Data Mining Finds 33 Unknown MXenes, Including Magnetic Semiconductors

Researchers at Linköping University in Sweden have identified 33 previously unknown MXenes, a family of ultrathin two-dimensional materials, by searching databases of compounds that were already on file. They made seven of those materials in the lab, then used machine learning to design 11 more that contain rare earth metals. Most of the new ones are semiconductors, and several are magnetic, two properties that conventional MXenes almost never have.

The work was published in the journal Science (DOI: 10.1126/science.aeh5957), with Ali Saffar Shamshirgar as first author, according to Chemistry World.

What MXenes are and why they were stuck

MXenes are sheets only a few atoms thick. The “M” is a metal and the “X” has traditionally been carbon or nitrogen. Yury Gogotsi and colleagues at Drexel University in Pennsylvania first made them in 2011, and they have been studied for uses such as supercapacitors and dry lubricants.

The field has been held back by how hard they are to make and by what they lack. Making MXenes usually means etching away part of a bulk crystal with harsh chemicals such as hydrofluoric acid. Known MXenes are not magnetic. Very few behave as semiconductors, which matters because semiconducting sheets could become parts of tiny transistors. Gogotsi told Chemistry World that whether semiconducting MXenes exist is one of the questions he hears most at conferences.

How the team found materials hiding in plain sight

Johanna Rosen, a computational materials physicist at Linköping, and her group suspected that some MXene structures had been made before 2011 without anyone recognizing them. They searched two large collections of crystal data, the Materials Project and the Open Quantum Materials Database, for compounds with MXene-like formulas.

The search was deliberately broad. The team let the metal position include certain rare earth elements alongside the usual transition metals, and let the X position hold other light elements such as oxygen and phosphorus. From the crystal structures, they picked out the 33 candidates. One of the seven they synthesized is thorium nitride oxide.

A faster way to make them

To produce the materials, the group developed what it calls self-propagating high-temperature synthesis. A reaction starts and then sustains itself, so it needs no continuous outside heating, and it yields the two-dimensional sheets directly instead of requiring a bulk crystal to be etched apart.

Shamshirgar said many of these compounds once took weeks of high-temperature processing to produce small samples. He told Chemistry World the new route lets the team test many chemistries quickly while using far less energy.

Machine learning and the rare earth MXenes

The researchers then fed the mined data to a machine learning model and asked it to suggest MXenes that had never been made. That step produced 11 new materials containing rare earth metals. The team checked them two ways: with density functional theory, a computer method for calculating how electrons behave in a material, and with lab measurements including Raman spectroscopy and x-ray diffraction. The predictions and the measurements agreed well.

Most of the 11 were semiconductors, and their magnetism varied by element. MXenes containing ytterbium were ferromagnetic. Those containing holmium were antiferromagnetic, meaning neighboring atoms point their magnetic moments in opposite directions.

Why it matters

Magnetic and semiconducting sheets this thin could be useful in spintronics, which stores and moves information using the spin of electrons, and in magnetic memory, according to the report. A simpler, lower-energy synthesis also makes it easier for other labs to try new combinations.

Gogotsi, who was not part of the study, called it “one of those rare papers” reporting several important advances at once. He said direct synthesis from relatively cheap starting materials is a big step. He expects the variety of MXenes and their properties to keep expanding beyond the ones his group found 15 years ago.

What is still unclear

This is not the only report of magnetic rare earth MXenes. Both Gogotsi and the Linköping team noted that a separate group, led by Q. Fang, recently described such materials independently in Nature.

Rosen said the next goal is to understand exactly how the synthesis reaction works. If the team can learn that, she said, it could make more materials from a wider range of chemistries.

Source: Chemistry World