A gas–liquid hollow fibre membrane contactor (HFMC) process has been introduced for carbon dioxide (CO2) separation from biogas where aqueous ammonia (NH3) is used to chemically enhance CO2 absorption and initiate heterogeneous nucleation of the reaction product ammonium bicarbonate at the membrane–solvent interface. Aqueous ammonia absorbents (2–7 M) were initially used in single pass for CO2 separation from a synthetic biogas where nucleation of ammonium bicarbonate crystals was observed at the perimeter of the micropores. Recirculation of the aqueous ammonia absorbent encouraged the growth of ammonium bicarbonate crystals on the shell-side of the membrane that measured several microns in diameter. However, at high aqueous NH3 concentrations (3–7 M), lumen side crystallisation occurred and obstructed gas flow through the lumen of the HFMC. The suggested mechanism for lumen-side crystallisation was absorbent breakthrough into the lumen due to pore wetting which was promoted by low absorbent surface tension at high NH3 concentration. Preferential shell-side nucleation can therefore be promoted by (1) raising surface tension of the absorbent and (2) selection of a membrane with a more regulated pore shape than the PTFE membrane used (d/L 0.065) as both actions can diminish solvent ingress into the pore. This was evidenced using 2 M NH3 absorbent where shell-side crystallisation was evidenced without the onset of lumen side crystallisation. Raising surface tension through the inclusion of salt into the chemical absorbent also promoted greater CO2 flux stability. Importantly, this study demonstrates that chemically enhanced HFMC are an attractive prospect for gas–liquid separation applications where reaction product recovery offers further economic value

Whilst the fouling of MBR membrane surfaces has been very extensively explored by the academic community, there is an increasingly widespread recognition by practitioners of the issue of clogging of membrane channels with sludge solids, sometimes termed "sludging". The study undertaken has quantified this phenomenon using a bespoke test cell allowing a flat sheet membrane channel to be viewed directly during operation and the accumulated solids determined by digital image processing. Sludging behaviour has then been correlated both with the sludge properties, from sludge samples taken from both an industrial and municipal MBR, and the permeability decline rate data. The work has revealed the expected trends in fouling propensity, as quantified by the exponent n of the Δp/Δt = m.exp(nJ) correlation from classical flux-step tests. With zero membrane aeration the industrial samples exhibited sludging, the filling of the complete thickness of the membrane channel with sludge solids, whereas for municipal sludge the solids formed a cake layer which did not fill the channel. In the absence of sludging the permeability decline followed the expected pattern of increasing at the elevated soluble COD and capillary suction time values of the industrial sludge, compared with municipal sludge at the same solids concentration range (8-12 g.L-1). However, there was no evident correlation between fouling (permeability decline without sludging) and sludging: incipient sludging did not appear to influence permeability, though can be assumed to negatively impact on long-term operation, or relate to the sCOD concentration. Sludging instead appeared to depend on the sludge physical properties, and primarily the viscosity: sludge samples at high viscosities were found to exhibit a different air-scour pattern to that at normal MLSS concentrations. Outcomes suggest that sludging is caused by rheological conditions promoting bubble coalescence and bubble stream constriction, reducing the exposure of the membrane surface to scouring air.