Bio-based monomers are a more sustainable alternative to conventional oil-based monomers [1]. The bis-epoxide limonene dioxide from the epoxidation of the terpene limonene has shown potential for different applications [2]. One of those applications is the use of limonene dioxide as a crosslinking agent to improve the solvent resistance of nanofiltration membranes. Epoxidation of terpenes is conventionally done using meta-chloroperoxybenzoic acid (m-CPBA), using metal complexes with metals such as Tungsten, Titanium, and Cobalt, or different hydroperoxides. A greener method of epoxidation explored is the use of in situ generated dimethyldioxirane from the reaction of acetone and potassium peroxy-monosulfate (Oxone) [3]. The reaction uses sodium bicarbonate buffer in aqueous solution with a mixture of limonene and acetone. This project aims to synthesize different bis-, and tris-epoxides from different bio-derived terpenes including limonene, gamma-terpinene, geraniol, farnesol, and nerol using the reported method using Oxone and ultrasonication. Epoxidation using m-CPBA is also investigated to compare it to the Oxone method. In general, epoxidation using m-CPBA results in higher amount of epoxide, but the Oxone method presents a more sustainable alternative with good results. Successfully synthesized epoxides are used to crosslink polybenzimidazole nanofiltration membranes. Solvent testing in dimethylacetamide is used to inspect whether crosslinking is successful. Polyethylene glycol diglycidyl ether is a commercial bis-epoxide that was used to validate the crosslinking method. Crosslinking was successful, as confirmed by solvent testing and FT-IR analysis. Filtration testing showed that the permeance of the membrane was not affected by crosslinking, while the membrane’s rejection was increased from 10.29 ± 1.01 % to 17.23 ± 2.49 % after crosslinking using polyethylene glycol diglycidyl ether. Nerol and limonene bis-epoxides were successfully synthesized with high purity and were tested as crosslinkers. However, crosslinking was unsuccessful, as demonstrated by solvent testing. This project successfully synthesized bis-epoxides from different terpenes using a greener method of epoxidation. The possibility of successful crosslinking using the terpene-based crosslinkers should be further investigated.

Thesis (Ph.D. (Chemical Engineering))--North-West University, Potchefstroom Campus, 2009.

Thesis (Ph.D. (Chemical Engineering))--North-West University, Potchefstroom Campus, 2009.

Thesis (Ph.D. (Chemical Engineering))--North-West University, Potchefstroom Campus, 2009.

The focus of this dissertation was the development, synthesis and modification of polymers for the preparation of membranes for organic solvent nanofiltration. High chemical stability in a wide range of solvents was a key requirement. Membranes prepared from synthesized polymers as well as from commercial polymers were designed and chemically modified to reach OSN requirements. A solvent stable thin-film composite (TFC) membrane is reported, which is fabricated on crosslinked polythiosemicarbazide (PTSC) as substrate. The membranes exhibited high fluxes towards solvents like THF, DMF and DMSO ranging around 20 L/m2 h at 5 bar with a MWCO of around 1000 g/mol. Ultrafiltration PTSC membranes were prepared by non-solvent induced phase separation and crosslinked with GPTMS. The crosslinking reaction was responsible for the formation of an inorganic-type-network that tuned the membrane pore size. The crosslinked membranes acquired high solvent stability in DMSO, DMF and THF with a MWCO above 1300 g/mol. Reaction Induced Phase Separation (RIPS) was introduced as a new method for the preparation of skinned asymmetric membranes. These membranes have two distinctive layers with different morphologies both from the same polymer. The top dense layer is composed of chemically crosslinked polymer chains while the bottom layer is a porous structure formed by non-crosslinked polymer chains. Such membranes were tested for vitamin B12 in solvents after either crosslinking the support or dissolving the support and fixing the freestanding membrane on alumina. Pebax® 1657 was utilized for the preparation of composite membranes by simple coating. Porous PAN membranes were coated with Pebax® 1657 which was then crosslinked using TDI. Crosslinked Pebax® membranes show high stability towards ethanol, propanol and acetone. The membranes were also stable in DMF once crosslinked PAN supports were used. Sodium alginate polymer was investigated for the preparation of thin film composite membranes. Composite membranes were prepared using PAN and crosslinked PAN supports; these membranes were tested for methanol and DMF. Freestanding nanofilms fixed on alumina were also tested for methanol and DMF as well as many other harsh solvents. The alginate composite membranes showed excellent solvent stability and good permeances and a MWCO of around 1300 g/mol.

