Preadsorption of hydrophilic polymers on hydrophobic ultrafiltration membranes can reduce the susceptibility of the modified membranes to protein fouling. The mechanisms of this anti-fouling action were investigated. Polysulfone and nuclear track-etched membranes with different average pore diameters were hydrophilized by preadsorption of two water-soluble polymers. The fouling of the unmodified and modified membranes due to filtration of a whey protein solution or to adsorption of the whey protein at the membrane surface was characterized by flux measurements and by electron microscopy. Adsorption of protein at the pore walls of ultrafiltration membranes, resulting in the narrowing of pores, is prevented by partly sealing off the pore entrances by polymer molecules presorbed at the external membrane surface. The observed blockage of pores of microfiltration membranes cannot be averted by the preadsorption technique.

The mechanisms of whey-protein fouling of polysulfone ultrafiltration (UF) membranes were studied. The results of whey-protein adsorption at a nonporous polysulfone model surface were compared with permeate flux measurements of polysulfone UF membranes fouled by filtration of a whey-protein solution. The effect of presorbing the model surface and the membranes with anionic, non-ionic and cationic polymers and surfactants on the adsorption of whey proteins and the membrane flux decline was established. Protein adsorption was found to be a major cause of the observed decline in membrane flux. Especially non-ionic, hydrophilic polymers were found to minimize protein adsorption as well as to decrease the membrane resistance during UF, while the application of surfactants and ionic polymers was generally less successful. Some of the observed effects of surface presorption can be explained by considering physico-chemical interactions between the protein and the modified surface.

Replacement of a considerable part of the traditional, aqueous reaction medium in biotechnology by an organic medium is a promising technique to broaden the scope and range of biotechnological processes. This seems especially to be true for the conversion of non-polar substances. The high capacity of solvents for sparingly water-soluble substrates and products could reduce the required volume of the reaction mixture significantly, and may also lead to less substrate and/or product inhibition in the aqueous biocatalyst phase, when these mechanisms are involved. Furthermore, the use of an organic solvent could shift reaction equilibria favourably and facilitate down-stream processing. In chapter 1 a general review is presented of non-aqueous solvent systems in biocatalytic processes. Special attention is paid to two-liquid-phase systems, involving water-immiscible solvents. Several facets of these biphasic systems have been studied in this thesis using the epoxidation of propene by gel-entrapped Mycobacterium cells as a model. After the description of the throughout this work employed techniques of gas-analysis automatization and of substrate-level control (chapter 2), the far-reaching consequences of the solvent choice are treated in chapter 3. Many solvents cause rapid inactivation of the free, propene-epoxidizing cells. This appears also to be the case if the cells are immobilized in calcium alginate. However, the support material prevents direct cell-organic solvent contact and the associated aggregation and clotting of cells, mostly accompanied with loss of activity. High activity retentions of the immobilized cells relate to low polarities and high molecular weights of the used solvents. The polarity, as expressed by the Hildebrand solubility parameter, is also useful for describing the solvent capacity for one of the two substrates, oxygen, and for the product, propene oxide. The capacity for propene is less well described by the Hildebrand solubility parameter, but also less relevant, as the capacity of the solvents for propene is always about two orders of magnitude higher than that of water, and thus limitation of the rate by unsufficient supply of propene is less likely to occur. It is stressed that optimization of the solvent polarity is necessary, as the requirement of a high activity retention conflicts with the need for a high solvent capacity for the polar propene oxide. Optimization of the polarity will also be likely in case of other types of two-liquid-phase bioconversions. External and internal-diffusion limitations, which are to be expected when using cells entrapped in a hydrophilic: gel, are quantified in chapters 4 and 5. With negligible product inhibition, satisfactory predictions of the mass-transfer effects on the intrinsic Michaelis-Menten kinetics of the immobilized cells are obtained by using a simple pore-diffusion model (chapter 4). Internal diffusion is found to severely limit the epoxidation rate. A more complex model for the intrinsic epoxidation kinetics has been derived for modelling of mass-transfer rates in case of product inhibition (chapter 5). The microkinetic model defined in chapter 4 is integrated in a macrokinetic model to describe the behaviour of a packed-bed immobilized-cell reactor (chapter 6). Depletion of the limiting substrate, oxygen, along the length of the bioreactor can be prevented by using an organic solvent, n-hexadecane, as the transport medium. It is argued that this finding may eliminate the need for a separate gas phase in the fixed-bed reactor. Model predictions of the oxygen conversion in the bioreactor at various degrees of external and internal-diffusion limitation, at various liquid space times and with water or n-hexadecane as the continuous phase are in good agreement with experimentally obtained values. In chapter 7 some other, main limitations of the epoxide production in the packed-bed organic-liquid-phase/immobilized-cell reactor are quantified. Product inhibition is reduced by absorption of the inhibitory epoxide in a cold di-n-octyl phthalate phase. The stability of the immobilized cells is increased by supplying the cells alternately with propene and a co-substrate (ethene). About 50 g dry weight of cells in a 1.7 dm 3 packed-bed reactor were used, which produced ~ 1.5 g chiral propene oxide; two third of the epoxide was absorbed in the octyl phthalate phase. Finally, in the last chapter of this thesis a general discussion is presented. The significance of optimization of the solvent polarity and of the interphase polarity, i.e. the polarity of the phase between biocatalyst and organic solvent is underlined. In case of entrapment in prepolymers, the hydrophobicity/hydrophilicity balance of the gel can be optimized with respect to polarities of substrates and products. Several features of hydrophilic and hydrophobic gels are compared. A quantitative illustration is given concerning the design on a technical scale of a fixed-bed organic-liquid-phase/immobilized-cell reactor. The advantages of using solvents with a high substrate capacity (often oxygen in case of aerobic processes) are demonstrated.

