Cross-Flow Filtration with Mixed Matrix Membranes for the Removal of Proteins from White Wines

Abstract

Bentonite fining is a process used in the wine making industry to eliminate protein instability in white wines. It is a time-consuming batch process that generates a large amount of waste products known as bentonite lees (up to 10% of wine volume). This thesis examines the feasibility of an alternative process to replace the current bentonite fining process via the use of bentonite-embedded mixed matrix membranes (MMM) in cross-flow filtration systems. This process aims to reduce the processing time and reduce or eliminate the generation of bentonite lees. While protein adsorber membranes made using the wet phase inversion technique had been studied before, this thesis, for the first time, applied the use of this technology on real wine. MMM of different polymer matrix materials, polysulphone (PS), polyethersulphone (PES) and polyvinylidene fluoride (PVDF), and three different bentonites (one laboratory grade bentonite and two commercial grade bentonites) as the embedded material were fabricated and characterised using scanning electron microscopy (SEM), lightbox imaging, confocal scanning laser microscopy (CLSM), thermal gravimetric analysis (TGA), and contact angle analysis to identify the impact of certain membrane characteristics on the effectiveness and efficiency of the process. Similarly, operating parameters such as pressure and pump speed were also analysed. Filtered wine was examined using heat stability tests, polyphenol content via gallic acid equivalent analysis (GAE), and colour analysis. The morphologies of all the fabricated membranes were characteristic of asymmetric phase inversion membranes. The more hydrophilic PS and PES membranes were found to provide higher flux and lower rejection of polyphenols than PVDF membranes of the same composition, up to 91% higher flux, and 13% lower rejection of polyphenols with 66.7% bentonite loading membranes operating at 4.5 bar pressure. Laboratory grade bentonite was found to be a more efficient bentonite to use in membranes than the commercial variants, offering double the adsorption capacity per unit mass. Preferred membrane morphological characteristics were found to be a low density skin layer for high flux and low selectivity based on size exclusion, predominantly sponge-like support layer for better contact between wine and bentonite particles, and few large cavities which are structural weak points. The most effective membrane fabricated in this study was made of 10 wt% PES, 25 wt% bentonite, 65 wt% solvent. This membrane produced the highest flux (87.1 L/mP2Ph after 30 min of filtration at 3.0 bar) and best selectivity (96% removal of proteins and less than 5% rejection of polyphenols). Wine filtered using this membrane was also found to have the least negative impact via polyphenol concentration, taste and aroma tests. The membrane rejected proteins via both size exclusion via the membrane morphology and skin layer and bentonite adsorption where bentonite adsorption accounted for 50% of the proteins removed. Operating parameters such as backflushing, pressure and flow rate were found to affect flux performance similar to typical cross-flow microfiltration system. Therefore, optimisation strategies should be identical to current systems. Scale up estimations using best conditions and PES membrane composition found in this study was performed. Membrane cartridges are capable of processing wine at a rate of 79 L/mP2Ph, with adsorption capacity capable of filtering 370 L/mP2P of Sauvignon Blanc wine with protein concentration of 0.3 g/L. This is comparable to some industrial wine cross-flow filtration modules. It is concluded that this new process is a feasible replacement for the traditional bentonite fining process. There is the potential of combining both the bentonite fining and filtration stages into one process. This was capable of producing protein stable wine with one filtration pass with low degree of negative impacts at a higher processing rate and an elimination of bentonite lees when compared to the traditional batch-wise bentonite fining process. The benefits will ultimately be balanced by the cost of the filtration modules. A brief cost analysis revealed that the proposed process can be more economical than the batch-wise process if the membrane modules can be fully regenerated 3 times.