Würzburg, Univ., Diss., 2009. / Zsfassung in engl. Sprache.

Th. doct.-ing.--Génie chimique--Toulouse--I.N.P., 1984. N°: 334.

Th. doct.-ing.--Nancy, I.N.P.L., 1981.

Th.--Génie chim.--Toulouse 3, 1982. N°: 1036.

Thesis (MTech (Chemical Engineering))--Cape Peninsula University of Technology, 2006 / A protocol already exists for fabrication of a capillary membrane having an internal ultrafiltration skin supported by a finger-like pore structure in the external capillary wall (Jacobs and Leukes, 1996; Jacobs and Sanderson, 1997). These membranes have been produced at the Institute of Polymer Science, University of Stellenbosch, South Africa. Two major applications emerged from the development of these internally skinned membranes. One application was in the production of potable water by Ultra-filtration (UF) from sources containing coloured water. A second application was in the immobilization of a white rot fungus in a ."gradostat" membrane bioreactor. Here a nutrient gradient through the membrane wall and fungal mat can be established and manipulated in order to stimulate continuous production of secondary metabolites (extra-cellular enzymes). These enzymes are useful in the degradation of polycyclic aromatic compounds, notably PCB species in contaminated water and soils (Jacobs and Sanderson, 1997). Two objectives emerged from experiences with the above applications. The first objective was to improve membrane performance in UF applications. In this case a reduction was sought in trans-membrane pressure differential required to attain a desired flux without sacrificing rejection. The pressure required for a given desired flux across a membrane depends on the resistance of the membrane skin layer and of its supporting sub-layer which together comprises the capillary wall and defmes its overall structure. If any of these resistances could be reduced, the overall resistance to transport of water would be reduced. Then it would be possible to operate the membrane at lower trans-membrane pressure differences. On the other hand, operation with higher pressure would also increase flux but require a thicker capillary wall to resist this pressure. In the attempt to optimise these properties of the capillary membrane, capillary membranes produced in the study reported here were tested to find the relationship of flux performance with the structures that resulted from varying key parameters affecting structure and integrity. The objective in the case of immobilizing fungi in membrane bioreactor applications was to attain thicker walls thus providing better support for the fungal mass. The internally skinned capillary membrane has finger-like microvoids that start next to the UF skin layer and extend across the capillary membrane wall and open at the external membrane periphery, giving an ideal structure for retaining the fungal biomass. The idea of a membrane with this type of morphology to immobilize white rot fungi was to anchor the growing fungus within these microvoids which imitate the natural environment in which these organisms live, that is, in the fibrous structure of decaying wood. The requirement to inoculate the microvoids with fungal spores (reproductive cells), implies that they need to be accessible from the outside, requiring a membrane wall that is externally unskinned. In the formation ofthe capillary membrane the processes of formation of the porous UP skin and the finger-like microvoids are mainly governed by diffusion of solvent out of a polymer dope (gel phase) and of non-solvent into the dope phase. Such exchanges are of primary importance between the bore fluid (containing non-solvent) and dope (containing solvent) or between the external spinning bath (high in solvent content) and dope. Diffusion effects also occur between the nascent pore voids and the precipitating polymer matrix. There are also expected to be some convection effects due to shear between the bore fluid and the moving dope gel phase and due to shrinkage ofthe gel phase. The variables selected for experimentation m the study reported here were: the dope extrusion rate (DER); dope composition (viscosity effects); bore fluid flow rate (BFF); bore fluid composition and wall thickness and diameter effects (determined largely by spinneret dimensions). Each of these has an expected effect on membrane structure and its resulting performance. Most were varied over narrow ranges indicated in the literature and by experience to be effective and critical. In addition, the effects of altering the walI thickness were investigated by using two different spinneret sizes. The external spinning bath composition (solvent content) was reported in the literature to be a particularly important parameter in the formation of externally unskinned membranes. Maintaining a high content of solvent in the external spinning bath could prevent skin formation. Too high a solvent content could, however, prevent phase transition and lead to later precipitation ofa dense skin on contact with the non-solvent in the later (humidification and rinsing) steps in the fmishing of the