AbstractThe production of propene oxide from propene and oxygen by non‐growing cells entrapped in calcium alginate was used to study the behaviour of a packed‐bed immobilized‐cell reactor operated with an organic solvent as the substrate reservoir. As a result of the high solubility of propene in the solvent used, n‐hexadecane, oxygen was considered to be the limiting substrate. Dilution of the biocatalyst bed with small glass particles appeared necessary to attain a high liquid/solid contacting efficiency between the hydrophilic gel particles and the hydrophobic solvent. The bed dilution had the advantage of avoiding bed compaction and reducing pressure drops. The use of an organic solvent as the transport medium prevented oxygen depletion along the length of the packed‐bed reactor. This eliminated the need for a separate gas phase in the bioreactor. A mathematical reactor model was developed to describe the combined effects of contacting pattern and external and internal diffusion limitations on the instrinsic kinetics of the immobilized cells. Experiments with the packed‐bed immobilized‐cell reactor were performed using an aqueous solution or n‐hexadecane as the reaction medium. Predicted oxygen conversions compared favourably to the observed values without the need for fitting factors.

Some major restrictions of the production of propene oxide in an organic liquid-phase immobilized cell packed-bed reactor were quantified, and techniques were investigated to enhance the epoxide production rates. Propene-epoxidizing Mycobacterium cells were entrapped in calcium alginate gel and contacted with the substrates, propene and oxygen, which were dissolved in a continuous organic phase, n-hexadecane. The effects of product inhibition by the toxic epoxide—microbial consumption of propene oxide and immobilized cell deactivation—restricted severely the accumulation of the epoxide in the recirculation reactor system and could be predicted using a simple mathematical model. Epoxide inhibition was reduced by absorbing the product in the gas phase in old di-n-octyl phthalate. The resulting increase in propene oxide production agreed with model calculations. An alternating supply of propene and a co-substrate (ethene) prolonged the half-life of the immobilized cells. Using 50 g dry weight of cells, 1.5 g stereospecific propene oxide was produced in two days, of which 1.0 g was absorbed in the di-n-octyl phthalate phase.

An existing model for the external and internal mass transfer effects on the kinetics of propene epoxidation of calcium alginate-entrapped cells was extended to account for inhibition of the produced propene oxide. The intrinsic kinetics of the product-inhibited epoxidation reaction were derived and its applicability was demonstrated using independent time course experiments in a circulation batch-reactor system. The problem of coupled product-inhibited epoxidation and oxygen diffusion to and in the spherical gel beads was solved applying the collocation method. The predicted and experimental propene and oxygen-consumption rates were found to be in agreement for various degrees of oxygen-diffusion limitation and propene oxide inhibition.

