Artificial Intelligence to identify new compounds

Hybrid perovskites form a very promising class of semiconductors for the design of low-cost solar cells with light conversion efficiencies now approaching those of silicon. Scientists from CNRS (The National Center for Scientific Research, better known by its acronym CNRS, is the largest…) have developed a tool (A tool is a finalized object used by a living being in order to increase its…) very advanced Machine Learning which makes it possible to instantly determine whether a new compound will be of the perovskite type or not. These results were published in the journal Advanced Materials.

A new technique for analyzing X-ray diffraction patterns based on very advanced machine learning makes it possible to instantly determine whether a new compound will be of the perovskite type or not.
© Florian Massuyeau

Halogenated hybrid perovskites represent a class of materials (A material is a material of natural or artificial origin that man shapes to…) semiconductors particularly studied in recent years for their remarkable photoelectric properties and their applications in photovoltaic systems. To be qualified as perovskites, these hybrid organic/inorganic materials based on lead (Lead is a chemical element of the family of crystallogens, symbol Pb and…) must have a network (A computer network is a set of equipment linked together to exchange information…) specific inorganic: halogenated lead octahedra connected only at their vertices. It is this particular structure which makes it possible to obtain good conduction properties which are useful for solar cells.

If it is easy to prepare many compositions by changinghalogen (Halogens are a chemical series consisting of the chemical elements of…) and/or of molecule (A molecule is an electrically neutral chemical assembly of at least two atoms, which…) organic (Organic chemistry is a branch of chemistry concerned with the description and study of a large…)it is impossible to determine, before a synthesis, what will be the structure of the material (A material is a material of natural or artificial origin that man shapes…) obtained. Only structural determination by X-ray diffraction on single crystal or powder (Powder is a fractional state of matter. It is a solid present…) with a high resolution makes it possible to know if a new material hybrid (In genetics, the hybrid is the crossing of two individuals of two varieties,…) is of the perovskite type. This time-consuming process requires specific expertise and therefore constitutes one of the main limiting factors of synthetic processes using broadband (The term broadband (or broadband by literal translation of the Anglo-Saxon expression…) of perovskites.

In this context (The context of an event includes the circumstances and conditions surrounding it; the…)scientists from theInstitute (An institute is a permanent organization created for a certain purpose. It is…) materials from Nantes Jean Rouxel (CNRS/Nantes University), in collaboration with chemists from the Institute for Research on catalysis (Catalysis is the action of a substance called a catalyst on a chemical transformation…) and theenvironment (The environment is everything that surrounds us. It is all the natural elements and…) of Lyon (CNRS/University Claude Bernard (Claude Bernard, born July 12, 1813 in Saint-Julien (Rhône) and died on…) Lyon 1), made a machine learning algorithm learn the important characteristics to identify perovskite-like structures in simple powder diffraction patterns of new materials. This new technique is not sufficient to determine completely (Completion or completely automatic, or by anglicism completion or…) a new structure but makes it possible to determine whether a new compound is of the perovskite type almost instantaneously. For this, they extracted from the database (In computing, a database (Abr.: “BD” or…) Cambridge Structural Database (CSD) all lead-based halogenated hybrid structures (about 1000), simulated their X-ray powder diffraction patterns and sorted these materials according to their perovskite character or not.

The two models (random forest and convolutional neural network) validated their approach with a prediction rate of around 90%. In addition, these models have allowed scientists to better understand the important characteristics in a diagram (A diagram is a simplified and structured visual representation of concepts, ideas,…) X-ray diffraction to differentiate perovskite structures from others. This new technique for analyzing X-ray diffraction patterns would also make it possible to detect other structural types. A new tool that is described in the review Advanced Materials.

Reference:
Perovskite or Not Perovskite? A Deep-Learning Approach to Automatically Identify New Hybrid Perovskites from X-ray Diffraction Patterns
Florian Massuyeau, Thibault Broux, Florent Coulet, Aude Demessence, Adel Mesbah and Romain Gautier, Advanced MaterialsAugust 13, 2022.

Did you like this article? Do you wish to support us ? Share it on social networks with your friends and/or comment on it, this will encourage us to publish more similar topics!

We wish to give thanks to the writer of this short article for this incredible material

Artificial Intelligence to identify new compounds


Our social media profiles here as well as additional related pages here.https://yaroos.com/related-pages/