capillary membrane product. The external bath composition was therefore varied so as to find the bath composition that would match the cloud point for the polymer dope employed. As expected, the thickness of the membranes increased with DER increase. However, it was found that there is a critical wall thickness where an external skin layer is formed as a result of increasing the DER. A certain volumetric ratio ofDER to BFF (1,5:1 for this study) was therefore maintained in order to produce externally unskinned membranes. This shows that although the final membrane structure is detennined by the casting dope formulation, the fabrication protocol plays an equally important role in controlling structural properties and perfonnance. There was no significant change with the membrane thickness as a result of changing BFF but the voids became longer and more in number as the BFF was increased. Too high solvent content (99% NMP in this study) resulted in an external skin layer being formed. According to Smolders et.al. (1992), when the solvent content in the external spinning bath is too high, the polymer at the surface of the newly fonned membrane slowly dissolves in the external spinning bath re-forming a dope-like solution. When the newly formed membrane passes through the humidifier, the dope-like solution solidifies to form an external skin. At the same instance, too low solvent (93% for this study) resulted in external skin being fonned. Externally unskinned membranes were formed at 94 and 96% NMP bath composition. The use of a small spinneret resulted in very thin walled externally unskinned membranes.

Thesis submitted in partial fulfillment for the Degree of Master of Technology: Chemical Engineering, Durban University of Technology, 2010. / THE OBJECTIVE of this study was to devise a purification process, using ultrafiltration membranes as the core technology, for the preparation of corn steep liquor (CSL) as a fermentation feedstock. This process inherently required the development of a pretreatment system for the ultrafiltration membranes for the removal of suspended solids and high fouling material from corn steep liquor. The ultrafiltration membrane system was required for the separation and removal of colloidal solids from corn steep liquor, and to fractionate and separate out unwanted proteins, to render the feedstock suitable for sterilisation and subsequent fermentation. THE CONCEPT of membrane technology was investigated in order to find a more practical alternative for what was deemed to be a difficult process problem. In particular, various pretreatment technologies were investigated to form a compact and robust process package. THE CORN STEEP LIQUOR, a by-product of the corn wet milling process, was obtained from African Products, Germiston, in the form of a concentrated slurry directly from an evaporator system. A diagnostic of the feedstock was carried out and from this information, it was decided that three pretreatment options would be investigated. The first option was the pH treatment of the corn liquor, by the addition of ammonia which induced the precipitation of solids. This was followed by liquid-solid separation, and the clarified liquor was fed to the membrane system. The second option looked at the separation of suspended solids from the liquor by the use of broth conditioning additives and separation of the solids by a decanter centrifuge. The third option investigated was the use of a gyratory screening system for the removal of all solids greater than 100 μ in size. IN THE pH TREATMENT of CSL, the process is effected by the addition of base to pH 7. The technology involves neutralisation of CSL in a mixing system, under predetermined conditions of temperature, agitation and rate of addition, followed by subsequent liquid-solid separation. Trials were conducted on a pilot plant to test the process. Initial trials, conducted on a small scale pilot filter press, proved to be successful for this application. A suspended solids removal of up to 98% was achieved. The average suspended solids in the filtrate was found to range between 0.1 to 0.25 %. Tests were also conducted on a hired “state of art” filtration plant under various conditions. A diaphragm membrane press was found to provide the best performance. Protein recoveries of above 95 % at fluxes of 35 L/m2h at temperatures above 50 °C, and an incremental application of feed pressure was most suited for the process. The removal of the colloidal solids by the above-mentioned process was found to improve the quality of sterilisation. A reduction of more than 90 % in coagulated solids was achieved. v i IT WAS OBSERVED that the separation of suspended solids from CSL is enhanced by the use of coagulation and flocculation. Although not commonly used for this purpose, it was felt that a decanter centrifuge was well suited for the subsequent separation of the flocs from the clarified liquor. This work describes the results of the trials with such a device and the impact of broth conditioning on the efficiency of the separation. Trials have been conducted using