The efficiency of membrane processes in the food industry is largely reliant onminimising fouling, optimising hydrodynamic conditions, improving cleaning, ordeveloping specialised membranes for a given purpose. A different strategy is to make an intuitive physical or chemical modification to readily available membranes.Presented here are experimental findings relating to the modification of commercial polysulfone (PS) membranes of varying nominal molecular weight cut-off (nMWCO) by primary alcohols using both a static and dynamic treatment methodology. Furthermore, an assessment of their performance is made by analysis of ultrafiltration (UF) experiments of model solutions, as well as an industrially relevant food stuff - black tea, for the purpose of haze removal. Previous research into black tea filtration showed that it’s fouling and rejection could be influenced strongly by membrane material and surface chemical aspects. This led to the hypothesis that modifying membranes with alcohols would aid in improving aspects of black tea UF.Modification of 25, 50, and 100 kDa membranes by primary alcohols (with the majority of focus on ethanol treatment) was found to dramatically increase membrane water flux after both 90 minutes of continuous alcohol permeation and 24 hours of treatment in a static environment. The effects on surface characteristics were equally as pronounced. Hydrophobicity of membranes was increased, and control membrane negative ζ potential was neutralised, indicating that selectivity and fouling propensity would be altered. Further characterisation of the membranes revealed that there was an enhancement in glycerol preservative removal from the membranes, and that pore forming agents used in membrane casting were being leached out. Additionally, macroscopic swelling caused expansion of the porous matrix as confirmed by dimensional and porosimetric measurements, a relaxation of the membrane active layer, and an elastic modulus decrease of active layer as measured by atomic force microscopy (AFM). For model protein dead-end filtration, the process duration for de-watering of a fixed amount of bovine serum albumin (BSA) solution was reduced, proving that flux uplifts were translated to the filtration environment. No differences in rejection of both 훽- lactoglobulin and BSA were observed. Sorption of protein was shown to be greater as deduced from adsorptive fouling measurements, which prompted an investigation into specific membrane-foulant interactions. AFM performed with a BSA-functionalised colloidal probe indicated that adherence of protein to treated membranes was indeed increased. This was caused by both a decrease in charge, which promoted deposition, and an increase in hydrophobicity, giving rise to additional hydrophobic interactions probably with the secondary protein structure.During filtration of black tea, flux improvements for treated membranes were observed over untreated membranes in the pressure controlled region of the generalised transmembrane pressure–flux relationship. Typically the rejection of solids and polyphenols for UF membranes was disproportional; the membrane in effect, stripping some polyphenols from the tea. In addition, solids transmission increased, though less so than the transmission of polyphenols. The disproportional increase in polyphenols transmission was attributed to the reduced surface charge of the membranes; the polyphenolic anions experiencing less repulsion from the membrane surface. This was reinforced by colour measurements which showed increased redness (an indicator of higher key polyphenols concentrations in tea).These findings were then applied under a constant dilution diafiltration regime to demonstrate the mitigation of polyphenols loss in a process simulation setting. It was found that ethanol treatment could improve yield of tea solids or use of untreated membranes gave way to stream fractionation. Also discussed is the potential treatment of waste streams based on the accumulation of polyphenols in the retentate side of the membrane process.An investigation into the longevity of alcohol treatment whilst filtering black tea over multiple cycles has been carried out. This investigation looked at 4/5 foul-clean cycles, and showed that some of the initial modifications made to the membrane such as increased PWF were apparent after fouling and cleaning with caustic solutions. Also, untreated membranes showed moderate cycle on cycle increases in performance (polyphenols transmission and flux) over multiple cycles, whereas treated membrane showed a decline from the first filtration cycle onwards. This indicated that whilst some modifications to membranes may be permanent, some may simply be an accelerated result of what may occur ‘naturally’ over multiple foul-clean cycles, regardless of ethanol treatment.Finally, an investigation into the merits of using microfiltration (MF) for black tea haze removal was made using three commercial polysulfone MF membranes with 0.5, 0.9 and 1.5 µm nominal pore sizes. This enabled a comparison of processing methods and showed that whilst UF was efficient in haze removal, MF could also be implemented to remove and prevent haze reformation at feed concentrations far in excess of those possible in UF. Fluxes were typically low, though results showed that they could be improved by increasing the cross flow velocity substantially, or perhaps using a different membrane module configuration (e.g. tubular ceramic). The 1.5 µm cut-off membrane offered the most acceptable fluxes, though haze removal and prevention efficiency after 4 weeks was not acceptable. A similar result was returned for the 0.9 µm membrane. For the 0.5 µm membrane, the lowest fluxes were recorded although haze remained below 10 NTU for up to 4 weeks after filtration under simulated in-bottle conditions.
|Date of Award
|24 Jun 2015
|Engineering and Physical Sciences Research Council
|Michael Bird (Supervisor)