Two-phase bioconversion product recovery by microfiltration I. Steady state studies.

Recovery of an aqueous bioconversion product from complex, two-phase Pseudomonas putida broths containing 20% (v/v) soybean oil presents a significant challenge for downstream processing. Although not used before in multiple-phase separation for complex biotech products, crossflow filtration employing ceramic filters is one of the most attractive options which allow the design of integrated, continuous bioconversion processes. As a first attempt, we studied multichannel, monolithic ceramic membranes of different nominal pore sizes and lumen diameters under steady-state conditions. The best performance was obtained with 0.2-microm-pore/3-mm-lumen membrane, which completely rejected both cells and oil droplets from the permeate, creating a clear aqueous product stream. Although the same separation was achieved, the 50K molecular weight cut-off (MWCO) ultrafilter showed greater irreversible but similar reversible resistance, in addition to an order-of-magnitude higher membrane resistance. Larger nominal pore microfilters, such as 0.45 and 1.0 microm, experienced both cell and oil leakage even at low transmembrane pressure (10 psig). Attributed to greater shear at the same recirculation rate, smaller lumen filters did provide greater permeate flux. However, for practical purposes, the 0. 2-microm-pore/4-mm-lumen ceramic membrane was chosen for further evaluation. Transmembrane pressures up to 50 psig provided only marginal gains in filtration performance, whereas increasing shear rate resulted in linear increases in steady-state flux, presumably due to formation of shear-sensitive, complex gel/oil/cell layer near the membrane surface. A nominal shear rate of 9200 s-1 and 20 psig transmembrane pressure were chosen as optimal operating conditions. Additional studies in a clean system revealed that as low as 5% (v/v) soybean oil in deionized (DI) water resulted in an order-of-magnitude decline in steady-state permeate flux. Breakthrough of oil droplets occurred at 35 psig transmembrane pressure. The severe fouling and breakthrough phenomena disappeared in the presence of washed cells for transmembrane pressure up to 43 psig, implying an oil/cell layer coating the membrane surface, thus preventing oil penetration. Serious membrane fouling was also experienced in microfiltration of oil-free, cell-free supernatant and oil-free whole broth. Consequently, soluble proteins/surfactants were suspected to be the major membrane foulants. Interestingly, soybean oil up to 30% (v/v) enhanced the flux, presumably through complicated interactions with the major foulants. Regeneration of membrane was best achieved with protease and hot caustic/bleach treatments, supporting the hypothesized fouling mechanisms mentioned above. This work provides process and system information for batch microfiltration runs in the future, to be reported elsewhere as Part II of this work.