Possibilities of Neural Network Powder Diffraction Analysis Crystal Structure of Chemical Compounds

Some possibilities of using convolutional artificial neural networks (ANN) for powder diffraction structural analysis of crystalline substances have been investigated. First, ANNs are used to classify crystalline systems and space groups according to calculated full-profile diffractograms calculated from the crystal structures of the ICSD database (2017 year). The ICSD database contains 192004 structures, of which 80% was used for in-depth network training, and 20% for independent testing of recognition accuracy. The accuracy of classification by a network of crystalline systems was 87.9%, and that of space groups was 77.2%. Secondly, the ANN is used for a similar classification of structural models generated by the stochastic genetic algorithm in the search processes for triclinic crystal structures of test compound K4SnO4 according to their full-profile diffraction patterns. The classification criterion was the entry of one or several atoms into their crystallographic positions in the structure of a substance. Independent deep network training was performed on 120 thousand structural models of the K4PbO4 triclinic structure generated in several runs of the genetic algorithm. The accuracy of the classification of K4SnO4 structural models exceeded 50%. The results show that deeply trained convolutional ANNs can be effective for classifying crystal structures according to the structural characteristics of their powder diffraction patterns

On the structural relations of malachite. I. The rosasite and ludwigite structure families

The crystal structures of malachite Cu2(OH)2CO3 and rosasite (Cu,Zn)2(OH)2CO3, though not isotypic, are closely related. A previously proposed approach explaining this relation via a common hypothetical parent structure is elaborated upon on the basis of group–subgroup considerations, leading to the conclusion that the aristotype of malachite and rosasite should crystallize in the space group Pbam (No. 55). An ICSD database search for actual representatives of this aristotype leads to the interesting observation that the structure type of ludwigite (Mg,Fe)2FeO2BO3, which is adopted by several natural and synthetic oxide orthoborates M 3O2BO3, is closely related to the proposed malachite–rosasite aristotype and thus to the malachite–rosasite family of hydroxide carbonates M 2(OH)2CO3 in general. Relations within both structure families and their analogies are summarized in a joint simplified Bärnighausen tree.

The use of an acoustic levitator to follow crystallization in small droplets by energy-dispersive X-ray diffraction

For the investigation of small sample volumes, the use of an acoustic levitator was tested as a `sample holder' for hovering droplets in a synchrotron beam. It might be advantageous to use levitated droplets instead of samples confined in solid holders, especially for the study of crystallization processes where the influence of containing walls has to be minimized. In a first experiment, the crystallization of sodium chloride in a small droplet of aqueous solution has been followed with a time resolution of 30 s. The collected diffraction peaks are compared with data in the ICSD database.

Inorganic crystal structures from powder data—A retrospection

More than 10 000 inorganic structures based on both X-ray and neutron powder diffraction data were extracted from the ICSD database and analyzed considering overall trends, cell symmetries, occurrence of the space groups, complexity, precision and reliability of reported data. It was found that the major amount of structures belong to higher symmetries and have 2–5 atoms in the asymmetric unit. Less than 35% of the e.s.d.'s of atomic coordinates is ≤10−3 and 10−4 or better was reached in only 5% of cases. Approximately one-quarter of papers contain possibly inaccurate atomic coordinates or interatomic distances. A short review of journals publishing structures derived from powders is also given.