Effect of fouling and cleaning upon the surface charge and porosity of polyethersulphone membranes during coffee brew decaffeination

Filtration processes could offer a viable alternative to use of solvents for the production of reduced caffeine coffee beverages (so called ‘half-decaf’) which have recently gained increased commercial attention. However fouling phenomena and membrane deterioration could significantly alter the process performance; hence fouling and cleaning mitigation strategies must be explored by monitoring key membrane properties such as porosity and surface charge. In this study, multiple cycle filtration of coffee brews for the selective reduction of caffeine has been performed. A commercially available synthetic tight ultrafiltration polyethersulphone (PES) membrane (GP95PP – Alfa Laval) and a bespoke loose nanofiltration mixed matrix PES membrane fabricated, were used for three consecutive fouling and cleaning filtration cycles. The impact and mechanisms of fouling and the cleaning protocol effectiveness were investigated using surface charge, porosity, membrane surface and membrane morphology characterization techniques. Streaming potential studies revealed negative surface zeta potential for the membranes in a pH range of ca. 3.5–8.0. The fouled membrane classes exhibited elevated negative charges, which were partially restored to the pristine levels when chemically cleaned, suggesting the presence of residual negatively charged foulants. Porosity studies revealed the presence of a porous cake layer with an increase of average pore size and pore volume, and a decrease of 10–20 % on the BET surface area of the fouled membranes. Presence of key compounds on the membrane surface and structure was confirmed via EDX and FT-IR spectroscopy. Surface roughness along with SEM microscopy verified the uniform deposition of the foulants with absence of bulky particles. The presence of a cake layer was confirmed via SEM cross-sectional images due to selective thick layer thickness increase of the fouled membranes. Flux differences between the membrane classes can be related to the thickness of the top layer. Flux decreases and flux recovery ratios were also reported over multiple cycles and linked with porosity and other surface properties.