an Alfa-Laval Model NX210 decanter, which was not specifically built for the work and therefore imposed several limitations on its performance. Despite these shortcomings, preliminary trials proved to be successful in achieving the separation objective. Tests were conducted using five different batches of CSL. With a maximum suspended solids loading of 4.3 % and a feed rate of 700 L/h, a solids recovery of 90 % was achieved. The clarified liquor contained residual solids between 0.5 and 0.8 %. The sludge had a solids concentration that ranged between 43 % and 65 %. COAGULATION AND FLOCCULATION dosages were kept within the limits of the laboratory evaluations. Flocculant dosages were controlled between 100 and 200 ppm, with the coagulant operating at higher dosages of between 400 and 2000 ppm. The only controllable parameter on the machine itself was the scroll differential speed. The best performance in terms of the cake dryness and centrate clarity was obtained at the lowest scroll differential speed of 4 rpm. THE USE OF GYRATORY SCREENS entailed passing the raw liquor through a set of two screens. The technology involves the use of a gyratory mechanism, which aids in the cleaning of the screens during continuous operation. Trials have been conducted on a pilot plant to test the system. Since the unit used was designed specifically for quick on-site screening exercises, it did not possess the added flexibility and robustness of a properly designed full scale unit. This imposed some limitations on its performance. However, despite these shortcomings, the trials conducted on the pilot plant proved to be successful in meeting the outlined objectives. A NUMBER OF TRIALS were performed on various batches of CSL. There was considerable batch to batch variation in the suspended solids content of the CSL and this was found to ultimately affect the throughput of the screening process. The feed suspended solids varied between 10 and 18 %. The highest throughput achieved was 400 L/h at a feed suspended solids loading of 14.5 %. It was found that temperature made a significant impact on the separation. The loss of heat in the feed stream caused excessive coagulation to occur thus increasing the suspended solids loading and lowering the throughput. The total solids in the sludge stream varied between 45 and 77 %. Protein loss in the sludge stream was around 1 %. Careful attention had to be given to the handling of the sludge stream. This stream displayed rheological characteristics typical of a non-Newtonian thixotropic fluid. The 100 μm screen operated best vi i when prior separation was done using a 180 to 200 μm screen. This reduced the solids loading on the tighter screen and increased the throughput by 10 to 15 %. The self cleaning mechanism also performed more efficiently under these conditions. THE SELECTED OPTION was then based on the influence the operation had on the ultrafiltration membranes, sterilisation of the product prior to fermentation and ultimately the fermentation performance. Subsequent testing of the pretreatment options were performed on an ultrafiltration membrane test cell. The product from the gyratory screens were found to produce the best overall results, where the highest fluxes and least amount of fouling occurred on the membranes tested. ONCE THE PRETREATMENT OPTION was decided, the development of the membane ultrafiltration system was then pursued. Trials were conducted on a laboratory scale, in a membrane test cell, to determine the preliminary screening of the membrane type, fouling effects and fluxes. It was found that polyvinylidene and polyacrylonitrile membranes produced the best overall fluxes of 11.25 and 10.96 L/m2h respectively. These membranes produced permeate protein concentrations of 121 and 115 g/L respectively. Sterilisation tests conducted on the permeate streams produced also showed that these two membranes had the lowest suspended solids concentrations. FERMENTABILITY tests conducted, showed that the ultrafiltered CSL, from these two membranes, produced increased cell counts and protein utilisation along with an increased product yield. Approximately 42 g/L of biomass was generated with lysine yields of 46 g/L. Further testwork revealed the non-Newtonian nature of CSL and its inherent viscosity effects. BENCH-SCALE testwork was conducted for various membrane configurations. With tubular membranes and hollow fibre membranes, average fluxes of 6.23 and 4.5 L/m2h were achieved respectively. Spiral wound membranes were found to be more consistent in their performance, with average fluxes of around 6.25 L/m2h. For the spiral wound membranes, it was found that the Desal-2 mesh spacer with a 80 mil thickness was most appropriate for the duty. PILOT PLANT testwork was conducted to scale-up the membrane system and to eliminate possible risks associated with the technology. The pilot plant studies showed up a number of principle design variables which needed careful attention. The flaws in the piloting system were subsequently rectified and this helped to improve the overall performance of the system.

