MXenes: linking old records to new ultrathin materials
· BehindLines newsroom
New materials can emerge from old databases too. On 29 October, Linköping University presented research combining machine learning searches with laboratory experiments to broaden the family of MXenes, materials arranged in very thin layers.
Finding materials before making them
These compounds combine metals with elements such as carbon or nitrogen. Linköping says the team searched older databases and found MXenes that had already been made but were unfamiliar to researchers in the field. Identifying them can therefore prevent an old composition from being presented as entirely new.
The abstract of a preliminary version submitted to arXiv on 17 August describes identifying candidates in repositories and then experimentally reproducing some of them. It also reports eleven new rare earth based MXenes. According to that abstract, experiments and theory reveal semiconducting behaviour and different magnetic states in this family.
These are separate steps: finding a promising database entry does not by itself establish that a material can be made, or that it will work in a device.
Assessing synthesis beyond the laboratory
The experimental method is another contribution of the project. The preliminary version describes a high temperature reaction that propagates without continuous external heating. Its abstract reports that the operation finishes in a few minutes. Doing without continuous external heating therefore does not mean manufacturing at low temperatures. The October statement highlights reductions in hazardous chemicals and water use.
However, the versions give different counts and durations. The August abstract mentions 38 candidates and an operation lasting a few minutes; the October communication counts 33 existing materials and refers to seconds. Those descriptions should not be combined into a single result.
The statement discusses possible applications including water purification, energy storage, telecommunications and biomedicine. These remain prospects, with no commercial performance demonstrated in this presentation. The immediate contribution is the connection between archived records, prediction and experimental verification.