Prediction of the formation of microporous aluminophosphates containing (6, 8)-rings using support vector machines

• SVM method is used on the database of AlPO synthesis.
• Predicting the formation of AlPOs containing (6, 8)-rings or not.
• Identifying which zeotype is formed among the known AlPOs with (6, 8)-rings.

A support vector machine (SVM) method has been used on the database of aluminophosphate (AlPO) synthesis with ca. 1700 reaction data to predict the formation of AlPOs with (6, 8)-rings based on 20 synthetic parameters associated with the gel composition, solvent and templates. Furthermore, four zeotype structures with different channel systems, such as AEN, AWO, ERI and CHA have been computational classified in the family of (6, 8)-ring-containing AlPOs. Analysis results reveal that several synthetic parameters, particular the molar ratio of solvent/Al2O3 (F3), and the Van der Waals (VDW) volume (F14), the greatest length (F11), and the number of free rotated single bond (F20) of templates have significant effects in directing the formation of target structure. Using single or combinations of several parameters, a better classifier can be trained to distinguish the (6, 8)-ring-containing AlPOs with desired zeotype structure or not, and the classification accuracy is about 85%. This work demonstrates that the computational method together with database of AlPO synthesis provides a possibility to establish the relationship between the synthetic factors and the resulting structures, which is important in guiding the rational synthesis of microporous aluminophosphate materials.

Figure optionsDownload as PowerPoint slide