Experimental verification of the internal pore and external film diffusion model is presented using the oxidation of propene to propene oxide by calcium alginate-entrapped Mycobacterium cells as a model system. Assuming pseudo-one-substrate Michaelis-Menten kinetics and negligible product inhibition, theoretical consumption rates for oxygen and propene the immobilized cells were compared to experimental values obtained in a circulation batch reactor system. The diameter of the spherical gel particles, immobilized cell density and the intrinsic kinetic parameters were measured independently. Internal pore diffusion of the limiting substrate was found the main mass-transfer resistance. A good agreement was observed between predicted and experimental epoxidation rates.

Several fundamental aspects of an organic-liquid-phase/ immobilized-cell system are studied using the epoxidation of propene as the model reaction. An automated experimental system, which facilitates the multiphase experiments, is described. Entrapment in calcium alginate gel provides high activity retention of the immobilized mycobacteria used. This gel is stable in the organic solvents tested and prevents aggregation and clotting of cells in the presence of these solvents, but does not seem to provide other protection against the solvents. High activity retention of the immobilized cells is favoured by low polarity and high molecular weight of the solvent. A bisubstrate kinetic model is used to describe the experimental consumption rates of oxygen and propene. Intraparticle substrate diffusion is found to limit the inherent kinetic reaction rate of the immobilized biocatalyst. The influence of particle diameter, catalytic activity and substrate level on epoxidation rates of pore-diffusion limited gel-entrapped cells is established under pseudo one-substrate conditions. Theoretical reaction rates, predicted by using an approximative pore diffusion model, correlate well with the experimental rates.

The microbial epoxidation of propene and 1-butene was used to study some fundamental aspects of two-liquid-phase biocatalytic conversions. Introduction of a water-immiscible organic solvent phase in a free-cell suspension gave rise to a series of undesired phenomena, e.g., inactivation by the solvent, clotting of biomass, and aggregation of cells at the liquid-liquid interface. Immobilization of the cells in hydrophilic gels, e.g., calcium alginate, prevented direct cell-organic solvent contact and the related clotting and aggregation of biomass. However, the gel entrapment did not seem to provide additional protection against the organic solvent. The influence of various organic solvents on the retention of immobilized-cell activity was related to solvent properties like the polarity (as expressed by the Hildebrand solubility parameter) and the molecular size (as expressed by the molecular weight or molar volume). High activity retention was favored by a low polarity in combination with a high molecular weight. The solubility parameter also proved useful to describe the capacity of various organic solvents for oxygen and alkene oxides. This facilitated the optimization of the solvent polarity.

Newly isolated and already available strains of alkene-utilizing bacteria were able to oxidize ethene, propene or 1-butene to the respective 1,2-epoxides. Resting-cell suspensions of organisms isolated on propene and butene, when grown on these substrates converted ethene quantitatively to epoxyethane. Some, but not all ethene-utilizing strains accumulated 1,2-epoxypropane or 1,2-epoxybutane when propene or butene was supplied, although not quantitatively because the epoxides produced were partially further metabolized. Suitable epoxide producers which eventually may be employed as biocatalysts in a biotechnological process were used for immobilization in calcium alginate and K-carrageenan; after immobilization, 60%–100% activity for epoxide production was retained.

An experimental set-up for automatic gas chromatographic analysis of circulation gas in a batch-reactor system is described. Gas sampling, substrate addition, data acquisition and data reduction are done with a coupled programmable integrator and microcomputer. On-line monitoring of the microbial oxidation of the gaseous alkenes, propene and 1- butene, to the corresponding epoxides is used to illustrate the operation of the experimental system. Measured gas concentrations of alkene and alkene oxide can be converted readily to quantify the reaction course in the liquid phase(s) of the circulation system. Results are presented for both one liquid phase (water) and two liquid phases (water and organic solvent) present in the reactor. The operational stability of the immobilized-cell system used for the epoxidations can be assessed by computer-controlled addition of gaseous alkene.