Bovine colostrum collected within 12h of parturition was de-fatted, de-caseinated, and ultrafiltered (UF) using a 5 kDa cut-off membrane; the resulting UF permeate was freeze dried to create a powder with possible use as a functional food ingredient. Samples representative of five lots of this powdered “colostrum low molecular weight fraction” (CLMWF) were analyzed for chemical composition and nutrient profile. The average contents of fat, moisture, and ash were 0.6%, 1.7%, and 8.3% w/w, respectively. Carbohydrate analysis showed an average of 58.2% w/w lactose monohydrate with no monosaccharides, other disaccharides, trioses, or tetroses detected. The total nitrogen content averaged 1.13% w/w, with 74% of this in the non-protein nitrogen fraction, producing a true protein content of 1.9% w/w. A significant mass fraction of the material (~29% w/w) remains to be characterized. The CLMWF powders were found to contain significant quantities of the minerals calcium (average 870 mg/100g), magnesium, (311 mg/100g), phosphorus (1473 mg/100g), potassium (1705 mg/100g) and sodium (690 mg/100g), the nutrients taurine (average 26.5 mg/100g), L-carnitine (40.5 mg/100g), thiamine (648 mcg/100g) and riboflavin (6991 mcg/100g), and the nucleos(t)ides uridine (55.2 mg/100g) and 5’UMP (18.8 mg/100g), cytidine (3.33 mg/100g) and 5’CMP (4.83 mg/100g) and guanosine (3.45 mg/100g) and 5’GMP (3.57 mg/100g).

Das Ausmaß der Proteinbindung eines Arzneistoffs wirkt sich auf viele unterschiedliche pharmakokinetische Parameter aus. So wird beispielsweise das Verteilungsvolumen, die Metabolisierung oder die Elimination des entsprechenden Stoffes durch die Höhe seiner Proteinbindung beeinflusst. Da nur der im Plasma frei vorliegende Anteil eines Arzneistoffs in der Lage ist biologische Membranen zu überwinden, können auch nur die freien Arzneistoffmoleküle eine pharmakologische Wirkung an Rezeptoren oder Enzymen auslösen. Dementsprechend ist auch die Intensität der hervorgerufenen Wirkung von der Größe des ungebundenen Anteils eines Arzneistoffs abhängig. Aufgrund dieser Zusammenhänge ist klar, dass die Proteinbindung eines Arzneistoffes letztendlich Einfluss auf die Dosisfindung hat. Zur Ermittlung der Proteinbindung stehen viele unterschiedliche Methoden, wie beispielsweise die HPLC, Kapillarelektrophorese, Ultrazentrifugation, Gleichgewichtsdialyse und Ultrafiltration zur Verfügung. In der vorliegenden Arbeit wurde die kontinuierliche Ultrafiltration zur Ermittlung der Proteinbindung von Arzneistoffen angewendet. Hier wird die Proteinbindung nicht nur anhand einer bestimmten Arzneistoff- bzw. Albuminkonzentration gemessen, sondern über einen weiteren Bereich von Wirkstoff-Protein-Verhältnissen beobachtet. Des Weiteren ist der apparative Aufwand im Vergleich zu vielen anderen Methoden als geringer einzustufen. Im Rahmen dieser Arbeit wurde, die auf der von Heinze[122] entwickelte Messanlage weiter optimiert und eine zweite Anlage mit einem Diodenarraydetektor aufgebaut. Für letztere musste eine Software-Anpassung vorgenommen werden. Folgende Projekte wurden durchgeführt: 1) Um den In-vivo-Bedingungen nahe zu kommen, wurde bei der Bestimmung der Proteinbindung der Sartane nicht nur BSA und HSA verwendet, sondern erstmals auch humanes Plasma. Die Plasmamessungen der Sartane verliefen insgesamt problemlos, allerdings ist eine erfolgreiche Messung stark von der Qualität des eingesetzten Plasmas abhängig, wie Messungen der Naphthylisochinoline gezeigt haben. Im Vergleich mit HSA und Plasma ergaben die Messungen der Sartane mit bovinem Serumalbumin geringfügig erniedrigte Proteinbindungswerte. Insgesamt sind alle Ergebnisse sehr gut mit den Literaturwerten vergleichbar. 2) Das Ausmaß der Proteinbindung von Naphthylisochinolinen war bislang unbekannt und lag im Bereich von ca. 30-70%. Erneut waren die Resultate aus den Messungen von BSA und HSA nahezu gleich. 