TPAOH:1536H2O at 95oC and atmospheric pressure. The synthesis solution was recirculated through the tubular alumina support with a flow rate of 6 ml/min for 72 h. The membranes were characterized by X-ray diffraction for phase identification and scanning electron microscopy for morphology determination. Single gas permeances of N2, H2, CH4, CO2, n-C4H10 and i-C4H10 were measured between 25 and 200oC. Mixtures of 5 wt% ethanol/water, 2-propanol/water and acetone/water were separated by pervaporation at different temperatures. The single gas permeances decreased with increasing temperature for weakly adsorbed gases. For n-C4H10 the permeance passed through a maximum and i-C4H10 permeance was nearly constant. For a membrane synthesized by two consecutive synthesis steps, the ideal selectivity for n-C4H10/i-C4H10 was 132 at 200oC. The selectivity in the pervaporation separation of ethanol-water mixture was 43 with a permeate flux of 0.2 kg/m2h at 25oC. With increasing temperature, selectivity decreased but the flux increased, the selectivity was 23 and the flux was 1.9 kg/m2h at 85oC. 2-propanol/water and acetone/water separation factors were 36 and 1024 with 0.2 and 0.1 kg/m2h fluxes, respectively. The separation factors and fluxes for membranes synthesized in the flow system were comparable with membranes synthesized in the batch system.

Zeolite A-type membranes are usually synthesized from hydrogels and rarely synthesized from clear solutions mostly in batch systems. Few studies were carried out using semi-continuous systems for zeolite A membrane synthesis. Zeolite A membranes are mainly used in pervaporation processes for separation of water from water/organic mixtures because of their hydrophilic property. In this study, zeolite A membranes were synthesized on -alumina supports from a clear solution with a molar composition of 49Na2O: 1Al2O: 5SiO2: 980H2O. Synthesis was done both in a batch system and in a flow system in which solution was circulated through the support under atmospheric pressure. Effects of synthesis temperature, time, flow rate and seeding on membrane formation were investigated. The membranes were characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), single gas permeation measurements and pervaporation tests. In batch system, pure zeolite A membranes having cubic form of zeolite A was obtained for the syntheses carried out at 60&deg / C for 24 h and 80&deg / C for 8 h. Thicknesses of the membranes synthesized at 80&deg / C and 60&deg / C were about 2 &micro / m and 4 &micro / m, respectively. N2 permeances were 2*10-8 mol/m2sPa and 8*10-8 mol/m2sPa for of the membranes synthesized in the batch system at 60&deg / C and 80&deg / C, respectively. When synthesis was carried out in flow system pure and continuous zeolite A membranes were obtained for all conditions. Membranes synthesized at 60&deg / C and 80&deg / C had thicknesses of about 1.5 and 2 &micro / m, respectively. Lower N2 permeations were obtained for the membranes synthesized in flow system. It was observed that flow rate and seeding did not significantly affect the thickness of the membrane layer. The membranes synthesized in this study are significantly thinner than the membranes reported in the literature. Single gas permeation tests at 25&deg / C for the membranes showed that comparable membranes with the ones in literature were obtained in this study. For a double layer membrane synthesized in flow system at 80&deg / C for 8h separation factor about 3700 was obtained for the separation of 92:8 (wt.%) ethanol/water mixture at 45&deg / C.

Zeolites are microporous hydrated aluminosilicate crystals containing alkali and/or alkali earth metal cations in their frameworks. Due to their molecular size pores, they can separate molecules according to their size and shape. Zeolites are mostly used in ion exchange, adsorption processes and catalytic applications. The hydrophilic/hydrophobic character of zeolites also makes them favorable materials for adsorption based separations. Recently the potential of zeolite/ceramic composite membranes have been shown in the separation of liquid and gas mixtures. Self-supported zeolite membranes with asymmetric structure can be an alternative to the composite zeolite membranes. Because asymmetric structure may eliminate the problems originated from the differences in thermal expansion coefficients of zeolites and ceramics. In this study tubular zeolite A membranes were prepared on binderless zeolite A supports. The supports were perepared by hydrothermal conversion of amorphous aluminosilicate tubes into zeolite A. The amorphous aluminosilicate powder, which was obtained by filtering the homogenous hydrogel with a composition of 2.5Na2O:1Al2O3:1.7SiO2:150H2O, was mixed with an organic binder (HEC-Hydroxyethyl Cellulose) and water to obtain the paste. The paste was then extruded through a home-made extruder into bars and tubes. These extrudates were dried at room temperature for 24 hours, calcined at 600oC for 2 hours to remove organic binder and finally synthesized at 80oC for 72 hours in hydrothermal conditions to convert amorphous aluminosilicate to zeolite. The effect of composition of the synthesis solution on the crystallinity and morphology of zeolite A tubes and bars were investigated. The crystallization field of zeolite A bars has been established and shown on a ternary phase diagram. Tubes were mechanically stable, typically had a crystallinity over 90% and a macroporosity of 35%. The tubes were composed of highly intergrown crystals of zeolite A. The average particle size was 3.5 &micro / m. The asymmetric membranes were synthesized by growing zeolite A films on binderless zeolite A supports with a geometry of disk, bar and tube. Continuous zeolite A films can only be obtained when the supports were saturated with water prior to synthesis. The film thicknesses were approximately 5 &micro / m on disks and approximately 10 &micro / m on tubes. A method was proposed to prepare self-supported tubular zeolite A membranes in this study.