| The aim of this research is to explore the structural reasons leading to the selectivity of collectors of chalcopyrite and to develop new reagents with high selectivity so as to investigate the probability of application of QSAR in development of flotation reagents. Through the research, two models with robust predictive abilities were obtained, with which it is revealed that the linearity of structures and the increase of LUMO density of double-bonded sulfur and charge on the single-bonded nitrogen/sulfur are favorable to the selectivity of collectors and a new collectors are synthesized and exhibited a much higher selectivity than conventional ones.Specifically, a group of experimental results was collected first and quantum calculations were carried out for the collectors involved in the experiments and two kinds of structural descriptors, which are chemical indices and topological indices, are extracted from the calculation results. To ensure the reliability of the values of such indices, the level of accuracy is adopted as fine in all of the calculations.Second, three regression methods are used to build up the model between experimental results and each kind of structural descriptors with six models obtained. Then internal and external validations were applied to investigate the statistical property of the six models. Finally, the best one for each kind of structural descriptors was obtained and in terms of predictive ability the one based on topological indices is superior to the other one.Third, both of the models were analyzed to try to explain the mechanism of selectivity in terms of structure. It can be concluded from the model of chemical indices that LUMO density of double-bonded sulfur and charge on the single-bonded sulfur/nitrogen are the two major factors affecting the selectivity of collectors, while the conclusion from the model of topological indices is that the linearity of molecules is the most significant determinant.Last, with both of the models, a new collector was designed and synthesized. It is indicated from its flotation experiments that this collector is significantly superior to the conventional ones, which further proved the predictive ability of the models obtainedThrough this research, we can draw the conclusion that it is totally possible to apply QSAR in development of flotation reagents, which will certainly accelerate the utilization of computer aided drug design methods in developing flotation reagents and enhance the production of new flotation reagents. |