3) Am Beispiel der Interaktion zwischen Phenprocoumon und Phenylbutazon wurden zwei unterschiedliche Ansätze getestet, um die Verdrängung aus der Proteinbindung zu simulieren. Die erste Methode entsprach hierbei einer Konkurrenz der beiden interagierenden Stoffe um die Proteinbindungsstellen. Durch den Einfluss des Phenylbutazon verringerte sich die Proteinbindung des Phenprocoumon um 1%, was allerdings als statistisch nicht signifikant betrachtet werden kann. Im zweiten Ansatz, der eine direktere Verdrängung aus der Proteinbindung simulieren sollte, fiel die Proteinbindung des Phenprocoumon gegenüber den Einzelmessungen um 2,5% ab. Unter physiologischen Konzentrationsverhältnissen sank sich die Proteinbindung des Phenprocoumon auf 93,3%. Der freie Anteil erhöhte sich dementsprechend von 1% auf 6,7%. Somit konnte der Einfluss des Phenylbutazon auf die Proteinbindung des Phenprocoumon erfolgreich nachgewiesen werden. Die unveränderte Proteinbindung des Phenylbutazon im inversen Ansatz und die ermittelten pK-Werte bestätigen diese Interaktion. Grundsätzlich ist es also möglich mit der kontinuierlichen Ultrafiltration solche Interaktionen zu simulieren. 4) Zuletzt sollte der Frage nachgegangen werden, ob es mit der kontinuierlichen Ultrafiltration auch möglich ist die Proteinbindung von wasserunlöslichen Stoffen, nämlich den Aziridinen, in Gegenwart steigender Mengen DMSO, zu bestimmen. Die erhaltenen Ergebnisse wurden mit Literaturwerten ohne DMSO-Zusatz verglichen. Abgesehen von Candesartan, das eine lineare Korrelation zwischen DMSO-Gehalt der Wirkstofflösung und Absinken der Proteinbindung zeigte, konnte kein Zusammenhang zwischen der DMSO-Konzentration und der gemessenen Proteinbindung festgestellt werden. Die Mittelwerte lagen im Bereich der Literaturwerte. Insgesamt zeigten alle Versuchsreihen, dass die kontinuierliche Ultrafiltration eine ausgezeichnete, schnelle und robuste Screeningmethode zur Bestimmung des Ausmaßes der Proteinbindung bekannter und neuer Wirkstoffe darstellt. / The extent of protein binding has an impact on many pharmacokinetic parameters, e.g. absorption, distribution volume, metabolism or elimination of a drug. Since only the unbound fraction of a drug can penetrate through biological membranes, only the free drug molecules can induce a pharmacological effect on receptors or enzymes. According to that, the intensity of the effect depends also on the extent of the free drug fraction. Finally, knowledge about the extent of protein binding of a drug is important for the dosage finding. There are many different methods described for the evaluation of the extent of protein binding of a drug, like HPLC, capillary electrophoresis, ultracentrifugation, equilibrium dialysis and ultrafiltration. In this study the continuous ultrafiltration was used to determine the extent of protein binding of drugs. Compared to the discontinuous ultrafiltration, the extent of protein binding was assessed over a wide range of drug-protein-ratios and not only with one defined drug respectively albumin concentration. Here the ultrafiltration instrument described by Heinze[122], was modified and a second system with a multiwavelength detector was established. In this context the software was adapted in a few details. The following experiments were performed: 1) To get close to in vivo conditions, the extent of protein binding of the sartans was determined for the first time by means of human plasma in addition to experiments with HSA and BSA. In general, the evaluation of the protein binding was not problematic. Nevertheless the experiments with the naphthylisoquinolines showed, that a successful experiment with human plasma depends on the quality of the plasma. Compared to the measurements with human serum albumin and plasma, the determination of the protein binding of the sartans with bovine serum albumin showed slightly lower protein binding values. However, the extent of the protein binding of the sartans with BSA, HSA and plasma was in good accordance to values reported in the literature. 