Polymer-based membrane filters are high-performance materials used in separation and purification applications such as water treatment, food processing and biomedicine. Synthetic membranes usually have high mechanical strength and chemical resistance but have low biocompatibility. Therefore, immobilization of biomolecules onto the surface of porous polymer membranes is a proven method to gain biocompatibility. Furthermore, the coupling of biomacromolecules such as proteins or enzymes enables the development of advanced bioactive membrane reactors (BMRs). In addition to the commonly applied chemical-based approaches for immobilizing biomolecules, a reagent-free method has been developed at the IOM that uses an electron beam (EB) for surface engineering. Briefly, high-energy electrons activate both the polymer material, and biomolecules dissolved in water, leading to the generation of reactive species that eventually form stable covalent bonds.
In this study, bovine serum albumin (BSA) protein was immobilized on a polymer microfiltration membrane using electron beam irradiation. Multiple processing parameters that could influence the immobilization result were selected as input features including numerical and categorical data. The aim was to predict the amount of immobilized protein on the membrane surface (grafting yield, GY) using a regression task. For this purpose, a response surface method within the design of experiments approach was employed. Multiple linear regression resulted in models with high predictive power (R²_pred = 0.9349 for the one-step method). The models were experimentally confirmed. Optimization with a desirability function led to significantly reduced processing costs (i.e., material, time, and energy) and thus, to an efficiency 230 times higher than the best literature value.