Several attempts have been made to combine the unique characteristics of graphene oxide (GO) and commercial polymers for successfully designing and fabricating next-generation membranes in filtration and separation technologies. The first part of the work develops a high flux polyethersulfone ultrafiltration membranes, by embedding GO sheets, starting from the polymer/GO solutions in ionic liquid and N, N dimethylformamide as co-solvents and promoting the pore formation via non-solvent induced phase separation. In the second part of the work, a protic ionic liquid was introduced as a solvent to disperse GO sheets and fabricate GO liquid crystal membranes for nanofiltration. The third part addresses the stability enhancement. GO membranes frequently disintegrate in aqueous environments due to swelling. Ethylenediamine was then used as a crosslinker, and the membranes were tested for organic solvent nanofiltration. Additionally, overcoming the permeation-rejection trade-off is challenging. Hence, the fourth work involved the intercalation of silica nanoparticles to form dual-sized nanochannels. In the final work, GO membranes were fabricated on the surface of hollow fibers to overcome scalability issues, by using a feasible spray coating method for efficient nanofiltration. Hollow fibers were crosslinked with hexamethylene diamine and GO was spray-coated on the crosslinked polymeric fibers for organic solvent nanofiltration. Overall, this study demonstrates the potential of GO in developing high-performance membranes for liquid separations relevant for industrial applications, such as wastewater treatment, food, chemical, petrochemical, and pharmaceutical processing.

Organic solvent nanofiltration (OSN) is a membrane-based sustainable alternative to conventional separation techniques because it is non-thermal and energy-efficient. The fabrication of membranes usually includes fossil-based polymers and toxic solvents that present significant challenges. For example, its declining availability, concerns about its degradability and cross-contamination that involve toxicity risks. Nowadays, there is an increasing interest in the development of more sustainable membranes that maintain an optimum performance even in harsh solvents. The aim of my thesis research is to develop stable OSN membranes from cellulose acetate and explore the use of deacetylation reactions. The effect of the degree of acetylation on the membrane performance and stability in different organic solvents was investigated. The chemical composition and morphology were investigated using Fourier Transform Infrared (FTIR), Scanning Electron Microscope (SEM), and Atomic Force Microscopy (AFM). It was found that cellulose acetate membranes with less than 22% acetylation present a satisfactory solvent resistance and rejection in harsh solvents, such as DMF and acetone. In the performance tests were identified two main trends: one for polar protic solvents and one for polar aprotic solvents. This was attributed to their capacity to interact with the membrane via H-bond formation. The molecular weight cutoff (MWCO) was in the range of 735–325 g mol–1 in aprotic solvents and higher than 885 g mol–1 for polar protic solvents. The results found in this research can be translated into a reduce in costs, waste generated, energy required, and time employed in the fabrication of membranes. Also, it opens potential areas in the industry as it can be implemented in harsh solvent environments.

Organic solvent nanofiltration (OSN) is a low-energy alternative for continuous separations in the chemical industry. As the pharmaceutical sector increasingly turns toward continuous manufacturing, OSN could become a sustainable solution for chiral separations. Here we present the first comprehensive theoretical assessment of enantioselective OSN processes. Lumped dynamic models were developed for various system configurations, including structurally diverse nanofiltration cascades and single-stage separations with side-stream recycling and in situ racemization. Enantiomer excess and recovery characteristics of the different processes were assessed in terms of the solute rejection values of the enantiomer pairs. The general feasibility of stereochemical resolution using OSN processes is discussed in detail. Fundamental connections between rejection selectivity, permeance selectivity, and enantiomer excess limitations are revealed. Quantitative process performance examples are presented based on theoretical rejection scenarios and cases from the literature on chiral membranes. A model-based prediction tool can be found on www.osndatabase.com to aid researchers in connecting materials development results with early-stage process performance assessments.

