Machine learning helped to simulate the synthesis of tetrahedral amorphous carbon

Machine learning helped to simulate the synthesis of tetrahedral amorphous carbon
Aalto University

With the help of computer modeling for the first time physicists were able to show how and by what means the formation of amorphous carbon films with high content of tetrahedral carbon, which makes its mechanical properties close to that of diamond. To simulate the process of obtaining such substances is failed, in particular, through the use of machine learning in calculating the potential of interaction of atoms among themselves, write scientists in Physical Review Letters.

In addition to the known crystalline modifications of carbon, the most stable at room temperature, diamond and graphite — carbon, there are amorphous phases of different structure. One of the most interesting from the point of view of possible applications — is tetrahedral amorphous carbon. As diamond for the majority of carbon atoms inside such a material is characterized by spThree-hybridization and, respectively, the tetrahedral environment, the atoms do not form an ordered crystal structure and are randomly distributed.

Experimental methods of producing films of tetrahedral amorphous carbon are now reasonably well developed: for example with the bombing of the crystalline carbon surface by high energy ions, it is possible to obtain an amorphous material, which contains up to 90 percent of the tetrahedral carbon. But why and by what mechanisms such films are formed so far to the end was not clear. To simulate the formation of tetrahedral amorphous carbon and the conditions necessary for this, is quite difficult due to the fact that carbon can exist in several different hybridization conditions, which in such a system, a very large number of degrees of freedom.

READ  Cryptomathic attacked Android in 2017

Physicists from Finland and the UK under the leadership of the Rates gábor (Gábor Csányi) from the University of Cambridge for the first time were able to simulate the formation of a film of amorphous carbon with a high content of atoms with a tetrahedral environment, which correspond to the known experimental data and explain the mechanism of its growth. To conduct such modeling with machine learning scientists picked up the interaction potential between the carbon atoms — this task would have been difficult to solve by classical methods of modeling because of the very large amount of necessary calculations. For learning algorithm, we used data obtained numerically using the density functional theory.

Then, using the obtained potential with the help of the traditional method of molecular dynamics, the scientists have simulated the deposition process of high-energy carbon ions on the carbon substrate. The simulated system consisted of several thousands of atoms and reproduced the plot of the real stuff have an area of approximately 15 square nanometers. As a result, scientists were able to accurately describe the known experimental results, in particular, data on the percentage of the tetrahedral amorphous carbon film depending on ion energy, and other material characteristics, such as its modulus of elasticity, the function of radial distribution of atoms in the film and the structure factor — indicators of the degree of orderliness of structure and its interaction with falling at her radiation.

Comparison of the experimental data (black solid line) with simulation results (red dashed line) radial distribution functions (top) and structure factors (bottom)

M. A. Caro et al./ Physical Review Letters, 2018

In addition to reproduce the experimental data, using a computer simulation physicists have managed to establish the mechanism by which the growth of such a film. It turned out that this mechanism is quite different from that previously assumed. Before this it was considered that the formation of amorphous carbon occurs according to the mechanism of “subplantation”: when hit by high-energy ion in an already formed film in a very small volume around it is a redistribution of bonds with the formation of a tetrahedral structure, as a result of relaxation of the structure after this comes the growth of an amorphous film. However, the simulations showed that tetrahedral amorphous carbon is formed on a different mechanism. In contact with the ion flying with great speed on the surface of the carbon material, the carbon atoms in the film “apart” to make room for the new atom, the resulting structure is compressed and flat spTwo-hybridized carbon becomes tetrahedral structure with spThree-a hybridization. Due to this displacement, the film also slightly increases its thickness, moving up and moving the planar hexagonal carbon to the surface.

READ  Booster with military satellite, cleanly separated from the booster
Machine learning helped to simulate the synthesis of tetrahedral amorphous carbon

Schematic representation of two mechanisms of growth of tetrahedral amorphous carbon films: subplantation (left) is the mechanism considered responsible for the growth of the film earlier, and the mechanism proposed in this work (right)

M. A. Caro et al./ Physical Review Letters, 2018

The authors note that the study was the first to obtain accurate quantitative data on the structure of the known amorphous material and the mechanism of its formation. So physicists assume that the machine learning methods in the near future will become an important tool for modeling crystalline and amorphous materials.

Tetrahedral amorphous carbon is not the only unusual material which can obtain from that item. Other forms of carbon, particularly graphene, carbyne, fullerenes and carbon nanotubes you can read in our material “Seven faces of carbon”.


Leave a Reply

Your email address will not be published. Required fields are marked *