Patrizia Marchetti

24 Aug 2012

Patrizia Marchetti is completing her PhD in Chemical Engineering as industrial-based student at Lonza AG (Visp, Switzerland), and Imperial College London (UK). She earned her master degree in Chemical Engineering at Politecnico di Milano (Italy) in 2009.  During her PhD, she has focused on the development of membrane technology for purification/separation of APIs (specifically peptides) and has investigated the fundamentals of solute and solvent transport through membranes.

PDF of presentation


Permeation through NF and UF membranes

Permeation of water, organic solvents and peptide mixtures through ceramic nanofiltration and ultrafiltration membranes was investigated.

Experimental results clearly indicated that the Hagen-Poiseuille equation fails in describing solvent permeation through NF membranes, whereas it is able to describe the experimental permeability through UF membranes.

A number of empirical models have been developed to approach this problem. These models are all characterized by the introduction of several solvent-membrane interaction parameters to describe the fluid transport confined in nanoscale pores. Unfortunately, these interaction parameters are often specific for the solvent-membrane couple and the generalization of the model to different solvent-membrane combinations is not possible.

In this work, a new model is proposed to describe solvent permeation through NF and UF membranes in the absence of solutes. The development of the model started from the Hagen-Poiseuille equation, which assumes the viscosity as the main influencing parameter for the permeation, and introduces several correction factors to account for the surface phenomena that arise in a nanotube due to solvent-membrane surface interactions. This approach has been initially introduced in the Washburn equation, where the Laplace pressure is used to describe the capillary rise due to surface tension. Additional correction factors were used in the proposed model to account for the effects due to solvent polarity and solvent molecular dimension.

The model parameters were regressed on experimental data for five solvent with different physical characteristics. Its predictive capability was tested on other six solvents, in both NF and UF range.

The model was also applied to selected aqueous and organic mixtures. It was observed that viscosity represents again the most influencing factor, being the permeability profiles very similar to the viscosity profiles of the corresponding mixtures. The model showed good predictions for all of them, apart from the singular exception of ACN / water mixture. The formation of complexes between water and ACN molecules was taken into account to explain the anomalous behavior of this mixture.

Finally, the work was extended to consider the influence of peptides on the permeability of the mixture. The effects of peptide concentration, pH, buffer, pressure and tangential velocity are presented.