2) The extent of the protein binding of the naphthylisoquinolines was unknown so far, and was found to be in the range between 30-70%. Once more, the results of the experiments using BSA were confirmed by the measurements with HSA. 3) To simulate the displacement from the albumin, two different methods have been developed. Due to their well known interaction, phenprocoumon and phenylbutazone were used as test substances. In the first method, the two substances compete for the protein binding sites. Due to the influence of phenylbutazone, the extent of protein binding of phenprocoumon decreased by 1%. However, this decrease is statistically not significant. The second method simulated a direct displacement out of the protein binding. Compared with the single measurements, in this experiments the extent of protein binding of phenprocoumon decreased by 2.5%. With use of a physiological concentration ratio, the protein binding of phenprocoumon decreased from 99.0 to 93.3%. Indicating, that the free fraction of phenprocoumon increased from 1% to 6.7%. Thus, the interaction between the two substances was demonstrated by this method. The constant protein binding of phenylbutazone in the inverse approach and the determined pK-values support this result. 4) Last, the question should be answered, if it is possible to determine the extent of protein binding of water insoluble substances, namely the aziridines, by means of continuous ultrafiltration. For this purpose, five test substances were dissolved in a buffer solution with 1-10% DMSO. The results of the experiments were compared to literature values without DMSO. Candesartan showed a linear correlation between the DMSO-concentration and the extent of protein binding. The results of the other four substances indicated no correlation between the content of DMSO in the solution and the protein binding values. However, in all cases the average values were in accordance to the literature data. Overall, every project showed, that the continuous ultrafiltration is an excellent, fast and robust screening method for the evaluation of the extent of protein binding of known as well as new substances.

The performance of membranes in various processes is largely dependent on their morphological properties. Thus, membrane structure has been continuously optimised for different applications. Anodic alumina membranes (AAMs) exhibit self-ordered pore structure and the pore size can be tuned in the sub-micrometre range. The aim of this PhD project is to propose and develop AAMs for the applications of membrane filtration and emulsification with potential for scale-up. In the project, the AAMs were initially fabricated in flat sheet form to optimise the process parameters to obtain membranes with a high quality of pore structure. The membrane pore diameter can be readily controlled by the anodization voltage. While AAMs are normally symmetric, by manipulating the anodization voltage, asymmetric AAMs consists of stem pores and active pores have been successfully made. After that, the flat AAMs with symmetric and homogeneous structure were used as a platform to study for surface modification and fluid transport in nano-channels. The surface chemistry and wettability of the membranes has been altered by grafting of silane molecules and carbon coating by chemical vapour deposition. Fluid flow measurement through pristine AAMs with pore diameter in the 20 nm to 100 nm range shows flow enhancement effect, experimentally for the first time, can occur in hydrophilic materials. Subsequently, tubular AAMs were fabricated using aluminium alloy tubes, to be assessed for ultrafiltration and membrane emulsification processes. The pore structure of the tubular AAMs was analogous to flat membranes. Despite the reduced pore circularity and hexagonal arrangement originated from the presence of impurities in the starting materials, the narrow pore size distribution was not compromised. In a selectivity-permeability analysis, the asymmetric tubular AAMs outperformed most of the commercial ceramic membranes but their flux was very low when compared to polymeric membranes. A bovine serum albumin filtration test showed that complete pore blocking-cake filtration model can be used to describe the fouling behaviour. Finally, symmetric tubular membranes were used to study dead-end and cross-flow emulsification processes. The resulting emulsions show low polydispersity. Using a membrane with 25 nm average pore diameter, the obtained average droplet size was as low as 120 nm during a cross-flow emulsification. This is by far the smallest achieved average droplet size by cross-flow membrane emulsification.

No description available.