The aim of the dissertation to study small organic solute rejection in organic solvent nanofiltration using cheminformatics and machine learning. Chemically diverse datasets were curated using a novel medium high-throughput methodology and literature data extraction. These datasets contained thousands of datapoints of small organic solute rejections and process parameters. Different chemical fingerprinting has been used for feature generation, such as molecular descriptors, Morgan fingerprints, and latent-vector representation. These features were used to train different machine learning models, such as graph neural networks, partial least squares regression, and boosting tree algorithms. The obtained models were used to predict small organic solute rejection for nanofiltration related applications. Correlation between rejection and the molecular descriptors have been studied to deepen the understanding of transport and rejection through polymeric membranes. Explainable artificial intelligence concept was used to study the effect of solute, solvent and membrane structure on solute rejection. The conclusion of the dissertation highlights the importance of the chemical structure effect in nanofiltration and provides a future perspective on data-driven approached for nanofiltration and organic solvent nanofiltration.

La synthèse de matériaux polymères stables en milieu organique a été réalisée sur la base d’une série de copolymères à blocs de type polyuréthane. La synthèse de membranes denses autosupportées a permis la caractérisation des différents polyuréthanes et la sélection des formulations les plus adaptées pour l’obtention de membranes composites à peau dense. Les propriétés des membranes ont été ajustées par modulation des réactifs de départ. Une bonne résistance aux solvants organiques a été notée et les flux de perméation les plus importants ont été obtenus avec les matériaux les plus souples. Une membrane rigide a été testée pour la récupération du catalyseur Grubbs-HoveydaII du toluène. Malgré un faible taux de gonflement, le taux de rejet était limité à 48%, permettant de rompre avec l’idée liant la rigidité du réseau à une bonne sélectivité. Une démarche compréhensive du mécanisme de transport en nanofiltration organique (NFO) a été menée sur des membranes en polydiméthylsiloxane (PDMS). L’effet de la pression sur le gonflement a été examiné à travers deux appareils mimant les conditions de pression en NFO. Il a été démontré que la pression n’affectait pas l’équilibre amont de sorption mais induisait une diminution du gonflement de l’interface aval de la membrane. Le gradient de gonflement a été proposé comme force motrice du transport des solvants. La perméation des solvants purs et des solutés a été ensuite étudiée. Le modèle de solution-diffusion a pu être proposé comme mécanisme de transport des solvants purs. La sélectivité des membranes s’est avérée être indépendante de l’affinité solvant-membrane mais dépendante de l’affinité soluté - solvant et soluté - membrane / The synthesis of polymeric solvent stable materials was carried out on the basis of block copolymers polyurethane. A series of self-supported dense membranes allowed characterizing the different polymers and the selection of the most suitable ones to prepare composite membranes with a dense top layer. The properties of the membranes were adjusted by tailoring the starting reagents. Good resistance of the membranes in organic solvents was observed and the most important permeation flows were obtained with the softer materials. A rigid membrane was tested for the recovery of the Grubbs-HoveydaII catalyst from toluene. Despite a low swelling rate, the selectivity of the membrane was limited to 48%. This result is in contradiction with the common idea stating that high selectivity is due to rigid polymer network. A comprehensive approach of the transport mechanism in organic solvent nanofiltration (OSN) was conducted using polydiméthylsiloxane membranes (PDMS). The effect of pressure on the swelling was examined using two devices mimicking the pressure conditions in OSN. It has been shown that the pressure does not affect the upstream equilibrium sorption but induced a decrease of the swelling of the downstream interface of the membrane. The swelling gradient between the two sides of the membrane was proposed as driving force of solvents transport. The nanofiltration of solvents and solutes were then studied. The results showed that the solution–diffusion model was fully valid for pure solvents transport. The selectivity of the membrane was found to be independent of the membrane-solvent interaction but affected by the solute-membrane affinity and solute-solvent interaction