Ion sensors with PVC-plasticized membranes are widely employed in routine analytical procedures as a convenient and inexpensive tool for activity assessment. Development of new sensor membrane compositions is a long and tedious process as it requires the search, the synthesis and characterization of potential new membrane active substances. The number of proposed anion-selective ligands for membrane ion sensor is much smaller than that for cations. This is due to the several reasons: large variability of inorganic anion shapes (unlike spherical cations), denser solvate shells of anions, narrow pH range of some forms. The particular problem relates to the development of the sensor with selectivity towards hydrophilic anions, like, e.g. carbonate. An attractive opportunity would be to create a model that could predict the selectivity of ligand towards carbonate without real synthesis and characterisation of the corresponding sensors. An appropriate tool for that can be QSPR (Quantitative Structure-Property Relationship) approach [1], where the chemical structure of organic ligand is encoded with a set of formal descriptors and related through the mathematical modelling to the target property (selectivity in our case).
In this work, based on 48 ligand structures suggested in literature for carbonate sensing we have developed a model for prediction of logK (HCO3-/Cl-). Substructural Molecular Fragments (SMF) were used to describe the structure of compounds, where fragments were considered as sequences of bonds and atoms. The Projection on Latent Structures (PLS) method was used to calculate the regression model. The obtained model was tested in prediction of selectivity of several newly synthesized ligands. The details on the results will be provided in the presenstation.
This study was supported by RFBR project #20-33-70272.
References:
- Katritzky A.R., Lobanov V.S. and Karelson M. Soc. Rev., 1995, 24, 279-287.
