Nanofiltration and molecularly imprinted membranes: a theoretical study based on quantum mechanics approach

Impurity removal as well as the recovery of molecules with low molecular weight and high value is a challenge of considerable interest. NanoFiltration (NF) and molecularly imprinted membranes (MIMs) can definitely be a reliable solution for this issue. Molecularly imprinted membranes can be considered as an intelligent improvement of the classical NF membranes. Progresses have been achieved in the field of removal or recovery of molecules with low molecular weight in relation to the material choice used for membrane preparation (material design). However, from a theoretical point of view, up to date, there is not a wide literature concerning the knowledge, at quantum level, on MIMs key properties. In order to begin to remedy this gap, in this contribution, some sub-nano properties such as binding energies and effective molecular sizes have been accurately calculated using a quantum mechanics approach. Energies of the noncovalent interactions involved among a template genotoxin (4,4'-methylendianiline, MDA) and the acrylonitrile (AN) copolymer with acrylic acid (AA), P(AN-co-AA) [1], were accurately calculated in the frame of Density Functional Theory. In addition, the noncovalent interactions among copolymer fragments were analysed. It is important to note that a quantum mechanics methodology is necessary to obtain reliable binding energies of hydrogen bonds found in the supramolecular complexes considered in this study. The water permeability of the blank and imprinted membranes is around of 12 L/m2 h bar, this value is typical for NF membranes. The permeability of the butanol is lower than water permeability. The blank and imprinted membranes retain a different amount of MDA. In particular, the specific binding capacity of the prepared MIM membrane is equal to 6.6 ?mol/gmem [2]. Starting from these experimental evidences, the energies of the main interactions occurring during the MIM formation were evaluated. The hydrogen bonding between the carboxylic hydrogen and carbonyl group is the main noncovalent interaction experimented by the copolymer chains in addition to the electrostatic interactions between the carboxylic and nitrile groups. Dispersion interactions among the triple bond of the nitriles have lower binding energy with respect to the above interactions. The noncovalent interactions, experimented by the template molecule, are those between the amino group of genotoxin and carboxylic or nitrile groups of copolymer. Quantum mechanical calculations show that both in dimethylformamide (DMF) and in vacuo the binding energies, related to the polymer...polymer and genotoxin...polymer bonds, are comparable. Bearing in mind that the noncovalent interactions between -CN and -COOH groups are the most probable, the corresponding binding energies were evaluated. These are equal to -4.7 kcal/mol in DMF and -13.8 kcal/mol in vacuo. These energies are comparable to the hydrogen bonding energy involved in the most probable genotoxin...polymer complex, that is -5.93 kcal/mol in DMF and -12.68 kcal in vacuo. Thus, the computational results suggest that the target molecule in the casting solution binds effectively to the carboxylic functional groups located along the copolymer chains. This would cause a greater availability of the -COOH functional groups once the genotoxin is removed from the imprinted membrane; otherwise, the carboxylic groups would interact by single hydrogen bond among them or with the -CN group. Once the template molecules were extracted, -COOH functional groups remain free and more available for successive interactions with amines or amides. The total free carboxylic groups, obtained by the random arrangement of the polymeric chains and by the imprinting effect, increases the membrane affinity. However, it is very important to keep this aspect distinct from that connected to the pore sizes or membrane free volume. These are linked to the evaporation time of the solvent during the membrane preparation or in general they depend on the membrane preparation procedure. For this reason, the effective diameters and the relative cross sections of the rose Bengal (dye used to estimate the membrane mean pore size; 92% of retention), dianiline and aniline were also calculated using the same level of quantum theory. The results show that less than two molecules of dianiline can pass through the pores of the membrane, whereas approximately four molecules of aniline can cross the membrane through its pores. Assuming that genotoxin should come into contact with the free and more available carboxylic groups in the MIM, as a result the aniline should be retained less than dianiline by the blank and imprinted membranes because the average size of the membrane pores is markedly larger than the effective size of aniline. Thus, the effect due to the more available carboxylic groups inside the pores should be less pronounced for the aniline. Although the molecular affinity and size exclusion are tightly coupled [3], it is necessary to dissociate the imprinting effect from the familiar molecular sieving or molecular ad-absorption in the blank membrane. The latter, in fact, are not directly linked to the membrane imprinting. Finally, in this contribution, the binding energies related to hydrogen bonds between single AN and AA monomers and MDA in water were also theoretically evaluated, in addition to the energy of the hydrogen bonding between an used cross-linker and the genotoxin. The comparison among the energies of the noncovalent bonds, experimented in the polymer fragment...MDA and single monomer...genotoxin, is useful to understand if the single monomers can interact more strongly with respect to the copolymer fragments. The research leading to these results has received funding from the European Community's Seventh Framework Programme under grant agreement n 214226 (New Molecular Purification Technology for Pharmaceutical Production, NEMOPURE). In addition, we are very grateful to CASPUR for the use of High Performance Computers. References [1] T. Kobayashi, T. Fukaya, M. Abe, and N. Fujii, Langmuir 2002, 18, 2866-2872 [2] G. De Luca, L. Donato, S. García Del Blanco, F. Tasselli, and E. Drioli, J. Phys. Chem. B 2011, 115, 9345-9351 dx.doi.org/10.1021/jp2006638 [3], J. Greens, A. Hillen, B. Bettens, B.Van der Bruggen and C. Vandecasteele, Journal of Chemical Technology and Biotechnology 2005, 80, 1371-1377