Classifying superheavy elements by machine learning
Among the 118 elements listed in the periodic table, there are nine superheavy elements (Mt, Ds, Mc, Rg, Nh, Fl, Lv, Ts, and Og) that have not yet been well studied experimentally because of their limited half-lives and production rates. How to classify these elements for further study remains an op...
| Main Authors: | , , , , , , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
American Physical Society
2019
|
| Online Access: | http://hdl.handle.net/1721.1/120709 https://orcid.org/0000-0002-7457-7959 |
| Summary: | Among the 118 elements listed in the periodic table, there are nine superheavy elements (Mt, Ds, Mc, Rg, Nh, Fl, Lv, Ts, and Og) that have not yet been well studied experimentally because of their limited half-lives and production rates. How to classify these elements for further study remains an open question. For superheavy elements, although relativistic quantum-mechanical calculations for the single atoms are more accurate and reliable than those for their molecules and crystals, there is no study reported to classify elements solely based on atomic properties. By using cutting-edge machine learning techniques, we find the relationship between atomic data and classification of elements, and further identify that Mt, Ds, Mc, Rg, Lv, Ts, and Og should be metals, while Nh and Fl should be metalloids. These findings not only highlight the significance of machine learning for superheavy atoms but also challenge the conventional belief that one can determine the characteristics of an element only by looking at its position in the table. |
|---|